diff --git a/CRAN-SUBMISSION b/CRAN-SUBMISSION
index 99e82aa..4128518 100644
--- a/CRAN-SUBMISSION
+++ b/CRAN-SUBMISSION
@@ -1,3 +1,3 @@
-Version: 1.1.0
-Date: 2023-10-04 21:17:30 UTC
-SHA: d49bad606380799b1b2c4f73a87ccc6585d05e5c
+Version: 1.1.1
+Date: 2024-02-07 17:45:58 UTC
+SHA: c69f5c5308ae5c0c96d955d46d309231299f9c2f
diff --git a/DESCRIPTION b/DESCRIPTION
index f422705..0eca603 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -1,6 +1,6 @@
 Package: jlmerclusterperm
 Title: Cluster-Based Permutation Analysis for Densely Sampled Time Data
-Version: 1.1.0
+Version: 1.1.1
 Authors@R: 
     person("June", "Choe", , "jchoe001@gmail.com", role = c("aut", "cre", "cph"),
            comment = c(ORCID = "0000-0002-0701-921X"))
diff --git a/NEWS.md b/NEWS.md
index 6e4d884..33833d0 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -1,4 +1,12 @@
-# jlmerclusterperm (development version)
+# jlmerclusterperm 1.1.1
+
+# jlmerclusterperm 1.1.1
+
+- More informative warnings when constructing a spec object
+
+- Fixed minor bugs when interfacing with Julia
+
+- Fixed bug with reporting convergence warnings when only one is encountered
 
 # jlmerclusterperm 1.1.0
 
diff --git a/R/aaa.R b/R/aaa.R
index 7d71394..1a3a5af 100644
--- a/R/aaa.R
+++ b/R/aaa.R
@@ -16,7 +16,7 @@ julia_cli <- function(x) {
   utils::tail(system2("julia", x, stdout = TRUE), 1L)
 }
 parse_julia_version <- function(version) {
-  gsub("^julia version .*(\\d+\\.\\d+\\.\\d+).*$", "\\1", version)
+  gsub("^julia.version .*(\\d+\\.\\d+\\.\\d+).*$", "\\1", version)
 }
 julia_version <- function() {
   parse_julia_version(julia_cli("--version"))
diff --git a/README.md b/README.md
index ee4f8f6..4cc83e9 100644
--- a/README.md
+++ b/README.md
@@ -273,7 +273,7 @@ following as you see fit.
 To cite jlmerclusterperm:
 
 - Choe, J. (2023). jlmerclusterperm: Cluster-Based Permutation Analysis
-  for Densely Sampled Time Data. R package version 1.1.0.
+  for Densely Sampled Time Data. R package version 1.1.1.
   <https://cran.r-project.org/package=jlmerclusterperm>.
 
 To cite the cluster-based permutation test:
diff --git a/cran-comments.md b/cran-comments.md
index 4d23855..0037a2f 100644
--- a/cran-comments.md
+++ b/cran-comments.md
@@ -1,7 +1,3 @@
-Fixes for CRAN:
-
-* Specified importing from backports which was missing previously
-
 ## R CMD check results
 
 0 errors | 0 warnings | 0 notes
diff --git a/docs/404.html b/docs/404.html
index 4f0090f..6cf2a37 100644
--- a/docs/404.html
+++ b/docs/404.html
@@ -33,7 +33,7 @@
     
     <a class="navbar-brand me-2" href="https://yjunechoe.github.io/jlmerclusterperm/index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/CODE_OF_CONDUCT.html b/docs/CODE_OF_CONDUCT.html
index 12edce3..b1496bf 100644
--- a/docs/CODE_OF_CONDUCT.html
+++ b/docs/CODE_OF_CONDUCT.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/CONTRIBUTING.html b/docs/CONTRIBUTING.html
index 09dc488..16a1924 100644
--- a/docs/CONTRIBUTING.html
+++ b/docs/CONTRIBUTING.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/LICENSE-text.html b/docs/LICENSE-text.html
index 66fae62..c3a9066 100644
--- a/docs/LICENSE-text.html
+++ b/docs/LICENSE-text.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/LICENSE.html b/docs/LICENSE.html
index c15eb5c..4dc3c4b 100644
--- a/docs/LICENSE.html
+++ b/docs/LICENSE.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/articles/Garrison-et-al-2020.html b/docs/articles/Garrison-et-al-2020.html
index d2536d1..1ef3e1d 100644
--- a/docs/articles/Garrison-et-al-2020.html
+++ b/docs/articles/Garrison-et-al-2020.html
@@ -35,7 +35,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -888,7 +888,7 @@ <h3 id="repeating-the-cpa-with-logistic-regression">Repeating the CPA with logis
 <span>  <span class="op">)</span></span>
 <span><span class="op">}</span><span class="op">)</span></span>
 <span><span class="co">#&gt;    user  system elapsed </span></span>
-<span><span class="co">#&gt;    0.04    0.07   11.46</span></span>
+<span><span class="co">#&gt;    0.05    0.11   10.97</span></span>
 <span><span class="va">CPA_agg_1000</span></span>
 <span><span class="co">#&gt; $null_cluster_dists</span></span>
 <span><span class="co">#&gt; ── <span style="font-weight: bold;">Null cluster-mass distribution</span> (t &gt; <span style="color: #0000BB;">2</span>) ────────────── <span style="color: #0000BB;">&lt;null_cluster_dists&gt;</span> ──</span></span>
@@ -912,7 +912,7 @@ <h3 id="repeating-the-cpa-with-logistic-regression">Repeating the CPA with logis
 <span>  <span class="op">)</span></span>
 <span><span class="op">}</span><span class="op">)</span></span>
 <span><span class="co">#&gt;    user  system elapsed </span></span>
-<span><span class="co">#&gt;    0.02    0.01   18.39</span></span>
+<span><span class="co">#&gt;    0.05    0.06   15.25</span></span>
 <span><span class="va">CPA_trial_1000</span></span>
 <span><span class="co">#&gt; $null_cluster_dists</span></span>
 <span><span class="co">#&gt; ── <span style="font-weight: bold;">Null cluster-mass distribution</span> (t &gt; <span style="color: #0000BB;">2.8</span>) ──────────── <span style="color: #0000BB;">&lt;null_cluster_dists&gt;</span> ──</span></span>
diff --git a/docs/articles/Geller-et-al-2020.html b/docs/articles/Geller-et-al-2020.html
index 4ade7c6..94cefe1 100644
--- a/docs/articles/Geller-et-al-2020.html
+++ b/docs/articles/Geller-et-al-2020.html
@@ -35,7 +35,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -479,7 +479,7 @@ <h2 id="d-specifying-random-effects">D) Specifying random effects<a class="ancho
 <span>  <span class="op">)</span></span>
 <span><span class="op">}</span><span class="op">)</span></span>
 <span><span class="co">#&gt;    user  system elapsed </span></span>
-<span><span class="co">#&gt;    0.00    0.08    8.12</span></span>
+<span><span class="co">#&gt;    0.02    0.10    9.42</span></span>
 <span><span class="va">re_CPA</span></span>
 <span><span class="co">#&gt; $null_cluster_dists</span></span>
 <span><span class="co">#&gt; ── <span style="font-weight: bold;">Null cluster-mass distribution</span> (chisq p &lt; <span style="color: #0000BB;">0.05</span>) ───── <span style="color: #0000BB;">&lt;null_cluster_dists&gt;</span> ──</span></span>
@@ -558,7 +558,7 @@ <h2 id="d-specifying-random-effects">D) Specifying random effects<a class="ancho
 <span>  <span class="op">)</span></span>
 <span><span class="op">}</span><span class="op">)</span></span>
 <span><span class="co">#&gt;    user  system elapsed </span></span>
-<span><span class="co">#&gt;    0.01    0.05   12.11</span></span>
+<span><span class="co">#&gt;    0.00    0.06   10.60</span></span>
 <span><span class="va">re_max_CPA</span></span>
 <span><span class="co">#&gt; $null_cluster_dists</span></span>
 <span><span class="co">#&gt; ── <span style="font-weight: bold;">Null cluster-mass distribution</span> (chisq p &lt; <span style="color: #0000BB;">0.05</span>) ───── <span style="color: #0000BB;">&lt;null_cluster_dists&gt;</span> ──</span></span>
@@ -714,7 +714,7 @@ <h2 id="e-contrast-coding">E) Contrast coding<a class="anchor" aria-label="ancho
 <span>  <span class="op">)</span></span>
 <span><span class="op">}</span><span class="op">)</span></span>
 <span><span class="co">#&gt;    user  system elapsed </span></span>
-<span><span class="co">#&gt;    0.09    0.06   12.41</span></span>
+<span><span class="co">#&gt;    0.00    0.01   11.68</span></span>
 <span><span class="va">final_CPA</span></span>
 <span><span class="co">#&gt; $null_cluster_dists</span></span>
 <span><span class="co">#&gt; ── <span style="font-weight: bold;">Null cluster-mass distribution</span> (chisq p &lt; <span style="color: #0000BB;">0.05</span>) ───── <span style="color: #0000BB;">&lt;null_cluster_dists&gt;</span> ──</span></span>
diff --git a/docs/articles/asynchronous-cpa.html b/docs/articles/asynchronous-cpa.html
index c258556..4d61181 100644
--- a/docs/articles/asynchronous-cpa.html
+++ b/docs/articles/asynchronous-cpa.html
@@ -35,7 +35,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -275,15 +275,15 @@ <h3 id="collecting-the-results">Collecting the results<a class="anchor" aria-lab
 inspection:</p>
 <div class="sourceCode" id="cb11"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">CPA_async</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://future.futureverse.org/reference/value.html" class="external-link">value</a></span><span class="op">(</span><span class="va">f</span><span class="op">)</span></span>
-<span><span class="co">#&gt; Connecting to Julia TCP server at localhost:11981 ...</span></span>
+<span><span class="co">#&gt; Connecting to Julia TCP server at localhost:11984 ...</span></span>
 <span><span class="co">#&gt; <span style="color: #00BBBB;">ℹ</span> Detecting empirical clusters and calculating cluster-mass statistics.</span></span>
-<span><span class="co">#&gt; <span style="color: #00BB00;">✔</span> Detecting empirical clusters and calculating cluster-mass statistics. <span style="color: #B2B2B2;">[57ms]</span></span></span>
+<span><span class="co">#&gt; <span style="color: #00BB00;">✔</span> Detecting empirical clusters and calculating cluster-mass statistics. <span style="color: #B2B2B2;">[71ms]</span></span></span>
 <span><span class="co">#&gt; </span></span>
 <span><span class="co">#&gt; <span style="color: #00BBBB;">ℹ</span> Sampling cluster-mass statistics from a bootstrapped null distribution.</span></span>
-<span><span class="co">#&gt; <span style="color: #00BB00;">✔</span> Sampling cluster-mass statistics from a bootstrapped null distribution. <span style="color: #B2B2B2;">[1.6s]</span></span></span>
+<span><span class="co">#&gt; <span style="color: #00BB00;">✔</span> Sampling cluster-mass statistics from a bootstrapped null distribution. <span style="color: #B2B2B2;">[1.7s]</span></span></span>
 <span><span class="co">#&gt; </span></span>
 <span><span class="co">#&gt; <span style="color: #00BBBB;">ℹ</span> Calculating the probability of the observed cluster-mass statistics.</span></span>
-<span><span class="co">#&gt; <span style="color: #00BB00;">✔</span> Calculating the probability of the observed cluster-mass statistics. <span style="color: #B2B2B2;">[24ms]</span></span></span>
+<span><span class="co">#&gt; <span style="color: #00BB00;">✔</span> Calculating the probability of the observed cluster-mass statistics. <span style="color: #B2B2B2;">[36ms]</span></span></span>
 <span><span class="co">#&gt; </span></span></code></pre></div>
 <p>Note how <code><a href="https://future.futureverse.org/reference/value.html" class="external-link">value()</a></code> also prints the messages encountered
 while executing the CPA code. We see this because we did not specify
diff --git a/docs/articles/deCarvalho-et-al-2021.html b/docs/articles/deCarvalho-et-al-2021.html
index 127672c..fa4e987 100644
--- a/docs/articles/deCarvalho-et-al-2021.html
+++ b/docs/articles/deCarvalho-et-al-2021.html
@@ -35,7 +35,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/articles/eyetrackingR-comparison.html b/docs/articles/eyetrackingR-comparison.html
index 9b401a8..793a196 100644
--- a/docs/articles/eyetrackingR-comparison.html
+++ b/docs/articles/eyetrackingR-comparison.html
@@ -35,7 +35,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -122,20 +122,21 @@ <h2 id="setup">Setup<a class="anchor" aria-label="anchor" href="#setup"></a>
 <span><span class="kw"><a href="https://rdrr.io/r/base/library.html" class="external-link">library</a></span><span class="op">(</span><span class="st"><a href="http://samforbes.me/eyetrackingR/" class="external-link">"eyetrackingR"</a></span><span class="op">)</span></span>
 <span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/utils/data.html" class="external-link">data</a></span><span class="op">(</span><span class="st">"word_recognition"</span><span class="op">)</span></span>
-<span><span class="va">data</span> <span class="op">&lt;-</span> <span class="fu"><a href="http://samforbes.me/eyetrackingR/reference/make_eyetrackingr_data.html" class="external-link">make_eyetrackingr_data</a></span><span class="op">(</span><span class="va">word_recognition</span>, </span>
-<span>                       participant_column <span class="op">=</span> <span class="st">"ParticipantName"</span>,</span>
-<span>                       trial_column <span class="op">=</span> <span class="st">"Trial"</span>,</span>
-<span>                       time_column <span class="op">=</span> <span class="st">"TimeFromTrialOnset"</span>,</span>
-<span>                       trackloss_column <span class="op">=</span> <span class="st">"TrackLoss"</span>,</span>
-<span>                       aoi_columns <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="st">'Animate'</span>,<span class="st">'Inanimate'</span><span class="op">)</span>,</span>
-<span>                       treat_non_aoi_looks_as_missing <span class="op">=</span> <span class="cn">TRUE</span></span>
-<span>                <span class="op">)</span></span>
+<span><span class="va">data</span> <span class="op">&lt;-</span> <span class="fu"><a href="http://samforbes.me/eyetrackingR/reference/make_eyetrackingr_data.html" class="external-link">make_eyetrackingr_data</a></span><span class="op">(</span><span class="va">word_recognition</span>,</span>
+<span>  participant_column <span class="op">=</span> <span class="st">"ParticipantName"</span>,</span>
+<span>  trial_column <span class="op">=</span> <span class="st">"Trial"</span>,</span>
+<span>  time_column <span class="op">=</span> <span class="st">"TimeFromTrialOnset"</span>,</span>
+<span>  trackloss_column <span class="op">=</span> <span class="st">"TrackLoss"</span>,</span>
+<span>  aoi_columns <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="st">"Animate"</span>, <span class="st">"Inanimate"</span><span class="op">)</span>,</span>
+<span>  treat_non_aoi_looks_as_missing <span class="op">=</span> <span class="cn">TRUE</span></span>
+<span><span class="op">)</span></span>
 <span></span>
 <span><span class="co"># subset to response window post word-onset</span></span>
-<span><span class="va">response_window</span> <span class="op">&lt;-</span> <span class="fu"><a href="http://samforbes.me/eyetrackingR/reference/subset_by_window.html" class="external-link">subset_by_window</a></span><span class="op">(</span><span class="va">data</span>, </span>
-<span>                                    window_start_time <span class="op">=</span> <span class="fl">15500</span>, </span>
-<span>                                    window_end_time <span class="op">=</span> <span class="fl">21000</span>, </span>
-<span>                                    rezero <span class="op">=</span> <span class="cn">FALSE</span><span class="op">)</span></span>
+<span><span class="va">response_window</span> <span class="op">&lt;-</span> <span class="fu"><a href="http://samforbes.me/eyetrackingR/reference/subset_by_window.html" class="external-link">subset_by_window</a></span><span class="op">(</span><span class="va">data</span>,</span>
+<span>  window_start_time <span class="op">=</span> <span class="fl">15500</span>,</span>
+<span>  window_end_time <span class="op">=</span> <span class="fl">21000</span>,</span>
+<span>  rezero <span class="op">=</span> <span class="cn">FALSE</span></span>
+<span><span class="op">)</span></span>
 <span></span>
 <span><span class="co"># analyze amount of trackloss by subjects and trials</span></span>
 <span><span class="op">(</span><span class="va">trackloss</span> <span class="op">&lt;-</span> <span class="fu"><a href="http://samforbes.me/eyetrackingR/reference/trackloss_analysis.html" class="external-link">trackloss_analysis</a></span><span class="op">(</span>data <span class="op">=</span> <span class="va">response_window</span><span class="op">)</span><span class="op">)</span></span>
@@ -156,19 +157,23 @@ <h2 id="setup">Setup<a class="anchor" aria-label="anchor" href="#setup"></a>
 <span><span class="co">#&gt; <span style="color: #949494;"># ℹ 1 more variable: TracklossForParticipant &lt;dbl&gt;</span></span></span>
 <span></span>
 <span><span class="co"># remove trials with &gt; 25% of trackloss</span></span>
-<span><span class="va">response_window_clean</span> <span class="op">&lt;-</span> <span class="fu"><a href="http://samforbes.me/eyetrackingR/reference/clean_by_trackloss.html" class="external-link">clean_by_trackloss</a></span><span class="op">(</span>data <span class="op">=</span> <span class="va">response_window</span>,</span>
-<span>                                            trial_prop_thresh <span class="op">=</span> <span class="fl">.25</span><span class="op">)</span></span>
+<span><span class="va">response_window_clean</span> <span class="op">&lt;-</span> <span class="fu"><a href="http://samforbes.me/eyetrackingR/reference/clean_by_trackloss.html" class="external-link">clean_by_trackloss</a></span><span class="op">(</span></span>
+<span>  data <span class="op">=</span> <span class="va">response_window</span>,</span>
+<span>  trial_prop_thresh <span class="op">=</span> <span class="fl">.25</span></span>
+<span><span class="op">)</span></span>
 <span></span>
 <span><span class="co"># create Target condition column</span></span>
-<span><span class="va">response_window_clean</span><span class="op">$</span><span class="va">Target</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/factor.html" class="external-link">as.factor</a></span><span class="op">(</span> <span class="fu"><a href="https://rdrr.io/r/base/ifelse.html" class="external-link">ifelse</a></span><span class="op">(</span>test <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/grep.html" class="external-link">grepl</a></span><span class="op">(</span><span class="st">'(Spoon|Bottle)'</span>, <span class="va">response_window_clean</span><span class="op">$</span><span class="va">Trial</span><span class="op">)</span>, </span>
-<span>                                       yes <span class="op">=</span> <span class="st">'Inanimate'</span>, </span>
-<span>                                       no  <span class="op">=</span> <span class="st">'Animate'</span><span class="op">)</span> <span class="op">)</span></span>
+<span><span class="va">response_window_clean</span><span class="op">$</span><span class="va">Target</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/factor.html" class="external-link">as.factor</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/ifelse.html" class="external-link">ifelse</a></span><span class="op">(</span>test <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/grep.html" class="external-link">grepl</a></span><span class="op">(</span><span class="st">"(Spoon|Bottle)"</span>, <span class="va">response_window_clean</span><span class="op">$</span><span class="va">Trial</span><span class="op">)</span>,</span>
+<span>  yes <span class="op">=</span> <span class="st">"Inanimate"</span>,</span>
+<span>  no  <span class="op">=</span> <span class="st">"Animate"</span></span>
+<span><span class="op">)</span><span class="op">)</span></span>
 <span></span>
 <span><span class="va">response_time</span> <span class="op">&lt;-</span> <span class="fu"><a href="http://samforbes.me/eyetrackingR/reference/make_time_sequence_data.html" class="external-link">make_time_sequence_data</a></span><span class="op">(</span><span class="va">response_window_clean</span>,</span>
-<span>                                  time_bin_size <span class="op">=</span> <span class="fl">100</span>, </span>
-<span>                                  predictor_columns <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="st">"Target"</span><span class="op">)</span>,</span>
-<span>                                  aois <span class="op">=</span> <span class="st">"Animate"</span>,</span>
-<span>                                  summarize_by <span class="op">=</span> <span class="st">"ParticipantName"</span> <span class="op">)</span></span></code></pre></div>
+<span>  time_bin_size <span class="op">=</span> <span class="fl">100</span>,</span>
+<span>  predictor_columns <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="st">"Target"</span><span class="op">)</span>,</span>
+<span>  aois <span class="op">=</span> <span class="st">"Animate"</span>,</span>
+<span>  summarize_by <span class="op">=</span> <span class="st">"ParticipantName"</span></span>
+<span><span class="op">)</span></span></code></pre></div>
 </details><p>We pick up from where the <code>response_time</code> dataframe is
 created. By this stage, the data has been filtered for track loss and
 aggregated by subject and by time bins, among other things.</p>
@@ -206,9 +211,11 @@ <h2 id="setup">Setup<a class="anchor" aria-label="anchor" href="#setup"></a>
 <p>For simplicity, we subset <code>response_time</code> to include only
 the columns of interest for the cluster-based permutation analysis.</p>
 <div class="sourceCode" id="cb2"><pre class="downlit sourceCode r">
-<code class="sourceCode R"><span><span class="va">response_time</span> <span class="op">&lt;-</span> <span class="va">response_time</span> <span class="op"><a href="https://magrittr.tidyverse.org/reference/pipe.html" class="external-link">%&gt;%</a></span> </span>
-<span>  <span class="fu"><a href="https://dplyr.tidyverse.org/reference/select.html" class="external-link">select</a></span><span class="op">(</span><span class="va">Prop</span>, <span class="va">Target</span>, <span class="va">TimeBin</span>, <span class="va">ParticipantName</span>,</span>
-<span>         <span class="va">Time</span>, <span class="va">AOI</span><span class="op">)</span></span>
+<code class="sourceCode R"><span><span class="va">response_time</span> <span class="op">&lt;-</span> <span class="va">response_time</span> <span class="op"><a href="https://magrittr.tidyverse.org/reference/pipe.html" class="external-link">%&gt;%</a></span></span>
+<span>  <span class="fu"><a href="https://dplyr.tidyverse.org/reference/select.html" class="external-link">select</a></span><span class="op">(</span></span>
+<span>    <span class="va">Prop</span>, <span class="va">Target</span>, <span class="va">TimeBin</span>, <span class="va">ParticipantName</span>,</span>
+<span>    <span class="va">Time</span>, <span class="va">AOI</span></span>
+<span>  <span class="op">)</span></span>
 <span><span class="fu">dplyr</span><span class="fu">::</span><span class="fu"><a href="https://pillar.r-lib.org/reference/glimpse.html" class="external-link">glimpse</a></span><span class="op">(</span><span class="va">response_time</span><span class="op">)</span></span>
 <span><span class="co">#&gt; Rows: 2,970</span></span>
 <span><span class="co">#&gt; Columns: 6</span></span>
@@ -221,9 +228,9 @@ <h2 id="setup">Setup<a class="anchor" aria-label="anchor" href="#setup"></a>
 <p>Lastly, we replicate the plot from the original vignette:</p>
 <div class="sourceCode" id="cb3"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="co"># visualize timecourse</span></span>
-<span><span class="fu"><a href="https://rdrr.io/r/graphics/plot.default.html" class="external-link">plot</a></span><span class="op">(</span><span class="va">response_time</span>, predictor_column <span class="op">=</span> <span class="st">"Target"</span><span class="op">)</span> <span class="op">+</span> </span>
+<span><span class="fu"><a href="https://rdrr.io/r/graphics/plot.default.html" class="external-link">plot</a></span><span class="op">(</span><span class="va">response_time</span>, predictor_column <span class="op">=</span> <span class="st">"Target"</span><span class="op">)</span> <span class="op">+</span></span>
 <span>  <span class="fu"><a href="https://ggplot2.tidyverse.org/reference/ggtheme.html" class="external-link">theme_light</a></span><span class="op">(</span><span class="op">)</span> <span class="op">+</span></span>
-<span>  <span class="fu"><a href="https://ggplot2.tidyverse.org/reference/coord_cartesian.html" class="external-link">coord_cartesian</a></span><span class="op">(</span>ylim <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>,<span class="fl">1</span><span class="op">)</span><span class="op">)</span></span>
+<span>  <span class="fu"><a href="https://ggplot2.tidyverse.org/reference/coord_cartesian.html" class="external-link">coord_cartesian</a></span><span class="op">(</span>ylim <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>, <span class="fl">1</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="co">#&gt; Warning: Removed 37 rows containing non-finite values (`stat_summary()`).</span></span>
 <span><span class="co">#&gt; Removed 37 rows containing non-finite values (`stat_summary()`).</span></span></code></pre></div>
 <p><img src="eyetrackingR-comparison_files/figure-html/unnamed-chunk-3-1.png" width="100%"></p>
@@ -239,8 +246,8 @@ <h2 id="cpa-in-eyetrackingr">CPA in <code>{eyetrackingR}</code><a class="anchor"
 </summary><p>The original vignette uses the following heuristic of choosing a
 threshold:</p>
 <div class="sourceCode" id="cb4"><pre class="downlit sourceCode r">
-<code class="sourceCode R"><span><span class="va">num_sub</span> <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/length.html" class="external-link">length</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/unique.html" class="external-link">unique</a></span><span class="op">(</span><span class="op">(</span><span class="va">response_window_clean</span><span class="op">$</span><span class="va">ParticipantName</span><span class="op">)</span><span class="op">)</span><span class="op">)</span></span>
-<span><span class="va">threshold_t</span> <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/stats/TDist.html" class="external-link">qt</a></span><span class="op">(</span>p <span class="op">=</span> <span class="fl">1</span> <span class="op">-</span> <span class="fl">.05</span><span class="op">/</span><span class="fl">2</span>, df <span class="op">=</span> <span class="va">num_sub</span><span class="op">-</span><span class="fl">1</span><span class="op">)</span></span>
+<code class="sourceCode R"><span><span class="va">num_sub</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/length.html" class="external-link">length</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/unique.html" class="external-link">unique</a></span><span class="op">(</span><span class="op">(</span><span class="va">response_window_clean</span><span class="op">$</span><span class="va">ParticipantName</span><span class="op">)</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="va">threshold_t</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/stats/TDist.html" class="external-link">qt</a></span><span class="op">(</span>p <span class="op">=</span> <span class="fl">1</span> <span class="op">-</span> <span class="fl">.05</span> <span class="op">/</span> <span class="fl">2</span>, df <span class="op">=</span> <span class="va">num_sub</span> <span class="op">-</span> <span class="fl">1</span><span class="op">)</span></span>
 <span><span class="va">threshold_t</span></span>
 <span><span class="co">#&gt; [1] 2.055529</span></span></code></pre></div>
 <p>How to choose a threshold is outside the scope of this article - we
@@ -251,10 +258,11 @@ <h2 id="cpa-in-eyetrackingr">CPA in <code>{eyetrackingR}</code><a class="anchor"
 <p>Prepare data for CPA with
 <code><a href="http://samforbes.me/eyetrackingR/reference/make_time_cluster_data.html" class="external-link">make_time_cluster_data()</a></code>:</p>
 <div class="sourceCode" id="cb5"><pre class="downlit sourceCode r">
-<code class="sourceCode R"><span><span class="va">df_timeclust</span> <span class="op">&lt;-</span> <span class="fu"><a href="http://samforbes.me/eyetrackingR/reference/make_time_cluster_data.html" class="external-link">make_time_cluster_data</a></span><span class="op">(</span><span class="va">response_time</span>, </span>
-<span>                                      test<span class="op">=</span> <span class="st">"t.test"</span>, paired<span class="op">=</span><span class="cn">TRUE</span>,</span>
-<span>                                      predictor_column <span class="op">=</span> <span class="st">"Target"</span>, </span>
-<span>                                      threshold <span class="op">=</span> <span class="va">threshold_t</span><span class="op">)</span></span></code></pre></div>
+<code class="sourceCode R"><span><span class="va">df_timeclust</span> <span class="op">&lt;-</span> <span class="fu"><a href="http://samforbes.me/eyetrackingR/reference/make_time_cluster_data.html" class="external-link">make_time_cluster_data</a></span><span class="op">(</span><span class="va">response_time</span>,</span>
+<span>  test <span class="op">=</span> <span class="st">"t.test"</span>, paired <span class="op">=</span> <span class="cn">TRUE</span>,</span>
+<span>  predictor_column <span class="op">=</span> <span class="st">"Target"</span>,</span>
+<span>  threshold <span class="op">=</span> <span class="va">threshold_t</span></span>
+<span><span class="op">)</span></span></code></pre></div>
 <p>This step computes the timewise statistics from the data and
 identifies the empirical clusters, which can be inspected with a
 <code><a href="https://rdrr.io/r/graphics/plot.default.html" class="external-link">plot()</a></code> and <code><a href="https://rdrr.io/r/base/summary.html" class="external-link">summary()</a></code> method:</p>
@@ -276,18 +284,16 @@ <h2 id="cpa-in-eyetrackingr">CPA in <code>{eyetrackingR}</code><a class="anchor"
 <code><a href="http://samforbes.me/eyetrackingR/reference/analyze_time_clusters.html" class="external-link">analyze_time_clusters()</a></code>:</p>
 <div class="sourceCode" id="cb8"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/system.time.html" class="external-link">system.time</a></span><span class="op">(</span><span class="op">{</span></span>
-<span></span>
 <span>  <span class="va">clust_analysis</span> <span class="op">&lt;-</span> <span class="fu"><a href="http://samforbes.me/eyetrackingR/reference/analyze_time_clusters.html" class="external-link">analyze_time_clusters</a></span><span class="op">(</span></span>
-<span>    <span class="va">df_timeclust</span>,</span>
-<span>    within_subj <span class="op">=</span> <span class="cn">TRUE</span>,</span>
-<span>    paired <span class="op">=</span> <span class="cn">TRUE</span>,</span>
-<span>    samples <span class="op">=</span> <span class="fl">150</span>,</span>
-<span>    quiet <span class="op">=</span> <span class="cn">TRUE</span></span>
+<span><span class="va">df_timeclust</span>,</span>
+<span>within_subj <span class="op">=</span> <span class="cn">TRUE</span>,</span>
+<span>paired <span class="op">=</span> <span class="cn">TRUE</span>,</span>
+<span>samples <span class="op">=</span> <span class="fl">150</span>,</span>
+<span>quiet <span class="op">=</span> <span class="cn">TRUE</span></span>
 <span>  <span class="op">)</span></span>
-<span></span>
 <span><span class="op">}</span><span class="op">)</span></span>
 <span><span class="co">#&gt;    user  system elapsed </span></span>
-<span><span class="co">#&gt;   13.72    0.10   34.46</span></span></code></pre></div>
+<span><span class="co">#&gt;   19.48    0.14   36.24</span></span></code></pre></div>
 <p>This simulates a null distribution of cluster-mass statistics. The
 output, when printed, is essentially the output of
 <code>summary(df_timeclust)</code> with p-values (the
@@ -322,13 +328,11 @@ <h2 id="cpa-in-jlmerclusterperm">CPA in <code>{jlmerclusterperm}</code><a class=
 <code class="sourceCode R"><span><span class="kw"><a href="https://rdrr.io/r/base/library.html" class="external-link">library</a></span><span class="op">(</span><span class="va"><a href="https://github.com/yjunechoe/jlmerclusterperm" class="external-link">jlmerclusterperm</a></span><span class="op">)</span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/base/options.html" class="external-link">options</a></span><span class="op">(</span><span class="st">"jlmerclusterperm.nthreads"</span> <span class="op">=</span> <span class="fl">1</span><span class="op">)</span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/base/system.time.html" class="external-link">system.time</a></span><span class="op">(</span><span class="op">{</span></span>
-<span>  </span>
 <span>  <span class="co"># Note the overhead for initializing the Julia session</span></span>
 <span>  <span class="fu"><a href="../reference/jlmerclusterperm_setup.html">jlmerclusterperm_setup</a></span><span class="op">(</span><span class="op">)</span></span>
-<span>  </span>
 <span><span class="op">}</span><span class="op">)</span></span>
 <span><span class="co">#&gt;    user  system elapsed </span></span>
-<span><span class="co">#&gt;    0.00    0.01   23.02</span></span></code></pre></div>
+<span><span class="co">#&gt;    0.04    0.01   26.13</span></span></code></pre></div>
 <p>In <code>{jlmerclusterpm}</code>, CPA can also be conducted in two
 steps:</p>
 <ol style="list-style-type: decimal">
@@ -377,7 +381,7 @@ <h2 id="cpa-in-jlmerclusterperm">CPA in <code>{jlmerclusterperm}</code><a class=
 <span>  <span class="va">cpa</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/clusterpermute.html">clusterpermute</a></span><span class="op">(</span><span class="va">spec</span>, threshold <span class="op">=</span> <span class="va">threshold_t</span>, nsim <span class="op">=</span> <span class="fl">150</span><span class="op">)</span></span>
 <span><span class="op">}</span><span class="op">)</span></span>
 <span><span class="co">#&gt;    user  system elapsed </span></span>
-<span><span class="co">#&gt;    0.02    0.00   15.63</span></span></code></pre></div>
+<span><span class="co">#&gt;    0.05    0.00   16.29</span></span></code></pre></div>
 <p>The same kinds of information are returned:</p>
 <div class="sourceCode" id="cb15"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">cpa</span></span>
@@ -430,7 +434,7 @@ <h2 id="cpa-in-jlmerclusterperm">CPA in <code>{jlmerclusterperm}</code><a class=
 replicated with a few lines of ggplot code:</p>
 <div class="sourceCode" id="cb18"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="co"># We flip the sign of the statistic here for visual equivalence</span></span>
-<span><span class="va">cpa_plot</span> <span class="op">&lt;-</span> <span class="va">null_distribution</span> <span class="op"><a href="https://magrittr.tidyverse.org/reference/pipe.html" class="external-link">%&gt;%</a></span> </span>
+<span><span class="va">cpa_plot</span> <span class="op">&lt;-</span> <span class="va">null_distribution</span> <span class="op"><a href="https://magrittr.tidyverse.org/reference/pipe.html" class="external-link">%&gt;%</a></span></span>
 <span>  <span class="fu"><a href="https://ggplot2.tidyverse.org/reference/ggplot.html" class="external-link">ggplot</a></span><span class="op">(</span><span class="fu"><a href="https://ggplot2.tidyverse.org/reference/aes.html" class="external-link">aes</a></span><span class="op">(</span><span class="op">-</span><span class="va">sum_statistic</span><span class="op">)</span><span class="op">)</span> <span class="op">+</span></span>
 <span>  <span class="fu"><a href="https://ggplot2.tidyverse.org/reference/geom_histogram.html" class="external-link">geom_histogram</a></span><span class="op">(</span></span>
 <span>    <span class="fu"><a href="https://ggplot2.tidyverse.org/reference/aes.html" class="external-link">aes</a></span><span class="op">(</span>y <span class="op">=</span> <span class="fu"><a href="https://ggplot2.tidyverse.org/reference/aes_eval.html" class="external-link">after_stat</a></span><span class="op">(</span><span class="va">density</span><span class="op">)</span><span class="op">)</span>,</span>
@@ -469,7 +473,7 @@ <h2 id="performance">Performance<a class="anchor" aria-label="anchor" href="#per
 <span>  <span class="fu"><a href="../reference/clusterpermute.html">clusterpermute</a></span><span class="op">(</span><span class="va">spec</span>, threshold <span class="op">=</span> <span class="va">threshold_t</span>, nsim <span class="op">=</span> <span class="fl">1000</span><span class="op">)</span></span>
 <span><span class="op">}</span><span class="op">)</span></span>
 <span><span class="co">#&gt;    user  system elapsed </span></span>
-<span><span class="co">#&gt;    0.04    0.02   10.59</span></span></code></pre></div>
+<span><span class="co">#&gt;    0.00    0.04   12.34</span></span></code></pre></div>
 <p>While <a href="http://samforbes.me/eyetrackingR/" class="external-link">eyetrackingR</a> does have an option for
 parallelization, it has limited support and is platform dependent. In
 contrast, <a href="https://github.com/yjunechoe/jlmerclusterperm" class="external-link">jlmerclusterperm</a> leverages Julia’s built-in
@@ -480,14 +484,14 @@ <h2 id="performance">Performance<a class="anchor" aria-label="anchor" href="#per
 <span>  <span class="fu"><a href="../reference/jlmerclusterperm_setup.html">jlmerclusterperm_setup</a></span><span class="op">(</span>restart <span class="op">=</span> <span class="cn">TRUE</span>, verbose <span class="op">=</span> <span class="cn">FALSE</span><span class="op">)</span></span>
 <span><span class="op">}</span><span class="op">)</span></span>
 <span><span class="co">#&gt;    user  system elapsed </span></span>
-<span><span class="co">#&gt;    0.09    0.03   20.78</span></span></code></pre></div>
+<span><span class="co">#&gt;    0.06    0.01   22.89</span></span></code></pre></div>
 <p>10,000 simulations with 7 threads:</p>
 <div class="sourceCode" id="cb22"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/system.time.html" class="external-link">system.time</a></span><span class="op">(</span><span class="op">{</span></span>
 <span>  <span class="fu"><a href="../reference/clusterpermute.html">clusterpermute</a></span><span class="op">(</span><span class="va">spec</span>, threshold <span class="op">=</span> <span class="va">threshold_t</span>, nsim <span class="op">=</span> <span class="fl">10000</span><span class="op">)</span></span>
 <span><span class="op">}</span><span class="op">)</span></span>
 <span><span class="co">#&gt;    user  system elapsed </span></span>
-<span><span class="co">#&gt;    0.22    0.32   48.99</span></span></code></pre></div>
+<span><span class="co">#&gt;    0.10    0.16   52.57</span></span></code></pre></div>
 </div>
   </main><aside class="col-md-3"><nav id="toc"><h2>On this page</h2>
     </nav></aside>
diff --git a/docs/articles/index.html b/docs/articles/index.html
index 7ad5d05..77cd0e0 100644
--- a/docs/articles/index.html
+++ b/docs/articles/index.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/articles/jlmerclusterperm.html b/docs/articles/jlmerclusterperm.html
index 336bb45..7bc9133 100644
--- a/docs/articles/jlmerclusterperm.html
+++ b/docs/articles/jlmerclusterperm.html
@@ -35,7 +35,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/articles/julia-interface.html b/docs/articles/julia-interface.html
index ea8158d..2cb8e73 100644
--- a/docs/articles/julia-interface.html
+++ b/docs/articles/julia-interface.html
@@ -35,7 +35,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/articles/reproducibility.html b/docs/articles/reproducibility.html
index 39f2d29..8daf434 100644
--- a/docs/articles/reproducibility.html
+++ b/docs/articles/reproducibility.html
@@ -35,7 +35,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/articles/tidying-output.html b/docs/articles/tidying-output.html
index 7581e58..9d05211 100644
--- a/docs/articles/tidying-output.html
+++ b/docs/articles/tidying-output.html
@@ -35,7 +35,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -373,7 +373,7 @@ <h3 id="in-other-contexts">In other contexts<a class="anchor" aria-label="anchor
 <span>  <span class="va">full_test</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/clusterpermute.html">clusterpermute</a></span><span class="op">(</span><span class="va">jlmer_spec</span>, threshold <span class="op">=</span> <span class="fl">1</span>, nsim <span class="op">=</span> <span class="fl">1000</span>, progress <span class="op">=</span> <span class="cn">FALSE</span><span class="op">)</span></span>
 <span><span class="op">}</span><span class="op">)</span></span>
 <span><span class="co">#&gt;    user  system elapsed </span></span>
-<span><span class="co">#&gt;    0.02    0.04    2.12</span></span>
+<span><span class="co">#&gt;    0.02    0.00    2.14</span></span>
 <span><span class="va">full_test</span></span>
 <span><span class="co">#&gt; $null_cluster_dists</span></span>
 <span><span class="co">#&gt; ── <span style="font-weight: bold;">Null cluster-mass distribution</span> (t &gt; <span style="color: #0000BB;">1</span>) ────────────── <span style="color: #0000BB;">&lt;null_cluster_dists&gt;</span> ──</span></span>
diff --git a/docs/authors.html b/docs/authors.html
index 54b4eeb..4837bda 100644
--- a/docs/authors.html
+++ b/docs/authors.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/index.html b/docs/index.html
index 3c51fb8..b48b8aa 100644
--- a/docs/index.html
+++ b/docs/index.html
@@ -35,7 +35,7 @@
     
     <a class="navbar-brand me-2" href="index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -108,7 +108,7 @@ <h2 id="installation-and-usage">Installation and usage<a class="anchor" aria-lab
 <div class="sourceCode" id="cb2"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="co"># install.packages("remotes")</span></span>
 <span><span class="fu">remotes</span><span class="fu">::</span><span class="fu"><a href="https://remotes.r-lib.org/reference/install_github.html" class="external-link">install_github</a></span><span class="op">(</span><span class="st">"yjunechoe/jlmerclusterperm"</span><span class="op">)</span></span></code></pre></div>
-<p>Using <code>jlmerclusterperm</code> requires a prior installation of the Julia programming language, which can be downloaded from either the <a href="https://julialang.org/" class="external-link">official website</a> or using the command line utility <a href="https://github.com/JuliaLang/juliaup" class="external-link">juliaup</a>. Julia version &gt;=1.8 is required and <a href="https://julialang.org/blog/2023/04/julia-1.9-highlights/#caching_of_native_code" class="external-link">1.9</a> is preferred for its substantial speed improvements.</p>
+<p>Using <code>jlmerclusterperm</code> requires a prior installation of the Julia programming language, which can be downloaded from either the <a href="https://julialang.org/" class="external-link">official website</a> or using the command line utility <a href="https://github.com/JuliaLang/juliaup" class="external-link">juliaup</a>. Julia version &gt;=1.8 is required and <a href="https://julialang.org/blog/2023/04/julia-1.9-highlights/#caching_of_native_code" class="external-link">1.9</a> or higher is preferred for the substantial speed improvements.</p>
 <p>Before using functions from <code>jlmerclusterperm</code>, an initial setup is required via calling <code><a href="reference/jlmerclusterperm_setup.html">jlmerclusterperm_setup()</a></code>. The very first call on a system will install necessary dependencies (this only happens once and takes around 10-15 minutes).</p>
 <p>Subsequent calls to <code><a href="reference/jlmerclusterperm_setup.html">jlmerclusterperm_setup()</a></code> incur a small overhead of around 30 seconds, plus slight delays for first-time function calls. You pay up front for start-up and warm-up costs and get blazingly-fast functions from the package.</p>
 <div class="sourceCode" id="cb3"><pre class="downlit sourceCode r">
@@ -288,7 +288,7 @@ <h2 id="citations">Citations<a class="anchor" aria-label="anchor" href="#citatio
 <p>If you use jlmerclusterperm for cluster-based permutation test with mixed-effects models in your research, please cite one (or more) of the following as you see fit.</p>
 <p>To cite jlmerclusterperm:</p>
 <ul>
-<li>Choe, J. (2023). jlmerclusterperm: Cluster-Based Permutation Analysis for Densely Sampled Time Data. R package version 1.1.0. <a href="https://cran.r-project.org/package=jlmerclusterperm" class="external-link uri">https://cran.r-project.org/package=jlmerclusterperm</a>.</li>
+<li>Choe, J. (2023). jlmerclusterperm: Cluster-Based Permutation Analysis for Densely Sampled Time Data. R package version 1.1.1. <a href="https://cran.r-project.org/package=jlmerclusterperm" class="external-link uri">https://cran.r-project.org/package=jlmerclusterperm</a>.</li>
 </ul>
 <p>To cite the cluster-based permutation test:</p>
 <ul>
diff --git a/docs/news/index.html b/docs/news/index.html
index d26c1f8..c4bc5d6 100644
--- a/docs/news/index.html
+++ b/docs/news/index.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -64,9 +64,15 @@ <h6 class="dropdown-header" data-toc-skip>Case Studies</h6>
     </div>
 
     <div class="section level2">
-<h2 class="pkg-version" data-toc-text="development version" id="jlmerclusterperm-development-version">jlmerclusterperm (development version)<a class="anchor" aria-label="anchor" href="#jlmerclusterperm-development-version"></a></h2>
+<h2 class="pkg-version" data-toc-text="1.1.1" id="jlmerclusterperm-111">jlmerclusterperm 1.1.1<a class="anchor" aria-label="anchor" href="#jlmerclusterperm-111"></a></h2><p class="text-muted">CRAN release: 2024-02-08</p>
 </div>
     <div class="section level2">
+<h2 class="pkg-version" data-toc-text="1.1.1" id="jlmerclusterperm-111-1">jlmerclusterperm 1.1.1<a class="anchor" aria-label="anchor" href="#jlmerclusterperm-111-1"></a></h2><p class="text-muted">CRAN release: 2024-02-08</p>
+<ul><li><p>More informative warnings when constructing a spec object</p></li>
+<li><p>Fixed minor bugs when interfacing with Julia</p></li>
+<li><p>Fixed bug with reporting convergence warnings when only one is encountered</p></li>
+</ul></div>
+    <div class="section level2">
 <h2 class="pkg-version" data-toc-text="1.1.0" id="jlmerclusterperm-110">jlmerclusterperm 1.1.0<a class="anchor" aria-label="anchor" href="#jlmerclusterperm-110"></a></h2><p class="text-muted">CRAN release: 2023-10-07</p>
 <p>Minor breaking change:</p>
 <ul><li>Slightly loosened the default convergence criterion for <code><a href="../reference/permute_timewise_statistics.html">permute_timewise_statistics()</a></code>
diff --git a/docs/pkgdown.yml b/docs/pkgdown.yml
index 56971f0..adf2c41 100644
--- a/docs/pkgdown.yml
+++ b/docs/pkgdown.yml
@@ -11,7 +11,7 @@ articles:
   reproducibility: reproducibility.html
   tidying-output: tidying-output.html
   jlmerclusterperm: jlmerclusterperm.html
-last_built: 2024-01-23T18:45Z
+last_built: 2024-02-08T13:14Z
 urls:
   reference: https://yjunechoe.github.io/jlmerclusterperm/reference
   article: https://yjunechoe.github.io/jlmerclusterperm/articles
diff --git a/docs/reference/calculate_clusters_pvalues.html b/docs/reference/calculate_clusters_pvalues.html
index 9e8aad7..7519372 100644
--- a/docs/reference/calculate_clusters_pvalues.html
+++ b/docs/reference/calculate_clusters_pvalues.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/cluster_permutation_tidiers.html b/docs/reference/cluster_permutation_tidiers.html
index a8548ff..769845f 100644
--- a/docs/reference/cluster_permutation_tidiers.html
+++ b/docs/reference/cluster_permutation_tidiers.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/clusterpermute.html b/docs/reference/clusterpermute.html
index 6df67a9..3aebe27 100644
--- a/docs/reference/clusterpermute.html
+++ b/docs/reference/clusterpermute.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/compute_timewise_statistics.html b/docs/reference/compute_timewise_statistics.html
index b5d08de..41cb516 100644
--- a/docs/reference/compute_timewise_statistics.html
+++ b/docs/reference/compute_timewise_statistics.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/extract_empirical_clusters.html b/docs/reference/extract_empirical_clusters.html
index 3793e9b..e6660ce 100644
--- a/docs/reference/extract_empirical_clusters.html
+++ b/docs/reference/extract_empirical_clusters.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/extract_null_cluster_dists.html b/docs/reference/extract_null_cluster_dists.html
index 028b596..fd4853c 100644
--- a/docs/reference/extract_null_cluster_dists.html
+++ b/docs/reference/extract_null_cluster_dists.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/index.html b/docs/reference/index.html
index 2a804af..1998f78 100644
--- a/docs/reference/index.html
+++ b/docs/reference/index.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/jlmer.html b/docs/reference/jlmer.html
index 0034efa..e790418 100644
--- a/docs/reference/jlmer.html
+++ b/docs/reference/jlmer.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/jlmerclusterperm-package.html b/docs/reference/jlmerclusterperm-package.html
index 8037a5e..7aec71e 100644
--- a/docs/reference/jlmerclusterperm-package.html
+++ b/docs/reference/jlmerclusterperm-package.html
@@ -14,7 +14,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/jlmerclusterperm_setup.html b/docs/reference/jlmerclusterperm_setup.html
index 8813ac7..2302030 100644
--- a/docs/reference/jlmerclusterperm_setup.html
+++ b/docs/reference/jlmerclusterperm_setup.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/julia_model_tidiers.html b/docs/reference/julia_model_tidiers.html
index 74dd6c8..b5555d3 100644
--- a/docs/reference/julia_model_tidiers.html
+++ b/docs/reference/julia_model_tidiers.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/julia_progress.html b/docs/reference/julia_progress.html
index fa92a84..ff21c12 100644
--- a/docs/reference/julia_progress.html
+++ b/docs/reference/julia_progress.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/julia_rng.html b/docs/reference/julia_rng.html
index 63bd3c8..92b347c 100644
--- a/docs/reference/julia_rng.html
+++ b/docs/reference/julia_rng.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/julia_setup_ok.html b/docs/reference/julia_setup_ok.html
index 293b4ae..412f127 100644
--- a/docs/reference/julia_setup_ok.html
+++ b/docs/reference/julia_setup_ok.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/make_jlmer_spec.html b/docs/reference/make_jlmer_spec.html
index 7eb097c..beb6e65 100644
--- a/docs/reference/make_jlmer_spec.html
+++ b/docs/reference/make_jlmer_spec.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -145,7 +145,7 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-in"><span>  <span class="va">weight</span> <span class="op">~</span> <span class="fl">1</span> <span class="op">+</span> <span class="va">Diet</span>, <span class="va">ChickWeight</span>,</span></span>
 <span class="r-in"><span>  time <span class="op">=</span> <span class="st">"Time"</span></span></span>
 <span class="r-in"><span><span class="op">)</span></span></span>
-<span class="r-msg co"><span class="r-pr">#&gt;</span> <span style="color: #BBBB00;">!</span> Grouping column <span style="color: #0000BB;">"Time"</span> does not uniquely identify rows in the data</span>
+<span class="r-msg co"><span class="r-pr">#&gt;</span> <span style="color: #BBBB00;">!</span> Grouping column <span style="color: #0000BB;">"Time"</span> does not uniquely identify rows in the data.</span>
 <span class="r-msg co"><span class="r-pr">#&gt;</span> <span style="color: #BBBB00;">!</span> Sampling rate for the `time` column <span style="color: #0000BB;">"Time"</span> is not constant - may affect interpretability of results.</span>
 <span class="r-in"><span><span class="fu"><a href="https://rdrr.io/r/base/unique.html" class="external-link">unique</a></span><span class="op">(</span><span class="va">ChickWeight</span><span class="op">$</span><span class="va">Time</span><span class="op">)</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span>  [1]  0  2  4  6  8 10 12 14 16 18 20 21</span>
diff --git a/docs/reference/permute_by_predictor.html b/docs/reference/permute_by_predictor.html
index 15800ad..b0537f4 100644
--- a/docs/reference/permute_by_predictor.html
+++ b/docs/reference/permute_by_predictor.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/permute_timewise_statistics.html b/docs/reference/permute_timewise_statistics.html
index ce8680d..b722821 100644
--- a/docs/reference/permute_timewise_statistics.html
+++ b/docs/reference/permute_timewise_statistics.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/reexports.html b/docs/reference/reexports.html
index 5e69fd8..788882f 100644
--- a/docs/reference/reexports.html
+++ b/docs/reference/reexports.html
@@ -32,7 +32,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/to_jlmer.html b/docs/reference/to_jlmer.html
index ce4ddd1..157383f 100644
--- a/docs/reference/to_jlmer.html
+++ b/docs/reference/to_jlmer.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/reference/walk_threshold_steps.html b/docs/reference/walk_threshold_steps.html
index 41f9797..275cac2 100644
--- a/docs/reference/walk_threshold_steps.html
+++ b/docs/reference/walk_threshold_steps.html
@@ -10,7 +10,7 @@
     
     <a class="navbar-brand me-2" href="../index.html">jlmerclusterperm</a>
 
-    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.0</small>
+    <small class="nav-text text-muted me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="">1.1.1</small>
 
     
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/docs/search.json b/docs/search.json
index 00d2cca..24af39b 100644
--- a/docs/search.json
+++ b/docs/search.json
@@ -1 +1 @@
-[{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/asynchronous-cpa.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Asynchronous CPA","text":"Minimal example data CPA specification: Example CPA (1000 simulations): details actual CPA skipped ’s focus vignette, example readme just nsim.","code":"library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE) chickweights <- ChickWeight chickweights$Time <- as.integer(factor(chickweights$Time)) chickweights_spec <- make_jlmer_spec(   formula = weight ~ 1 + Diet,   data = chickweights,   subject = \"Chick\", time = \"Time\" ) set_rng_state(123L) CPA <- clusterpermute(   chickweights_spec,   threshold = 2.5,   nsim = 1000,   progress = FALSE )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/asynchronous-cpa.html","id":"basic-idea","dir":"Articles","previous_headings":"","what":"Basic idea","title":"Asynchronous CPA","text":"basic idea behind asynchronous CPA strategy start background R process whose sole job send instructions CPA Julia. ’s asynchronous running CPA way block interactive R session. Since work done Julia anyways, parallelization virtually impact performance evaluating R code. future package allows asynchronous evaluation R code. ’s big package implementing complex topic. can read project futureverse just show bare minimum . advanced users, note ’re use parallelization asynchronous properties (non-blocking evaluation R code background process). recommended start multiple R processes running CPA Julia session shared already multithreaded (can spare cores, set options(\"jlmerclusterperm.nthreads\") calling jlmerclusterperm_setup()). Later updates jlmerclusterperm may wrap workflow principled way, now vignette serves minimally working example asynchronously running CPA.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/asynchronous-cpa.html","id":"setup-for-async-cpa","dir":"Articles","previous_headings":"Walkthrough","what":"Setup for async CPA","title":"Asynchronous CPA","text":"Three things order workflow: Load future package Initialize multisession future Grab options jlmerclusterperm package environment Please treat .jlmerclusterperm internal variable read-object - ’s unexported meant manipulated.","code":"library(future) plan(multisession) pkgopts <- as.list(jlmerclusterperm:::.jlmerclusterperm)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/asynchronous-cpa.html","id":"creating-the-future-object","dir":"Articles","previous_headings":"Walkthrough","what":"Creating the future object","title":"Asynchronous CPA","text":"start creating special object class <Future> using future::future(). can simply pass clusterpermute() code first argument future(), slight modification ensure background process connects Julia session. template follows. First, make pkgopts object (defined ) available future (via globals argument). , inside future expression ensure background process shares Julia session (via list2env(...)). future object replicating 1000-simulation CPA look like following: future created, immediately starts executing code background process. important evaluate future object (f) directly. Instead, query future::resolved() - simply tells whether background evaluation completed : , use loop show background CPA non-blocking. evaluating R code interactive session simultaneously CPA running: reach end loop 2 seconds, approximately long CPA took complete (confirm next section). point future completed result available collection. Although R code just crude progress alert using Sys.sleep(), can freely evaluate R code except functions call Julia. can make plots, clean data, write analyses, etc., just don’t run another CPA one already running Julia process shared.","code":"# Not run future(   {     list2env(pkgopts, jlmerclusterperm:::.jlmerclusterperm)     ## Your CPA code below ##   },   globals = structure(TRUE, add = \"pkgopts\") ) f <- future(   {     list2env(as.list(pkgopts), jlmerclusterperm:::.jlmerclusterperm)     ## Your CPA code below ##     set_rng_state(123L)     clusterpermute(       chickweights_spec,       threshold = 2.5,       nsim = 1000     )   },   globals = structure(TRUE, add = \"pkgopts\") ) resolved(f) #> [1] FALSE i <- 0 while (!resolved(f)) {   Sys.sleep(0.5)   i <<- i + 1   cat(sprintf(\"Elapsed: %.01fs\", i / 2), \"\\n\") } #> Elapsed: 0.5s  #> Elapsed: 1.0s  #> Elapsed: 1.5s  #> Elapsed: 2.0s resolved(f) #> [1] TRUE"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/asynchronous-cpa.html","id":"collecting-the-results","dir":"Articles","previous_headings":"Walkthrough","what":"Collecting the results","title":"Asynchronous CPA","text":"can collect output background process future::value(). first, double check make sure background process indeed finished evaluating: ’s confirmed, can use value() collect results future assign variable inspection: Note value() also prints messages encountered executing CPA code. see specify progress = FALSE clusterpermute() call passed future. purposes, just note adding times messages similar saw loop (2 seconds). output asynchronous CPA (CPA_async) identical initial CPA ran beginning (CPA) ran default seed 1 RNG counter value 123L: separate vignette covers Julia RNG. practice, may want use random seed (via set_rng_seed()) background CPA.","code":"resolved(f) #> [1] TRUE CPA_async <- value(f) #> Connecting to Julia TCP server at localhost:11981 ... #> ℹ Detecting empirical clusters and calculating cluster-mass statistics. #> ✔ Detecting empirical clusters and calculating cluster-mass statistics. [57ms] #>  #> ℹ Sampling cluster-mass statistics from a bootstrapped null distribution. #> ✔ Sampling cluster-mass statistics from a bootstrapped null distribution. [1.6s] #>  #> ℹ Calculating the probability of the observed cluster-mass statistics. #> ✔ Calculating the probability of the observed cluster-mass statistics. [24ms] #> identical(CPA_async, CPA) #> [1] TRUE"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"background","dir":"Articles","previous_headings":"","what":"Background","title":"de Carvalho et al. 2021","text":"data comes eye-tracking study de Carvalho, Dautriche, Fiévet, & Christophe (2021) “Toddlers exploit referential syntactic cues flexibly adapt interpretation novel verb meanings.” article reproduces expands Experiment 2 analysis, used CPA (t-tests) conduct pairwise comparison three conditions (relevant Figure 7 paper ).  data comes mostly ready CPA straight study’s OSF repository. collapsible chunk contains code prepare data reproduce figure. Data cleaning Code Figure 7 plots data original study (close reproduce ). E2_data_agg dataframe following four columns: Subject: Unique identifier subjects Condition: -participant factor variable three levels (\"Intransitive\", \"RightDislocated\", \"Transitive\") Time: continuous measure time 0-8000ms 50ms intervals Prop: Proportion looks target (averaged across trials within condition, subject) following reproduction Figure 7 original paper:  original study used pairwise comparisons analyze relationship three conditions. smaller figures (B, C, D) plot following relationships: (B) looks target \"Transitive\" compared \"RightDislocated\" (C) looks target \"RightDislocated\" compared \"Intransitive\" (D) looks target \"Transitive\" compared \"Intransitive\" rest vignette organized follows: First, replicate individual pairwise CPAs original paper. Next, consider parsimonious analysis reformulates research hypothesis choice regression contrast. Lastly, build model (2) consider issues around model diagnostics complexity (linear vs. logistic; fixed vs. mixed) context CPA Load package start Julia instance jlmerclusterperm_setup() proceeding.","code":"library(dplyr) library(forcats) filepath <- \"https://osf.io/download/var2h/?view_only=2a06746ba360446e9857df73307d21be\" E2_data_raw <- readr::read_delim(filepath)  E2_data <- E2_data_raw %>%   filter(away != 1) %>%   mutate(Target = as.integer(bin_2P == 1)) %>%   mutate(Condition = fct_recode(     factor(Condition, levels = c(\"intrans\", \"pros\", \"trans\")),     \"Intransitive\" = \"intrans\",     \"RightDislocated\" = \"pros\",     \"Transitive\" = \"trans\"   )) %>%   select(Subject, Trial, Condition, Time, Target) E2_data_agg <- E2_data %>%   group_by(Subject, Condition, Time) %>%   summarize(Prop = mean(Target), .groups = \"drop\")  # Unaggregated trial-level data of 1s and 0s E2_data #> # A tibble: 69,123 × 5 #>    Subject Trial Condition        Time Target #>    <chr>   <dbl> <fct>           <dbl>  <int> #>  1 200.asc     0 RightDislocated     0      0 #>  2 200.asc     0 RightDislocated   400      0 #>  3 200.asc     0 RightDislocated   450      1 #>  4 200.asc     0 RightDislocated   500      1 #>  5 200.asc     0 RightDislocated   550      1 #>  6 200.asc     0 RightDislocated   600      1 #>  7 200.asc     0 RightDislocated   650      1 #>  8 200.asc     0 RightDislocated   700      1 #>  9 200.asc     0 RightDislocated   750      1 #> 10 200.asc     0 RightDislocated   800      1 #> # ℹ 69,113 more rows  # Aggregated subject-mean proportions data used in original study E2_data_agg #> # A tibble: 11,540 × 4 #>    Subject Condition        Time  Prop #>    <chr>   <fct>           <dbl> <dbl> #>  1 200.asc RightDislocated     0 0.5   #>  2 200.asc RightDislocated    50 1     #>  3 200.asc RightDislocated   100 1     #>  4 200.asc RightDislocated   150 1     #>  5 200.asc RightDislocated   200 1     #>  6 200.asc RightDislocated   250 0.5   #>  7 200.asc RightDislocated   300 0.5   #>  8 200.asc RightDislocated   350 0.4   #>  9 200.asc RightDislocated   400 0.286 #> 10 200.asc RightDislocated   450 0.429 #> # ℹ 11,530 more rows library(ggplot2) make_fig7_plot <- function(conditions) {   E2_data %>%     group_by(Condition, Time) %>%     summarize(       Prop = mean(Target), se = sqrt(var(Target) / n()),       lower = Prop - se, upper = Prop + se,       .groups = \"drop\"     ) %>%     filter(Condition %in% conditions) %>%     ggplot(aes(Time, Prop, color = Condition, fill = Condition)) +     geom_hline(aes(yintercept = .5)) +     geom_ribbon(       aes(ymin = lower, ymax = upper),       alpha = .2,       show.legend = FALSE,     ) +     geom_line(linewidth = 1) +     scale_color_manual(       aesthetics = c(\"color\", \"fill\"),       values = setNames(scales::hue_pal()(3)[c(2, 1, 3)], levels(E2_data$Condition))     ) +     scale_y_continuous(limits = c(.2, .8), oob = scales::oob_keep) +     labs(y = NULL, x = NULL) +     theme_minimal() +     theme(axis.title.y = element_text(angle = 0, vjust = .5, hjust = 0)) } fig7_comparisons <- list(   \"A\" = c(\"Intransitive\", \"RightDislocated\", \"Transitive\"),   \"B\" = c(\"RightDislocated\", \"Transitive\"),   \"C\" = c(\"Intransitive\", \"RightDislocated\"),   \"D\" = c(\"Intransitive\", \"Transitive\") ) fig7 <- lapply(fig7_comparisons, make_fig7_plot)  # Figure 7 combined plots library(patchwork) p_top <- fig7$A +   scale_x_continuous(breaks = scales::breaks_width(1000)) +   theme(legend.position = \"bottom\") p_bot <- (fig7$B + fig7$C + fig7$D) & guides(color = guide_none()) fig7_combined <- p_top / guide_area() / p_bot +   plot_layout(guides = \"collect\", heights = c(3, .1, 1)) +   plot_annotation(tag_levels = \"A\") E2_data_agg #> # A tibble: 11,540 × 4 #>    Subject Condition        Time  Prop #>    <chr>   <fct>           <dbl> <dbl> #>  1 200.asc RightDislocated     0 0.5   #>  2 200.asc RightDislocated    50 1     #>  3 200.asc RightDislocated   100 1     #>  4 200.asc RightDislocated   150 1     #>  5 200.asc RightDislocated   200 1     #>  6 200.asc RightDislocated   250 0.5   #>  7 200.asc RightDislocated   300 0.5   #>  8 200.asc RightDislocated   350 0.4   #>  9 200.asc RightDislocated   400 0.286 #> 10 200.asc RightDislocated   450 0.429 #> # ℹ 11,530 more rows fig7_combined library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"replicating-the-pairwise-cpas","dir":"Articles","previous_headings":"","what":"Replicating the pairwise CPAs","title":"de Carvalho et al. 2021","text":"three conditions experiment levels Condition factor variable: begin replication \"Transitive\" vs. \"RightDislocated\" comparison shown Figure 7B apply logic two pairwise comparisons 7C 7D.","code":"levels(E2_data_agg$Condition) #> [1] \"Intransitive\"    \"RightDislocated\" \"Transitive\""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"transitive-vs--rightdislocated","dir":"Articles","previous_headings":"Replicating the pairwise CPAs","what":"Transitive vs. RightDislocated","title":"de Carvalho et al. 2021","text":"reproduced Figure 7B compares \"Transitive\" \"RightDislocated\" conditions.  paper (caption Figure 7; emphasis mine) reports: transitive right-dislocated conditions differed second repetition novel verbs (~6400 ms onset test trials end trials). now replicate analysis. First, prepare specification object. Two things note : express original t-test regression model Condition predictor drop third, unused condition data factor representation Next, fit global model sanity check structure model output. get one estimate ConditionTransitive positive coefficient, ’d expect: Finally, call clusterpermute() threshold = 1.5 (original study) simulate 100 permutations: detect largest empirical cluster spanning 6400-8000ms reported original paper. converges around p=0.02 separate 10,000-simulation run (shown ).","code":"fig7$B spec_7B <- make_jlmer_spec(   formula = Prop ~ Condition,   data = E2_data_agg %>%     filter(Condition %in% c(\"Transitive\", \"RightDislocated\")) %>%     mutate(Condition = droplevels(Condition)), # or forcats::fct_drop()   subject = \"Subject\", time = \"Time\" ) spec_7B #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: Prop ~ 1 + ConditionTransitive #> Predictors: #>   Condition: ConditionTransitive #> Groupings: #>   Subject: Subject #>   Trial: #>   Time: Time #> Data: #> # A tibble: 7,688 × 4 #>    Prop ConditionTransitive Subject  Time #>   <dbl>               <dbl> <chr>   <dbl> #> 1   0.5                   0 200.asc     0 #> 2   1                     0 200.asc    50 #> 3   1                     0 200.asc   100 #> # ℹ 7,685 more rows #> ──────────────────────────────────────────────────────────────────────────────── jlmer(spec_7B) #> <Julia object of type StatsModels.TableRegressionModel> #> ────────────────────────────────────────────────────────────────────────────────── #>                          Coef.  Std. Error       z  Pr(>|z|)  Lower 95%  Upper 95% #> ────────────────────────────────────────────────────────────────────────────────── #> (Intercept)          0.592999   0.00364424  162.72    <1e-99   0.585856  0.600141 #> ConditionTransitive  0.0569017  0.0051591    11.03    <1e-27   0.04679   0.0670133 #> ────────────────────────────────────────────────────────────────────────────────── clusterpermute(spec_7B, threshold = 1.5, nsim = 100L, progress = FALSE) #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1.5) ──────────── <null_cluster_dists> ── #> ConditionTransitive (n = 100) #>   Mean (SD): -2.674 (27.83) #>   Coverage intervals: 95% [-61.740, 46.439] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1.5) ──────────────────────── <empirical_clusters> ── #> ConditionTransitive #>   [300, 1150]: 40.643 (p=0.1287) #>   [2700, 2750]: -4.068 (p=0.9010) #>   [6150, 6200]: 3.249 (p=0.9505) #>   [6400, 8000]: 88.742 (p=0.0198) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"rightdislocated-vs--intransitive","dir":"Articles","previous_headings":"Replicating the pairwise CPAs","what":"RightDislocated vs. Intransitive","title":"de Carvalho et al. 2021","text":"reproduced Figure 7C compares \"RightDislocated\" \"Intransitive\" conditions.  paper reports: intransitive right-dislocated conditions differed first repetition novel verbs (2100 ms 3500 ms beginning test trials). repeat CPA procedure pairwise comparison: largest empirical cluster detect spans 2150-3650ms, slightly different cluster reported original paper (2100-3500ms). relatively less “extreme” cluster converges around p=0.05 10,000-simulation run.","code":"fig7$C spec_7C <- make_jlmer_spec(   formula = Prop ~ Condition,   data = E2_data_agg %>%     filter(Condition %in% c(\"RightDislocated\", \"Intransitive\")) %>%     mutate(Condition = droplevels(Condition)),   subject = \"Subject\", time = \"Time\" )  clusterpermute(spec_7C, threshold = 1.5, nsim = 100L, progress = FALSE) #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1.5) ──────────── <null_cluster_dists> ── #> ConditionRightDislocated (n = 100) #>   Mean (SD): 0.619 (33.26) #>   Coverage intervals: 95% [-63.734, 67.326] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1.5) ──────────────────────── <empirical_clusters> ── #> ConditionRightDislocated #>   [150, 200]: 3.562 (p=0.9010) #>   [2150, 3650]: 72.842 (p=0.0297) #>   [4550, 5050]: 23.836 (p=0.3960) #>   [5150, 5350]: 8.589 (p=0.7723) #>   [5700, 5750]: 3.332 (p=0.9208) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"transitive-vs--intransitive","dir":"Articles","previous_headings":"Replicating the pairwise CPAs","what":"Transitive vs. Intransitive","title":"de Carvalho et al. 2021","text":"reproduced Figure 7D compares \"Transitive\" \"Intransitive\" conditions.  paper reports: transitive intransitive conditions differed slightly offset first sentence test trials (4500 ms beginning test trials end trials). repeat CPA procedure pairwise comparison: largest empirical cluster detect spans 4600-8000ms, slightly different cluster reported original paper (4500-8000ms). converges around p=0.001 separate 10,000-simulation run.","code":"fig7$D spec_7D <- make_jlmer_spec(   formula = Prop ~ Condition,   data = E2_data_agg %>%     filter(Condition %in% c(\"Transitive\", \"Intransitive\")) %>%     mutate(Condition = droplevels(Condition)),   subject = \"Subject\", time = \"Time\" )  clusterpermute(spec_7D, threshold = 1.5, nsim = 100L, progress = FALSE) #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1.5) ──────────── <null_cluster_dists> ── #> ConditionTransitive (n = 100) #>   Mean (SD): 2.024 (41.73) #>   Coverage intervals: 95% [-70.510, 88.501] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1.5) ──────────────────────── <empirical_clusters> ── #> ConditionTransitive #>   [300, 1000]: 25.928 (p=0.4356) #>   [1300, 1400]: 4.808 (p=0.8713) #>   [2800, 2900]: 5.356 (p=0.8218) #>   [3000, 4300]: 52.638 (p=0.2079) #>   [4600, 8000]: 172.635 (p=0.0198) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"expressed-as-regression-contrasts","dir":"Articles","previous_headings":"","what":"Expressed as regression contrasts","title":"de Carvalho et al. 2021","text":"now consider parsimonious analysis translates research hypothesis contrast coding avoid multiple testing. Specifically, exploit fact original paper specifies hypothesis Intransitive < RightDislocated < Transitive.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"helmert-deviation-coding","dir":"Articles","previous_headings":"Expressed as regression contrasts","what":"Helmert (deviation) coding","title":"de Carvalho et al. 2021","text":"Testing ordinal relationship levels category require possible pairwise comparisons; instead, can approximated via Helmert coding (.k.. deviance coding) K levels expressed K-1 contrasts, contrast successively comparing level vs. average previous (typically lower) levels. Critically, Helmert contrasts orthogonal, can test simultaneously single model. data, test ordinal relationship Intransitive < RightDislocated < Transitive via two contrasts: \"RightDislocated\" vs. \"Intransitive\" \"Transitive\" vs. average \"RightDislocated\" \"Intransitive\" corresponding numerical coding following: practice, Helmert contrasts often standardized deviations expressed unit 1. also comparison \"RightDislocated\" vs. \"Intransitive\" expressed 1 unit RvsI comparison \"Transitive\" vs. average \"RightDislocated\" \"Intransitive\" expressed 1 unit TvsRI: contrast matrix, make new column original data called ConditionHelm copying Condition column, apply contrasts new column: Lastly, build new specification making use full data. , predict Prop ConditionHelm estimate effect contrasts single model. sanity check, fit global model - expect estimate contrast indeed find RvsI TvsRI output positive coefficients. suggests ordinal relationship three conditions hold least globally. Note coefficient RvsI contrast exactly pairwise model using spec_7C earlier, also compared \"RightDislocated\" \"Intransitive\": spec_helm two conditions RvsI coded -0.5 0.5, spec_7C’s treatment coding coded 0 1. Since express difference \"Intransitive\" \"RightDislocated\" unit +1, two coefficients equal magnitude sign.","code":"condition_helm <- contr.helmert(3) colnames(condition_helm) <- c(\"RvsI\", \"TvsRI\") condition_helm #>   RvsI TvsRI #> 1   -1    -1 #> 2    1    -1 #> 3    0     2 condition_helm[, 1] <- condition_helm[, 1] / 2 condition_helm[, 2] <- condition_helm[, 2] / 3 condition_helm #>   RvsI      TvsRI #> 1 -0.5 -0.3333333 #> 2  0.5 -0.3333333 #> 3  0.0  0.6666667 E2_data_agg$ConditionHelm <- E2_data_agg$Condition contrasts(E2_data_agg$ConditionHelm) <- condition_helm  # For pretty-printing as fractions MASS::fractions(contrasts(E2_data_agg$ConditionHelm)) #>                 RvsI TvsRI #> Intransitive    -1/2 -1/3  #> RightDislocated  1/2 -1/3  #> Transitive         0  2/3 spec_helm <- make_jlmer_spec(   formula = Prop ~ ConditionHelm,   data = E2_data_agg,   subject = \"Subject\", time = \"Time\" ) spec_helm #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: Prop ~ 1 + ConditionHelmRvsI + ConditionHelmTvsRI #> Predictors: #>   ConditionHelm: ConditionHelmRvsI, ConditionHelmTvsRI #> Groupings: #>   Subject: Subject #>   Trial: #>   Time: Time #> Data: #> # A tibble: 11,540 × 5 #>    Prop ConditionHelmRvsI ConditionHelmTvsRI Subject  Time #>   <dbl>             <dbl>              <dbl> <chr>   <dbl> #> 1   0.5               0.5             -0.333 200.asc     0 #> 2   1                 0.5             -0.333 200.asc    50 #> 3   1                 0.5             -0.333 200.asc   100 #> # ℹ 11,537 more rows #> ──────────────────────────────────────────────────────────────────────────────── jlmer(spec_helm) %>%   tidy(effects = \"fixed\") #> # A tibble: 3 × 5 #>   term               estimate std.error statistic  p.value #>   <chr>                 <dbl>     <dbl>     <dbl>    <dbl> #> 1 (Intercept)          0.591    0.00215     275.  0        #> 2 ConditionHelmRvsI    0.0640   0.00526      12.2 5.67e-34 #> 3 ConditionHelmTvsRI   0.0889   0.00457      19.5 2.02e-84 jlmer(spec_7C) %>%   tidy(effects = \"fixed\") #> # A tibble: 2 × 5 #>   term                     estimate std.error statistic  p.value #>   <chr>                       <dbl>     <dbl>     <dbl>    <dbl> #> 1 (Intercept)                0.529    0.00379     140.  0        #> 2 ConditionRightDislocated   0.0640   0.00536      11.9 8.30e-33"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"interpreting-cpa-results","dir":"Articles","previous_headings":"Expressed as regression contrasts","what":"Interpreting CPA results","title":"de Carvalho et al. 2021","text":"proceed normal clusterpermute() using new spec_helm: ’s summary find CPA_helm: RvsI clusters similar detected previous “partial” CPA spec_7C. slight differences cluster-mass due part due Helmert-coded model simultaneously estimating TvsRI contrast. importantly, CPA_helm detects largest cluster \"RightDislocated\" \"Intransitive\" (2150-3650ms). converges around p=0.05 10,000-simulation run. TvsRI clusters new, largest cluster predictor spans 5800ms-8000ms. converges around p=0.01 separate 10,000-simulation run. cluster effectively captures region relationship Transitive > (RightDislocated & Intransitive) emerges robust. Essentially, TvsRI comparison line \"Transitive\" invisible line runs lines \"Intransitive\" \"RightDislocated\". conclude visualizing clusters two Helmert-coded terms, annotated empirical data. “invisible” line RI Helmert coding drawn dashed line.","code":"reset_rng_state() CPA_helm <- clusterpermute(spec_helm, threshold = 1.5, nsim = 100L, progress = FALSE) CPA_helm #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1.5) ──────────── <null_cluster_dists> ── #> ConditionHelmRvsI (n = 100) #>   Mean (SD): 4.019 (28.80) #>   Coverage intervals: 95% [-45.805, 67.827] #> ConditionHelmTvsRI (n = 100) #>   Mean (SD): 1.213 (24.59) #>   Coverage intervals: 95% [-50.218, 47.462] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1.5) ──────────────────────── <empirical_clusters> ── #> ConditionHelmRvsI #>   [150, 200]: 3.654 (p=0.8911) #>   [2150, 3650]: 71.141 (p=0.0297) #>   [4550, 5350]: 36.043 (p=0.2079) #>   [5700, 5750]: 3.328 (p=0.9010) #> ConditionHelmTvsRI #>   [300, 1050]: 35.863 (p=0.1881) #>   [3500, 3700]: 8.348 (p=0.6436) #>   [3850, 4250]: 15.985 (p=0.4455) #>   [4800, 5600]: 36.143 (p=0.1782) #>   [5800, 8000]: 123.631 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── fig7A_v2 <- fig7$A +   geom_line(     aes(Time, Prop, linetype = \"RI\"),     inherit.aes = FALSE,     data = . %>%       filter(Condition %in% c(\"RightDislocated\", \"Intransitive\")) %>%       group_by(Time) %>%       summarize(Prop = mean(Prop)),   ) +   scale_linetype_manual(values = \"41\", guide = guide_legend(\"ConditionHelm\")) +   guides(x = guide_none(\"\"))  clusters_annotation <- tidy(CPA_helm$empirical_clusters) %>%   mutate(contrast = gsub(\".*([TR])vs([RI]+)\", \"\\\\1 vs. \\\\2\", predictor)) %>%   ggplot(aes(y = fct_rev(contrast))) +   geom_segment(     aes(       x = start, xend = end, yend = contrast,       color = pvalue < 0.05     ),     linewidth = 8   ) +   scale_color_manual(values = c(\"grey70\", \"steelblue\")) +   scale_y_discrete() +   scale_x_continuous(n.breaks = 9, limits = range(E2_data_agg$Time)) +   theme_minimal() +   theme(     axis.title = element_blank(),     axis.text.y = element_text(face = \"bold\"),     panel.border = element_rect(fill = NA),     panel.grid.major.y = element_blank()   )  fig7A_v2 / clusters_annotation +   plot_layout(heights = c(4, 1))"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"model-complexity","dir":"Articles","previous_headings":"","what":"Model complexity","title":"de Carvalho et al. 2021","text":"wrap case study considering complex CPA uses logistic mixed effects models trial-level data fixations target (1s 0s).","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"logistic-mixed-model","dir":"Articles","previous_headings":"Model complexity","what":"Logistic mixed model","title":"de Carvalho et al. 2021","text":"un-aggregated trial-level data use section E2_data, comes initial data preparation code chunk: apply Helmert/deviation-coded contrast matrix: specification object E2_data, add trial = \"Trial\" predict Target instead Prop. also add -subject random intercept formula: , CPA family = \"binomial\": now visualize results CPA_helm_complex CPA_helm side side:  results largely , except largest, significant cluster identified TvsRI extends much complex CPA simple CPA. examine difference next.","code":"E2_data #> # A tibble: 69,123 × 5 #>    Subject Trial Condition        Time Target #>    <chr>   <dbl> <fct>           <dbl>  <int> #>  1 200.asc     0 RightDislocated     0      0 #>  2 200.asc     0 RightDislocated   400      0 #>  3 200.asc     0 RightDislocated   450      1 #>  4 200.asc     0 RightDislocated   500      1 #>  5 200.asc     0 RightDislocated   550      1 #>  6 200.asc     0 RightDislocated   600      1 #>  7 200.asc     0 RightDislocated   650      1 #>  8 200.asc     0 RightDislocated   700      1 #>  9 200.asc     0 RightDislocated   750      1 #> 10 200.asc     0 RightDislocated   800      1 #> # ℹ 69,113 more rows E2_data$ConditionHelm <- E2_data$Condition contrasts(E2_data$ConditionHelm) <- condition_helm MASS::fractions(contrasts(E2_data$ConditionHelm)) #>                 RvsI TvsRI #> Intransitive    -1/2 -1/3  #> RightDislocated  1/2 -1/3  #> Transitive         0  2/3 spec_helm_complex <- make_jlmer_spec(   formula = Target ~ ConditionHelm + (1 | Subject),   data = E2_data,   subject = \"Subject\", trial = \"Trial\", time = \"Time\" ) spec_helm_complex #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: Target ~ 1 + ConditionHelmRvsI + ConditionHelmTvsRI + (1 | Subject) #> Predictors: #>   ConditionHelm: ConditionHelmRvsI, ConditionHelmTvsRI #> Groupings: #>   Subject: Subject #>   Trial: Trial #>   Time: Time #> Data: #> # A tibble: 69,123 × 6 #>   Target ConditionHelmRvsI ConditionHelmTvsRI Subject Trial  Time #>    <int>             <dbl>              <dbl> <chr>   <dbl> <dbl> #> 1      0               0.5             -0.333 200.asc     0     0 #> 2      0               0.5             -0.333 200.asc     0   400 #> 3      1               0.5             -0.333 200.asc     0   450 #> # ℹ 69,120 more rows #> ──────────────────────────────────────────────────────────────────────────────── reset_rng_state() CPA_helm_complex <- clusterpermute(   spec_helm_complex,   family = \"binomial\",   threshold = 1.5, nsim = 100,   progress = FALSE ) CPA_helm_complex #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1.5) ──────────── <null_cluster_dists> ── #> ConditionHelmRvsI (n = 100) #>   Mean (SD): 1.791 (31.12) #>   Coverage intervals: 95% [-60.956, 80.602] #> ConditionHelmTvsRI (n = 100) #>   Mean (SD): 2.704 (28.90) #>   Coverage intervals: 95% [-54.636, 62.925] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1.5) ──────────────────────── <empirical_clusters> ── #> ConditionHelmRvsI #>   [2100, 3450]: 69.563 (p=0.0495) #>   [4200, 5050]: 40.460 (p=0.1584) #>   [5150, 5350]: 8.379 (p=0.7228) #> ConditionHelmTvsRI #>   [350, 800]: 19.358 (p=0.4158) #>   [900, 1050]: 7.519 (p=0.6535) #>   [1250, 1350]: 4.980 (p=0.7228) #>   [3100, 4450]: 56.770 (p=0.0792) #>   [4650, 8000]: 181.165 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── clusters_annotation2 <- tidy(CPA_helm_complex$empirical_clusters) %>%   mutate(contrast = gsub(\".*([TR])vs([RI]+)\", \"\\\\1 vs. \\\\2\", predictor)) %>%   ggplot(aes(y = fct_rev(contrast))) +   geom_segment(     aes(       x = start, xend = end, yend = contrast,       color = pvalue < 0.05     ),     linewidth = 8   ) +   scale_color_manual(values = c(\"grey70\", \"steelblue\")) +   scale_y_discrete() +   scale_x_continuous(n.breaks = 9, limits = range(E2_data$Time)) +   theme_minimal() +   theme(     legend.position = \"bottom\",     axis.title = element_blank(),     axis.text.y = element_text(face = \"bold\"),     panel.border = element_rect(fill = NA),     panel.grid.major.y = element_blank()   )  clusters_annotation / clusters_annotation2 &   guides(color = guide_none()) &   plot_annotation(tag_levels = list(c(\"Simple\", \"Complex\")))"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"comparison-of-cpas","dir":"Articles","previous_headings":"Model complexity","what":"Comparison of CPAs","title":"de Carvalho et al. 2021","text":"Looking timewise statistics computed simple vs. complex CPA tells us . figure plots information calls compute_timewise_statistics():  Whereas largest cluster starts emerge 5800ms CPA_helm, emerges much earlier 4650ms CPA_helm_complex. zoom region around 5800ms, see timewise statistics T vs. RI CPA_helm suddenly dip 1.5 threshold 5750ms:  CPA better? existence dips spikes indicate problem, ’s consistent expectation simple CPA less robust variance. can inspect time-point model 5750ms two CPAs fitting : ’s standard way comparing goodness fit linear fixed-effects model logistic mixed-effects model fitted different data. complex model outperforms simple model classic metrics inspect glance(). doesn’t come surprise, differences largely driven number observations (nobs). Determining appropriate degree model complexity CPA beyond scope vignette, pursue discussion . Instead, conclude old wisdom: chain strong weakest link.","code":"# Compute the timewise statistics from the CPA specifications empirical_statistics <- bind_rows(   tidy(compute_timewise_statistics(spec_helm)),   tidy(compute_timewise_statistics(spec_helm_complex, family = \"binomial\")),   .id = \"spec\" ) %>%   mutate(     CPA = c(\"Simple\", \"Complex\")[as.integer(spec)],     Contrast = gsub(\".*([TR])vs([RI]+)\", \"\\\\1 vs. \\\\2\", predictor)   )  # Time series plot of the statistics, with a line for each Helmert contrasts empirical_statistics_fig <- ggplot(empirical_statistics, aes(time, statistic)) +   geom_line(aes(color = Contrast, linetype = CPA), linewidth = 1, alpha = .7) +   geom_hline(yintercept = c(-1.5, 1.5), linetype = 2) +   theme_classic()  empirical_statistics_fig empirical_statistics_fig +   geom_vline(xintercept = 5750) +   coord_cartesian(xlim = 5750 + c(-500, 500), ylim = c(1, 2.5)) jlmer_simple_5750 <- to_jlmer(   formula = Prop ~ ConditionHelm,   data = E2_data_agg %>% filter(Time == 5750) ) jlmer_complex_5750 <- to_jlmer(   formula = Target ~ ConditionHelm + (1 | Subject),   family = \"binomial\",   data = E2_data %>% filter(Time == 5750) ) glance(jlmer_simple_5750) #> # A tibble: 1 × 8 #>    nobs    df sigma logLik   AIC   BIC deviance df.residual #>   <dbl> <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <dbl> #> 1    72     4 0.223   7.51 -7.02  2.08     3.42          68 glance(jlmer_complex_5750) #> # A tibble: 1 × 8 #>    nobs    df sigma logLik   AIC   BIC deviance df.residual #>   <int> <int> <lgl>  <dbl> <dbl> <dbl>    <dbl>       <int> #> 1   448     4 NA     -301.  609.  626.     601.         444"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/eyetrackingR-comparison.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Comparison to eyetrackingR","text":"following collapsed code chunk runs code “Prerequisites” section original eyetrackingR vignette: pick response_time dataframe created. stage, data filtered track loss aggregated subject time bins, among things. fuller description experiment research hypothesis can found original eyetrackingR vignette. purposes exercise, note following minimal facts analysis: hypothesis Prop ~ Target. t-test used estimate effect Target Prop Target experiment condition. two-level factor (\"Animate\" vs. \"Inanimate\") Prop response variable. continuous measure proportion (0-1) looks area interest ParticipantName column identifies participant TimeBin represents time, binned 100ms windows Additionally, keep following columns necessary metadata eyetrackingR approach, though can dropped jlmerclusterperm approach. Time continuous, raw (.e., un-binned) measure time AOI identity area interest simplicity, subset response_time include columns interest cluster-based permutation analysis. Lastly, replicate plot original vignette:","code":"set.seed(42)  library(\"Matrix\") library(\"lme4\") library(\"ggplot2\") library(\"eyetrackingR\")  data(\"word_recognition\") data <- make_eyetrackingr_data(word_recognition,                         participant_column = \"ParticipantName\",                        trial_column = \"Trial\",                        time_column = \"TimeFromTrialOnset\",                        trackloss_column = \"TrackLoss\",                        aoi_columns = c('Animate','Inanimate'),                        treat_non_aoi_looks_as_missing = TRUE                 )  # subset to response window post word-onset response_window <- subset_by_window(data,                                      window_start_time = 15500,                                      window_end_time = 21000,                                      rezero = FALSE)  # analyze amount of trackloss by subjects and trials (trackloss <- trackloss_analysis(data = response_window)) #> # A tibble: 155 × 6 #>    ParticipantName Trial          Samples TracklossSamples TracklossForTrial #>    <fct>           <fct>            <dbl>            <dbl>             <dbl> #>  1 ANCAT139        FamiliarBottle     330              161            0.488  #>  2 ANCAT18         FamiliarBird       330               74            0.224  #>  3 ANCAT18         FamiliarBottle     330               43            0.130  #>  4 ANCAT18         FamiliarCow        330              159            0.482  #>  5 ANCAT18         FamiliarDog        330               95            0.288  #>  6 ANCAT18         FamiliarHorse      330              165            0.5    #>  7 ANCAT18         FamiliarSpoon      330               95            0.288  #>  8 ANCAT22         FamiliarBird       330               14            0.0424 #>  9 ANCAT22         FamiliarBottle     330                8            0.0242 #> 10 ANCAT22         FamiliarDog        330               55            0.167  #> # ℹ 145 more rows #> # ℹ 1 more variable: TracklossForParticipant <dbl>  # remove trials with > 25% of trackloss response_window_clean <- clean_by_trackloss(data = response_window,                                             trial_prop_thresh = .25)  # create Target condition column response_window_clean$Target <- as.factor( ifelse(test = grepl('(Spoon|Bottle)', response_window_clean$Trial),                                         yes = 'Inanimate',                                         no  = 'Animate') )  response_time <- make_time_sequence_data(response_window_clean,                                   time_bin_size = 100,                                    predictor_columns = c(\"Target\"),                                   aois = \"Animate\",                                   summarize_by = \"ParticipantName\" ) response_time <- response_time %>%    select(Prop, Target, TimeBin, ParticipantName,          Time, AOI) dplyr::glimpse(response_time) #> Rows: 2,970 #> Columns: 6 #> $ Prop            <dbl> NaN, NaN, NaN, NaN, NaN, NaN, 0, 0, 1, 1, 1, 1, 1, 1, … #> $ Target          <fct> Animate, Animate, Animate, Animate, Animate, Animate, … #> $ TimeBin         <dbl> 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165,… #> $ ParticipantName <fct> ANCAT18, ANCAT18, ANCAT18, ANCAT18, ANCAT18, ANCAT18, … #> $ Time            <dbl> 15500, 15600, 15700, 15800, 15900, 16000, 16100, 16200… #> $ AOI             <chr> \"Animate\", \"Animate\", \"Animate\", \"Animate\", \"Animate\",… # visualize timecourse plot(response_time, predictor_column = \"Target\") +    theme_light() +   coord_cartesian(ylim = c(0,1)) #> Warning: Removed 37 rows containing non-finite values (`stat_summary()`). #> Removed 37 rows containing non-finite values (`stat_summary()`)."},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/eyetrackingR-comparison.html","id":"cpa-in-eyetrackingr","dir":"Articles","previous_headings":"","what":"CPA in {eyetrackingR}","title":"Comparison to eyetrackingR","text":"Continuing response_time dataframe, jump “Bootstrapped cluster-based permutation analysis” original vignette. original vignette uses following heuristic choosing threshold: choose threshold outside scope article - simply use value approaches. eyetrackingR, CPA conducted two steps: Prepare data CPA make_time_cluster_data(): step computes timewise statistics data identifies empirical clusters, can inspected plot() summary() method:  Run permutation test cluster data analyze_time_clusters(): simulates null distribution cluster-mass statistics. output, printed, essentially output summary(df_timeclust) p-values (Probability column) null distribution (extremety empirical clusters context) can visualized plot() method:","code":"num_sub = length(unique((response_window_clean$ParticipantName))) threshold_t = qt(p = 1 - .05/2, df = num_sub-1) threshold_t #> [1] 2.055529 df_timeclust <- make_time_cluster_data(response_time,                                        test= \"t.test\", paired=TRUE,                                       predictor_column = \"Target\",                                        threshold = threshold_t) plot(df_timeclust) summary(df_timeclust) #> Test Type:    t.test  #> Predictor:    Target  #> Formula:  Pair(Prop[Target == \"Animate\"], Prop[Target == \"Inanimate\"]) ~      1  #> Summary of Clusters ====== #>   Cluster Direction SumStatistic StartTime EndTime #> 1       1  Positive    132.29900     16100   19300 #> 2       2  Positive     42.31067     19400   20800 system.time({    clust_analysis <- analyze_time_clusters(     df_timeclust,     within_subj = TRUE,     paired = TRUE,     samples = 150,     quiet = TRUE   )  }) #>    user  system elapsed  #>   13.72    0.10   34.46 clust_analysis #> Test Type:    t.test  #> Predictor:    Target  #> Formula:  Pair(Prop[Target == \"Animate\"], Prop[Target == \"Inanimate\"]) ~      1  #> Null Distribution   ======  #>  Mean:        -1.4815  #>  2.5%:        -31.1268  #> 97.5%:        17.9841  #> Summary of Clusters ====== #>   Cluster Direction SumStatistic StartTime EndTime Probability #> 1       1  Positive    132.29900     16100   19300 0.000000000 #> 2       2  Positive     42.31067     19400   20800 0.006666667 plot(clust_analysis)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/eyetrackingR-comparison.html","id":"cpa-in-jlmerclusterperm","dir":"Articles","previous_headings":"","what":"CPA in {jlmerclusterperm}","title":"Comparison to eyetrackingR","text":"First, set jlmerclusterperm. comparability, use single thread: {jlmerclusterpm}, CPA can also conducted two steps: Make specification object make_jlmer_spec() first specify formula, data, grouping columns data. Instead paired t-test, specify linear model formula Prop ~ Target. output prepares data CPA subsetting applying contrast coding schemes, among things: Run permutation test specification using clusterpermute() kinds information returned: different pieces information available inspection using tidy(), returns dataframe underlying summary: contrast eyetrackingR, jlmerclusterperm provide custom plot() method. However, kinds plots can replicated lines ggplot code:  Note sign clusters (originally) flipped two approaches, trivial difference simply reflects default choice baseline t.text() vs. regression contrasts. can addressed setting contrasts prior running CPA (see related issue another article). Lastly, side--side comparison results:","code":"library(jlmerclusterperm) options(\"jlmerclusterperm.nthreads\" = 1) system.time({      # Note the overhead for initializing the Julia session   jlmerclusterperm_setup()    }) #>    user  system elapsed  #>    0.00    0.01   23.02 spec <- make_jlmer_spec(   formula = Prop ~ Target,   data = response_time,   subject = \"ParticipantName\",   trial = \"Target\",   time = \"TimeBin\" ) #> ! Dropping 37 rows with missing values. spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: Prop ~ 1 + TargetInanimate #> Predictors: #>   Target: TargetInanimate #> Groupings: #>   Subject: ParticipantName #>   Trial: Target #>   Time: TimeBin #> Data: #> # A tibble: 2,933 × 5 #>    Prop TargetInanimate ParticipantName Target  TimeBin #>   <dbl>           <dbl> <fct>           <fct>     <dbl> #> 1     0               0 ANCAT18         Animate     161 #> 2     0               0 ANCAT18         Animate     162 #> 3     1               0 ANCAT18         Animate     163 #> # ℹ 2,930 more rows #> ──────────────────────────────────────────────────────────────────────────────── system.time({   cpa <- clusterpermute(spec, threshold = threshold_t, nsim = 150) }) #>    user  system elapsed  #>    0.02    0.00   15.63 cpa #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2.05552943864287) ─────────────────────── #> TargetInanimate (n = 150) #>   Mean (SD): -1.009 (11.03) #>   Coverage intervals: 95% [-25.856, 27.183] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2.05552943864287) ─────────── <empirical_clusters> ── #> TargetInanimate #>   [161, 191]: -130.192 (p=0.0066) #>   [194, 209]: -45.791 (p=0.0066) #> ──────────────────────────────────────────────────────────────────────────────── null_distribution <- tidy(cpa$null_cluster_dists) null_distribution #> # A tibble: 150 × 6 #>    predictor       start   end length sum_statistic sim   #>    <chr>           <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 TargetInanimate    NA    NA     NA          0    001   #>  2 TargetInanimate   208   209      2          5.09 002   #>  3 TargetInanimate   197   200      4         -8.94 003   #>  4 TargetInanimate   205   206      2         -4.32 004   #>  5 TargetInanimate    NA    NA     NA          0    005   #>  6 TargetInanimate   193   196      4        -10.8  006   #>  7 TargetInanimate   187   193      7         19.3  007   #>  8 TargetInanimate   177   180      4          9.34 008   #>  9 TargetInanimate   159   161      3         -7.57 009   #> 10 TargetInanimate    NA    NA     NA          0    010   #> # ℹ 140 more rows empirical_clusters <- tidy(cpa$empirical_clusters) empirical_clusters #> # A tibble: 2 × 7 #>   predictor       id    start   end length sum_statistic  pvalue #>   <chr>           <fct> <dbl> <dbl>  <dbl>         <dbl>   <dbl> #> 1 TargetInanimate 1       161   191     31        -130.  0.00662 #> 2 TargetInanimate 2       194   209     16         -45.8 0.00662 # We flip the sign of the statistic here for visual equivalence cpa_plot <- null_distribution %>%    ggplot(aes(-sum_statistic)) +   geom_histogram(     aes(y = after_stat(density)),     binwidth = 3   ) +   geom_density() +   geom_vline(     aes(xintercept = -sum_statistic, color = id),     data = empirical_clusters,     linetype = 2,     linewidth = 1   ) cpa_plot library(patchwork) plot(clust_analysis) + cpa_plot"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/eyetrackingR-comparison.html","id":"performance","dir":"Articles","previous_headings":"","what":"Performance","title":"Comparison to eyetrackingR","text":"jlmerclusterperm scales easily higher number simulations, especially fixed-effects models: eyetrackingR option parallelization, limited support platform dependent. contrast, jlmerclusterperm leverages Julia’s built-multi-threading support performant consistent: 10,000 simulations 7 threads:","code":"system.time({   clusterpermute(spec, threshold = threshold_t, nsim = 1000) }) #>    user  system elapsed  #>    0.04    0.02   10.59 options(\"jlmerclusterperm.nthreads\" = 7) system.time({   jlmerclusterperm_setup(restart = TRUE, verbose = FALSE) }) #>    user  system elapsed  #>    0.09    0.03   20.78 system.time({   clusterpermute(spec, threshold = threshold_t, nsim = 10000) }) #>    user  system elapsed  #>    0.22    0.32   48.99"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"background","dir":"Articles","previous_headings":"","what":"Background","title":"Garrison et al. 2020","text":"data comes eye-tracking study Garrison, Baudet, Breitfield, Aberman, & Bergelson (2020) “Familiarity plays small role noun comprehension 12-18 months”, way peekbankr corpus package. original study used growth curve model analyze children’s looks object familiar (“”) vs. unfamiliar (“”) different ages (median-split “younger” “older” groups). replicating specific analysis . Instead, follow Katie Von Holzen’s tutorial used dataset conduct simpler, cluster-based permutation analysis (CPA) differences two age groups proportion looks target. tutorial used eyetrackingR package data preparation CPA, vignette use dplyr jlmerclusterperm tasks, respectively. following set-code reads data prepares analysis: data ts_data_agg ’ll use replicate tutorial just 4 columns relevant cluster-permutation analysis: subject_id: Unique identifier subjects age: -participant factor variable (\"Younger\", \"Older\") time: continuous measure time 0-3975ms 25ms intervals prop: -subject proportion looks target object following replicates figure tutorial introduces cluster-permutation analysis:","code":"library(dplyr) remotes::install_github(\"langcog/peekbankr\", quiet = TRUE) library(peekbankr)  # Load data aoi_timepoints <- get_aoi_timepoints(dataset_name = \"garrison_bergelson_2020\") administrations <- get_administrations(dataset_name = \"garrison_bergelson_2020\")  # Pre-processing ## Filter for age groups of interest ps_data <- aoi_timepoints %>%   left_join(administrations, by = \"administration_id\") %>%   filter(!between(age, 14, 16)) %>%   mutate(age_binned = ifelse(age < 14, \"Younger\", \"Older\")) ## Filter for time window of interest ts_window <- ps_data %>%   filter(t_norm >= 0, t_norm < 4000) ## Identify trials to exclude (trackloss in >25% of samples) to_exclude <- ts_window %>%   group_by(subject_id, trial_id) %>%   summarize(prop = mean(aoi == \"missing\"), .groups = \"drop\") %>%   filter(prop > 0.25) ## Exclude disqualifying trials and keep information relevant for CPA ts_data <- ts_window %>%   anti_join(to_exclude, by = c(\"subject_id\", \"trial_id\")) %>%   mutate(     target = aoi == \"target\",     missing = aoi == \"missing\",     age = factor(age_binned, c(\"Younger\", \"Older\"))   ) %>%   select(subject_id, age, trial_id, time = t_norm, target, missing)  # Data of subject mean proportions of fixations to target ts_data_agg <- ts_data %>%   group_by(subject_id, age, time) %>%   summarize(prop = sum(target) / sum(!missing), .groups = \"drop\")  # Data of trial-level fixations to target (0s and 1s) ts_data_trials <- ts_data %>%   filter(!missing) %>%   mutate(target = as.integer(target)) %>%   select(subject_id, age, trial_id, time, target) ts_data_agg #> # A tibble: 2,560 × 4 #>    subject_id age    time  prop #>         <int> <fct> <dbl> <dbl> #>  1       1413 Older     0 0.7   #>  2       1413 Older    25 0.733 #>  3       1413 Older    50 0.7   #>  4       1413 Older    75 0.7   #>  5       1413 Older   100 0.677 #>  6       1413 Older   125 0.677 #>  7       1413 Older   150 0.645 #>  8       1413 Older   175 0.645 #>  9       1413 Older   200 0.581 #> 10       1413 Older   225 0.548 #> # ℹ 2,550 more rows library(ggplot2) tutorial_fig <- ggplot(ts_data_agg, aes(x = time, y = prop, color = age)) +   stat_smooth(method = \"loess\", span = 0.1, se = TRUE, aes(fill = age), alpha = 0.3) +   theme_bw() +   labs(y = \"Prop. of Target Looks\") +   geom_hline(yintercept = .5, color = \"black\") +   geom_vline(xintercept = 0, color = \"black\") +   coord_cartesian(ylim = c(0, 1)) +   theme(     legend.position = \"bottom\",     legend.title = element_blank()   ) tutorial_fig"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"outline","dir":"Articles","previous_headings":"","what":"Outline","title":"Garrison et al. 2020","text":"case study vignette showcases four features CPA jlmerclusterperm: Prepping data CPA using make_jlmer_spec() CPA one go clusterpermute() modular, step--step approach CPA Using logistic regression trial-level data 1s 0s Load package start Julia instance jlmerclusterperm_setup() proceeding.","code":"library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"a-prepping-a-specification-object","dir":"Articles","previous_headings":"","what":"A) Prepping a specification object","title":"Garrison et al. 2020","text":"Conducting cluster-based permutation analysis jlmerclusterm starts creating special specification object, compiles information necessary CPA. function make_jlmer_spec() returns specification object class <jlmer_spec> minimally requires two arguments: formula: R regression model formula data: data frame example, model prop using age predictor ts_data_agg data: printed display includes four pieces information specification object: Formula: model formula Julia syntax. , looks similar formula provided, contrasts “spelled ” (age becomes ageOlder) Predictors: list predictors form ( original : expanded ). case, one predictor age expanded ageOlder ’s treatment coded “Younger” reference level. Groupings: specified call make_jlmer_spec(). empty simple_spec provided formula data. Data: transformed dataframe whose columns terms expanded formula. Note default dummy coding applied discrete variable age - now ageOlder 0 “Younger” 1 “Older”. start bare specification object , lacks parts CPA, can use jlmerclusterperm’s simple interface Julia regression models. function jlmer() takes specification object input returns Julia regression model. model represents “global” model fitted entire data, collapsing across time. equivalent lm() model specifications: raise tangent jlmer() recommend sanity-checking design output global model prior using model specifications CPA. regression models fit Julia “hood”, want make sure output comparable expect R. global model can also tell upper bound model complexity. example, global model singular fit, time-wise models fitted subsets data likely . Functions CPA also take <jlmer_spec> object require information time (calculate time-wise statistics) subject/trial (bootstrapped permutation). ts_data_agg data trial-level information summarized subject, just leave . trial argument can omitted <jlmer_spec> object time sample/bin uniquely identified subject. Otherwise, trial column data uniquely identifies time values within subject (example, column items counterbalanced designed participant sees every item one variants).","code":"simple_spec <- make_jlmer_spec(formula = prop ~ 1 + age, data = ts_data_agg) simple_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: prop ~ 1 + ageOlder #> Predictors: #>   age: ageOlder #> Data: #> # A tibble: 2,560 × 2 #>    prop ageOlder #>   <dbl>    <dbl> #> 1 0.7          1 #> 2 0.733        1 #> 3 0.7          1 #> # ℹ 2,557 more rows #> ──────────────────────────────────────────────────────────────────────────────── global_jmod <- jlmer(simple_spec) tidy(global_jmod) #> # A tibble: 2 × 5 #>   term        estimate std.error statistic  p.value #>   <chr>          <dbl>     <dbl>     <dbl>    <dbl> #> 1 (Intercept)   0.523    0.00387     135.  0        #> 2 ageOlder      0.0984   0.00548      18.0 3.45e-72 library(broom) # for the `tidy()` method for `lm()` global_rmod <- lm(formula = prop ~ 1 + age, data = ts_data_agg) tidy(global_rmod) #> # A tibble: 2 × 5 #>   term        estimate std.error statistic  p.value #>   <chr>          <dbl>     <dbl>     <dbl>    <dbl> #> 1 (Intercept)   0.523    0.00387     135.  0        #> 2 ageOlder      0.0984   0.00548      18.0 4.45e-68 ts_data_agg_spec <- make_jlmer_spec(   formula = prop ~ 1 + age, data = ts_data_agg,   subject = \"subject_id\", time = \"time\",   trial = NULL # This is the default value ) ts_data_agg_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: prop ~ 1 + ageOlder #> Predictors: #>   age: ageOlder #> Groupings: #>   Subject: subject_id #>   Trial: #>   Time: time #> Data: #> # A tibble: 2,560 × 4 #>    prop ageOlder subject_id  time #>   <dbl>    <dbl>      <int> <dbl> #> 1 0.7          1       1413     0 #> 2 0.733        1       1413    25 #> 3 0.7          1       1413    50 #> # ℹ 2,557 more rows #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"b-cpa-with-clusterpermute","dir":"Articles","previous_headings":"","what":"B) CPA with clusterpermute()","title":"Garrison et al. 2020","text":"clusterpermute() function runs CPA one go, using information <jlmer_spec> object. addition specification object, must also supply (t-value) threshold number simulations run: result list two elements: <null_cluster_dists> object holds information distribution bootstrapped cluster-mass statistics random permutations data: <empirical_clusters> object holds per-predictor information clusters detected, p-values significance testing null: output explained next section, presented purposes simply note results similar reported Katie Von Holzen’s tutorial, copied . minor numerical differences due stochastic nature permutation test ’re comparing t-tests R regression models Julia.","code":"CPA_agg <- clusterpermute(   ts_data_agg_spec,   threshold = 2,   nsim = 100,   progress = FALSE # Suppress printing of progress for vignette ) CPA_agg #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> ageOlder (n = 100) #>   Mean (SD): -0.844 (29.71) #>   Coverage intervals: 95% [-46.664, 71.698] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 74.724 (p=0.0396) #>   [2675, 2775]: 11.017 (p=0.4356) #>   [3500, 3900]: 46.295 (p=0.0990) #> ──────────────────────────────────────────────────────────────────────────────── #> ## Summary of Clusters ====== #> ##   Cluster Direction SumStatistic StartTime EndTime Probability #> ## 1       1  Positive    74.723687       850    1550       0.051 #> ## 2       2  Positive     6.905622      2725    2800       0.405 #> ## 3       3  Positive    46.294644      3500    3925       0.111"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"c-a-step-by-step-approach","dir":"Articles","previous_headings":"","what":"C) A step-by-step approach","title":"Garrison et al. 2020","text":"clusterpermute() function showcased consists five parts, also standalone functions package: compute_timewise_statistics() extract_empirical_clusters() permute_timewise_statistics() extract_null_cluster_dists() calculate_clusters_pvalues() functions correspond algorithmic steps statistical test. clusterpermute() obviates need think individual components, knowing gives greater flexibility procedure. section walks CPA step--step using functions.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"empirical-clusters","dir":"Articles","previous_headings":"C) A step-by-step approach","what":"1) Empirical clusters","title":"Garrison et al. 2020","text":"compute_timewise_statistics() function takes <jlmer_spec> object returns timewise statistics form matrix row predictor column time point: example see 25ms, t-value age -0.127. consistent figure tutorial replicated show two lines mostly overlapping. Just demonstration purposes, use to_jlmer() quickly fit model using formula just data 25ms. see t-value age model exactly saw : construct empirical clusters timewise statistics. Clusters defined continuous span statistics sign whose absolute value passes certain threshold. extract_empirical_clusters() function takes timewise statistics threshold just : t-value threshold 2, detect three clusters data main effect age (ageOlder). numbers brackets show span cluster number right show sum t-values cluster (.k.. cluster-mass statistic). can collect <empirical_clusters> object data frame tidy(): add figure :  point know whether size clusters observe (un)likely emerge chance. calculate probability, need simulate null distribution.","code":"empirical_statistics <- compute_timewise_statistics(ts_data_agg_spec) dim(empirical_statistics) #> [1]   1 160 empirical_statistics[1, 1:5, drop = FALSE] # First 5 time points #>           Time #> Predictor          0         25         50         75        100 #>   ageOlder 0.1166025 -0.1273179 -0.2387778 -0.4022095 -0.2804397 to_jlmer(prop ~ 1 + age, data = filter(ts_data_agg, time == 25)) %>%   tidy() %>%   filter(term == \"ageOlder\") %>%   pull(statistic) #> [1] -0.1273179 empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 2) empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 74.724 #>   [2675, 2775]: 11.017 #>   [3500, 3900]: 46.295 #> ──────────────────────────────────────────────────────────────────────────────── empirical_clusters_df <- tidy(empirical_clusters) empirical_clusters_df #> # A tibble: 3 × 6 #>   predictor id    start   end length sum_statistic #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl> #> 1 ageOlder  1       850  1525     28          74.7 #> 2 ageOlder  2      2675  2775      5          11.0 #> 3 ageOlder  3      3500  3900     17          46.3 tutorial_fig +   geom_segment(     aes(x = start, xend = end, y = -Inf, yend = -Inf),     linewidth = 10, color = \"steelblue\",     inherit.aes = FALSE,     data = empirical_clusters_df   ) +   geom_text(     aes(       y = -Inf, x = start + (end - start) / 2,       label = paste(\"Cluster\", id)     ),     vjust = -2,     inherit.aes = FALSE,     data = empirical_clusters_df   )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"null-distribution","dir":"Articles","previous_headings":"C) A step-by-step approach","what":"2) Null distribution","title":"Garrison et al. 2020","text":"permute_timewise_statistics() function takes specification object number bootstrapped permutation data simulate (via nsim argument). returned 3-dimensional array like output compute_timewise_statistics() except additional dimension representing simulation: permutation algorithm preserves grouping structures data specified specification object. example, testing -subject predictor like age, age groups participants random swapped preserving temporal structure data. null distribution collection largest cluster-mass statistic simulated data. clusters detected simulation, contributes cluster-mass zero null. use extract_null_cluster_dists() construct null distribution, using threshold value 2 allow comparison empirical clusters: printed, returned <null_cluster_dists> object shows summary statistics. can use tidy() collect samples null data frame: null distribution can plotted like :  Now left test probability observing cluster-mass statistics extreme detected clusters empirical_clusters, using null_cluster_dists reference.","code":"null_statistics <- permute_timewise_statistics(ts_data_agg_spec, nsim = 100) # simulation by time by predictor dim(null_statistics) #> [1] 100 160   1 # First 5 time points of the first three simulations null_statistics[1:3, 1:5, 1, drop = FALSE] #> , , Predictor = ageOlder #>  #>      Time #> Sim            0        25        50        75        100 #>   001 -0.1037696 0.1387838 0.4583545 0.4337538 0.69136613 #>   002  1.4358330 1.2745381 1.2985888 0.9111359 1.16401311 #>   003  0.2545470 0.3885999 0.3374939 0.3200497 0.05603617 null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 2) null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> ageOlder (n = 100) #>   Mean (SD): -6.150 (40.29) #>   Coverage intervals: 95% [-114.564, 74.425] #> ──────────────────────────────────────────────────────────────────────────────── null_cluster_dists_df <- tidy(null_cluster_dists) null_cluster_dists_df #> # A tibble: 100 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 ageOlder     NA    NA     NA          0    001   #>  2 ageOlder     NA    NA     NA          0    002   #>  3 ageOlder   2175  2225      3         -6.37 003   #>  4 ageOlder     NA    NA     NA          0    004   #>  5 ageOlder   3425  3475      3         -6.67 005   #>  6 ageOlder     NA    NA     NA          0    006   #>  7 ageOlder     NA    NA     NA          0    007   #>  8 ageOlder     NA    NA     NA          0    008   #>  9 ageOlder     NA    NA     NA          0    009   #> 10 ageOlder   1525  2475     39       -114.   010   #> # ℹ 90 more rows plot(   density(null_cluster_dists_df$sum_statistic),   main = \"Null distribution of cluster-mass statistics\" )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"significance-test","dir":"Articles","previous_headings":"C) A step-by-step approach","what":"3) Significance test","title":"Garrison et al. 2020","text":"calculate_clusters_pvalues() function computes p-value empirical cluster returns augmented <empirical_clusters> object: , p-values slightly different test stochastic. separate vignette covers issues replicability, seed, rng. collected tidy(), p-values added column: Note reporting pvalues, ’s customary add 1 numerator denominator. observed data, however unlikely, one particular arrangement (permutation) data. controlled via add1 argument - false default calculate_clusters_pvalues() true clusterpermute(). now come full circle: pvalue-augmented <empirical_clusters> object, along <null_cluster_dists> object, also clusterpermute() function returned .","code":"empirical_clusters_tested <- calculate_clusters_pvalues(empirical_clusters, null_cluster_dists) empirical_clusters_tested #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 74.724 (p=0.1100) #>   [2675, 2775]: 11.017 (p=0.4000) #>   [3500, 3900]: 46.295 (p=0.1700) #> ──────────────────────────────────────────────────────────────────────────────── tidy(empirical_clusters_tested) #> # A tibble: 3 × 7 #>   predictor id    start   end length sum_statistic pvalue #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>  <dbl> #> 1 ageOlder  1       850  1525     28          74.7   0.11 #> 2 ageOlder  2      2675  2775      5          11.0   0.4  #> 3 ageOlder  3      3500  3900     17          46.3   0.17 calculate_clusters_pvalues(empirical_clusters, null_cluster_dists, add1 = TRUE) #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 74.724 (p=0.1188) #>   [2675, 2775]: 11.017 (p=0.4059) #>   [3500, 3900]: 46.295 (p=0.1782) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"aggregated-vs--trial-level-data","dir":"Articles","previous_headings":"D) As logistic regression","what":"Aggregated vs. trial-level data","title":"Garrison et al. 2020","text":"data using far aggregated data collapsed across trials within subject. words, exactly one measure proportion looks target per time point within subject: possible also fit logistic regression model trial-level measure “hits” (1s 0s) target area interest. un-aggregated, trial-level data comes data preparation code top vignette. demonstrate relationship two data sets: specification object fitting logistic regression models trial-level data requires two changes original: formula must predict binary target variable, instead prop trial grouping must specified uniquely identify time point within subject following appropriate <jlmer_spec> object CPA using logistic regression: , sanity check global model using family = \"binomial\": estimates similar global_jmod previous section, log-odds.","code":"nrow(ts_data_agg) #> [1] 2560 ts_data_agg %>%   distinct(subject_id, time) %>%   nrow() #> [1] 2560 ts_data_trials #> # A tibble: 62,466 × 5 #>    subject_id age   trial_id  time target #>         <int> <fct>    <int> <dbl>  <int> #>  1       1413 Older     2659     0      1 #>  2       1413 Older     2659    25      1 #>  3       1413 Older     2659    50      1 #>  4       1413 Older     2659    75      1 #>  5       1413 Older     2659   100      1 #>  6       1413 Older     2659   125      1 #>  7       1413 Older     2659   150      1 #>  8       1413 Older     2659   175      1 #>  9       1413 Older     2659   200      1 #> 10       1413 Older     2659   225      1 #> # ℹ 62,456 more rows identical(   ts_data_agg,   ts_data_trials %>%     group_by(subject_id, age, time) %>%     summarize(prop = mean(target), .groups = \"drop\") ) #> [1] TRUE ts_data_trial_spec <- make_jlmer_spec(   formula = target ~ 1 + age, data = ts_data_trials,   subject = \"subject_id\", trial = \"trial_id\", time = \"time\" ) ts_data_trial_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: target ~ 1 + ageOlder #> Predictors: #>   age: ageOlder #> Groupings: #>   Subject: subject_id #>   Trial: trial_id #>   Time: time #> Data: #> # A tibble: 62,466 × 5 #>   target ageOlder subject_id trial_id  time #>    <int>    <dbl>      <int>    <int> <dbl> #> 1      1        1       1413     2659     0 #> 2      1        1       1413     2659    25 #> 3      1        1       1413     2659    50 #> # ℹ 62,463 more rows #> ──────────────────────────────────────────────────────────────────────────────── global_jmod_binom <- jlmer(ts_data_trial_spec, family = \"binomial\") tidy(global_jmod_binom) #> # A tibble: 2 × 5 #>   term        estimate std.error statistic   p.value #>   <chr>          <dbl>     <dbl>     <dbl>     <dbl> #> 1 (Intercept)    0.113    0.0121      9.31 1.24e- 20 #> 2 ageOlder       0.392    0.0164     23.9  1.08e-126"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"repeating-the-cpa-with-logistic-regression","dir":"Articles","previous_headings":"D) As logistic regression","what":"Repeating the CPA with logistic regression","title":"Garrison et al. 2020","text":"repeat procedure use new specification object set family = \"binomial\": Instead 3 clusters, now detect 5. can annotate detected clusters time series data :  differences see aggregated vs. trial-level data completely expected follow straightforwardly following two facts: cluster-mass statistic derived size t-values time point, effect size. virtue unaggregated ts_data_trials data number observations, t-values overall inflated difference means (also see output global_jmod vs. global_jmod_binom well). Whereas aggregated data reflects complete pooling trial-level variation, binomial model trial-level data performs pooling treats trial independent. Thus, certain item experiment show outlier effect one time point, example, strongly influence mass cluster spanning given time point. first issue (inflated statistics) trivial just takes matter adjusting threshold. fact, good “stretching ” t-statistics, get higher resolution shape timewise statistics. However. second issue (pooling) troubling can warp fundamental shape timewise statistics. diagnose factors visualizing timewise statistics. also plot timewise statistics aggregated data comparison:  see trial-level binomial model overall inflated t-values, well pronounced “peaks” (specifically maximum 1300ms). middle-ground, partial pooling approaching using mixed-effects models CPA showcased case study vignettes. present purposes, raise threshold value (2.8) gets us close detecting least largest cluster empirical_statistics, spanning 850ms-1525ms: run cluster permutation test threshold just largest cluster: gives us pvalue cluster spanning 850ms 1525ms similar one reported aggregated data: appear converge higher nsim data set. also time 1000-simulation runs showcase speed jlmerclusterperm. Note use set_rng_state(), makes two CPAs generate permutations observed data.","code":"empirical_statistics_binom <- compute_timewise_statistics(ts_data_trial_spec, family = \"binomial\") empirical_clusters_binom <- extract_empirical_clusters(empirical_statistics_binom, threshold = 2) empirical_clusters_binom #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [650, 725]: 8.727 #>   [800, 1575]: 112.199 #>   [1775, 1875]: 12.764 #>   [2600, 2825]: 23.733 #>   [3475, 3975]: 62.714 #> ──────────────────────────────────────────────────────────────────────────────── empirical_clusters_binom_df <- tidy(empirical_clusters_binom) tutorial_fig +   geom_segment(     aes(x = start, xend = end, y = -Inf, yend = -Inf),     linewidth = 10, color = \"steelblue\",     inherit.aes = FALSE,     data = empirical_clusters_binom_df   ) +   geom_text(     aes(       y = -Inf, x = start + (end - start) / 2,       label = paste(\"Cluster\", id)     ),     vjust = -2,     inherit.aes = FALSE,     data = empirical_clusters_binom_df   ) timewise_statistics_fig <- ggplot(mapping = aes(time, statistic)) +   geom_hline(aes(yintercept = 0)) +   geom_line(     aes(color = \"aggregated\"),     linewidth = 1.5,     data = tidy(empirical_statistics)   ) +   geom_line(     aes(color = \"trial-level\"),     linewidth = 1.5,     data = tidy(empirical_statistics_binom)   ) timewise_statistics_fig +   geom_vline(     aes(xintercept = time),     data = tidy(empirical_statistics_binom) %>%       slice(which.max(abs(statistic)))   ) empirical_clusters_binom2 <- extract_empirical_clusters(empirical_statistics_binom, threshold = 2.8) empirical_clusters_binom2 #> ── Empirical clusters (t > 2.8) ──────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 102.450 #>   [3500, 3875]: 51.365 #> ──────────────────────────────────────────────────────────────────────────────── CPA_trial <- clusterpermute(   ts_data_trial_spec,   family = \"binomial\",   threshold = 2.8,   nsim = 100,   top_n = 1, # Just test the largest cluster   progress = FALSE # Suppress printing of progress for vignette ) CPA_trial$empirical_clusters #> ── Empirical clusters (t > 2.8) ──────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 102.450 (p=0.0297) #> ────────────────────────────────────────────────────────────────────────────────  CPA_agg$empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 74.724 (p=0.0396) #>   [2675, 2775]: 11.017 (p=0.4356) #>   [3500, 3900]: 46.295 (p=0.0990) #> ──────────────────────────────────────────────────────────────────────────────── system.time({   set_rng_state(123L)   CPA_agg_1000 <- clusterpermute(     ts_data_agg_spec,     threshold = 2,     nsim = 1000, top_n = 1, progress = FALSE   ) }) #>    user  system elapsed  #>    0.04    0.07   11.46 CPA_agg_1000 #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> ageOlder (n = 1000) #>   Mean (SD): 0.078 (40.29) #>   Coverage intervals: 95% [-76.748, 82.717] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 74.724 (p=0.0629) #> ──────────────────────────────────────────────────────────────────────────────── system.time({   set_rng_state(123L)   CPA_trial_1000 <- clusterpermute(     ts_data_trial_spec,     family = \"binomial\", threshold = 2.8,     nsim = 1000, top_n = 1, progress = FALSE   ) }) #>    user  system elapsed  #>    0.02    0.01   18.39 CPA_trial_1000 #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2.8) ──────────── <null_cluster_dists> ── #> ageOlder (n = 1000) #>   Mean (SD): -0.045 (55.09) #>   Coverage intervals: 95% [-119.239, 125.506] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2.8) ──────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 102.450 (p=0.0689) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"background","dir":"Articles","previous_headings":"","what":"Background","title":"Geller et al. 2020","text":"data comes lexical decision task using pupilometry (specifically, measure pupil dilation) study differences processing difficulty print vs. cursive script. data available part gazer package Geller, Winn, Mahr, & Mirman 2020. follow Jason Geller’s tutorial used clusterperm package conduct cluster-based permutation analysis (CPA) differences pupil size print cursive conditions. data package (cursive_agg_data) looks slightly different used tutorial, ensure full reproducibility use exact data tutorial used: data comes prepared analysis box. following columns relevant analysis: subject: Unique identifier subjects script: within-participant factor variable (\"cursive\", \"print\") time: continuous measure time 0-2500ms 100ms intervals aggbaseline: response variable representing normalized (“baseline-corrected”) pupil size name cusrive_agg_data variable suggests, data aggregated within subject, collapsing across trials. following reproduces figure tutorial:  within-participant predictors like script, permutation algorithm randomly swap labels condition trials within subject. preserves temporal structure trial-level data (swapping trial-level grouping) well subject-level grouping structure (swapping trials across participants).","code":"library(dplyr) cursive_agg_data <- as_tibble(read.csv(\"https://raw.githubusercontent.com/jgeller112/drjasongeller/main/content/blog/2020-07-10-CBPT/blog_data.csv\")) cursive_agg_data #> # A tibble: 2,184 × 5 #>        X subject script  timebins aggbaseline #>    <int> <chr>   <chr>      <int>       <dbl> #>  1     1 10b.edf cursive        0      36.9   #>  2     2 10b.edf cursive      100      27.5   #>  3     3 10b.edf cursive      200      19.6   #>  4     4 10b.edf cursive      300      12.6   #>  5     5 10b.edf cursive      400       5.60  #>  6     6 10b.edf cursive      500       0.968 #>  7     7 10b.edf cursive      600      -5.98  #>  8     8 10b.edf cursive      700      -5.96  #>  9     9 10b.edf cursive      800       3.01  #> 10    10 10b.edf cursive      900      12.9   #> # ℹ 2,174 more rows library(ggplot2) script_fig <- ggplot(cursive_agg_data, aes(timebins, aggbaseline)) +   stat_summary(aes(linetype = script), geom = \"line\", linewidth = 1) script_fig"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"outline","dir":"Articles","previous_headings":"","what":"Outline","title":"Geller et al. 2020","text":"case study vignette showcases five features CPA jlmerclusterperm: Prepping data CPA using make_jlmer_spec() clusterpermute() default statistic = \"t\" comparison “t” “chisq” Specifying random effects Contrast coding Load package start Julia instance jlmerclusterperm_setup() proceeding.","code":"library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"a-prepping-a-specification-object","dir":"Articles","previous_headings":"","what":"A) Prepping a specification object","title":"Geller et al. 2020","text":"start simple specification object model aggbaseline using script predictor script_fig data: sanity check, fit global model data… …check ’s comparable expect lm(): full specification object CPA must also declare grouping structures present data. rule thumb every observation (row) data must uniquely identified combination columns subject, trial, time. CPA time series data time must always specified, subject must also always specified permutation algorithm respect subject-level grouping data. cursive_agg_data data collapses across trials within subject, column trial. However, leave trial argument unspecified observations uniquely identified subject time alone. instead 2 rows per subject-time combination, one script condition: Therefore need “dummy” column trial distinctly mark “cursive” vs. “script” trials. save new data cursive_agg: makes observations uniquely identified columns subject, trial, time: final specification object looks like following:","code":"simple_spec <- make_jlmer_spec(   formula = aggbaseline ~ 1 + script,   data = cursive_agg_data ) simple_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: aggbaseline ~ 1 + scriptprint #> Predictors: #>   script: scriptprint #> Data: #> # A tibble: 2,184 × 2 #>   aggbaseline scriptprint #>         <dbl>       <dbl> #> 1        36.9           0 #> 2        27.5           0 #> 3        19.6           0 #> # ℹ 2,181 more rows #> ──────────────────────────────────────────────────────────────────────────────── jlmer(simple_spec) #> <Julia object of type StatsModels.TableRegressionModel> #> ──────────────────────────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|)  Lower 95%  Upper 95% #> ──────────────────────────────────────────────────────────────────────── #> (Intercept)   46.7221     2.79622  16.71    <1e-61    41.2417    52.2026 #> scriptprint  -10.2582     3.95445  -2.59    0.0095   -18.0088    -2.5076 #> ──────────────────────────────────────────────────────────────────────── summary(lm(formula = aggbaseline ~ 1 + script, data = cursive_agg_data))$coefficients #>              Estimate Std. Error   t value     Pr(>|t|) #> (Intercept)  46.72214   2.796221 16.709029 4.512338e-59 #> scriptprint -10.25818   3.954454 -2.594083 9.547816e-03 cursive_agg_data %>%   count(subject, timebins) #> # A tibble: 1,092 × 3 #>    subject timebins     n #>    <chr>      <int> <int> #>  1 10b.edf        0     2 #>  2 10b.edf      100     2 #>  3 10b.edf      200     2 #>  4 10b.edf      300     2 #>  5 10b.edf      400     2 #>  6 10b.edf      500     2 #>  7 10b.edf      600     2 #>  8 10b.edf      700     2 #>  9 10b.edf      800     2 #> 10 10b.edf      900     2 #> # ℹ 1,082 more rows cursive_agg <- cursive_agg_data %>%   mutate(trial_type = paste0(script, \"_agg\")) cursive_agg #> # A tibble: 2,184 × 6 #>        X subject script  timebins aggbaseline trial_type  #>    <int> <chr>   <chr>      <int>       <dbl> <chr>       #>  1     1 10b.edf cursive        0      36.9   cursive_agg #>  2     2 10b.edf cursive      100      27.5   cursive_agg #>  3     3 10b.edf cursive      200      19.6   cursive_agg #>  4     4 10b.edf cursive      300      12.6   cursive_agg #>  5     5 10b.edf cursive      400       5.60  cursive_agg #>  6     6 10b.edf cursive      500       0.968 cursive_agg #>  7     7 10b.edf cursive      600      -5.98  cursive_agg #>  8     8 10b.edf cursive      700      -5.96  cursive_agg #>  9     9 10b.edf cursive      800       3.01  cursive_agg #> 10    10 10b.edf cursive      900      12.9   cursive_agg #> # ℹ 2,174 more rows cursive_agg %>%   count(subject, timebins, trial_type) %>%   distinct(n) #> # A tibble: 1 × 1 #>       n #>   <int> #> 1     1 cursive_agg_spec <- make_jlmer_spec(   formula = aggbaseline ~ 1 + script,   data = cursive_agg,   subject = \"subject\", trial = \"trial_type\", time = \"timebins\" ) cursive_agg_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: aggbaseline ~ 1 + scriptprint #> Predictors: #>   script: scriptprint #> Groupings: #>   Subject: subject #>   Trial: trial_type #>   Time: timebins #> Data: #> # A tibble: 2,184 × 5 #>   aggbaseline scriptprint subject trial_type  timebins #>         <dbl>       <dbl> <chr>   <chr>          <int> #> 1        36.9           0 10b.edf cursive_agg        0 #> 2        27.5           0 10b.edf cursive_agg      100 #> 3        19.6           0 10b.edf cursive_agg      200 #> # ℹ 2,181 more rows #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"b-cpa-with-default-statistic-t","dir":"Articles","previous_headings":"","what":"B) CPA with default statistic = \"t\"","title":"Geller et al. 2020","text":"CPA output original tutorial (using 100 simulations) copied comparison: Using threshold 2 default statistic = \"t\", clusterpermute() returns following: results look different original. due following: cluster-mass statistic (cms) lower. ran clusterpermute() default statistic = \"t\". Different clusters identified. simple model account subject-level variation, whereas original error term formula aggbaseline ~ script + Error(subject). sign cluster reversed. defaults aov() used original tutorial different default contrast coding regression model. now address issue turn, building CPA closely replicate results tutorial.","code":"#> #> effect   b0     b1  sign        cms           p #> #> script    0    100     1  10.121215   0.1584158 #> #> script  900   1000    -1  8.756152    0.1782178 #> #> script 1900   2500     1  82.198279   0.0099010 set_rng_state(123L) clusterpermute(   cursive_agg_spec,   statistic = \"t\", # Default value spelled out   threshold = 2,   nsim = 100,   progress = FALSE ) #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> scriptprint (n = 100) #>   Mean (SD): 0.000 (0.00) #>   Coverage intervals: 95% [0.000, 0.000] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> scriptprint #>   [2000, 2300]: -8.748 (p=0.0099) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"c-comparing-t-vs--chisq","dir":"Articles","previous_headings":"","what":"C) Comparing “t” vs. “chisq”","title":"Geller et al. 2020","text":"original tutorial used clusterperm::cluster_nhds() conduct CPA, fits ANOVA models time. , timewise statistic used chi-squared threshold determined p-value chi-squared statistic. Using chi-squared statistics p-value threshold also supported clusterpermute() using statistic = \"chisq\" (instead default \"t\"): returns cluster now numerically larger cluster-mass statistic, expected (chi-squared asymptotic t^2). , compare shape timewise statistics “t” threshold 2 “chisq” threshold p=0.05:  find clusters identified 2000ms-2300ms “t” “chisq”, peaks pronounced “chisq”. differences two inconsequential example, may produce different results cases. chi-squared statistic (jlmerclusterperm computes via likelihood ratio test) often preferred testing single parameters makes less assumptions tend robust (glmmFAQ). goodness--fit tests level predictor model formula, “chisq” less interpretable multi-level predictors (k-1 > 1). teasing apart contribution individual levels multi-level predictor, using statistic = \"t\" appropriate.","code":"set_rng_state(123L) clusterpermute(   cursive_agg_spec,   statistic = \"chisq\",   threshold = 0.05, # Threshold is now the p-value of the chi-squared statistic   nsim = 100,   progress = FALSE ) #> $null_cluster_dists #> ── Null cluster-mass distribution (chisq p < 0.05) ───── <null_cluster_dists> ── #> script (n = 100, df = 1) #>   Mean (SD): 0.000 (0.00) #>   Coverage intervals: 95% [0.000, 0.000] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (chisq p < 0.05) ───────────────── <empirical_clusters> ── #> script (df = 1) #>   [2000, 2300]: -19.085 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── timewise_ts <- compute_timewise_statistics(cursive_agg_spec, statistic = \"t\") timewise_chisqs <- compute_timewise_statistics(cursive_agg_spec, statistic = \"chisq\") library(ggplot2) timewise_fig <- ggplot(mapping = aes(x = time, y = statistic)) +   geom_line(     aes(color = \"fixed-t\"),     linewidth = 1.5,     data = tidy(timewise_ts)   ) +   geom_line(     aes(color = \"fixed-chisq\"),     linewidth = 1.5,     data = tidy(timewise_chisqs)   ) +   geom_hline(     aes(yintercept = c(-2, 2, qchisq(.95, df = 1), -qchisq(.95, df = 1))),     color = rep(c(\"#00BFC4\", \"#F8766D\"), each = 2), linetype = 2   ) +   scale_color_manual(     values = setNames(       c(\"#E69F00\", \"#56B4E9\", \"#009E73\", \"#F0E442\"),       c(\"fixed-t\", \"fixed-chisq\", \"re-intercept-chisq\", \"re-max-chisq\")     )   ) timewise_fig"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"d-specifying-random-effects","dir":"Articles","previous_headings":"","what":"D) Specifying random effects","title":"Geller et al. 2020","text":"biggest missing component point subject random effects, original tutorial captures via error term Error(subject). isn’t strict equivalent regression, specifying random intercepts subject (1 | subject) gets us close: repeat results original tutorial comparison: now plot timewise statistics random intercept models:  point may wonder whether results change much used maximal model random effects structure (1 + script | subject). first create another specification object maximal formula: compute timewise statistics: find chisq statistics maximal model virtually identical parsimonious, intercept-model (two lines overlap):  Parsimony incredibly important simulation - jlmerclusterperm fast, model complexity still major bottleneck. adding extra random effect terms (random slope correlation) negligible effects statistic, removing likely inconsequential CPA . show following CPA run, uses maximal model. Notice results identical intercept-re_CPA substantially slower: run maximal mixed model CPA purely demonstration purposes. random effects structure actually unidentifiable given aggregated data, point time 2 observations subject (mean condition). Lastly, back figure data, annotate clusters detected chisq p-value threshold 0.05 random intercept CPA:","code":"cursive_agg_spec_re <- make_jlmer_spec(   formula = aggbaseline ~ 1 + script + (1 | subject),   data = cursive_agg,   subject = \"subject\", time = \"timebins\", trial = \"trial_type\" ) set_rng_state(123L) system.time({   re_CPA <- clusterpermute(     cursive_agg_spec_re,     statistic = \"chisq\",     threshold = 0.05,     nsim = 100,     progress = FALSE   ) }) #>    user  system elapsed  #>    0.00    0.08    8.12 re_CPA #> $null_cluster_dists #> ── Null cluster-mass distribution (chisq p < 0.05) ───── <null_cluster_dists> ── #> script (n = 100, df = 1) #>   Mean (SD): 1.132 (12.58) #>   Coverage intervals: 95% [-21.081, 32.649] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (chisq p < 0.05) ───────────────── <empirical_clusters> ── #> script (df = 1) #>   [0, 100]: -9.770 (p=0.1881) #>   [900, 1000]: 8.521 (p=0.2475) #>   [1900, 2500]: -73.440 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── #> #> effect   b0     b1  sign        cms           p #> #> script    0    100     1  10.121215   0.1584158 #> #> script  900   1000    -1  8.756152    0.1782178 #> #> script 1900   2500     1  82.198279   0.0099010 timewise_chisqs_re <- compute_timewise_statistics(cursive_agg_spec_re, statistic = \"chisq\") timewise_fig_re <- timewise_fig +   geom_line(     aes(color = \"re-intercept-chisq\"),     linewidth = 1.5,     data = tidy(timewise_chisqs_re)   ) timewise_fig_re cursive_agg_spec_re_max <- make_jlmer_spec(   formula = aggbaseline ~ 1 + script + (1 + script | subject),   data = cursive_agg,   subject = \"subject\", time = \"timebins\", trial = \"trial_type\" ) timewise_chisqs_re_max <- compute_timewise_statistics(cursive_agg_spec_re_max, statistic = \"chisq\") #> ℹ 1 singular fit (3.85%). timewise_fig_re +   geom_line(     aes(color = \"re-max-chisq\"),     linewidth = 1.5,     data = tidy(timewise_chisqs_re_max)   ) set_rng_state(123L) system.time({   re_max_CPA <- clusterpermute(     cursive_agg_spec_re_max,     statistic = \"chisq\",     threshold = 0.05,     nsim = 100,     progress = FALSE   ) }) #>    user  system elapsed  #>    0.01    0.05   12.11 re_max_CPA #> $null_cluster_dists #> ── Null cluster-mass distribution (chisq p < 0.05) ───── <null_cluster_dists> ── #> script (n = 100, df = 1) #>   Mean (SD): 1.132 (12.58) #>   Coverage intervals: 95% [-21.081, 32.649] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (chisq p < 0.05) ───────────────── <empirical_clusters> ── #> script (df = 1) #>   [0, 100]: -9.770 (p=0.1881) #>   [900, 1000]: 8.521 (p=0.2475) #>   [1900, 2500]: -73.440 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── empirical_clusters_df <- tidy(extract_empirical_clusters(timewise_chisqs_re, threshold = 0.05)) script_fig +   geom_segment(     aes(       x = start, xend = end, y = -Inf, yend = -Inf,       color = factor(sign(sum_statistic))     ),     linewidth = 10,     inherit.aes = FALSE,     data = empirical_clusters_df   ) +   geom_text(     aes(       y = -Inf, x = start + (end - start) / 2,       label = paste(\"Cluster\", id)     ),     vjust = -2,     inherit.aes = FALSE,     data = empirical_clusters_df   ) +   labs(color = \"Sign of cluster\")"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"e-contrast-coding","dir":"Articles","previous_headings":"","what":"E) Contrast coding","title":"Geller et al. 2020","text":"last piece puzzle flipped sign effect. Whereas detect negative cluster line cursive line print, original tutorial reports positive cluster (output repeated ): Fixing trivial - just takes different choice contrast. example, can flip levels factor make “print” reference level: also sum-code set “print” -1 “cursive” 1: CPA operates domain timewise statistics (effect size), whether difference “print” “cursive” expressed unit 1 (treatment coding) 2 (contrast coding) bearing CPA. conclude, re-run time 1000-simulation CPA using intercept-model new treatment coding script “print” reference level. Finally, annotating just significant cluster figure:","code":"#> #> effect   b0     b1  sign        cms           p #> #> script    0    100     1  10.121215   0.1584158 #> #> script  900   1000    -1  8.756152    0.1782178 #> #> script 1900   2500     1  82.198279   0.0099010 rev_contrast_df <- cursive_agg rev_contrast_df$script <- factor(rev_contrast_df$script, levels = c(\"print\", \"cursive\")) contrasts(rev_contrast_df$script) #>         cursive #> print         0 #> cursive       1 reverse_contrast_spec <- make_jlmer_spec(   formula = aggbaseline ~ 1 + script + (1 | subject),   data = rev_contrast_df,   subject = \"subject\", time = \"timebins\", trial = \"trial_type\" ) set_rng_state(123L) clusterpermute(   reverse_contrast_spec,   statistic = \"chisq\",   threshold = 0.05,   nsim = 100,   progress = FALSE ) #> $null_cluster_dists #> ── Null cluster-mass distribution (chisq p < 0.05) ───── <null_cluster_dists> ── #> script (n = 100, df = 1) #>   Mean (SD): -1.132 (12.58) #>   Coverage intervals: 95% [-32.649, 21.081] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (chisq p < 0.05) ───────────────── <empirical_clusters> ── #> script (df = 1) #>   [0, 100]: 9.770 (p=0.1881) #>   [900, 1000]: -8.521 (p=0.2475) #>   [1900, 2500]: 73.440 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── sum_contrast_df <- cursive_agg sum_contrast_df$script <- factor(sum_contrast_df$script) contrasts(sum_contrast_df$script) <- contr.sum(2) contrasts(sum_contrast_df$script) #>         [,1] #> cursive    1 #> print     -1 sum_contrast_spec <- make_jlmer_spec(   formula = aggbaseline ~ 1 + script + (1 | subject),   data = sum_contrast_df,   subject = \"subject\", time = \"timebins\", trial = \"trial_type\" ) set_rng_state(123L) clusterpermute(   sum_contrast_spec,   statistic = \"chisq\",   threshold = 0.05,   nsim = 100,   progress = FALSE ) #> $null_cluster_dists #> ── Null cluster-mass distribution (chisq p < 0.05) ───── <null_cluster_dists> ── #> script (n = 100, df = 1) #>   Mean (SD): -1.132 (12.58) #>   Coverage intervals: 95% [-32.649, 21.081] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (chisq p < 0.05) ───────────────── <empirical_clusters> ── #> script (df = 1) #>   [0, 100]: 9.770 (p=0.1881) #>   [900, 1000]: -8.521 (p=0.2475) #>   [1900, 2500]: 73.440 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── set_rng_state(123L) system.time({   final_CPA <- clusterpermute(     reverse_contrast_spec,     statistic = \"chisq\",     threshold = 0.05,     nsim = 1000,     progress = FALSE   ) }) #>    user  system elapsed  #>    0.09    0.06   12.41 final_CPA #> $null_cluster_dists #> ── Null cluster-mass distribution (chisq p < 0.05) ───── <null_cluster_dists> ── #> script (n = 1000, df = 1) #>   Mean (SD): -0.107 (16.50) #>   Coverage intervals: 95% [-36.966, 41.530] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (chisq p < 0.05) ───────────────── <empirical_clusters> ── #> script (df = 1) #>   [0, 100]: 9.770 (p=0.2108) #>   [900, 1000]: -8.521 (p=0.2358) #>   [1900, 2500]: 73.440 (p=0.0090) #> ──────────────────────────────────────────────────────────────────────────────── signif_clusters_df <- tidy(final_CPA$empirical_clusters) %>%   filter(pvalue < 0.05) signif_clusters_df #> # A tibble: 1 × 7 #>   predictor id    start   end length sum_statistic  pvalue #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>   <dbl> #> 1 script    3      1900  2500      7          73.4 0.00899 script_fig +   geom_segment(     aes(x = start, xend = end, y = -Inf, yend = -Inf),     color = \"steelblue\", linewidth = 10,     inherit.aes = FALSE,     data = signif_clusters_df   ) +   geom_text(     aes(       y = -Inf, x = start + (end - start) / 2,       label = paste(\"p =\", signif(pvalue, 3))     ),     vjust = -2,     inherit.aes = FALSE,     data = signif_clusters_df   )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/jlmerclusterperm.html","id":"overview-of-cpa","dir":"Articles","previous_headings":"","what":"Overview of CPA","title":"Introduction to jlmerclusterperm","text":"Cluster-based permutation analysis (CPA) simulation-based, non-parametric statistical test difference groups time series. suitable analyzing densely-sampled time data (EEG eye-tracking research) research hypothesis often specified existence effect (e.g., predicted higher-order cognitive mechanisms) agnostic temporal details effect (precise moment emergence). CPA formalizes two intuitions means difference groups: countable unit difference (.e., cluster) contiguous, uninterrupted span sufficiently large differences time point (determined via threshold). degree extremity cluster (.e., cluster-mass statistic) measure sensitive magnitude difference, variability, sample size (e.g., t-statistic regression). CPA procedure identifies empirical clusters time series data tests significance observed cluster-mass statistics bootstrapped permutations data. bootstrapping procedure samples “null” (via random shuffling condition labels), yielding distribution cluster-mass statistics emerging chance. statistical significance cluster probability observing cluster-mass statistic extreme cluster’s simulated null distribution.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/jlmerclusterperm.html","id":"package-design","dir":"Articles","previous_headings":"","what":"Package design","title":"Introduction to jlmerclusterperm","text":"jlmerclusterperm provides wholesale piecemeal approach conducting CPA. main workhorse function package clusterpermute(), composed five smaller functions called internally succession. smaller functions representing algorithmic steps CPA also exported, allow control procedure (e.g., debugging diagnosing CPA run). See function documentation .","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/jlmerclusterperm.html","id":"organization-of-vignettes","dir":"Articles","previous_headings":"","what":"Organization of vignettes","title":"Introduction to jlmerclusterperm","text":"package vignettes roughly divided two groups: topics case studies. researcher data ready analysis, recommended go case studies first pluck relevant bits desired analysis order increasing complexity, case studies : Garrison et al. 2020, introduces package’s functions running CPA, demonstrating wholesale piecemeal approaches. also compares use linear vs. logistic regression calculation timewise statistics. Geller et al. 2020, demonstrates use random effects regression contrasts CPA specification. also compares use t vs. chisq timewise statistics. de Carvalho et al. 2021, showcases using custom regression contrasts test relationship multiple levels factor. also discusses issues surrounding complexity timewise regression model. topics cover general package features: Tidying output: tidy() ways collecting jlmerclusterperm objects tidy data. Julia interface: interacting Julia objects using JuliaConnectoR package. Reproducibility: using Julia RNG make CPA runs reproducible. Asynchronous CPA: running CPA background process using future package. Comparison eyetrackingR: Translating eyetrackingR code CPA jlmerclusterperm.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/jlmerclusterperm.html","id":"compared-to-other-implementations","dir":"Articles","previous_headings":"","what":"Compared to other implementations","title":"Introduction to jlmerclusterperm","text":"R packages cluster-based permutation analysis, clusterperm, eyetrackingR, permuco, permutes. Compared existing implementations, jlmerclusterperm package optimized CPAs based mixed-effects regression models, suitable typical experimental research data multiple, crossed grouping structures (e.g., subjects items). also package interface individual algorithmic steps CPA. Lastly, thanks Julia back-end, bootstrapping fast even mixed models.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/jlmerclusterperm.html","id":"further-readings","dir":"Articles","previous_headings":"","what":"Further readings","title":"Introduction to jlmerclusterperm","text":"original method paper CPA Maris & Oostenveld 2007 recent overview paper Meyer et al. 2021 (CPA EEG research, applies broadly) critical paper Sassenhagen & Draschkow 2019 DOs DON’Ts interpreting reporting CPA results. review paper eyetracking analysis methods Ito & Knoeferle 2022 compares CPA statistical techniques testing difference groups time series data. JOSS paper permuco package (Frossard & Renaud 2021) references therein. Benedikt V. Ehinger’s blog post, includes detailed graphical explainer CPA procedure.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/julia-interface.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Julia interface","text":"","code":"library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/julia-interface.html","id":"interacting-with-julia-objects","dir":"Articles","previous_headings":"","what":"Interacting with Julia objects","title":"Julia interface","text":"jlmerclusterperm functions collect Julia objects R objects, except jlmer() to_jlmer() return GLM.jl MixedModels.jl fitted model objects. can use functions JuliaConnectoR interact (pointers ) Julia objects: can also call functions Julia packages already installed. example, Effects.jl marginaleffects/emmeans-like features: Note jlmer() to_lmer() just conveniences sanity checking steps cluster-based permutation test, thus used general interface fitting regression models Julia serious analyses.","code":"jmod <- to_jlmer(Reaction ~ Days + (Days | Subject), lme4::sleepstudy) jmod #> <Julia object of type LinearMixedModel{Float64}> #> Variance components: #>             Column    Variance Std.Dev.   Corr. #> Subject  (Intercept)  565.51067 23.78047 #>          Days          32.68212  5.71683 +0.08 #> Residual              654.94145 25.59182 #> ────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|) #> ────────────────────────────────────────────────── #> (Intercept)  251.405      6.63226  37.91    <1e-99 #> Days          10.4673     1.50224   6.97    <1e-11 #> ────────────────────────────────────────────────── library(JuliaConnectoR) juliaLet(\"x.rePCA\", x = jmod) #> <Julia object of type NamedTuple{(:Subject,), Tuple{Vector{Float64}}}> #> (Subject = [0.5406660682947617, 1.0],) juliaCall(\"issingular\", jmod) #> [1] FALSE # juliaEval('using Pkg; Pkg.activate(); Pkg.add(\"Effects\")') juliaEval(\"using Effects\") juliaLet(\"effects(x.namedelements, y)\", x = list(Days = 2:5), y = jmod) #> <Julia object of type DataFrame> #> 4×5 DataFrame #>  Row │ Days   Reaction  err      lower    upper #>      │ Int64  Float64   Float64  Float64  Float64 #> ─────┼──────────────────────────────────────────── #>    1 │     2   272.34   7.09427  265.245  279.434 #>    2 │     3   282.807  7.70544  275.102  290.512 #>    3 │     4   293.274  8.55557  284.719  301.83 #>    4 │     5   303.742  9.58128  294.16   313.323"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/julia-interface.html","id":"julia-global-options","dir":"Articles","previous_headings":"","what":"Julia global options","title":"Julia interface","text":"functions involved cluster-based permutation analysis print progress bars update real time console. function julia_progress() controls whether show progress bar width printed progress bar.","code":"# Set Julia progress options and save old state old_progress_opts <- julia_progress(show = FALSE, width = 30) old_progress_opts #> $show #> [1] TRUE #> #> $width #> [1] 50 julia_progress() #> $show #> [1] FALSE #> #> $width #> [1] 30  # Restored old state julia_progress(old_progress_opts) identical(old_progress_opts, julia_progress()) #> [1] TRUE  # The syntax to reset progress options julia_progress(show = TRUE, width = \"auto\") julia_progress()"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/julia-interface.html","id":"the-julia-environment","dir":"Articles","previous_headings":"","what":"The Julia environment","title":"Julia interface","text":"following packages loaded Julia environment via jlmerclusterperm_setup():","code":"cat(readLines(system.file(\"julia/Project.toml\", package = \"jlmerclusterperm\")), sep = \"\\n\") #> [deps] #> DataFrames = \"a93c6f00-e57d-5684-b7b6-d8193f3e46c0\" #> Distributions = \"31c24e10-a181-5473-b8eb-7969acd0382f\" #> GLM = \"38e38edf-8417-5370-95a0-9cbb8c7f171a\" #> JlmerClusterPerm = \"2a59065a-9564-4903-82dd-0e42ce19d0e1\" #> MixedModels = \"ff71e718-51f3-5ec2-a782-8ffcbfa3c316\" #> ProgressMeter = \"92933f4c-e287-5a05-a399-4b506db050ca\" #> Random = \"9a3f8284-a2c9-5f02-9a11-845980a1fd5c\" #> Random123 = \"74087812-796a-5b5d-8853-05524746bad3\" #> StatsBase = \"2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91\" #> StatsModels = \"3eaba693-59b7-5ba5-a881-562e759f1c8d\" #> Suppressor = \"fd094767-a336-5f1f-9728-57cf17d0bbfb\" #>  #> # add DataFrames@1.3, GLM@1.8, MixedModels@4, ProgressMeter@1.7, Random123@1.6, StatsBase@0.33, StatsModels@0.7, Suppressor@0.2 #> # add Distributions, Random #> # dev JlmerClusterPerm #>  #> [compat] #> DataFrames = \"< 1.4\" #> GLM = \"1.8\" #> JlmerClusterPerm = \"=0.1.1\" #> MixedModels = \"4\" #> ProgressMeter = \"1.7\" #> Random123 = \"1.6\" #> StatsBase = \"0.33\" #> StatsModels = \"0.7\" #> Suppressor = \"0.2\" #> julia = \"1.8\""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/julia-interface.html","id":"optional-configurations","dir":"Articles","previous_headings":"","what":"Optional configurations","title":"Julia interface","text":"speed-ups model fitting, install MKL.jl (Intel core) AppleAccelerate.jl (Mac) load start modifying startup.jl file. example, Windows machine startup file located : load MKL.jl:","code":"JuliaConnectoR::juliaCall(\"Base._local_julia_startup_file\") #> \"C:/Users/jchoe/.julia/config/startup.jl\" if contains(first(Sys.cpu_info()).model, \"Intel\")   using MKL end"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/reproducibility.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Reproducibility","text":"Minimal example data:  jlmer specification object used showcase:","code":"library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE) library(dplyr, warn.conflicts = FALSE) library(ggplot2) chickweights <- as_tibble(ChickWeight) %>%   mutate(diet = forcats::fct_collapse(Diet, A = c(1, 3), B = c(2, 4))) %>%   filter(Time %in% 0:20) ggplot(chickweights, aes(Time, weight, color = diet)) +   stat_summary(fun.args = list(mult = 1.96)) +   stat_summary(geom = \"line\") +   theme_minimal(base_size = 18) jlmer_spec <- make_jlmer_spec(   formula = weight ~ diet + (1 | Chick), data = chickweights,   subject = \"Chick\", time = \"Time\" ) jlmer_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + dietB + (1 | Chick) #> Predictors: #>   diet: dietB #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 4 #>   weight dietB Chick  Time #>    <dbl> <dbl> <chr> <dbl> #> 1     42     0 1         0 #> 2     51     0 1         2 #> 3     59     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/reproducibility.html","id":"rng-seed-vs--counter","dir":"Articles","previous_headings":"","what":"RNG seed vs. counter","title":"Reproducibility","text":"CPA powered bootstrapping, inherently stochastic procedure. Reproducibility guaranteed clusterpermute() calls ran seed counter values. properties Julia random number generator used jlmerclusterperm refer two slightly different concepts: seed conceptually equivalent “seed” talked R, set.seed(). seed used RNG’s initialization, locking specific state persists session. jlmerclusterperm, default seed 1 primarily controlled via options(\"jlmerclusterperm.seed\") must set calling jlmerclusterperm_setup(). counter initializes 0 increments value every time RNG consumed produce random number. equivalent concept base R, though ’s handy intuitive way interfacing RNG’s state interactive session. counter can manipulated via functions reset_rng_state(), set_rng_state(), get_rng_state(). rest vignette showcase interacting counter state, useful practice.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/reproducibility.html","id":"behavior-of-rng-counter","dir":"Articles","previous_headings":"","what":"Behavior of RNG counter","title":"Reproducibility","text":"start fresh session, counter initialized zero. specify seed, ’s one: start simple clusterpermute() 100 simulations. reproducibility profile run seed 1 counter 0: run, counter now incremented 3712: call clusterpermute() now runs new counter value, produces different result. , expected given stochastic nature CPA (thus need run tens thousands simulations reporting results): second run also increments counter state, note ’s exactly amount (possible predict): difference two runs specifically null distribution cluster-mass statistics. 100-simulation samples drawn share underlying (null) distribution, though identical one another (rarely ).","code":"get_rng_state() #> [1] 0 get_rng_seed() #> [1] 1 CPA_1_0 <- clusterpermute(jlmer_spec, threshold = 1, nsim = 100, progress = FALSE) CPA_1_0 #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 100) #>   Mean (SD): 0.724 (5.54) #>   Coverage intervals: 95% [-10.847, 12.592] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 (p=0.0198) #> ──────────────────────────────────────────────────────────────────────────────── get_rng_state() #> [1] 3712 CPA_1_3712 <- clusterpermute(jlmer_spec, threshold = 1, nsim = 100, progress = FALSE) CPA_1_3712 #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 100) #>   Mean (SD): -0.618 (8.51) #>   Coverage intervals: 95% [-18.328, 19.076] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 (p=0.0792) #> ──────────────────────────────────────────────────────────────────────────────── get_rng_state() #> [1] 7427 ggplot(mapping = aes(sum_statistic)) +   geom_density(     aes(fill = \"1\"),     alpha = .5,     data = tidy(CPA_1_0$null_cluster_dists)   ) +   geom_density(     aes(fill = \"3712\"),     alpha = .5,     data = tidy(CPA_1_3712$null_cluster_dists)   ) +   guides(fill = guide_legend(\"Counter\")) +   theme_classic()"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/reproducibility.html","id":"manipulating-rng-counter","dir":"Articles","previous_headings":"","what":"Manipulating RNG counter","title":"Reproducibility","text":"Using set_rng_counter(), possible reproduce prior run. example, reproduce second clusterpermute() call, set counter back 3712 set_rng_state(): results subsequent run counter state identical previous run: reproduce first CPA ran counter 0, can use set_rng_state() simply use reset_rng_state() alias set_rng_state(0): behavior can observed specific step simulating null distribution well. function corresponding step permute_timewise_statistics() run 100 simulations step counter zero: Running cluster detection algorithm result using extract_null_cluster_dists() returns null cluster-mass distribution identical CPA_1_0:","code":"set_rng_state(3712) get_rng_state() #> [1] 3712 identical(   CPA_1_3712,   clusterpermute(jlmer_spec, threshold = 1, nsim = 100, progress = FALSE) ) #> [1] TRUE reset_rng_state() get_rng_state() #> [1] 0 identical(   CPA_1_0,   clusterpermute(jlmer_spec, threshold = 1, nsim = 100, progress = FALSE) ) #> [1] TRUE reset_rng_state() get_rng_state() #> [1] 0 null_statistics <- permute_timewise_statistics(jlmer_spec, nsim = 100) get_rng_state() #> [1] 3712 null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 1) null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 100) #>   Mean (SD): 0.724 (5.54) #>   Coverage intervals: 95% [-10.847, 12.592] #> ──────────────────────────────────────────────────────────────────────────────── identical(null_cluster_dists, CPA_1_0$null_cluster_dists) #> [1] TRUE"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/reproducibility.html","id":"technical-details-and-notes","dir":"Articles","previous_headings":"","what":"Technical details and notes","title":"Reproducibility","text":"Julia back-end uses thread-safe, counter-based RNG called Threefry (specifically, Random123.Threefry2x) comes Random123.jl Julia library adaptation original algorithm developed Salmon, Moraes, Dror, Shaw (2011). RNG initialized “rounds” value 20, also default. Threefry used speed stability across Julia versions. Additionally, noted earlier, RNG initializes explicit seed 1. behavior different case R, random seed chosen user every new session. consequence absence explicit seeding via options(\"jlmerclusterperm.seed\"), first run every new session (.e., runs counter value zero) identical. Given interface RNG counter, many unique advantages using different seed manipulating RNG counter. interface RNG seed also provided via set_rng_seed() get_rng_seed(). Lastly, please aware changing seed R session (e.g., via set.seed()) effect jlmerclusterperm functions powered Julia back-end.","code":"# Restart Julia session jlmerclusterperm_setup(restart = TRUE, verbose = FALSE)  # RNG seed is again 1 get_rng_seed() #> [1] 1  # So results of the first CPA of the session (with counter = 0) is the same get_rng_state() #> [1] 0 identical(   CPA_1_0,   clusterpermute(jlmer_spec, threshold = 1, nsim = 100, progress = FALSE) ) #> [1] TRUE"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/reproducibility.html","id":"miscellaneous","dir":"Articles","previous_headings":"","what":"Miscellaneous","title":"Reproducibility","text":"may wondering whether two CPAs overlap range counter values also share overlapping samples null. actually case practice shuffling CPA implemented:  two colored lines represent two 100-simulation bootstrap permutations differ initial RNG counter value 1. point shape two lines look different - one just shifted form another. formal investigations (e.g., simulating power CPA), ’s course best practice generate random seeds. counter convenience balances interactivity reproducibility.","code":"# 100 simulations of timewise statistics ## First CPA using counter at 1 set_rng_state(1) x <- permute_timewise_statistics(jlmer_spec, nsim = 100) ## Second CPA using counter at 2 set_rng_state(2) y <- permute_timewise_statistics(jlmer_spec, nsim = 100)  # The null cluster-mass statistics x_null <- extract_null_cluster_dists(x, threshold = 1) x_null_stats <- tidy(x_null)$sum_statistic y_null <- extract_null_cluster_dists(y, threshold = 1) y_null_stats <- tidy(y_null)$sum_statistic  # Relationship between two CPAs with similar initial counters matplot(cbind(x_null_stats, y_null_stats), type = \"b\", pch = 1)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Tidying output","text":"Minimal example data:  jlmer specification object used showcase:","code":"library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) library(dplyr, warn.conflicts = FALSE) library(ggplot2) chickweights <- as_tibble(ChickWeight) %>%   mutate(diet = forcats::fct_collapse(Diet, A = c(1, 3), B = c(2, 4))) %>%   filter(Time %in% 0:20) ggplot(chickweights, aes(Time, weight, color = diet)) +   stat_summary(fun.args = list(mult = 1.96)) +   stat_summary(geom = \"line\") +   theme_minimal(base_size = 18) jlmer_spec <- make_jlmer_spec(   formula = weight ~ diet + (1 | Chick), data = chickweights,   subject = \"Chick\", time = \"Time\" ) jlmer_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + dietB + (1 | Chick) #> Predictors: #>   diet: dietB #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 4 #>   weight dietB Chick  Time #>    <dbl> <dbl> <chr> <dbl> #> 1     42     0 1         0 #> 2     51     0 1         2 #> 3     59     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"a-tour-of-tidy-methods","dir":"Articles","previous_headings":"","what":"A tour of tidy() methods","title":"Tidying output","text":"non-<data.frame> objects returned jlmerclusterperm functions tidy() method return data frame representation object.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"timewise_statistics-objects","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"<timewise_statistics> objects","title":"Tidying output","text":"Functions compute_timewise_statistics() permute_timewise_statistics() return object class <timewise_statistics>, array statistics. Note isn’t stylized print() method class. Instead, recommended use tidy() inspect results data frame statistic represented row. case compute_timewise_statistics(), output matrix (2D array) predictor time. dimension becomes column tidy() data: case compute_timewise_statistics(), output 3D array simulation, time, predictor. , dimension becomes column tidy() data:","code":"empirical_statistics <- compute_timewise_statistics(jlmer_spec) class(empirical_statistics) #> [1] \"timewise_statistics\" dim(empirical_statistics) # predictor x time #> [1]  1 11 tidy(empirical_statistics) #> # A tibble: 11 × 3 #>    predictor  time statistic #>    <chr>     <dbl>     <dbl> #>  1 dietB         0    -1.09  #>  2 dietB         2     2.29  #>  3 dietB         4     3.14  #>  4 dietB         6     4.01  #>  5 dietB         8     2.90  #>  6 dietB        10     2.44  #>  7 dietB        12     2.18  #>  8 dietB        14     1.26  #>  9 dietB        16     0.673 #> 10 dietB        18     0.532 #> 11 dietB        20     0.809 null_statistics <- permute_timewise_statistics(jlmer_spec, nsim = 1000) class(null_statistics) #> [1] \"timewise_statistics\" dim(null_statistics) # simulation x time x predictor #> [1] 1000   11    1 tidy(null_statistics) #> # A tibble: 11,000 × 4 #>    predictor  time statistic sim   #>    <chr>     <dbl>     <dbl> <fct> #>  1 dietB         0   -0.309  0001  #>  2 dietB         0    2.10   0002  #>  3 dietB         0   -2.83   0003  #>  4 dietB         0   -0.568  0004  #>  5 dietB         0    2.10   0005  #>  6 dietB         0   -1.09   0006  #>  7 dietB         0   -0.830  0007  #>  8 dietB         0   -0.0515 0008  #>  9 dietB         0   -0.0515 0009  #> 10 dietB         0    0.206  0010  #> # ℹ 10,990 more rows"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"empirical_clusters-objects","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"<empirical_clusters> objects","title":"Tidying output","text":"<empirical_clusters> object returned extract_empirical_clusters() print method defined, makes easier eye: tidy() data frame representation also contains information like length, number time points cluster spans. id variable uniquely identifies cluster within predictor.","code":"empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 1) class(empirical_clusters) #> [1] \"empirical_clusters\" empirical_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 #> ──────────────────────────────────────────────────────────────────────────────── tidy(empirical_clusters) #> # A tibble: 1 × 6 #>   predictor id    start   end length sum_statistic #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl> #> 1 dietB     1         2    14      7          18.2"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"null_cluster_dists-objects","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"<null_cluster_dists> objects","title":"Tidying output","text":"<null_cluster_dists> object returned extract_null_cluster_dists() also print method defined, displays summary statistics: null distribution constructed largest cluster-mass statistic observed bootstrapped permutations original data. words, simulation contributes one sample null. tidy() representation, simulation predictor represents row. sum_statistic column represents size largest cluster simulation. clusters found, column zero values start, end, length NA. output tidy(null_statistics) tidy(null_cluster_dists) form relational data. example, can use left_join() find time wise statistics permuted data extreme cluster-mass statistic: Note sum_statistic simply sum statistic values span largest cluster:","code":"null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 1) class(null_cluster_dists) #> [1] \"null_cluster_dists\" null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 1000) #>   Mean (SD): 0.014 (7.11) #>   Coverage intervals: 95% [-15.360, 15.902] #> ──────────────────────────────────────────────────────────────────────────────── tidy(null_cluster_dists) #> # A tibble: 1,000 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 dietB         8    20      7         11.1  0001  #>  2 dietB         6    10      3          3.75 0002  #>  3 dietB        18    20      2          2.07 0003  #>  4 dietB         2    10      5          6.00 0004  #>  5 dietB         4    10      4          5.91 0005  #>  6 dietB        NA    NA     NA          0    0006  #>  7 dietB         8    20      7         12.2  0007  #>  8 dietB         2     8      4          5.29 0008  #>  9 dietB        NA    NA     NA          0    0009  #> 10 dietB        NA    NA     NA          0    0010  #> # ℹ 990 more rows largest_permuted_cluster <- tidy(null_cluster_dists) %>%   slice(which.max(abs(sum_statistic))) %>%   left_join(tidy(null_statistics), by = c(\"predictor\", \"sim\")) largest_permuted_cluster #> # A tibble: 11 × 8 #>    predictor start   end length sum_statistic sim    time statistic #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> <dbl>     <dbl> #>  1 dietB         2    20     10          29.4 0319      0     -1.09 #>  2 dietB         2    20     10          29.4 0319      2      1.67 #>  3 dietB         2    20     10          29.4 0319      4      2.56 #>  4 dietB         2    20     10          29.4 0319      6      2.97 #>  5 dietB         2    20     10          29.4 0319      8      3.47 #>  6 dietB         2    20     10          29.4 0319     10      3.54 #>  7 dietB         2    20     10          29.4 0319     12      3.49 #>  8 dietB         2    20     10          29.4 0319     14      3.54 #>  9 dietB         2    20     10          29.4 0319     16      3.05 #> 10 dietB         2    20     10          29.4 0319     18      2.65 #> 11 dietB         2    20     10          29.4 0319     20      2.48 largest_permuted_cluster %>%   filter(between(time, start, end)) %>%   summarize(sum_statistic = unique(sum_statistic), sum(statistic)) #> # A tibble: 1 × 2 #>   sum_statistic `sum(statistic)` #>           <dbl>            <dbl> #> 1          29.4             29.4"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"in-other-contexts","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"In other contexts","title":"Tidying output","text":"output calculate_clusters_pvalues() just <empirical_clusters> object augmented p-values: available, p-values returned column tidy() data: Note p-value cluster 0.027 27 samples null extreme observed: case one-fell-swoop function clusterpermute(), <null_cluster_dists> object <empirical_clusters> object returned list: can tidy() elements list lapply() purrr::map():","code":"tested_clusters <- calculate_clusters_pvalues(empirical_clusters, null_cluster_dists) tested_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 (p=0.0270) #> ──────────────────────────────────────────────────────────────────────────────── tidy(tested_clusters) #> # A tibble: 1 × 7 #>   predictor id    start   end length sum_statistic pvalue #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>  <dbl> #> 1 dietB     1         2    14      7          18.2  0.027 tidy(null_cluster_dists) %>%   filter(abs(sum_statistic) >= abs(tidy(tested_clusters)$sum_statistic)) #> # A tibble: 27 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 dietB         0    20     11         -18.7 0072  #>  2 dietB         0    20     11         -19.4 0114  #>  3 dietB         6    20      8          19.0 0118  #>  4 dietB         0    20     11         -23.6 0137  #>  5 dietB         4    20      9         -21.7 0142  #>  6 dietB         2    20     10          24.7 0268  #>  7 dietB         4    20      9          22.7 0289  #>  8 dietB         2    20     10          29.4 0319  #>  9 dietB         2    20     10          26.4 0325  #> 10 dietB         4    20      9          26.5 0384  #> # ℹ 17 more rows # Timing the entire process this time system.time({   full_test <- clusterpermute(jlmer_spec, threshold = 1, nsim = 1000, progress = FALSE) }) #>    user  system elapsed  #>    0.02    0.04    2.12 full_test #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 1000) #>   Mean (SD): 0.066 (7.61) #>   Coverage intervals: 95% [-16.959, 17.216] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 (p=0.0420) #> ──────────────────────────────────────────────────────────────────────────────── lapply(full_test, tidy) #> $null_cluster_dists #> # A tibble: 1,000 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 dietB        14    20      4         -5.26 0001  #>  2 dietB        14    20      4          9.22 0002  #>  3 dietB         6    20      8         11.8  0003  #>  4 dietB        NA    NA     NA          0    0004  #>  5 dietB        NA    NA     NA          0    0005  #>  6 dietB        16    20      3         -3.69 0006  #>  7 dietB        14    20      4          6.23 0007  #>  8 dietB        NA    NA     NA          0    0008  #>  9 dietB        12    20      5          5.38 0009  #> 10 dietB        NA    NA     NA          0    0010  #> # ℹ 990 more rows #>  #> $empirical_clusters #> # A tibble: 1 × 7 #>   predictor id    start   end length sum_statistic pvalue #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>  <dbl> #> 1 dietB     1         2    14      7          18.2 0.0420"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"tidying-julia-model-objects","dir":"Articles","previous_headings":"","what":"Tidying Julia model objects","title":"Tidying output","text":"Functions jlmer() to_jlmer() return object class <jlmer_mod>, JuliaConnectoR pointer Julia model object: can use tidy() glance() return information Julia model data frame: mimic behavior methods broom/broom.mixed defined (g)lm/(g)lmer models: Note Julia MixedModels.jl defaults REML = FALSE, whereas TRUE default lme4.","code":"jlmer_mod <- to_jlmer(weight ~ diet + (1 | Chick), chickweights) class(jlmer_mod) #> [1] \"jlmer_mod\"        \"JuliaStructProxy\" \"JuliaProxy\" jlmer_mod #> <Julia object of type LinearMixedModel> #> Variance components: #>             Column    Variance Std.Dev. #> Chick    (Intercept)   305.1809 17.4694 #> Residual              3854.0809 62.0812 #>  ────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|) #> ────────────────────────────────────────────────── #> (Intercept)  108.16       4.76331  22.71    <1e-99 #> dietB         12.2644     7.45319   1.65    0.0999 #> ────────────────────────────────────────────────── tidy(jlmer_mod) #> # A tibble: 4 × 7 #>   effect   group    term            estimate std.error statistic    p.value #>   <chr>    <chr>    <chr>              <dbl>     <dbl>     <dbl>      <dbl> #> 1 fixed    NA       (Intercept)        108.       4.76     22.7   3.83e-114 #> 2 fixed    NA       dietB               12.3      7.45      1.65  9.99e-  2 #> 3 ran_pars Chick    sd__(Intercept)     17.5     NA        NA    NA         #> 4 ran_pars Residual sd__Observation     62.1     NA        NA    NA     glance(jlmer_mod) #> # A tibble: 1 × 8 #>    nobs    df sigma logLik   AIC   BIC deviance df.residual #>   <int> <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <int> #> 1   533     4  62.1 -2972. 5952. 5969.    5944.         529 library(lme4) library(broom.mixed) lmer_mod <- lmer(weight ~ diet + (1 | Chick), chickweights, REML = FALSE) tidy(lmer_mod) #> # A tibble: 4 × 6 #>   effect   group    term            estimate std.error statistic #>   <chr>    <chr>    <chr>              <dbl>     <dbl>     <dbl> #> 1 fixed    NA       (Intercept)        108.       4.76     22.7  #> 2 fixed    NA       dietB               12.3      7.45      1.65 #> 3 ran_pars Chick    sd__(Intercept)     17.5     NA        NA    #> 4 ran_pars Residual sd__Observation     62.1     NA        NA glance(lmer_mod) #> # A tibble: 1 × 7 #>    nobs sigma logLik   AIC   BIC deviance df.residual #>   <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <int> #> 1   533  62.1 -2972. 5952. 5969.    5944.         529"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Tidying output","text":"Minimal example data:  jlmer specification object used vignette:","code":"library(jlmerclusterperm) system.time(jlmerclusterperm_setup(verbose = FALSE)) #>    user  system elapsed  #>    0.03    0.03   24.11 julia_progress(show = FALSE) library(dplyr, warn.conflicts = FALSE) library(ggplot2) chickweights <- as_tibble(ChickWeight) %>%    mutate(diet = forcats::fct_collapse(Diet, A = c(1, 3), B = c(2, 4))) %>%    filter(Time %in% 0:20) ggplot(chickweights, aes(Time, weight, color = diet)) +   stat_summary(fun.args = list(mult = 1.96)) +   stat_summary(geom = \"line\") +   theme_minimal(base_size = 18) jlmer_spec <- make_jlmer_spec(   formula = weight ~ diet + (1 | Chick), data = chickweights,   subject = \"Chick\", time = \"Time\" ) jlmer_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + dietB + (1 | Chick) #> Predictors: #>   diet: dietB #> Specials: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 4 #>   weight dietB Chick  Time #>    <dbl> <dbl> <chr> <dbl> #> 1     42     0 1         0 #> 2     51     0 1         2 #> 3     59     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"a-tour-of-tidy-methods","dir":"Articles","previous_headings":"","what":"A tour of tidy() methods","title":"Tidying output","text":"non-<data.frame> objects returned jlmerclusterperm functions tidy() method return data frame representation object.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"timewise_statistics-objects","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"<timewise_statistics> objects","title":"Tidying output","text":"Functions compute_timewise_statistics() permute_timewise_statistics() return object class <timewise_statistics>, array statistics. Note isn’t stylized print() method class. Instead, recommended use tidy() inspect results data frame statistic represented row. case compute_timewise_statistics(), output matrix (2D array) predictor time. dimension becomes column tidy() data: case compute_timewise_statistics(), output 3D array simulation, time, predictor. , dimension becomes column tidy() data:","code":"empirical_statistics <- compute_timewise_statistics(jlmer_spec) class(empirical_statistics) #> [1] \"timewise_statistics\" dim(empirical_statistics) # predictor x time #> [1]  1 11 tidy(empirical_statistics) #> # A tibble: 11 × 3 #>    predictor  time statistic #>    <chr>     <dbl>     <dbl> #>  1 dietB         0    -1.09  #>  2 dietB         2     2.29  #>  3 dietB         4     3.14  #>  4 dietB         6     4.01  #>  5 dietB         8     2.90  #>  6 dietB        10     2.44  #>  7 dietB        12     2.18  #>  8 dietB        14     1.26  #>  9 dietB        16     0.673 #> 10 dietB        18     0.532 #> 11 dietB        20     0.809 null_statistics <- permute_timewise_statistics(jlmer_spec, nsim = 1000) class(null_statistics) #> [1] \"timewise_statistics\" dim(null_statistics) # simulation x time x predictor #> [1] 1000   11    1 tidy(null_statistics) #> # A tibble: 11,000 × 4 #>    predictor  time statistic sim   #>    <chr>     <dbl>     <dbl> <fct> #>  1 dietB         0   -0.309  0001  #>  2 dietB         0    2.10   0002  #>  3 dietB         0   -2.83   0003  #>  4 dietB         0   -0.568  0004  #>  5 dietB         0    2.10   0005  #>  6 dietB         0   -1.09   0006  #>  7 dietB         0   -0.830  0007  #>  8 dietB         0   -0.0515 0008  #>  9 dietB         0   -0.0515 0009  #> 10 dietB         0    0.206  0010  #> # ℹ 10,990 more rows"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"empirical_clusters-objects","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"<empirical_clusters> objects","title":"Tidying output","text":"<empirical_clusters> object returned extract_empirical_clusters() print method defined, makes easier eye: tidy() data frame representation also contains information like length, number time points cluster spans. id variable uniquely identifies cluster within predictor.","code":"empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 1) class(empirical_clusters) #> [1] \"empirical_clusters\" empirical_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 #> ──────────────────────────────────────────────────────────────────────────────── tidy(empirical_clusters) #> # A tibble: 1 × 6 #>   predictor id    start   end length sum_statistic #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl> #> 1 dietB     1         2    14      7          18.2"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"null_cluster_dists-objects","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"<null_cluster_dists> objects","title":"Tidying output","text":"<null_cluster_dists> object returned extract_null_cluster_dists() also print method defined, displays summary statistics: null distribution constructed largest cluster-mass statistic observed bootstrapped permutations original data. words, simulation contributes one sample null. tidy() representation, simulation predictor represents row. sum_statistic column represents size largest cluster simulation. clusters found, column zero values start, end, length NA. output tidy(null_statistics) tidy(null_cluster_dists) form relational data. example, can use left_join() find time wise statistics permuted data extreme cluster-mass statistic: Note sum_statistic simply sum statistic values span largest cluster:","code":"null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 1) class(null_cluster_dists) #> [1] \"null_cluster_dists\" null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 1000) #>   Mean (SD): 0.014 (7.11) #>   Coverage intervals: 95% [-15.360, 15.902] #> ──────────────────────────────────────────────────────────────────────────────── tidy(null_cluster_dists) #> # A tibble: 1,000 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 dietB         8    20      7         11.1  0001  #>  2 dietB         6    10      3          3.75 0002  #>  3 dietB        18    20      2          2.07 0003  #>  4 dietB         2    10      5          6.00 0004  #>  5 dietB         4    10      4          5.91 0005  #>  6 dietB        NA    NA     NA          0    0006  #>  7 dietB         8    20      7         12.2  0007  #>  8 dietB         2     8      4          5.29 0008  #>  9 dietB        NA    NA     NA          0    0009  #> 10 dietB        NA    NA     NA          0    0010  #> # ℹ 990 more rows largest_permuted_cluster <- tidy(null_cluster_dists) %>%    slice(which.max(abs(sum_statistic))) %>%    left_join(tidy(null_statistics), by = c(\"predictor\", \"sim\")) largest_permuted_cluster #> # A tibble: 11 × 8 #>    predictor start   end length sum_statistic sim    time statistic #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> <dbl>     <dbl> #>  1 dietB         2    20     10          29.4 0319      0     -1.09 #>  2 dietB         2    20     10          29.4 0319      2      1.67 #>  3 dietB         2    20     10          29.4 0319      4      2.56 #>  4 dietB         2    20     10          29.4 0319      6      2.97 #>  5 dietB         2    20     10          29.4 0319      8      3.47 #>  6 dietB         2    20     10          29.4 0319     10      3.54 #>  7 dietB         2    20     10          29.4 0319     12      3.49 #>  8 dietB         2    20     10          29.4 0319     14      3.54 #>  9 dietB         2    20     10          29.4 0319     16      3.05 #> 10 dietB         2    20     10          29.4 0319     18      2.65 #> 11 dietB         2    20     10          29.4 0319     20      2.48 largest_permuted_cluster %>%    filter(between(time, start, end)) %>%    summarize(sum_statistic = unique(sum_statistic), sum(statistic)) #> # A tibble: 1 × 2 #>   sum_statistic `sum(statistic)` #>           <dbl>            <dbl> #> 1          29.4             29.4"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"in-other-contexts","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"In other contexts","title":"Tidying output","text":"output calculate_clusters_pvalues() just <empirical_clusters> object augmented p-values: available, p-values returned column tidy() data: Note p-value cluster 0.027 27 samples null extreme observed: case one-fell-swoop function clusterpermute(), <null_cluster_dists> object <empirical_clusters> object returned list: can tidy() elements list lapply() purrr::map():","code":"tested_clusters <- calculate_clusters_pvalues(empirical_clusters, null_cluster_dists) tested_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 (p=0.0270) #> ──────────────────────────────────────────────────────────────────────────────── tidy(tested_clusters) #> # A tibble: 1 × 7 #>   predictor id    start   end length sum_statistic pvalue #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>  <dbl> #> 1 dietB     1         2    14      7          18.2  0.027 tidy(null_cluster_dists) %>%    filter(abs(sum_statistic) >= abs(tidy(tested_clusters)$sum_statistic)) #> # A tibble: 27 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 dietB         0    20     11         -18.7 0072  #>  2 dietB         0    20     11         -19.4 0114  #>  3 dietB         6    20      8          19.0 0118  #>  4 dietB         0    20     11         -23.6 0137  #>  5 dietB         4    20      9         -21.7 0142  #>  6 dietB         2    20     10          24.7 0268  #>  7 dietB         4    20      9          22.7 0289  #>  8 dietB         2    20     10          29.4 0319  #>  9 dietB         2    20     10          26.4 0325  #> 10 dietB         4    20      9          26.5 0384  #> # ℹ 17 more rows # Timing it for fun this time system.time({   full_test <- clusterpermute(jlmer_spec, threshold = 1, nsim = 1000, progress = FALSE) }) #>    user  system elapsed  #>    0.00    0.01    6.53 full_test$null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 1000) #>   Mean (SD): 0.066 (7.61) #>   Coverage intervals: 95% [-16.959, 17.216] #> ──────────────────────────────────────────────────────────────────────────────── full_test$empirical_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 (p=0.0420) #> ──────────────────────────────────────────────────────────────────────────────── lapply(full_test, tidy) #> $null_cluster_dists #> # A tibble: 1,000 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 dietB        14    20      4         -5.26 0001  #>  2 dietB        14    20      4          9.22 0002  #>  3 dietB         6    20      8         11.8  0003  #>  4 dietB        NA    NA     NA          0    0004  #>  5 dietB        NA    NA     NA          0    0005  #>  6 dietB        16    20      3         -3.69 0006  #>  7 dietB        14    20      4          6.23 0007  #>  8 dietB        NA    NA     NA          0    0008  #>  9 dietB        12    20      5          5.38 0009  #> 10 dietB        NA    NA     NA          0    0010  #> # ℹ 990 more rows #>  #> $empirical_clusters #> # A tibble: 1 × 7 #>   predictor id    start   end length sum_statistic pvalue #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>  <dbl> #> 1 dietB     1         2    14      7          18.2 0.0420"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"tidying-julia-model-objects","dir":"Articles","previous_headings":"","what":"Tidying Julia model objects","title":"Tidying output","text":"Functions jlmer() to_jlmer() return object class <jlmer_mod>, JuliaConnectoR pointer Julia model object: can use tidy() glance() return information Julia model data frame: mimic behavior methods broom/broom.mixed defined (g)lm/(g)lmer models: Note Julia MixedModels.jl defaults REML = FALSE, whereas TRUE default lme4.","code":"jlmer_mod <- to_jlmer(weight ~ diet + (1 | Chick), chickweights) class(jlmer_mod) #> [1] \"jlmer_mod\"        \"JuliaStructProxy\" \"JuliaProxy\" jlmer_mod #> <Julia object of type LinearMixedModel{Float64}> #> Variance components: #>             Column    Variance Std.Dev. #> Chick    (Intercept)   305.1809 17.4694 #> Residual              3854.0809 62.0812 #>  ────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|) #> ────────────────────────────────────────────────── #> (Intercept)  108.16       4.76331  22.71    <1e-99 #> dietB         12.2644     7.45319   1.65    0.0999 #> ────────────────────────────────────────────────── tidy(jlmer_mod) #> # A tibble: 4 × 7 #>   effect   group    term            estimate std.error statistic    p.value #>   <chr>    <chr>    <chr>              <dbl>     <dbl>     <dbl>      <dbl> #> 1 fixed    NA       (Intercept)        108.       4.76     22.7   3.83e-114 #> 2 fixed    NA       dietB               12.3      7.45      1.65  9.99e-  2 #> 3 ran_pars Chick    sd__(Intercept)     17.5     NA        NA    NA         #> 4 ran_pars Residual sd__Observation     62.1     NA        NA    NA     glance(jlmer_mod) #> # A tibble: 1 × 7 #>    nobs sigma logLik   AIC   BIC deviance df.residual #>   <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <int> #> 1   533  62.1 -2972. 5952. 5969.    5944.         529 library(lme4) library(broom.mixed) lmer_mod <- lmer(weight ~ diet + (1 | Chick), chickweights, REML = FALSE) tidy(lmer_mod) #> # A tibble: 4 × 6 #>   effect   group    term            estimate std.error statistic #>   <chr>    <chr>    <chr>              <dbl>     <dbl>     <dbl> #> 1 fixed    NA       (Intercept)        108.       4.76     22.7  #> 2 fixed    NA       dietB               12.3      7.45      1.65 #> 3 ran_pars Chick    sd__(Intercept)     17.5     NA        NA    #> 4 ran_pars Residual sd__Observation     62.1     NA        NA glance(lmer_mod) #> # A tibble: 1 × 7 #>    nobs sigma logLik   AIC   BIC deviance df.residual #>   <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <int> #> 1   533  62.1 -2972. 5952. 5969.    5944.         529"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/authors.html","id":null,"dir":"","previous_headings":"","what":"Authors","title":"Authors and Citation","text":"June Choe. Author, maintainer, copyright holder.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/authors.html","id":"citation","dir":"","previous_headings":"","what":"Citation","title":"Authors and Citation","text":"Choe J (2024). jlmerclusterperm: Cluster-Based Permutation Analysis Densely Sampled Time Data. https://github.com/yjunechoe/jlmerclusterperm, https://yjunechoe.github.io/jlmerclusterperm/.","code":"@Manual{,   title = {jlmerclusterperm: Cluster-Based Permutation Analysis for Densely Sampled Time Data},   author = {June Choe},   year = {2024},   note = {https://github.com/yjunechoe/jlmerclusterperm, https://yjunechoe.github.io/jlmerclusterperm/}, }"},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"our-pledge","dir":"","previous_headings":"","what":"Our Pledge","title":"Contributor Covenant Code of Conduct","text":"members, contributors, leaders pledge make participation community harassment-free experience everyone, regardless age, body size, visible invisible disability, ethnicity, sex characteristics, gender identity expression, level experience, education, socio-economic status, nationality, personal appearance, race, caste, color, religion, sexual identity orientation. pledge act interact ways contribute open, welcoming, diverse, inclusive, healthy community.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"our-standards","dir":"","previous_headings":"","what":"Our Standards","title":"Contributor Covenant Code of Conduct","text":"Examples behavior contributes positive environment community include: Demonstrating empathy kindness toward people respectful differing opinions, viewpoints, experiences Giving gracefully accepting constructive feedback Accepting responsibility apologizing affected mistakes, learning experience Focusing best just us individuals, overall community Examples unacceptable behavior include: use sexualized language imagery, sexual attention advances kind Trolling, insulting derogatory comments, personal political attacks Public private harassment Publishing others’ private information, physical email address, without explicit permission conduct reasonably considered inappropriate professional setting","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"enforcement-responsibilities","dir":"","previous_headings":"","what":"Enforcement Responsibilities","title":"Contributor Covenant Code of Conduct","text":"Community leaders responsible clarifying enforcing standards acceptable behavior take appropriate fair corrective action response behavior deem inappropriate, threatening, offensive, harmful. Community leaders right responsibility remove, edit, reject comments, commits, code, wiki edits, issues, contributions aligned Code Conduct, communicate reasons moderation decisions appropriate.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"scope","dir":"","previous_headings":"","what":"Scope","title":"Contributor Covenant Code of Conduct","text":"Code Conduct applies within community spaces, also applies individual officially representing community public spaces. Examples representing community include using official e-mail address, posting via official social media account, acting appointed representative online offline event.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"enforcement","dir":"","previous_headings":"","what":"Enforcement","title":"Contributor Covenant Code of Conduct","text":"Instances abusive, harassing, otherwise unacceptable behavior may reported community leaders responsible enforcement codeofconduct@rstudio.com. complaints reviewed investigated promptly fairly. community leaders obligated respect privacy security reporter incident.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"enforcement-guidelines","dir":"","previous_headings":"","what":"Enforcement Guidelines","title":"Contributor Covenant Code of Conduct","text":"Community leaders follow Community Impact Guidelines determining consequences action deem violation Code Conduct:","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"id_1-correction","dir":"","previous_headings":"Enforcement Guidelines","what":"1. Correction","title":"Contributor Covenant Code of Conduct","text":"Community Impact: Use inappropriate language behavior deemed unprofessional unwelcome community. Consequence: private, written warning community leaders, providing clarity around nature violation explanation behavior inappropriate. public apology may requested.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"id_2-warning","dir":"","previous_headings":"Enforcement Guidelines","what":"2. Warning","title":"Contributor Covenant Code of Conduct","text":"Community Impact: violation single incident series actions. Consequence: warning consequences continued behavior. interaction people involved, including unsolicited interaction enforcing Code Conduct, specified period time. includes avoiding interactions community spaces well external channels like social media. Violating terms may lead temporary permanent ban.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"id_3-temporary-ban","dir":"","previous_headings":"Enforcement Guidelines","what":"3. Temporary Ban","title":"Contributor Covenant Code of Conduct","text":"Community Impact: serious violation community standards, including sustained inappropriate behavior. Consequence: temporary ban sort interaction public communication community specified period time. public private interaction people involved, including unsolicited interaction enforcing Code Conduct, allowed period. Violating terms may lead permanent ban.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"id_4-permanent-ban","dir":"","previous_headings":"Enforcement Guidelines","what":"4. Permanent Ban","title":"Contributor Covenant Code of Conduct","text":"Community Impact: Demonstrating pattern violation community standards, including sustained inappropriate behavior, harassment individual, aggression toward disparagement classes individuals. Consequence: permanent ban sort public interaction within community.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"attribution","dir":"","previous_headings":"","what":"Attribution","title":"Contributor Covenant Code of Conduct","text":"Code Conduct adapted Contributor Covenant, version 2.1, available https://www.contributor-covenant.org/version/2/1/code_of_conduct.html. Community Impact Guidelines inspired [Mozilla’s code conduct enforcement ladder][https://github.com/mozilla/inclusion]. answers common questions code conduct, see FAQ https://www.contributor-covenant.org/faq. Translations available https://www.contributor-covenant.org/translations.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CONTRIBUTING.html","id":null,"dir":"","previous_headings":"","what":"Contributing to jlmerclusterperm","title":"Contributing to jlmerclusterperm","text":"outlines propose change jlmerclusterperm.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CONTRIBUTING.html","id":"fixing-typos","dir":"","previous_headings":"","what":"Fixing typos","title":"Contributing to jlmerclusterperm","text":"can fix typos, spelling mistakes, grammatical errors documentation directly using GitHub web interface, long changes made source file. generally means ’ll need edit roxygen2 comments .R, .Rd file. can find .R file generates .Rd reading comment first line.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CONTRIBUTING.html","id":"bigger-changes","dir":"","previous_headings":"","what":"Bigger changes","title":"Contributing to jlmerclusterperm","text":"want make bigger change, ’s good idea first file issue make sure someone team agrees ’s needed. ’ve found bug, please file issue illustrates bug minimal reprex (also help write unit test, needed).","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CONTRIBUTING.html","id":"pull-request-process","dir":"","previous_headings":"Bigger changes","what":"Pull request process","title":"Contributing to jlmerclusterperm","text":"Fork package clone onto computer. haven’t done , recommend using usethis::create_from_github(\"yjunechoe/jlmerclusterperm\", fork = TRUE). Install development dependencies devtools::install_dev_deps(), make sure package passes R CMD check running devtools::check(). R CMD check doesn’t pass cleanly, ’s good idea ask help continuing. Create Git branch pull request (PR). recommend using usethis::pr_init(\"brief-description--change\"). Make changes, commit git, create PR running usethis::pr_push(), following prompts browser. title PR briefly describe change. body PR contain Fixes #issue-number. user-facing changes, add bullet top NEWS.md (.e. just first header). Follow style described https://style.tidyverse.org/news.html.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CONTRIBUTING.html","id":"code-style","dir":"","previous_headings":"Bigger changes","what":"Code style","title":"Contributing to jlmerclusterperm","text":"New code follow tidyverse style guide. can use styler package apply styles, please don’t restyle code nothing PR. use roxygen2, Markdown syntax, documentation. use testthat unit tests. Contributions test cases included easier accept.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CONTRIBUTING.html","id":"code-of-conduct","dir":"","previous_headings":"","what":"Code of Conduct","title":"Contributing to jlmerclusterperm","text":"Please note jlmerclusterperm project released Contributor Code Conduct. contributing project agree abide terms.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/index.html","id":"jlmerclusterperm-","dir":"","previous_headings":"","what":"jlmerclusterperm","title":"jlmerclusterperm","text":"Julia GLM.jl MixedModels.jl based implementation cluster-based permutation test time series data, powered JuliaConnectoR.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/index.html","id":"installation-and-usage","dir":"","previous_headings":"","what":"Installation and usage","title":"jlmerclusterperm","text":"Install released version jlmerclusterperm CRAN: install development version GitHub : Using jlmerclusterperm requires prior installation Julia programming language, can downloaded either official website using command line utility juliaup. Julia version >=1.8 required 1.9 preferred substantial speed improvements. using functions jlmerclusterperm, initial setup required via calling jlmerclusterperm_setup(). first call system install necessary dependencies (happens takes around 10-15 minutes). Subsequent calls jlmerclusterperm_setup() incur small overhead around 30 seconds, plus slight delays first-time function calls. pay front start-warm-costs get blazingly-fast functions package. See Get Started page package website background tutorials.","code":"install.packages(\"jlmerclusterperm\") # install.packages(\"remotes\") remotes::install_github(\"yjunechoe/jlmerclusterperm\") # Both lines must be run at the start of each new session library(jlmerclusterperm) jlmerclusterperm_setup()"},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/index.html","id":"wholesale-cpa-with-clusterpermute","dir":"","previous_headings":"Quick tour of package functionalities","what":"Wholesale CPA with clusterpermute()","title":"jlmerclusterperm","text":"time series data:  Preparing specification object make_jlmer_spec(): Cluster-based permutation test clusterpermute(): Including random effects:","code":"chickweights <- ChickWeight chickweights$Time <- as.integer(factor(chickweights$Time)) matplot(   tapply(chickweights$weight, chickweights[c(\"Time\", \"Diet\")], mean),   type = \"b\", lwd = 3, ylab = \"Weight\", xlab = \"Time\" ) chickweights_spec <- make_jlmer_spec(   formula = weight ~ 1 + Diet,   data = chickweights,   subject = \"Chick\", time = \"Time\" ) chickweights_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #>   weight Diet2 Diet3 Diet4 Chick Time #> 1     42     0     0     0     1    1 #> 2     51     0     0     0     1    2 #> 3     59     0     0     0     1    3 #>  [ reached 'max' / getOption(\"max.print\") -- omitted 575 rows ] #> ──────────────────────────────────────────────────────────────────────────────── set_rng_state(123L) clusterpermute(   chickweights_spec,   threshold = 2.5,   nsim = 100 ) #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2.5) ──────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): -0.039 (1.89) #>   Coverage intervals: 95% [-2.862, 0.000] #> Diet3 (n = 100) #>   Mean (SD): -0.129 (2.02) #>   Coverage intervals: 95% [0.000, 0.000] #> Diet4 (n = 100) #>   Mean (SD): 0.296 (3.21) #>   Coverage intervals: 95% [0.000, 5.797] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2.5) ──────────────────────── <empirical_clusters> ── #> Diet2 #>   [3, 4]: 6.121 (p=0.0495) #> Diet3 #>   [3, 12]: 35.769 (p=0.0099) #> Diet4 #>   [2, 8]: 32.442 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── chickweights_re_spec <- make_jlmer_spec(   formula = weight ~ 1 + Diet + (1 | Chick),   data = chickweights,   subject = \"Chick\", time = \"Time\" ) set_rng_state(123L) clusterpermute(   chickweights_re_spec,   threshold = 2.5,   nsim = 100 )$empirical_clusters #> ── Empirical clusters (t > 2.5) ──────────────────────── <empirical_clusters> ── #> Diet2 #>   [3, 4]: 6.387 (p=0.0594) #> Diet3 #>   [2, 12]: 39.919 (p=0.0099) #> Diet4 #>   [2, 8]: 33.853 (p=0.0099) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/index.html","id":"piecemeal-approach-to-cpa","dir":"","previous_headings":"Quick tour of package functionalities","what":"Piecemeal approach to CPA","title":"jlmerclusterperm","text":"Computing time-wise statistics observed data:  Identifying empirical clusters: Simulating null distribution: Significance testing cluster-mass statistic: Iterating range threshold values:","code":"empirical_statistics <- compute_timewise_statistics(chickweights_spec) matplot(t(empirical_statistics), type = \"b\", pch = 1, lwd = 3, ylab = \"t-statistic\") abline(h = 2.5, lty = 3) empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 2.5) empirical_clusters #> ── Empirical clusters (t > 2.5) ──────────────────────── <empirical_clusters> ── #> Diet2 #>   [3, 4]: 6.121 #> Diet3 #>   [3, 12]: 35.769 #> Diet4 #>   [2, 8]: 32.442 #> ──────────────────────────────────────────────────────────────────────────────── set_rng_state(123L) null_statistics <- permute_timewise_statistics(chickweights_spec, nsim = 100) null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 2.5) null_cluster_dists #> ── Null cluster-mass distribution (t > 2.5) ──────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): -0.039 (1.89) #>   Coverage intervals: 95% [-2.862, 0.000] #> Diet3 (n = 100) #>   Mean (SD): -0.129 (2.02) #>   Coverage intervals: 95% [0.000, 0.000] #> Diet4 (n = 100) #>   Mean (SD): 0.296 (3.21) #>   Coverage intervals: 95% [0.000, 5.797] #> ──────────────────────────────────────────────────────────────────────────────── calculate_clusters_pvalues(empirical_clusters, null_cluster_dists, add1 = TRUE) #> ── Empirical clusters (t > 2.5) ──────────────────────── <empirical_clusters> ── #> Diet2 #>   [3, 4]: 6.121 (p=0.0495) #> Diet3 #>   [3, 12]: 35.769 (p=0.0099) #> Diet4 #>   [2, 8]: 32.442 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── walk_threshold_steps(empirical_statistics, null_statistics, steps = c(2, 2.5, 3)) #>    threshold predictor id start end length sum_statistic     pvalue #> 1        2.0     Diet2  1     3   5      3      8.496376 0.07920792 #> 2        2.0     Diet3  1     2  12     11     38.216035 0.00990099 #> 3        2.0     Diet4  1     2  12     11     41.651468 0.00990099 #> 4        2.5     Diet2  1     3   4      2      6.121141 0.04950495 #> 5        2.5     Diet3  1     3  12     10     35.768957 0.00990099 #> 6        2.5     Diet4  1     2   8      7     32.442352 0.00990099 #> 31       3.0     Diet3  1     3   5      3     12.719231 0.00990099 #> 21       3.0     Diet3  2     9  12      4     14.037622 0.00990099 #> 41       3.0     Diet4  1     2   7      6     29.659402 0.00990099"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/index.html","id":"acknowledgments","dir":"","previous_headings":"","what":"Acknowledgments","title":"jlmerclusterperm","text":"paper Maris & Oostenveld (2007) originally proposed cluster-based permutation analysis. JuliaConnectoR package powering R interface Julia. Julia packages GLM.jl MixedModels.jl fast implementations (mixed effects) regression models. Existing implementations CPA R (permuco, permutes, etc.) whose designs inspired CPA interface jlmerclusterperm.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/index.html","id":"citations","dir":"","previous_headings":"","what":"Citations","title":"jlmerclusterperm","text":"use jlmerclusterperm cluster-based permutation test mixed-effects models research, please cite one () following see fit. cite jlmerclusterperm: Choe, J. (2023). jlmerclusterperm: Cluster-Based Permutation Analysis Densely Sampled Time Data. R package version 1.1.0. https://cran.r-project.org/package=jlmerclusterperm. cite cluster-based permutation test: Maris, E., & Oostenveld, R. (2007). Nonparametric statistical testing EEG- MEG-data. Journal Neuroscience Methods, 164, 177–190. doi: 10.1016/j.jneumeth.2007.03.024. cite Julia programming language: Bezanson, J., Edelman, ., Karpinski, S., & Shah, V. B. (2017). Julia: Fresh Approach Numerical Computing. SIAM Review, 59(1), 65–98. doi: 10.1137/141000671. cite GLM.jl MixedModels.jl Julia libraries, consult Zenodo pages: GLM: https://doi.org/10.5281/zenodo.3376013 MixedModels: https://zenodo.org/badge/latestdoi/9106942","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/LICENSE.html","id":null,"dir":"","previous_headings":"","what":"MIT License","title":"MIT License","text":"Copyright (c) 2023 June Choe Permission hereby granted, free charge, person obtaining copy software associated documentation files (“Software”), deal Software without restriction, including without limitation rights use, copy, modify, merge, publish, distribute, sublicense, /sell copies Software, permit persons Software furnished , subject following conditions: copyright notice permission notice shall included copies substantial portions Software. SOFTWARE PROVIDED “”, WITHOUT WARRANTY KIND, EXPRESS IMPLIED, INCLUDING LIMITED WARRANTIES MERCHANTABILITY, FITNESS PARTICULAR PURPOSE NONINFRINGEMENT. EVENT SHALL AUTHORS COPYRIGHT HOLDERS LIABLE CLAIM, DAMAGES LIABILITY, WHETHER ACTION CONTRACT, TORT OTHERWISE, ARISING , CONNECTION SOFTWARE USE DEALINGS SOFTWARE.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/calculate_clusters_pvalues.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate bootstrapped p-values of cluster-mass statistics — calculate_clusters_pvalues","title":"Calculate bootstrapped p-values of cluster-mass statistics — calculate_clusters_pvalues","text":"Calculate bootstrapped p-values cluster-mass statistics","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/calculate_clusters_pvalues.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate bootstrapped p-values of cluster-mass statistics — calculate_clusters_pvalues","text":"","code":"calculate_clusters_pvalues(   empirical_clusters,   null_cluster_dists,   add1 = FALSE )  clusters_are_comparable(empirical_clusters, null_cluster_dists, error = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/calculate_clusters_pvalues.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate bootstrapped p-values of cluster-mass statistics — calculate_clusters_pvalues","text":"empirical_clusters empirical_clusters object null_cluster_dists null_cluster_dists object add1 Whether add 1 numerator denominator calculating p-value. Use TRUE effectively count observed statistic part permuted null distribution (recommended larger nsim prior publishing results). error Whether throw error incompatible","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/calculate_clusters_pvalues.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate bootstrapped p-values of cluster-mass statistics — calculate_clusters_pvalues","text":"empirical_clusters object augmented p-values.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/calculate_clusters_pvalues.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculate bootstrapped p-values of cluster-mass statistics — calculate_clusters_pvalues","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Make empirical clusters empirical_statistics <- compute_timewise_statistics(spec) empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 2) empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 #> Diet3 #>   [2, 20]: 34.628 #> Diet4 #>   [2, 20]: 39.371 #> ────────────────────────────────────────────────────────────────────────────────  # Make null cluster-mass distribution reset_rng_state() null_statistics <- permute_timewise_statistics(spec, nsim = 100) null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 2)  # Significance test the empirical cluster(s) from each predictor against the simulated null calculate_clusters_pvalues(empirical_clusters, null_cluster_dists) #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 (p=0.0200) #> Diet3 #>   [2, 20]: 34.628 (p=0.0000) #> Diet4 #>   [2, 20]: 39.371 (p=0.0000) #> ────────────────────────────────────────────────────────────────────────────────  # Set `add1 = TRUE` to normalize by adding 1 to numerator and denominator calculate_clusters_pvalues(empirical_clusters, null_cluster_dists, add1 = TRUE) #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 (p=0.0297) #> Diet3 #>   [2, 20]: 34.628 (p=0.0099) #> Diet4 #>   [2, 20]: 39.371 (p=0.0099) #> ────────────────────────────────────────────────────────────────────────────────  # This sequence of procedures is equivalent to `clusterpermute()` reset_rng_state() clusterpermute(spec, threshold = 2, nsim = 100, progress = FALSE) #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): 0.266 (2.13) #>   Coverage intervals: 95% [0.000, 6.579] #> Diet3 (n = 100) #>   Mean (SD): -0.124 (2.75) #>   Coverage intervals: 95% [-4.777, 5.707] #> Diet4 (n = 100) #>   Mean (SD): 0.007 (2.50) #>   Coverage intervals: 95% [0.000, 0.000] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 (p=0.0297) #> Diet3 #>   [2, 20]: 34.628 (p=0.0099) #> Diet4 #>   [2, 20]: 39.371 (p=0.0099) #> ────────────────────────────────────────────────────────────────────────────────  # The empirical clusters and the null cluster-mass distribution must be comparable empirical_clusters2 <- extract_empirical_clusters(empirical_statistics, threshold = 3) # For example, below code errors because thresholds are different (2 vs. 3) try( calculate_clusters_pvalues(empirical_clusters2, null_cluster_dists) ) #> Error in clusters_are_comparable(empirical_clusters, null_cluster_dists,  :  #>   Cluster-mass statistics between empirical and null are not comparable. #> ℹ `empirical_clusters` and `null_cluster_dists` must share the same `statistic` #>   and `threshold`. #> ✖ threshold: empirical uses 3 but null uses 2.  # Check for compatibility with `clusters_are_comparable()` clusters_are_comparable(empirical_clusters, null_cluster_dists) #> [1] TRUE clusters_are_comparable(empirical_clusters2, null_cluster_dists) #> [1] FALSE  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/clusterpermute.html","id":null,"dir":"Reference","previous_headings":"","what":"Conduct a cluster-based permutation test — clusterpermute","title":"Conduct a cluster-based permutation test — clusterpermute","text":"Conduct cluster-based permutation test","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/clusterpermute.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Conduct a cluster-based permutation test — clusterpermute","text":"","code":"clusterpermute(   jlmer_spec,   family = c(\"gaussian\", \"binomial\"),   statistic = c(\"t\", \"chisq\"),   threshold,   nsim = 100L,   predictors = NULL,   binned = FALSE,   top_n = Inf,   add1 = TRUE,   ...,   progress = TRUE )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/clusterpermute.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Conduct a cluster-based permutation test — clusterpermute","text":"jlmer_spec Data prepped jlmer make_jlmer_spec() family GLM family. Currently supports \"gaussian\" \"binomial\". statistic Test statistic calculating cluster mass. Can one \"t\" (default) regression model output \"chisq\" likelihood ratio test (takes twice long calculate). threshold threshold value statistic must pass contribute cluster mass. Interpretation differs choice statistic (): statistic = \"t\", threshold t-value (beta/std.err) regression model. statistic = \"chisq\", threshold p-value chi-squared statistics likelihood ratio tests. nsim Number simulations description predictors (Optional) subset predictors test. Defaults NULL tests predictors. binned Whether data aggregated/collapsed time bins. Defaults FALSE, requires cluster span least two time points. TRUE, allows length-1 clusters exist. top_n many clusters return, order size cluster-mass statistic. Defaults Inf return detected clusters. add1 Whether add 1 numerator denominator calculating p-value. Use TRUE effectively count observed statistic part permuted null distribution (recommended larger nsim prior publishing results). ... Optional arguments passed Julia model fitting. Defaults fast = TRUE (family = \"binomial\") progress = FALSE. progress Defaults TRUE, prints progress step cluster permutation test.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/clusterpermute.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Conduct a cluster-based permutation test — clusterpermute","text":"list null_cluster_dists empirical_clusters p-values","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/clusterpermute.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Conduct a cluster-based permutation test — clusterpermute","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Should minimally provide `threshold` and `nsim`, in addition to the spec object reset_rng_state() CPA <- clusterpermute(spec, threshold = 2, nsim = 100, progress = FALSE) CPA #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): 0.266 (2.13) #>   Coverage intervals: 95% [0.000, 6.579] #> Diet3 (n = 100) #>   Mean (SD): -0.124 (2.75) #>   Coverage intervals: 95% [-4.777, 5.707] #> Diet4 (n = 100) #>   Mean (SD): 0.007 (2.50) #>   Coverage intervals: 95% [0.000, 0.000] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 (p=0.0297) #> Diet3 #>   [2, 20]: 34.628 (p=0.0099) #> Diet4 #>   [2, 20]: 39.371 (p=0.0099) #> ────────────────────────────────────────────────────────────────────────────────  # CPA is a list of `<null_cluster_dists>` and `<empirical_clusters>` objects sapply(CPA, class) #>   null_cluster_dists   empirical_clusters  #> \"null_cluster_dists\" \"empirical_clusters\"   # You can extract the individual components for further inspection CPA$null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): 0.266 (2.13) #>   Coverage intervals: 95% [0.000, 6.579] #> Diet3 (n = 100) #>   Mean (SD): -0.124 (2.75) #>   Coverage intervals: 95% [-4.777, 5.707] #> Diet4 (n = 100) #>   Mean (SD): 0.007 (2.50) #>   Coverage intervals: 95% [0.000, 0.000] #> ──────────────────────────────────────────────────────────────────────────────── CPA$empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 (p=0.0297) #> Diet3 #>   [2, 20]: 34.628 (p=0.0099) #> Diet4 #>   [2, 20]: 39.371 (p=0.0099) #> ────────────────────────────────────────────────────────────────────────────────  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/cluster_permutation_tidiers.html","id":null,"dir":"Reference","previous_headings":"","what":"Tidiers for cluster permutation test objects — cluster_permutation_tidiers","title":"Tidiers for cluster permutation test objects — cluster_permutation_tidiers","text":"Tidiers cluster permutation test objects","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/cluster_permutation_tidiers.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tidiers for cluster permutation test objects — cluster_permutation_tidiers","text":"","code":"# S3 method for timewise_statistics tidy(x, ...)  # S3 method for empirical_clusters tidy(x, ...)  # S3 method for null_cluster_dists tidy(x, ...)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/cluster_permutation_tidiers.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tidiers for cluster permutation test objects — cluster_permutation_tidiers","text":"x object class <timewise_statistics>, <empirical_clusters>, <null_cluster_dists> ... Unused","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/cluster_permutation_tidiers.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Tidiers for cluster permutation test objects — cluster_permutation_tidiers","text":"data frame","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/cluster_permutation_tidiers.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tidiers for cluster permutation test objects — cluster_permutation_tidiers","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Method for `<timewise_statistics>` empirical_statistics <- compute_timewise_statistics(spec) class(empirical_statistics) #> [1] \"timewise_statistics\" tidy(empirical_statistics) #> # A tibble: 33 × 3 #>    predictor  time statistic #>    <chr>     <dbl>     <dbl> #>  1 Diet2         0    -1.60  #>  2 Diet3         0    -1.37  #>  3 Diet4         0    -0.916 #>  4 Diet2         2     1.67  #>  5 Diet3         2     2.45  #>  6 Diet4         2     3.53  #>  7 Diet2         4     2.60  #>  8 Diet3         4     4.48  #>  9 Diet4         4     6.27  #> 10 Diet2         6     3.52  #> # ℹ 23 more rows  reset_rng_state() null_statistics <- permute_timewise_statistics(spec, nsim = 100) class(null_statistics) #> [1] \"timewise_statistics\" tidy(null_statistics) #> # A tibble: 3,300 × 4 #>    predictor  time statistic sim   #>    <chr>     <dbl>     <dbl> <fct> #>  1 Diet2         0    -0.898 001   #>  2 Diet2         0     1.40  002   #>  3 Diet2         0    -1.44  003   #>  4 Diet2         0    -0.449 004   #>  5 Diet2         0     1.38  005   #>  6 Diet2         0    -0.692 006   #>  7 Diet2         0     0.113 007   #>  8 Diet2         0     0.334 008   #>  9 Diet2         0     0.558 009   #> 10 Diet2         0    -0.449 010   #> # ℹ 3,290 more rows  # Method for `<empirical_clusters>` empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 2) class(empirical_clusters) #> [1] \"empirical_clusters\" tidy(empirical_clusters) #> # A tibble: 3 × 6 #>   predictor id    start   end length sum_statistic #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl> #> 1 Diet2     1         4     8      3          8.50 #> 2 Diet3     1         2    20     10         34.6  #> 3 Diet4     1         2    20     10         39.4   # Method for `<null_cluster_dists>` null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 2) class(null_cluster_dists) #> [1] \"null_cluster_dists\" tidy(null_cluster_dists) #> # A tibble: 300 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 Diet2        10    20      6          14.8 001   #>  2 Diet2        NA    NA     NA           0   002   #>  3 Diet2        NA    NA     NA           0   003   #>  4 Diet2        NA    NA     NA           0   004   #>  5 Diet2        NA    NA     NA           0   005   #>  6 Diet2        NA    NA     NA           0   006   #>  7 Diet2        NA    NA     NA           0   007   #>  8 Diet2        NA    NA     NA           0   008   #>  9 Diet2        NA    NA     NA           0   009   #> 10 Diet2        NA    NA     NA           0   010   #> # ℹ 290 more rows  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/compute_timewise_statistics.html","id":null,"dir":"Reference","previous_headings":"","what":"Fit Julia regression models to each time point of a time series data — compute_timewise_statistics","title":"Fit Julia regression models to each time point of a time series data — compute_timewise_statistics","text":"Fit Julia regression models time point time series data","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/compute_timewise_statistics.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fit Julia regression models to each time point of a time series data — compute_timewise_statistics","text":"","code":"compute_timewise_statistics(   jlmer_spec,   family = c(\"gaussian\", \"binomial\"),   statistic = c(\"t\", \"chisq\"),   ... )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/compute_timewise_statistics.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fit Julia regression models to each time point of a time series data — compute_timewise_statistics","text":"jlmer_spec Data prepped jlmer make_jlmer_spec() family GLM family. Currently supports \"gaussian\" \"binomial\". statistic Test statistic calculating cluster mass. Can one \"t\" (default) regression model output \"chisq\" likelihood ratio test (takes twice long calculate). ... Optional arguments passed Julia model fitting. Defaults fast = TRUE (family = \"binomial\") progress = FALSE.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/compute_timewise_statistics.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fit Julia regression models to each time point of a time series data — compute_timewise_statistics","text":"predictor--time matrix cluster statistics, class timewise_statistics.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/compute_timewise_statistics.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Fit Julia regression models to each time point of a time series data — compute_timewise_statistics","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Predictor x Time matrix of t-statistics from regression output empirical_statistics <- compute_timewise_statistics(spec) round(empirical_statistics, 2) #>          Time #> Predictor     0    2    4    6    8   10   12   14   16   18   20 #>     Diet2 -1.60 1.67 2.60 3.52 2.38 1.91 1.94 1.34 1.16 1.40 1.51 #>     Diet3 -1.37 2.45 4.48 4.54 3.70 2.97 3.06 2.98 3.05 3.60 3.80 #>     Diet4 -0.92 3.53 6.27 7.00 5.12 4.07 3.66 2.78 2.16 2.13 2.64  # Collect as dataframe with `tidy()` empirical_statistics_df <- tidy(empirical_statistics) empirical_statistics_df #> # A tibble: 33 × 3 #>    predictor  time statistic #>    <chr>     <dbl>     <dbl> #>  1 Diet2         0    -1.60  #>  2 Diet3         0    -1.37  #>  3 Diet4         0    -0.916 #>  4 Diet2         2     1.67  #>  5 Diet3         2     2.45  #>  6 Diet4         2     3.53  #>  7 Diet2         4     2.60  #>  8 Diet3         4     4.48  #>  9 Diet4         4     6.27  #> 10 Diet2         6     3.52  #> # ℹ 23 more rows  # Timewise statistics are from regression models fitted to each time point # - Note the identical statistics at `Time == 0` empirical_statistics_df %>%   filter(time == 0) #> # A tibble: 3 × 3 #>   predictor  time statistic #>   <chr>     <dbl>     <dbl> #> 1 Diet2         0    -1.60  #> 2 Diet3         0    -1.37  #> 3 Diet4         0    -0.916 to_jlmer(weight ~ 1 + Diet, filter(ChickWeight, Time == 0)) %>%   tidy() %>%   select(term, statistic) #> # A tibble: 4 × 2 #>   term        statistic #>   <chr>           <dbl> #> 1 (Intercept)   164.    #> 2 Diet2          -1.60  #> 3 Diet3          -1.37  #> 4 Diet4          -0.916  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_empirical_clusters.html","id":null,"dir":"Reference","previous_headings":"","what":"Detect largest clusters from a time sequence of predictor statistics — extract_empirical_clusters","title":"Detect largest clusters from a time sequence of predictor statistics — extract_empirical_clusters","text":"Detect largest clusters time sequence predictor statistics","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_empirical_clusters.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Detect largest clusters from a time sequence of predictor statistics — extract_empirical_clusters","text":"","code":"extract_empirical_clusters(   empirical_statistics,   threshold,   binned = FALSE,   top_n = Inf )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_empirical_clusters.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Detect largest clusters from a time sequence of predictor statistics — extract_empirical_clusters","text":"empirical_statistics predictor--time matrix empirical timewise statistics. threshold threshold value statistic must pass contribute cluster mass. Interpretation differs choice statistic (): statistic = \"t\", threshold t-value (beta/std.err) regression model. statistic = \"chisq\", threshold p-value chi-squared statistics likelihood ratio tests. binned Whether data aggregated/collapsed time bins. Defaults FALSE, requires cluster span least two time points. TRUE, allows length-1 clusters exist. top_n many clusters return, order size cluster-mass statistic. Defaults Inf return detected clusters.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_empirical_clusters.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Detect largest clusters from a time sequence of predictor statistics — extract_empirical_clusters","text":"empirical_clusters object.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_empirical_clusters.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Detect largest clusters from a time sequence of predictor statistics — extract_empirical_clusters","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Empirical clusters are derived from the timewise statistics empirical_statistics <- compute_timewise_statistics(spec) empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 2) empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 #> Diet3 #>   [2, 20]: 34.628 #> Diet4 #>   [2, 20]: 39.371 #> ────────────────────────────────────────────────────────────────────────────────  # Collect as dataframe with `tidy()` empirical_clusters_df <- tidy(empirical_clusters) empirical_clusters_df #> # A tibble: 3 × 6 #>   predictor id    start   end length sum_statistic #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl> #> 1 Diet2     1         4     8      3          8.50 #> 2 Diet3     1         2    20     10         34.6  #> 3 Diet4     1         2    20     10         39.4   # Changing the `threshold` value identifies different clusters extract_empirical_clusters(empirical_statistics, threshold = 1) #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [2, 20]: 19.427 #> Diet3 #>   [2, 20]: 34.628 #> Diet4 #>   [2, 20]: 39.371 #> ────────────────────────────────────────────────────────────────────────────────  # A predictor can have zero or multiple clusters associated with it extract_empirical_clusters(empirical_statistics, threshold = 3) #> ── Empirical clusters (t > 3) ────────────────────────── <empirical_clusters> ── #> Diet3 #>   [4, 8]: 12.719 #>   [16, 20]: 10.449 #> Diet4 #>   [2, 12]: 29.659 #> ──────────────────────────────────────────────────────────────────────────────── #> ! No clusters found for Diet2  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_null_cluster_dists.html","id":null,"dir":"Reference","previous_headings":"","what":"Construct a null distribution of cluster-mass statistics — extract_null_cluster_dists","title":"Construct a null distribution of cluster-mass statistics — extract_null_cluster_dists","text":"Construct null distribution cluster-mass statistics","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_null_cluster_dists.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Construct a null distribution of cluster-mass statistics — extract_null_cluster_dists","text":"","code":"extract_null_cluster_dists(null_statistics, threshold, binned = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_null_cluster_dists.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Construct a null distribution of cluster-mass statistics — extract_null_cluster_dists","text":"null_statistics simulation--time--predictor 3D array null (permuted) timewise statistics. threshold threshold value statistic must pass contribute cluster mass. Interpretation differs choice statistic (): statistic = \"t\", threshold t-value (beta/std.err) regression model. statistic = \"chisq\", threshold p-value chi-squared statistics likelihood ratio tests. binned Whether data aggregated/collapsed time bins. Defaults FALSE, requires cluster span least two time points. TRUE, allows length-1 clusters exist.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_null_cluster_dists.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Construct a null distribution of cluster-mass statistics — extract_null_cluster_dists","text":"null_cluster_dists object.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_null_cluster_dists.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Construct a null distribution of cluster-mass statistics — extract_null_cluster_dists","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Null cluster-mass distributions are derived from the permuted timewise statistics reset_rng_state() null_statistics <- permute_timewise_statistics(spec, nsim = 100) null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 2) null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): 0.266 (2.13) #>   Coverage intervals: 95% [0.000, 6.579] #> Diet3 (n = 100) #>   Mean (SD): -0.124 (2.75) #>   Coverage intervals: 95% [-4.777, 5.707] #> Diet4 (n = 100) #>   Mean (SD): 0.007 (2.50) #>   Coverage intervals: 95% [0.000, 0.000] #> ────────────────────────────────────────────────────────────────────────────────  # Collect as dataframe with `tidy()` # - Each simulation contributes one (largest) cluster-mass statistic to the null # - When no clusters are found, the `sum_statistic` value is zero null_cluster_dists_df <- tidy(null_cluster_dists) null_cluster_dists_df #> # A tibble: 300 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 Diet2        10    20      6          14.8 001   #>  2 Diet2        NA    NA     NA           0   002   #>  3 Diet2        NA    NA     NA           0   003   #>  4 Diet2        NA    NA     NA           0   004   #>  5 Diet2        NA    NA     NA           0   005   #>  6 Diet2        NA    NA     NA           0   006   #>  7 Diet2        NA    NA     NA           0   007   #>  8 Diet2        NA    NA     NA           0   008   #>  9 Diet2        NA    NA     NA           0   009   #> 10 Diet2        NA    NA     NA           0   010   #> # ℹ 290 more rows  # Changing the `threshold` value changes the shape of the null extract_null_cluster_dists(null_statistics, threshold = 1) #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): 0.565 (5.52) #>   Coverage intervals: 95% [-9.681, 12.061] #> Diet3 (n = 100) #>   Mean (SD): 0.089 (6.27) #>   Coverage intervals: 95% [-15.847, 12.185] #> Diet4 (n = 100) #>   Mean (SD): 0.715 (5.23) #>   Coverage intervals: 95% [-7.843, 11.728] #> ──────────────────────────────────────────────────────────────────────────────── extract_null_cluster_dists(null_statistics, threshold = 3) #> ── Null cluster-mass distribution (t > 3) ────────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): 0.000 (0.00) #>   Coverage intervals: 95% [0.000, 0.000] #> Diet3 (n = 100) #>   Mean (SD): 0.000 (0.00) #>   Coverage intervals: 95% [0.000, 0.000] #> Diet4 (n = 100) #>   Mean (SD): 0.000 (0.00) #>   Coverage intervals: 95% [0.000, 0.000] #> ────────────────────────────────────────────────────────────────────────────────  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmer.html","id":null,"dir":"Reference","previous_headings":"","what":"Fit a Julia regression model using jlmer specifications — jlmer","title":"Fit a Julia regression model using jlmer specifications — jlmer","text":"Fit Julia regression model using jlmer specifications","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fit a Julia regression model using jlmer specifications — jlmer","text":"","code":"jlmer(jlmer_spec, family = c(\"gaussian\", \"binomial\"), ..., progress = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fit a Julia regression model using jlmer specifications — jlmer","text":"jlmer_spec Data prepped jlmer make_jlmer_spec() family GLM family. Currently supports \"gaussian\" \"binomial\". ... Optional arguments passed Julia model fitting. progress TRUE, prints timing iterations.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fit a Julia regression model using jlmer specifications — jlmer","text":"jlmer_mod object.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmer.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Fit a Julia regression model using jlmer specifications — jlmer","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  # Fitting a regression model with a specification object spec <- make_jlmer_spec(weight ~ 1 + Diet, ChickWeight) jlmer(spec) #> <Julia object of type StatsModels.TableRegressionModel> #> ──────────────────────────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|)  Lower 95%  Upper 95% #> ──────────────────────────────────────────────────────────────────────── #> (Intercept)  102.645      4.67395  21.96    <1e-99   93.4847    111.806 #> Diet2         19.9712     7.86744   2.54    0.0111    4.55132    35.3911 #> Diet3         40.3045     7.86744   5.12    <1e-06   24.8847     55.7244 #> Diet4         32.6173     7.91046   4.12    <1e-04   17.113      48.1215 #> ────────────────────────────────────────────────────────────────────────  # `lm()` equivalent summary(lm(weight ~ 1 + Diet, ChickWeight))$coef #>              Estimate Std. Error   t value     Pr(>|t|) #> (Intercept) 102.64545   4.673954 21.961161 4.712762e-78 #> Diet2        19.97121   7.867437  2.538465 1.139708e-02 #> Diet3        40.30455   7.867437  5.122958 4.113938e-07 #> Diet4        32.61726   7.910461  4.123307 4.286352e-05  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm-package.html","id":null,"dir":"Reference","previous_headings":"","what":"jlmerclusterperm: Cluster-Based Permutation Analysis for Densely Sampled Time Data — jlmerclusterperm-package","title":"jlmerclusterperm: Cluster-Based Permutation Analysis for Densely Sampled Time Data — jlmerclusterperm-package","text":"implementation fast cluster-based permutation analysis (CPA) densely-sampled time data developed Maris & Oostenveld, 2007 doi:10.1016/j.jneumeth.2007.03.024 . Supports (generalized, mixed-effects) regression models calculation timewise statistics. Provides wholesale piecemeal interface CPA procedure emphasis interpretability diagnostics. Integrates 'Julia' libraries 'MixedModels.jl' 'GLM.jl' performance improvements, additional functionalities interfacing 'Julia' 'R' powered 'JuliaConnectoR' package.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm-package.html","id":"author","dir":"Reference","previous_headings":"","what":"Author","title":"jlmerclusterperm: Cluster-Based Permutation Analysis for Densely Sampled Time Data — jlmerclusterperm-package","text":"Maintainer: June Choe jchoe001@gmail.com (ORCID) [copyright holder]","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm_setup.html","id":null,"dir":"Reference","previous_headings":"","what":"Initial setup for the jlmerclusterperm package — jlmerclusterperm_setup","title":"Initial setup for the jlmerclusterperm package — jlmerclusterperm_setup","text":"Initial setup jlmerclusterperm package","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm_setup.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initial setup for the jlmerclusterperm package — jlmerclusterperm_setup","text":"","code":"jlmerclusterperm_setup(..., restart = TRUE, verbose = TRUE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm_setup.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initial setup for the jlmerclusterperm package — jlmerclusterperm_setup","text":"... Ignored. restart Whether set fresh Julia session, given one already running. FALSE jlmerclusterperm_setup() already called, nothing happens. verbose Print progress messages Julia console","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm_setup.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initial setup for the jlmerclusterperm package — jlmerclusterperm_setup","text":"TRUE","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm_setup.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initial setup for the jlmerclusterperm package — jlmerclusterperm_setup","text":"","code":"# \\donttest{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE)  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_model_tidiers.html","id":null,"dir":"Reference","previous_headings":"","what":"Tidier methods for Julia regression models — julia_model_tidiers","title":"Tidier methods for Julia regression models — julia_model_tidiers","text":"Tidier methods Julia regression models","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_model_tidiers.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tidier methods for Julia regression models — julia_model_tidiers","text":"","code":"# S3 method for jlmer_mod tidy(x, effects = c(\"var_model\", \"ran_pars\", \"fixed\"), ...)  # S3 method for jlmer_mod glance(x, ...)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_model_tidiers.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tidier methods for Julia regression models — julia_model_tidiers","text":"x object class jlmer_mod effects One \"var_model\", \"ran_pars\", \"fixed\" ... Unused","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_model_tidiers.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Tidier methods for Julia regression models — julia_model_tidiers","text":"data frame","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_model_tidiers.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tidier methods for Julia regression models — julia_model_tidiers","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  # Fixed-effects only model mod1 <- to_jlmer(weight ~ 1 + Diet, ChickWeight) tidy(mod1) #> # A tibble: 4 × 5 #>   term        estimate std.error statistic   p.value #>   <chr>          <dbl>     <dbl>     <dbl>     <dbl> #> 1 (Intercept)    103.       4.67     22.0  6.77e-107 #> 2 Diet2           20.0      7.87      2.54 1.11e-  2 #> 3 Diet3           40.3      7.87      5.12 3.01e-  7 #> 4 Diet4           32.6      7.91      4.12 3.73e-  5 glance(mod1) #> # A tibble: 1 × 8 #>    nobs    df sigma logLik   AIC   BIC deviance df.residual #>   <dbl> <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <dbl> #> 1   578     5  69.3 -3268. 6546. 6568. 2758693.         573  # Mixed model mod2 <- to_jlmer(weight ~ 1 + Diet + (1 | Chick), ChickWeight) tidy(mod2) #> # A tibble: 6 × 7 #>   effect   group    term            estimate std.error statistic   p.value #>   <chr>    <chr>    <chr>              <dbl>     <dbl>     <dbl>     <dbl> #> 1 fixed    NA       (Intercept)        102.       5.78     17.6   2.74e-69 #> 2 fixed    NA       Diet2               20.8      9.79      2.13  3.34e- 2 #> 3 fixed    NA       Diet3               41.2      9.79      4.20  2.63e- 5 #> 4 fixed    NA       Diet4               33.3      9.83      3.39  7.08e- 4 #> 5 ran_pars Chick    sd__(Intercept)     15.7     NA        NA    NA        #> 6 ran_pars Residual sd__Observation     67.3     NA        NA    NA        glance(mod2) #> # A tibble: 1 × 8 #>    nobs    df sigma logLik   AIC   BIC deviance df.residual #>   <int> <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <int> #> 1   578     6  67.3 -3265. 6542. 6568.    6530.         572  # Select which of fixed/random effects to return tidy(mod2, effects = \"fixed\") #> # A tibble: 4 × 5 #>   term        estimate std.error statistic  p.value #>   <chr>          <dbl>     <dbl>     <dbl>    <dbl> #> 1 (Intercept)    102.       5.78     17.6  2.74e-69 #> 2 Diet2           20.8      9.79      2.13 3.34e- 2 #> 3 Diet3           41.2      9.79      4.20 2.63e- 5 #> 4 Diet4           33.3      9.83      3.39 7.08e- 4 tidy(mod2, effects = \"ran_pars\") #> # A tibble: 2 × 3 #>   group    term            estimate #>   <chr>    <chr>              <dbl> #> 1 Chick    sd__(Intercept)     15.7 #> 2 Residual sd__Observation     67.3  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_progress.html","id":null,"dir":"Reference","previous_headings":"","what":"Set/get options for Julia progress bar — julia_progress","title":"Set/get options for Julia progress bar — julia_progress","text":"Set/get options Julia progress bar","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_progress.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set/get options for Julia progress bar — julia_progress","text":"","code":"julia_progress(show, width)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_progress.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set/get options for Julia progress bar — julia_progress","text":"show Whether show progress bar. may also pass list \"show\" \"width\". width Width progress bar. \"auto\", adjusts progress bar width fit console.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_progress.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set/get options for Julia progress bar — julia_progress","text":"Previous values show width","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_progress.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set/get options for Julia progress bar — julia_progress","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) # }  # Show current progress options julia_progress() #> $show #> [1] TRUE #>  #> $width #> [1] 36 #>   # Set options and save previous options old_progress_opts <- julia_progress(show = FALSE, width = 100) julia_progress() #> $show #> [1] FALSE #>  #> $width #> [1] 100 #>   # Restoring progress settings by passing a list of old options old_progress_opts #> $show #> [1] TRUE #>  #> $width #> [1] 36 #>  julia_progress(old_progress_opts) identical(julia_progress(), old_progress_opts) #> [1] TRUE  # Alternatively, reset to default settings using this syntax: julia_progress(show = TRUE, width = \"auto\")  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_rng.html","id":null,"dir":"Reference","previous_headings":"","what":"Interface to the Julia RNG — julia_rng","title":"Interface to the Julia RNG — julia_rng","text":"Interface Julia RNG","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_rng.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Interface to the Julia RNG — julia_rng","text":"","code":"set_rng_state(i)  reset_rng_state()  get_rng_state()  set_rng_seed(seed)  get_rng_seed()"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_rng.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Interface to the Julia RNG — julia_rng","text":"Counter number seed Seed","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_rng.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Interface to the Julia RNG — julia_rng","text":"current seed counter","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_rng.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Interface to the Julia RNG — julia_rng","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  # RNG initializes to seed=1 counter=0 get_rng_seed() #> [1] 1 get_rng_state() #> [1] 0  # setter/getter for RNG counter set_rng_state(123) get_rng_state() #> [1] 123  # setter/getter for RNG seed set_rng_seed(2) get_rng_seed() #> [1] 2  # restore to initial setting (seed=1, counter=0) set_rng_seed(1) set_rng_state(0)  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_setup_ok.html","id":null,"dir":"Reference","previous_headings":"","what":"Check Julia setup requirements for jlmerclusterperm — julia_setup_ok","title":"Check Julia setup requirements for jlmerclusterperm — julia_setup_ok","text":"Check Julia setup requirements jlmerclusterperm","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_setup_ok.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check Julia setup requirements for jlmerclusterperm — julia_setup_ok","text":"","code":"julia_setup_ok()"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_setup_ok.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check Julia setup requirements for jlmerclusterperm — julia_setup_ok","text":"Boolean","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_setup_ok.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Check Julia setup requirements for jlmerclusterperm — julia_setup_ok","text":"","code":"julia_setup_ok() #> [1] TRUE"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/make_jlmer_spec.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a specifications object for fitting regression models in Julia — make_jlmer_spec","title":"Create a specifications object for fitting regression models in Julia — make_jlmer_spec","text":"Create specifications object fitting regression models Julia","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/make_jlmer_spec.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a specifications object for fitting regression models in Julia — make_jlmer_spec","text":"","code":"make_jlmer_spec(   formula,   data,   subject = NULL,   trial = NULL,   time = NULL,   drop_terms = NULL,   ... )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/make_jlmer_spec.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a specifications object for fitting regression models in Julia — make_jlmer_spec","text":"formula Model formula R syntax data data frame subject Column subjects data. trial Column trials data. Must uniquely identify time series within subject (example, column items counterbalanced design subject sees exactly one item). time Column time data. drop_terms (Optional) terms drop reconstructed model formula ... Unused, extensibility.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/make_jlmer_spec.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a specifications object for fitting regression models in Julia — make_jlmer_spec","text":"object class jlmer_spec.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/make_jlmer_spec.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a specifications object for fitting regression models in Julia — make_jlmer_spec","text":"","code":"# Bare specification object (minimal spec for fitting a global model) spec <- make_jlmer_spec(weight ~ 1 + Diet, ChickWeight) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Data: #> # A tibble: 578 × 4 #>   weight Diet2 Diet3 Diet4 #>    <dbl> <dbl> <dbl> <dbl> #> 1     42     0     0     0 #> 2     51     0     0     0 #> 3     59     0     0     0 #> # ℹ 575 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Constraints on specification for CPA: # 1) The combination of `subject`, `trial`, and `time` must uniquely identify rows in the data # 2) `time` must have constant sampling rate (i.e., evenly spaced values) spec_wrong <- make_jlmer_spec(   weight ~ 1 + Diet, ChickWeight,   time = \"Time\" ) #> ! Grouping column \"Time\" does not uniquely identify rows in the data #> ! Sampling rate for the `time` column \"Time\" is not constant - may affect interpretability of results. unique(ChickWeight$Time) #>  [1]  0  2  4  6  8 10 12 14 16 18 20 21  # Corrected specification for the above spec_correct <- make_jlmer_spec(   weight ~ 1 + Diet, subset(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec_correct #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_by_predictor.html","id":null,"dir":"Reference","previous_headings":"","what":"Permute data while respecting grouping structure(s) of observations — permute_by_predictor","title":"Permute data while respecting grouping structure(s) of observations — permute_by_predictor","text":"Permute data respecting grouping structure(s) observations","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_by_predictor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Permute data while respecting grouping structure(s) of observations — permute_by_predictor","text":"","code":"permute_by_predictor(   jlmer_spec,   predictors,   predictor_type = c(\"guess\", \"between_participant\", \"within_participant\"),   n = 1L )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_by_predictor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Permute data while respecting grouping structure(s) of observations — permute_by_predictor","text":"jlmer_spec Data prepped jlmer make_jlmer_spec() predictors vector terms model. multiple, must form levels one predictor. predictor_type Whether predictor \"between_participant\" \"within_participant\". Defaults \"guess\". n Number permuted samples data generate. Defaults 1L.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_by_predictor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Permute data while respecting grouping structure(s) of observations — permute_by_predictor","text":"long dataframe permuted re-samples .id column representing replication IDs.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_by_predictor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Permute data while respecting grouping structure(s) of observations — permute_by_predictor","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  # Example data setup chickweights_df <- ChickWeight chickweights_df <- chickweights_df[chickweights_df$Time <= 20, ] chickweights_df$DietInt <- as.integer(chickweights_df$Diet) head(chickweights_df) #> Grouped Data: weight ~ Time | Chick #>   weight Time Chick Diet DietInt #> 1     42    0     1    1       1 #> 2     51    2     1    1       1 #> 3     59    4     1    1       1 #> 4     64    6     1    1       1 #> 5     76    8     1    1       1 #> 6     93   10     1    1       1  # Example 1: Spec object using the continuous `DietInt` predictor chickweights_spec1 <- make_jlmer_spec(   formula = weight ~ 1 + DietInt,   data = chickweights_df,   subject = \"Chick\", time = \"Time\" ) chickweights_spec1 #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + DietInt #> Predictors: #>   DietInt: DietInt #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 4 #>   weight DietInt Chick  Time #>    <dbl>   <dbl> <ord> <dbl> #> 1     42       1 1         0 #> 2     51       1 1         2 #> 3     59       1 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Shuffling `DietInt` values guesses `predictor_type = \"between_participant\"` reset_rng_state() spec1_perm1 <- permute_by_predictor(chickweights_spec1, predictors = \"DietInt\") #> ℹ Guessed `predictor_type` to be \"between_participant\" # This calls the same shuffling algorithm for CPA in Julia, so counter is incremented get_rng_state() #> [1] 38  # Shuffling under shared RNG state reproduces the same permutation of the data reset_rng_state() spec1_perm2 <- permute_by_predictor(chickweights_spec1, predictors = \"DietInt\") #> ℹ Guessed `predictor_type` to be \"between_participant\" identical(spec1_perm1, spec1_perm2) #> [1] TRUE  # Example 2: Spec object using the multilevel `Diet` predictor chickweights_spec2 <- make_jlmer_spec(   formula = weight ~ 1 + Diet,   data = chickweights_df,   subject = \"Chick\", time = \"Time\" ) chickweights_spec2 #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Levels of a category are automatically shuffled together reset_rng_state() spec2_perm1 <- permute_by_predictor(chickweights_spec2, predictors = \"Diet2\") #> ℹ Guessed `predictor_type` to be \"between_participant\" #> ℹ Shuffling all levels of the factor together (\"Diet2\", \"Diet3\", and \"Diet4\") reset_rng_state() spec2_perm2 <- permute_by_predictor(chickweights_spec2, predictors = c(\"Diet2\", \"Diet3\", \"Diet4\")) #> ℹ Guessed `predictor_type` to be \"between_participant\" identical(spec2_perm1, spec2_perm2) #> [1] TRUE  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_timewise_statistics.html","id":null,"dir":"Reference","previous_headings":"","what":"Simulate cluster-mass statistics via bootstrapped permutations — permute_timewise_statistics","title":"Simulate cluster-mass statistics via bootstrapped permutations — permute_timewise_statistics","text":"Simulate cluster-mass statistics via bootstrapped permutations","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_timewise_statistics.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Simulate cluster-mass statistics via bootstrapped permutations — permute_timewise_statistics","text":"","code":"permute_timewise_statistics(   jlmer_spec,   family = c(\"gaussian\", \"binomial\"),   statistic = c(\"t\", \"chisq\"),   nsim = 100L,   predictors = NULL,   ... )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_timewise_statistics.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Simulate cluster-mass statistics via bootstrapped permutations — permute_timewise_statistics","text":"jlmer_spec Data prepped jlmer make_jlmer_spec() family GLM family. Currently supports \"gaussian\" \"binomial\". statistic Test statistic calculating cluster mass. Can one \"t\" (default) regression model output \"chisq\" likelihood ratio test (takes twice long calculate). nsim Number simulations description predictors (Optional) subset predictors test. Defaults NULL tests predictors. ... Optional arguments passed Julia model fitting. Defaults fast = TRUE (family = \"binomial\") progress = FALSE.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_timewise_statistics.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Simulate cluster-mass statistics via bootstrapped permutations — permute_timewise_statistics","text":"simulation--time--predictor 3D array cluster statistics, class timewise_statistics.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_timewise_statistics.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Simulate cluster-mass statistics via bootstrapped permutations — permute_timewise_statistics","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Simulation x Time x Predictor array of t-statistics from regression output reset_rng_state() null_statistics <- permute_timewise_statistics(spec, nsim = 3) round(null_statistics, 2) #> , , Predictor = Diet2 #>  #>    Time #> Sim     0    2    4    6    8   10   12   14   16   18   20 #>   1 -0.90 0.41 0.39 1.12 1.58 2.13 2.34 2.58 2.47 2.67 2.61 #>   2  1.40 0.90 0.96 1.44 1.61 1.45 1.23 0.90 0.80 0.78 0.76 #>   3 -1.44 1.23 0.73 0.88 0.93 0.69 0.58 0.61 0.77 1.04 1.43 #>  #> , , Predictor = Diet3 #>  #>    Time #> Sim    0     2     4     6     8    10    12    14    16    18    20 #>   1 0.00 -0.21  0.11 -0.68 -0.66 -0.38 -0.01 -0.02 -0.17 -0.20 -0.12 #>   2 0.70  0.56 -0.10  0.17  0.37  0.37  0.55  0.21  0.46  0.44  0.02 #>   3 0.72  1.90  0.85  0.20  0.19 -0.09 -0.34 -0.24  0.34  0.44  0.83 #>  #> , , Predictor = Diet4 #>  #>    Time #> Sim     0     2     4     6     8    10    12   14    16    18    20 #>   1  0.45 -0.82 -0.33 -0.68 -0.36 -0.35 -0.41 0.19 -0.04 -0.16 -0.16 #>   2  2.09 -0.69 -0.10  0.34  0.56  0.63  0.35 0.05 -0.05 -0.26 -0.43 #>   3 -2.16 -1.38 -1.35 -0.85 -0.02 -0.12  0.01 0.59  0.71  0.67  0.53 #>   # Collect as dataframe with `tidy()` permuted_timewise_stats_df <- tidy(null_statistics) permuted_timewise_stats_df #> # A tibble: 99 × 4 #>    predictor  time statistic sim   #>    <chr>     <dbl>     <dbl> <fct> #>  1 Diet2         0 -8.98e- 1 1     #>  2 Diet2         0  1.40e+ 0 2     #>  3 Diet2         0 -1.44e+ 0 3     #>  4 Diet3         0  2.34e-15 1     #>  5 Diet3         0  6.98e- 1 2     #>  6 Diet3         0  7.21e- 1 3     #>  7 Diet4         0  4.49e- 1 1     #>  8 Diet4         0  2.09e+ 0 2     #>  9 Diet4         0 -2.16e+ 0 3     #> 10 Diet2         2  4.12e- 1 1     #> # ℹ 89 more rows  # Permutations ran under the same RNG state are identical reset_rng_state() null_statistics2 <- permute_timewise_statistics(spec, nsim = 3) identical(null_statistics, null_statistics2) #> [1] TRUE  get_rng_state() #> [1] 111 null_statistics3 <- permute_timewise_statistics(spec, nsim = 3) identical(null_statistics, null_statistics3) #> [1] FALSE  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/reexports.html","id":null,"dir":"Reference","previous_headings":"","what":"Objects exported from other packages — reexports","title":"Objects exported from other packages — reexports","text":"objects imported packages. Follow links see documentation. generics glance, tidy JuliaConnectoR stopJulia","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/to_jlmer.html","id":null,"dir":"Reference","previous_headings":"","what":"Fit a Julia regression model using lme4 syntax — to_jlmer","title":"Fit a Julia regression model using lme4 syntax — to_jlmer","text":"Fit Julia regression model using lme4 syntax","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/to_jlmer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fit a Julia regression model using lme4 syntax — to_jlmer","text":"","code":"to_jlmer(   formula,   data,   family = c(\"gaussian\", \"binomial\"),   jlmer_spec_opts = list(),   ...,   progress = FALSE )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/to_jlmer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fit a Julia regression model using lme4 syntax — to_jlmer","text":"formula Model formula R syntax data data frame family GLM family. Currently supports \"gaussian\" \"binomial\". jlmer_spec_opts List options passed make_jlmer_spec() ... Optional arguments passed Julia model fitting. progress TRUE, prints timing iterations.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/to_jlmer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fit a Julia regression model using lme4 syntax — to_jlmer","text":"jlmer_mod object.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/to_jlmer.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Fit a Julia regression model using lme4 syntax — to_jlmer","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  # Fitting a regression model with R formula syntax to_jlmer(weight ~ 1 + Diet, ChickWeight) #> <Julia object of type StatsModels.TableRegressionModel> #> ──────────────────────────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|)  Lower 95%  Upper 95% #> ──────────────────────────────────────────────────────────────────────── #> (Intercept)  102.645      4.67395  21.96    <1e-99   93.4847    111.806 #> Diet2         19.9712     7.86744   2.54    0.0111    4.55132    35.3911 #> Diet3         40.3045     7.86744   5.12    <1e-06   24.8847     55.7244 #> Diet4         32.6173     7.91046   4.12    <1e-04   17.113      48.1215 #> ────────────────────────────────────────────────────────────────────────  # `lm()` equivalent summary(lm(weight ~ 1 + Diet, ChickWeight))$coef #>              Estimate Std. Error   t value     Pr(>|t|) #> (Intercept) 102.64545   4.673954 21.961161 4.712762e-78 #> Diet2        19.97121   7.867437  2.538465 1.139708e-02 #> Diet3        40.30455   7.867437  5.122958 4.113938e-07 #> Diet4        32.61726   7.910461  4.123307 4.286352e-05  # Fitting a mixed model with {lme4} syntax to_jlmer(weight ~ 1 + Diet + (1 | Chick), ChickWeight) #> <Julia object of type LinearMixedModel> #> Variance components: #>             Column    Variance Std.Dev. #> Chick    (Intercept)   247.2559 15.7244 #> Residual              4527.9661 67.2902 #>  ────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|) #> ────────────────────────────────────────────────── #> (Intercept)  101.781      5.78499  17.59    <1e-68 #> Diet2         20.8353     9.79412   2.13    0.0334 #> Diet3         41.1687     9.79412   4.20    <1e-04 #> Diet4         33.284      9.82881   3.39    0.0007 #> ──────────────────────────────────────────────────  # Passing MixedModels.jl fit options to the `...` to_jlmer(weight ~ 1 + Diet + (1 | Chick), ChickWeight, REML = TRUE) #> <Julia object of type LinearMixedModel> #> Variance components: #>             Column    Variance Std.Dev. #> Chick    (Intercept)   305.7755 17.4864 #> Residual              4526.5647 67.2797 #>  ────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|) #> ────────────────────────────────────────────────── #> (Intercept)  101.634      6.04044  16.83    <1e-62 #> Diet2         20.9832    10.2365    2.05    0.0404 #> Diet3         41.3165    10.2365    4.04    <1e-04 #> Diet4         33.4041    10.27      3.25    0.0011 #> ──────────────────────────────────────────────────  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/walk_threshold_steps.html","id":null,"dir":"Reference","previous_headings":"","what":"Test the probability of cluster-mass statistics over a range of threshold values — walk_threshold_steps","title":"Test the probability of cluster-mass statistics over a range of threshold values — walk_threshold_steps","text":"Test probability cluster-mass statistics range threshold values","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/walk_threshold_steps.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test the probability of cluster-mass statistics over a range of threshold values — walk_threshold_steps","text":"","code":"walk_threshold_steps(   empirical_statistics,   null_statistics,   steps,   top_n = Inf,   binned = FALSE,   add1 = TRUE,   progress = TRUE )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/walk_threshold_steps.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test the probability of cluster-mass statistics over a range of threshold values — walk_threshold_steps","text":"empirical_statistics predictor--time matrix empirical timewise statistics. null_statistics simulation--time--predictor 3D array null (permuted) timewise statistics. steps vector threshold values test top_n many clusters return, order size cluster-mass statistic. Defaults Inf return detected clusters. binned Whether data aggregated/collapsed time bins. Defaults FALSE, requires cluster span least two time points. TRUE, allows length-1 clusters exist. add1 Whether add 1 numerator denominator calculating p-value. Use TRUE effectively count observed statistic part permuted null distribution (recommended larger nsim prior publishing results). progress Whether display progress bar","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/walk_threshold_steps.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test the probability of cluster-mass statistics over a range of threshold values — walk_threshold_steps","text":"data frame predictor clusters-mass statistics threshold.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/walk_threshold_steps.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Test the probability of cluster-mass statistics over a range of threshold values — walk_threshold_steps","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, subset(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Compute timewise statistics for the observed and permuted data empirical_statistics <- compute_timewise_statistics(spec) null_statistics <- permute_timewise_statistics(spec, nsim = 100)  # Test cluster mass/probability under different threshold values walk_threshold_steps(empirical_statistics, null_statistics, steps = 1:3,                      progress = FALSE) #> # A tibble: 9 × 8 #>   threshold predictor id    start   end length sum_statistic  pvalue #>       <int> <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>   <dbl> #> 1         1 Diet2     1         2    20     10         19.4  0.00990 #> 2         1 Diet3     1         2    20     10         34.6  0.00990 #> 3         1 Diet4     1         2    20     10         39.4  0.00990 #> 4         2 Diet2     1         4     8      3          8.50 0.0297  #> 5         2 Diet3     1         2    20     10         34.6  0.00990 #> 6         2 Diet4     1         2    20     10         39.4  0.00990 #> 7         3 Diet3     1         4     8      3         12.7  0.00990 #> 8         3 Diet3     2        16    20      3         10.4  0.00990 #> 9         3 Diet4     1         2    12      6         29.7  0.00990  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-110","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.1.0","title":"jlmerclusterperm 1.1.0","text":"CRAN release: 2023-10-07 Minor breaking change: Slightly loosened default convergence criterion permute_timewise_statistics() Fixes CRAN: Specified importing R_user_dir() backports missing previously","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-106","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.6","title":"jlmerclusterperm 1.0.6","text":"CRAN release: 2023-09-19 Fixed bug add1 argument walk_threshold_steps() passed properly R < 4.0 falls back using backports R_user_dir() Fixes CRAN: Downgrade DataFrames.jl dependency 1.3 avoid mysterious pre-compilation failure CRAN windows checks","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-105","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.5","title":"jlmerclusterperm 1.0.5","text":"CRAN release: 2023-09-10 user-visible changes Fixes CRAN: Ensure minimum version requirements Tables.jl DataAPI.jl met","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-104","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.4","title":"jlmerclusterperm 1.0.4","text":"CRAN release: 2023-09-06 user-visible changes Restructured internal Julia code module (JlmerClusterPerm.jl) Upon activating project, Manifest.toml cached tools::R_user_dir() speed pre-compilation.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-103","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.3","title":"jlmerclusterperm 1.0.3","text":"CRAN release: 2023-07-19 Fixes CRAN: Use string package version comparison","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-102","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.2","title":"jlmerclusterperm 1.0.2","text":"CRAN release: 2023-07-15","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"bug-fixes-1-0-2","dir":"Changelog","previous_headings":"","what":"Bug fixes","title":"jlmerclusterperm 1.0.2","text":"jlmerclusterperm_setup() now works rc alpha builds Julia Fixes CRAN: exclusion Manifest.toml now specified .Rbuildignore - Project.toml bundled intended.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-101","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.1","title":"jlmerclusterperm 1.0.1","text":"CRAN release: 2023-07-02 jlmerclusterperm_setup() now exits early Julia version requirement (>=1.8) met. Fixes CRAN: Check Julia version requirement proceeding examples tests","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-100","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.0","title":"jlmerclusterperm 1.0.0","text":"CRAN release: 2023-06-27","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"breaking-changes-1-0-0","dir":"Changelog","previous_headings":"","what":"Breaking changes","title":"jlmerclusterperm 1.0.0","text":"threshold_steps argument walk_threshold_steps() renamed steps.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"new-features-1-0-0","dir":"Changelog","previous_headings":"","what":"New features","title":"jlmerclusterperm 1.0.0","text":"New functions interface Julia RNG seed: get_rng_seed() set_rng_seed()","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"other-improvements-1-0-0","dir":"Changelog","previous_headings":"","what":"Other improvements","title":"jlmerclusterperm 1.0.0","text":"jlmerclusterperm_setup() now echos Pkg.instantiate() print precompilation information upon first setup call","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-020","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 0.2.0","title":"jlmerclusterperm 0.2.0","text":"Added vignettes. Significant usability improvements","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-010","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 0.1.0","title":"jlmerclusterperm 0.1.0","text":"Initial release","code":""}]
+[{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/asynchronous-cpa.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Asynchronous CPA","text":"Minimal example data CPA specification: Example CPA (1000 simulations): details actual CPA skipped ’s focus vignette, example readme just nsim.","code":"library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE) chickweights <- ChickWeight chickweights$Time <- as.integer(factor(chickweights$Time)) chickweights_spec <- make_jlmer_spec(   formula = weight ~ 1 + Diet,   data = chickweights,   subject = \"Chick\", time = \"Time\" ) set_rng_state(123L) CPA <- clusterpermute(   chickweights_spec,   threshold = 2.5,   nsim = 1000,   progress = FALSE )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/asynchronous-cpa.html","id":"basic-idea","dir":"Articles","previous_headings":"","what":"Basic idea","title":"Asynchronous CPA","text":"basic idea behind asynchronous CPA strategy start background R process whose sole job send instructions CPA Julia. ’s asynchronous running CPA way block interactive R session. Since work done Julia anyways, parallelization virtually impact performance evaluating R code. future package allows asynchronous evaluation R code. ’s big package implementing complex topic. can read project futureverse just show bare minimum . advanced users, note ’re use parallelization asynchronous properties (non-blocking evaluation R code background process). recommended start multiple R processes running CPA Julia session shared already multithreaded (can spare cores, set options(\"jlmerclusterperm.nthreads\") calling jlmerclusterperm_setup()). Later updates jlmerclusterperm may wrap workflow principled way, now vignette serves minimally working example asynchronously running CPA.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/asynchronous-cpa.html","id":"setup-for-async-cpa","dir":"Articles","previous_headings":"Walkthrough","what":"Setup for async CPA","title":"Asynchronous CPA","text":"Three things order workflow: Load future package Initialize multisession future Grab options jlmerclusterperm package environment Please treat .jlmerclusterperm internal variable read-object - ’s unexported meant manipulated.","code":"library(future) plan(multisession) pkgopts <- as.list(jlmerclusterperm:::.jlmerclusterperm)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/asynchronous-cpa.html","id":"creating-the-future-object","dir":"Articles","previous_headings":"Walkthrough","what":"Creating the future object","title":"Asynchronous CPA","text":"start creating special object class <Future> using future::future(). can simply pass clusterpermute() code first argument future(), slight modification ensure background process connects Julia session. template follows. First, make pkgopts object (defined ) available future (via globals argument). , inside future expression ensure background process shares Julia session (via list2env(...)). future object replicating 1000-simulation CPA look like following: future created, immediately starts executing code background process. important evaluate future object (f) directly. Instead, query future::resolved() - simply tells whether background evaluation completed : , use loop show background CPA non-blocking. evaluating R code interactive session simultaneously CPA running: reach end loop 2 seconds, approximately long CPA took complete (confirm next section). point future completed result available collection. Although R code just crude progress alert using Sys.sleep(), can freely evaluate R code except functions call Julia. can make plots, clean data, write analyses, etc., just don’t run another CPA one already running Julia process shared.","code":"# Not run future(   {     list2env(pkgopts, jlmerclusterperm:::.jlmerclusterperm)     ## Your CPA code below ##   },   globals = structure(TRUE, add = \"pkgopts\") ) f <- future(   {     list2env(as.list(pkgopts), jlmerclusterperm:::.jlmerclusterperm)     ## Your CPA code below ##     set_rng_state(123L)     clusterpermute(       chickweights_spec,       threshold = 2.5,       nsim = 1000     )   },   globals = structure(TRUE, add = \"pkgopts\") ) resolved(f) #> [1] FALSE i <- 0 while (!resolved(f)) {   Sys.sleep(0.5)   i <<- i + 1   cat(sprintf(\"Elapsed: %.01fs\", i / 2), \"\\n\") } #> Elapsed: 0.5s  #> Elapsed: 1.0s  #> Elapsed: 1.5s  #> Elapsed: 2.0s resolved(f) #> [1] TRUE"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/asynchronous-cpa.html","id":"collecting-the-results","dir":"Articles","previous_headings":"Walkthrough","what":"Collecting the results","title":"Asynchronous CPA","text":"can collect output background process future::value(). first, double check make sure background process indeed finished evaluating: ’s confirmed, can use value() collect results future assign variable inspection: Note value() also prints messages encountered executing CPA code. see specify progress = FALSE clusterpermute() call passed future. purposes, just note adding times messages similar saw loop (2 seconds). output asynchronous CPA (CPA_async) identical initial CPA ran beginning (CPA) ran default seed 1 RNG counter value 123L: separate vignette covers Julia RNG. practice, may want use random seed (via set_rng_seed()) background CPA.","code":"resolved(f) #> [1] TRUE CPA_async <- value(f) #> Connecting to Julia TCP server at localhost:11984 ... #> ℹ Detecting empirical clusters and calculating cluster-mass statistics. #> ✔ Detecting empirical clusters and calculating cluster-mass statistics. [71ms] #>  #> ℹ Sampling cluster-mass statistics from a bootstrapped null distribution. #> ✔ Sampling cluster-mass statistics from a bootstrapped null distribution. [1.7s] #>  #> ℹ Calculating the probability of the observed cluster-mass statistics. #> ✔ Calculating the probability of the observed cluster-mass statistics. [36ms] #> identical(CPA_async, CPA) #> [1] TRUE"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"background","dir":"Articles","previous_headings":"","what":"Background","title":"de Carvalho et al. 2021","text":"data comes eye-tracking study de Carvalho, Dautriche, Fiévet, & Christophe (2021) “Toddlers exploit referential syntactic cues flexibly adapt interpretation novel verb meanings.” article reproduces expands Experiment 2 analysis, used CPA (t-tests) conduct pairwise comparison three conditions (relevant Figure 7 paper ).  data comes mostly ready CPA straight study’s OSF repository. collapsible chunk contains code prepare data reproduce figure. Data cleaning Code Figure 7 plots data original study (close reproduce ). E2_data_agg dataframe following four columns: Subject: Unique identifier subjects Condition: -participant factor variable three levels (\"Intransitive\", \"RightDislocated\", \"Transitive\") Time: continuous measure time 0-8000ms 50ms intervals Prop: Proportion looks target (averaged across trials within condition, subject) following reproduction Figure 7 original paper:  original study used pairwise comparisons analyze relationship three conditions. smaller figures (B, C, D) plot following relationships: (B) looks target \"Transitive\" compared \"RightDislocated\" (C) looks target \"RightDislocated\" compared \"Intransitive\" (D) looks target \"Transitive\" compared \"Intransitive\" rest vignette organized follows: First, replicate individual pairwise CPAs original paper. Next, consider parsimonious analysis reformulates research hypothesis choice regression contrast. Lastly, build model (2) consider issues around model diagnostics complexity (linear vs. logistic; fixed vs. mixed) context CPA Load package start Julia instance jlmerclusterperm_setup() proceeding.","code":"library(dplyr) library(forcats) filepath <- \"https://osf.io/download/var2h/?view_only=2a06746ba360446e9857df73307d21be\" E2_data_raw <- readr::read_delim(filepath)  E2_data <- E2_data_raw %>%   filter(away != 1) %>%   mutate(Target = as.integer(bin_2P == 1)) %>%   mutate(Condition = fct_recode(     factor(Condition, levels = c(\"intrans\", \"pros\", \"trans\")),     \"Intransitive\" = \"intrans\",     \"RightDislocated\" = \"pros\",     \"Transitive\" = \"trans\"   )) %>%   select(Subject, Trial, Condition, Time, Target) E2_data_agg <- E2_data %>%   group_by(Subject, Condition, Time) %>%   summarize(Prop = mean(Target), .groups = \"drop\")  # Unaggregated trial-level data of 1s and 0s E2_data #> # A tibble: 69,123 × 5 #>    Subject Trial Condition        Time Target #>    <chr>   <dbl> <fct>           <dbl>  <int> #>  1 200.asc     0 RightDislocated     0      0 #>  2 200.asc     0 RightDislocated   400      0 #>  3 200.asc     0 RightDislocated   450      1 #>  4 200.asc     0 RightDislocated   500      1 #>  5 200.asc     0 RightDislocated   550      1 #>  6 200.asc     0 RightDislocated   600      1 #>  7 200.asc     0 RightDislocated   650      1 #>  8 200.asc     0 RightDislocated   700      1 #>  9 200.asc     0 RightDislocated   750      1 #> 10 200.asc     0 RightDislocated   800      1 #> # ℹ 69,113 more rows  # Aggregated subject-mean proportions data used in original study E2_data_agg #> # A tibble: 11,540 × 4 #>    Subject Condition        Time  Prop #>    <chr>   <fct>           <dbl> <dbl> #>  1 200.asc RightDislocated     0 0.5   #>  2 200.asc RightDislocated    50 1     #>  3 200.asc RightDislocated   100 1     #>  4 200.asc RightDislocated   150 1     #>  5 200.asc RightDislocated   200 1     #>  6 200.asc RightDislocated   250 0.5   #>  7 200.asc RightDislocated   300 0.5   #>  8 200.asc RightDislocated   350 0.4   #>  9 200.asc RightDislocated   400 0.286 #> 10 200.asc RightDislocated   450 0.429 #> # ℹ 11,530 more rows library(ggplot2) make_fig7_plot <- function(conditions) {   E2_data %>%     group_by(Condition, Time) %>%     summarize(       Prop = mean(Target), se = sqrt(var(Target) / n()),       lower = Prop - se, upper = Prop + se,       .groups = \"drop\"     ) %>%     filter(Condition %in% conditions) %>%     ggplot(aes(Time, Prop, color = Condition, fill = Condition)) +     geom_hline(aes(yintercept = .5)) +     geom_ribbon(       aes(ymin = lower, ymax = upper),       alpha = .2,       show.legend = FALSE,     ) +     geom_line(linewidth = 1) +     scale_color_manual(       aesthetics = c(\"color\", \"fill\"),       values = setNames(scales::hue_pal()(3)[c(2, 1, 3)], levels(E2_data$Condition))     ) +     scale_y_continuous(limits = c(.2, .8), oob = scales::oob_keep) +     labs(y = NULL, x = NULL) +     theme_minimal() +     theme(axis.title.y = element_text(angle = 0, vjust = .5, hjust = 0)) } fig7_comparisons <- list(   \"A\" = c(\"Intransitive\", \"RightDislocated\", \"Transitive\"),   \"B\" = c(\"RightDislocated\", \"Transitive\"),   \"C\" = c(\"Intransitive\", \"RightDislocated\"),   \"D\" = c(\"Intransitive\", \"Transitive\") ) fig7 <- lapply(fig7_comparisons, make_fig7_plot)  # Figure 7 combined plots library(patchwork) p_top <- fig7$A +   scale_x_continuous(breaks = scales::breaks_width(1000)) +   theme(legend.position = \"bottom\") p_bot <- (fig7$B + fig7$C + fig7$D) & guides(color = guide_none()) fig7_combined <- p_top / guide_area() / p_bot +   plot_layout(guides = \"collect\", heights = c(3, .1, 1)) +   plot_annotation(tag_levels = \"A\") E2_data_agg #> # A tibble: 11,540 × 4 #>    Subject Condition        Time  Prop #>    <chr>   <fct>           <dbl> <dbl> #>  1 200.asc RightDislocated     0 0.5   #>  2 200.asc RightDislocated    50 1     #>  3 200.asc RightDislocated   100 1     #>  4 200.asc RightDislocated   150 1     #>  5 200.asc RightDislocated   200 1     #>  6 200.asc RightDislocated   250 0.5   #>  7 200.asc RightDislocated   300 0.5   #>  8 200.asc RightDislocated   350 0.4   #>  9 200.asc RightDislocated   400 0.286 #> 10 200.asc RightDislocated   450 0.429 #> # ℹ 11,530 more rows fig7_combined library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"replicating-the-pairwise-cpas","dir":"Articles","previous_headings":"","what":"Replicating the pairwise CPAs","title":"de Carvalho et al. 2021","text":"three conditions experiment levels Condition factor variable: begin replication \"Transitive\" vs. \"RightDislocated\" comparison shown Figure 7B apply logic two pairwise comparisons 7C 7D.","code":"levels(E2_data_agg$Condition) #> [1] \"Intransitive\"    \"RightDislocated\" \"Transitive\""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"transitive-vs--rightdislocated","dir":"Articles","previous_headings":"Replicating the pairwise CPAs","what":"Transitive vs. RightDislocated","title":"de Carvalho et al. 2021","text":"reproduced Figure 7B compares \"Transitive\" \"RightDislocated\" conditions.  paper (caption Figure 7; emphasis mine) reports: transitive right-dislocated conditions differed second repetition novel verbs (~6400 ms onset test trials end trials). now replicate analysis. First, prepare specification object. Two things note : express original t-test regression model Condition predictor drop third, unused condition data factor representation Next, fit global model sanity check structure model output. get one estimate ConditionTransitive positive coefficient, ’d expect: Finally, call clusterpermute() threshold = 1.5 (original study) simulate 100 permutations: detect largest empirical cluster spanning 6400-8000ms reported original paper. converges around p=0.02 separate 10,000-simulation run (shown ).","code":"fig7$B spec_7B <- make_jlmer_spec(   formula = Prop ~ Condition,   data = E2_data_agg %>%     filter(Condition %in% c(\"Transitive\", \"RightDislocated\")) %>%     mutate(Condition = droplevels(Condition)), # or forcats::fct_drop()   subject = \"Subject\", time = \"Time\" ) spec_7B #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: Prop ~ 1 + ConditionTransitive #> Predictors: #>   Condition: ConditionTransitive #> Groupings: #>   Subject: Subject #>   Trial: #>   Time: Time #> Data: #> # A tibble: 7,688 × 4 #>    Prop ConditionTransitive Subject  Time #>   <dbl>               <dbl> <chr>   <dbl> #> 1   0.5                   0 200.asc     0 #> 2   1                     0 200.asc    50 #> 3   1                     0 200.asc   100 #> # ℹ 7,685 more rows #> ──────────────────────────────────────────────────────────────────────────────── jlmer(spec_7B) #> <Julia object of type StatsModels.TableRegressionModel> #> ────────────────────────────────────────────────────────────────────────────────── #>                          Coef.  Std. Error       z  Pr(>|z|)  Lower 95%  Upper 95% #> ────────────────────────────────────────────────────────────────────────────────── #> (Intercept)          0.592999   0.00364424  162.72    <1e-99   0.585856  0.600141 #> ConditionTransitive  0.0569017  0.0051591    11.03    <1e-27   0.04679   0.0670133 #> ────────────────────────────────────────────────────────────────────────────────── clusterpermute(spec_7B, threshold = 1.5, nsim = 100L, progress = FALSE) #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1.5) ──────────── <null_cluster_dists> ── #> ConditionTransitive (n = 100) #>   Mean (SD): -2.674 (27.83) #>   Coverage intervals: 95% [-61.740, 46.439] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1.5) ──────────────────────── <empirical_clusters> ── #> ConditionTransitive #>   [300, 1150]: 40.643 (p=0.1287) #>   [2700, 2750]: -4.068 (p=0.9010) #>   [6150, 6200]: 3.249 (p=0.9505) #>   [6400, 8000]: 88.742 (p=0.0198) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"rightdislocated-vs--intransitive","dir":"Articles","previous_headings":"Replicating the pairwise CPAs","what":"RightDislocated vs. Intransitive","title":"de Carvalho et al. 2021","text":"reproduced Figure 7C compares \"RightDislocated\" \"Intransitive\" conditions.  paper reports: intransitive right-dislocated conditions differed first repetition novel verbs (2100 ms 3500 ms beginning test trials). repeat CPA procedure pairwise comparison: largest empirical cluster detect spans 2150-3650ms, slightly different cluster reported original paper (2100-3500ms). relatively less “extreme” cluster converges around p=0.05 10,000-simulation run.","code":"fig7$C spec_7C <- make_jlmer_spec(   formula = Prop ~ Condition,   data = E2_data_agg %>%     filter(Condition %in% c(\"RightDislocated\", \"Intransitive\")) %>%     mutate(Condition = droplevels(Condition)),   subject = \"Subject\", time = \"Time\" )  clusterpermute(spec_7C, threshold = 1.5, nsim = 100L, progress = FALSE) #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1.5) ──────────── <null_cluster_dists> ── #> ConditionRightDislocated (n = 100) #>   Mean (SD): 0.619 (33.26) #>   Coverage intervals: 95% [-63.734, 67.326] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1.5) ──────────────────────── <empirical_clusters> ── #> ConditionRightDislocated #>   [150, 200]: 3.562 (p=0.9010) #>   [2150, 3650]: 72.842 (p=0.0297) #>   [4550, 5050]: 23.836 (p=0.3960) #>   [5150, 5350]: 8.589 (p=0.7723) #>   [5700, 5750]: 3.332 (p=0.9208) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"transitive-vs--intransitive","dir":"Articles","previous_headings":"Replicating the pairwise CPAs","what":"Transitive vs. Intransitive","title":"de Carvalho et al. 2021","text":"reproduced Figure 7D compares \"Transitive\" \"Intransitive\" conditions.  paper reports: transitive intransitive conditions differed slightly offset first sentence test trials (4500 ms beginning test trials end trials). repeat CPA procedure pairwise comparison: largest empirical cluster detect spans 4600-8000ms, slightly different cluster reported original paper (4500-8000ms). converges around p=0.001 separate 10,000-simulation run.","code":"fig7$D spec_7D <- make_jlmer_spec(   formula = Prop ~ Condition,   data = E2_data_agg %>%     filter(Condition %in% c(\"Transitive\", \"Intransitive\")) %>%     mutate(Condition = droplevels(Condition)),   subject = \"Subject\", time = \"Time\" )  clusterpermute(spec_7D, threshold = 1.5, nsim = 100L, progress = FALSE) #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1.5) ──────────── <null_cluster_dists> ── #> ConditionTransitive (n = 100) #>   Mean (SD): 2.024 (41.73) #>   Coverage intervals: 95% [-70.510, 88.501] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1.5) ──────────────────────── <empirical_clusters> ── #> ConditionTransitive #>   [300, 1000]: 25.928 (p=0.4356) #>   [1300, 1400]: 4.808 (p=0.8713) #>   [2800, 2900]: 5.356 (p=0.8218) #>   [3000, 4300]: 52.638 (p=0.2079) #>   [4600, 8000]: 172.635 (p=0.0198) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"expressed-as-regression-contrasts","dir":"Articles","previous_headings":"","what":"Expressed as regression contrasts","title":"de Carvalho et al. 2021","text":"now consider parsimonious analysis translates research hypothesis contrast coding avoid multiple testing. Specifically, exploit fact original paper specifies hypothesis Intransitive < RightDislocated < Transitive.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"helmert-deviation-coding","dir":"Articles","previous_headings":"Expressed as regression contrasts","what":"Helmert (deviation) coding","title":"de Carvalho et al. 2021","text":"Testing ordinal relationship levels category require possible pairwise comparisons; instead, can approximated via Helmert coding (.k.. deviance coding) K levels expressed K-1 contrasts, contrast successively comparing level vs. average previous (typically lower) levels. Critically, Helmert contrasts orthogonal, can test simultaneously single model. data, test ordinal relationship Intransitive < RightDislocated < Transitive via two contrasts: \"RightDislocated\" vs. \"Intransitive\" \"Transitive\" vs. average \"RightDislocated\" \"Intransitive\" corresponding numerical coding following: practice, Helmert contrasts often standardized deviations expressed unit 1. also comparison \"RightDislocated\" vs. \"Intransitive\" expressed 1 unit RvsI comparison \"Transitive\" vs. average \"RightDislocated\" \"Intransitive\" expressed 1 unit TvsRI: contrast matrix, make new column original data called ConditionHelm copying Condition column, apply contrasts new column: Lastly, build new specification making use full data. , predict Prop ConditionHelm estimate effect contrasts single model. sanity check, fit global model - expect estimate contrast indeed find RvsI TvsRI output positive coefficients. suggests ordinal relationship three conditions hold least globally. Note coefficient RvsI contrast exactly pairwise model using spec_7C earlier, also compared \"RightDislocated\" \"Intransitive\": spec_helm two conditions RvsI coded -0.5 0.5, spec_7C’s treatment coding coded 0 1. Since express difference \"Intransitive\" \"RightDislocated\" unit +1, two coefficients equal magnitude sign.","code":"condition_helm <- contr.helmert(3) colnames(condition_helm) <- c(\"RvsI\", \"TvsRI\") condition_helm #>   RvsI TvsRI #> 1   -1    -1 #> 2    1    -1 #> 3    0     2 condition_helm[, 1] <- condition_helm[, 1] / 2 condition_helm[, 2] <- condition_helm[, 2] / 3 condition_helm #>   RvsI      TvsRI #> 1 -0.5 -0.3333333 #> 2  0.5 -0.3333333 #> 3  0.0  0.6666667 E2_data_agg$ConditionHelm <- E2_data_agg$Condition contrasts(E2_data_agg$ConditionHelm) <- condition_helm  # For pretty-printing as fractions MASS::fractions(contrasts(E2_data_agg$ConditionHelm)) #>                 RvsI TvsRI #> Intransitive    -1/2 -1/3  #> RightDislocated  1/2 -1/3  #> Transitive         0  2/3 spec_helm <- make_jlmer_spec(   formula = Prop ~ ConditionHelm,   data = E2_data_agg,   subject = \"Subject\", time = \"Time\" ) spec_helm #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: Prop ~ 1 + ConditionHelmRvsI + ConditionHelmTvsRI #> Predictors: #>   ConditionHelm: ConditionHelmRvsI, ConditionHelmTvsRI #> Groupings: #>   Subject: Subject #>   Trial: #>   Time: Time #> Data: #> # A tibble: 11,540 × 5 #>    Prop ConditionHelmRvsI ConditionHelmTvsRI Subject  Time #>   <dbl>             <dbl>              <dbl> <chr>   <dbl> #> 1   0.5               0.5             -0.333 200.asc     0 #> 2   1                 0.5             -0.333 200.asc    50 #> 3   1                 0.5             -0.333 200.asc   100 #> # ℹ 11,537 more rows #> ──────────────────────────────────────────────────────────────────────────────── jlmer(spec_helm) %>%   tidy(effects = \"fixed\") #> # A tibble: 3 × 5 #>   term               estimate std.error statistic  p.value #>   <chr>                 <dbl>     <dbl>     <dbl>    <dbl> #> 1 (Intercept)          0.591    0.00215     275.  0        #> 2 ConditionHelmRvsI    0.0640   0.00526      12.2 5.67e-34 #> 3 ConditionHelmTvsRI   0.0889   0.00457      19.5 2.02e-84 jlmer(spec_7C) %>%   tidy(effects = \"fixed\") #> # A tibble: 2 × 5 #>   term                     estimate std.error statistic  p.value #>   <chr>                       <dbl>     <dbl>     <dbl>    <dbl> #> 1 (Intercept)                0.529    0.00379     140.  0        #> 2 ConditionRightDislocated   0.0640   0.00536      11.9 8.30e-33"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"interpreting-cpa-results","dir":"Articles","previous_headings":"Expressed as regression contrasts","what":"Interpreting CPA results","title":"de Carvalho et al. 2021","text":"proceed normal clusterpermute() using new spec_helm: ’s summary find CPA_helm: RvsI clusters similar detected previous “partial” CPA spec_7C. slight differences cluster-mass due part due Helmert-coded model simultaneously estimating TvsRI contrast. importantly, CPA_helm detects largest cluster \"RightDislocated\" \"Intransitive\" (2150-3650ms). converges around p=0.05 10,000-simulation run. TvsRI clusters new, largest cluster predictor spans 5800ms-8000ms. converges around p=0.01 separate 10,000-simulation run. cluster effectively captures region relationship Transitive > (RightDislocated & Intransitive) emerges robust. Essentially, TvsRI comparison line \"Transitive\" invisible line runs lines \"Intransitive\" \"RightDislocated\". conclude visualizing clusters two Helmert-coded terms, annotated empirical data. “invisible” line RI Helmert coding drawn dashed line.","code":"reset_rng_state() CPA_helm <- clusterpermute(spec_helm, threshold = 1.5, nsim = 100L, progress = FALSE) CPA_helm #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1.5) ──────────── <null_cluster_dists> ── #> ConditionHelmRvsI (n = 100) #>   Mean (SD): 4.019 (28.80) #>   Coverage intervals: 95% [-45.805, 67.827] #> ConditionHelmTvsRI (n = 100) #>   Mean (SD): 1.213 (24.59) #>   Coverage intervals: 95% [-50.218, 47.462] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1.5) ──────────────────────── <empirical_clusters> ── #> ConditionHelmRvsI #>   [150, 200]: 3.654 (p=0.8911) #>   [2150, 3650]: 71.141 (p=0.0297) #>   [4550, 5350]: 36.043 (p=0.2079) #>   [5700, 5750]: 3.328 (p=0.9010) #> ConditionHelmTvsRI #>   [300, 1050]: 35.863 (p=0.1881) #>   [3500, 3700]: 8.348 (p=0.6436) #>   [3850, 4250]: 15.985 (p=0.4455) #>   [4800, 5600]: 36.143 (p=0.1782) #>   [5800, 8000]: 123.631 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── fig7A_v2 <- fig7$A +   geom_line(     aes(Time, Prop, linetype = \"RI\"),     inherit.aes = FALSE,     data = . %>%       filter(Condition %in% c(\"RightDislocated\", \"Intransitive\")) %>%       group_by(Time) %>%       summarize(Prop = mean(Prop)),   ) +   scale_linetype_manual(values = \"41\", guide = guide_legend(\"ConditionHelm\")) +   guides(x = guide_none(\"\"))  clusters_annotation <- tidy(CPA_helm$empirical_clusters) %>%   mutate(contrast = gsub(\".*([TR])vs([RI]+)\", \"\\\\1 vs. \\\\2\", predictor)) %>%   ggplot(aes(y = fct_rev(contrast))) +   geom_segment(     aes(       x = start, xend = end, yend = contrast,       color = pvalue < 0.05     ),     linewidth = 8   ) +   scale_color_manual(values = c(\"grey70\", \"steelblue\")) +   scale_y_discrete() +   scale_x_continuous(n.breaks = 9, limits = range(E2_data_agg$Time)) +   theme_minimal() +   theme(     axis.title = element_blank(),     axis.text.y = element_text(face = \"bold\"),     panel.border = element_rect(fill = NA),     panel.grid.major.y = element_blank()   )  fig7A_v2 / clusters_annotation +   plot_layout(heights = c(4, 1))"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"model-complexity","dir":"Articles","previous_headings":"","what":"Model complexity","title":"de Carvalho et al. 2021","text":"wrap case study considering complex CPA uses logistic mixed effects models trial-level data fixations target (1s 0s).","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"logistic-mixed-model","dir":"Articles","previous_headings":"Model complexity","what":"Logistic mixed model","title":"de Carvalho et al. 2021","text":"un-aggregated trial-level data use section E2_data, comes initial data preparation code chunk: apply Helmert/deviation-coded contrast matrix: specification object E2_data, add trial = \"Trial\" predict Target instead Prop. also add -subject random intercept formula: , CPA family = \"binomial\": now visualize results CPA_helm_complex CPA_helm side side:  results largely , except largest, significant cluster identified TvsRI extends much complex CPA simple CPA. examine difference next.","code":"E2_data #> # A tibble: 69,123 × 5 #>    Subject Trial Condition        Time Target #>    <chr>   <dbl> <fct>           <dbl>  <int> #>  1 200.asc     0 RightDislocated     0      0 #>  2 200.asc     0 RightDislocated   400      0 #>  3 200.asc     0 RightDislocated   450      1 #>  4 200.asc     0 RightDislocated   500      1 #>  5 200.asc     0 RightDislocated   550      1 #>  6 200.asc     0 RightDislocated   600      1 #>  7 200.asc     0 RightDislocated   650      1 #>  8 200.asc     0 RightDislocated   700      1 #>  9 200.asc     0 RightDislocated   750      1 #> 10 200.asc     0 RightDislocated   800      1 #> # ℹ 69,113 more rows E2_data$ConditionHelm <- E2_data$Condition contrasts(E2_data$ConditionHelm) <- condition_helm MASS::fractions(contrasts(E2_data$ConditionHelm)) #>                 RvsI TvsRI #> Intransitive    -1/2 -1/3  #> RightDislocated  1/2 -1/3  #> Transitive         0  2/3 spec_helm_complex <- make_jlmer_spec(   formula = Target ~ ConditionHelm + (1 | Subject),   data = E2_data,   subject = \"Subject\", trial = \"Trial\", time = \"Time\" ) spec_helm_complex #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: Target ~ 1 + ConditionHelmRvsI + ConditionHelmTvsRI + (1 | Subject) #> Predictors: #>   ConditionHelm: ConditionHelmRvsI, ConditionHelmTvsRI #> Groupings: #>   Subject: Subject #>   Trial: Trial #>   Time: Time #> Data: #> # A tibble: 69,123 × 6 #>   Target ConditionHelmRvsI ConditionHelmTvsRI Subject Trial  Time #>    <int>             <dbl>              <dbl> <chr>   <dbl> <dbl> #> 1      0               0.5             -0.333 200.asc     0     0 #> 2      0               0.5             -0.333 200.asc     0   400 #> 3      1               0.5             -0.333 200.asc     0   450 #> # ℹ 69,120 more rows #> ──────────────────────────────────────────────────────────────────────────────── reset_rng_state() CPA_helm_complex <- clusterpermute(   spec_helm_complex,   family = \"binomial\",   threshold = 1.5, nsim = 100,   progress = FALSE ) CPA_helm_complex #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1.5) ──────────── <null_cluster_dists> ── #> ConditionHelmRvsI (n = 100) #>   Mean (SD): 1.791 (31.12) #>   Coverage intervals: 95% [-60.956, 80.602] #> ConditionHelmTvsRI (n = 100) #>   Mean (SD): 2.704 (28.90) #>   Coverage intervals: 95% [-54.636, 62.925] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1.5) ──────────────────────── <empirical_clusters> ── #> ConditionHelmRvsI #>   [2100, 3450]: 69.563 (p=0.0495) #>   [4200, 5050]: 40.460 (p=0.1584) #>   [5150, 5350]: 8.379 (p=0.7228) #> ConditionHelmTvsRI #>   [350, 800]: 19.358 (p=0.4158) #>   [900, 1050]: 7.519 (p=0.6535) #>   [1250, 1350]: 4.980 (p=0.7228) #>   [3100, 4450]: 56.770 (p=0.0792) #>   [4650, 8000]: 181.165 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── clusters_annotation2 <- tidy(CPA_helm_complex$empirical_clusters) %>%   mutate(contrast = gsub(\".*([TR])vs([RI]+)\", \"\\\\1 vs. \\\\2\", predictor)) %>%   ggplot(aes(y = fct_rev(contrast))) +   geom_segment(     aes(       x = start, xend = end, yend = contrast,       color = pvalue < 0.05     ),     linewidth = 8   ) +   scale_color_manual(values = c(\"grey70\", \"steelblue\")) +   scale_y_discrete() +   scale_x_continuous(n.breaks = 9, limits = range(E2_data$Time)) +   theme_minimal() +   theme(     legend.position = \"bottom\",     axis.title = element_blank(),     axis.text.y = element_text(face = \"bold\"),     panel.border = element_rect(fill = NA),     panel.grid.major.y = element_blank()   )  clusters_annotation / clusters_annotation2 &   guides(color = guide_none()) &   plot_annotation(tag_levels = list(c(\"Simple\", \"Complex\")))"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/deCarvalho-et-al-2021.html","id":"comparison-of-cpas","dir":"Articles","previous_headings":"Model complexity","what":"Comparison of CPAs","title":"de Carvalho et al. 2021","text":"Looking timewise statistics computed simple vs. complex CPA tells us . figure plots information calls compute_timewise_statistics():  Whereas largest cluster starts emerge 5800ms CPA_helm, emerges much earlier 4650ms CPA_helm_complex. zoom region around 5800ms, see timewise statistics T vs. RI CPA_helm suddenly dip 1.5 threshold 5750ms:  CPA better? existence dips spikes indicate problem, ’s consistent expectation simple CPA less robust variance. can inspect time-point model 5750ms two CPAs fitting : ’s standard way comparing goodness fit linear fixed-effects model logistic mixed-effects model fitted different data. complex model outperforms simple model classic metrics inspect glance(). doesn’t come surprise, differences largely driven number observations (nobs). Determining appropriate degree model complexity CPA beyond scope vignette, pursue discussion . Instead, conclude old wisdom: chain strong weakest link.","code":"# Compute the timewise statistics from the CPA specifications empirical_statistics <- bind_rows(   tidy(compute_timewise_statistics(spec_helm)),   tidy(compute_timewise_statistics(spec_helm_complex, family = \"binomial\")),   .id = \"spec\" ) %>%   mutate(     CPA = c(\"Simple\", \"Complex\")[as.integer(spec)],     Contrast = gsub(\".*([TR])vs([RI]+)\", \"\\\\1 vs. \\\\2\", predictor)   )  # Time series plot of the statistics, with a line for each Helmert contrasts empirical_statistics_fig <- ggplot(empirical_statistics, aes(time, statistic)) +   geom_line(aes(color = Contrast, linetype = CPA), linewidth = 1, alpha = .7) +   geom_hline(yintercept = c(-1.5, 1.5), linetype = 2) +   theme_classic()  empirical_statistics_fig empirical_statistics_fig +   geom_vline(xintercept = 5750) +   coord_cartesian(xlim = 5750 + c(-500, 500), ylim = c(1, 2.5)) jlmer_simple_5750 <- to_jlmer(   formula = Prop ~ ConditionHelm,   data = E2_data_agg %>% filter(Time == 5750) ) jlmer_complex_5750 <- to_jlmer(   formula = Target ~ ConditionHelm + (1 | Subject),   family = \"binomial\",   data = E2_data %>% filter(Time == 5750) ) glance(jlmer_simple_5750) #> # A tibble: 1 × 8 #>    nobs    df sigma logLik   AIC   BIC deviance df.residual #>   <dbl> <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <dbl> #> 1    72     4 0.223   7.51 -7.02  2.08     3.42          68 glance(jlmer_complex_5750) #> # A tibble: 1 × 8 #>    nobs    df sigma logLik   AIC   BIC deviance df.residual #>   <int> <int> <lgl>  <dbl> <dbl> <dbl>    <dbl>       <int> #> 1   448     4 NA     -301.  609.  626.     601.         444"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/eyetrackingR-comparison.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Comparison to eyetrackingR","text":"following collapsed code chunk runs code “Prerequisites” section original eyetrackingR vignette: pick response_time dataframe created. stage, data filtered track loss aggregated subject time bins, among things. fuller description experiment research hypothesis can found original eyetrackingR vignette. purposes exercise, note following minimal facts analysis: hypothesis Prop ~ Target. t-test used estimate effect Target Prop Target experiment condition. two-level factor (\"Animate\" vs. \"Inanimate\") Prop response variable. continuous measure proportion (0-1) looks area interest ParticipantName column identifies participant TimeBin represents time, binned 100ms windows Additionally, keep following columns necessary metadata eyetrackingR approach, though can dropped jlmerclusterperm approach. Time continuous, raw (.e., un-binned) measure time AOI identity area interest simplicity, subset response_time include columns interest cluster-based permutation analysis. Lastly, replicate plot original vignette:","code":"set.seed(42)  library(\"Matrix\") library(\"lme4\") library(\"ggplot2\") library(\"eyetrackingR\")  data(\"word_recognition\") data <- make_eyetrackingr_data(word_recognition,   participant_column = \"ParticipantName\",   trial_column = \"Trial\",   time_column = \"TimeFromTrialOnset\",   trackloss_column = \"TrackLoss\",   aoi_columns = c(\"Animate\", \"Inanimate\"),   treat_non_aoi_looks_as_missing = TRUE )  # subset to response window post word-onset response_window <- subset_by_window(data,   window_start_time = 15500,   window_end_time = 21000,   rezero = FALSE )  # analyze amount of trackloss by subjects and trials (trackloss <- trackloss_analysis(data = response_window)) #> # A tibble: 155 × 6 #>    ParticipantName Trial          Samples TracklossSamples TracklossForTrial #>    <fct>           <fct>            <dbl>            <dbl>             <dbl> #>  1 ANCAT139        FamiliarBottle     330              161            0.488  #>  2 ANCAT18         FamiliarBird       330               74            0.224  #>  3 ANCAT18         FamiliarBottle     330               43            0.130  #>  4 ANCAT18         FamiliarCow        330              159            0.482  #>  5 ANCAT18         FamiliarDog        330               95            0.288  #>  6 ANCAT18         FamiliarHorse      330              165            0.5    #>  7 ANCAT18         FamiliarSpoon      330               95            0.288  #>  8 ANCAT22         FamiliarBird       330               14            0.0424 #>  9 ANCAT22         FamiliarBottle     330                8            0.0242 #> 10 ANCAT22         FamiliarDog        330               55            0.167  #> # ℹ 145 more rows #> # ℹ 1 more variable: TracklossForParticipant <dbl>  # remove trials with > 25% of trackloss response_window_clean <- clean_by_trackloss(   data = response_window,   trial_prop_thresh = .25 )  # create Target condition column response_window_clean$Target <- as.factor(ifelse(test = grepl(\"(Spoon|Bottle)\", response_window_clean$Trial),   yes = \"Inanimate\",   no  = \"Animate\" ))  response_time <- make_time_sequence_data(response_window_clean,   time_bin_size = 100,   predictor_columns = c(\"Target\"),   aois = \"Animate\",   summarize_by = \"ParticipantName\" ) response_time <- response_time %>%   select(     Prop, Target, TimeBin, ParticipantName,     Time, AOI   ) dplyr::glimpse(response_time) #> Rows: 2,970 #> Columns: 6 #> $ Prop            <dbl> NaN, NaN, NaN, NaN, NaN, NaN, 0, 0, 1, 1, 1, 1, 1, 1, … #> $ Target          <fct> Animate, Animate, Animate, Animate, Animate, Animate, … #> $ TimeBin         <dbl> 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165,… #> $ ParticipantName <fct> ANCAT18, ANCAT18, ANCAT18, ANCAT18, ANCAT18, ANCAT18, … #> $ Time            <dbl> 15500, 15600, 15700, 15800, 15900, 16000, 16100, 16200… #> $ AOI             <chr> \"Animate\", \"Animate\", \"Animate\", \"Animate\", \"Animate\",… # visualize timecourse plot(response_time, predictor_column = \"Target\") +   theme_light() +   coord_cartesian(ylim = c(0, 1)) #> Warning: Removed 37 rows containing non-finite values (`stat_summary()`). #> Removed 37 rows containing non-finite values (`stat_summary()`)."},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/eyetrackingR-comparison.html","id":"cpa-in-eyetrackingr","dir":"Articles","previous_headings":"","what":"CPA in {eyetrackingR}","title":"Comparison to eyetrackingR","text":"Continuing response_time dataframe, jump “Bootstrapped cluster-based permutation analysis” original vignette. original vignette uses following heuristic choosing threshold: choose threshold outside scope article - simply use value approaches. eyetrackingR, CPA conducted two steps: Prepare data CPA make_time_cluster_data(): step computes timewise statistics data identifies empirical clusters, can inspected plot() summary() method:  Run permutation test cluster data analyze_time_clusters(): simulates null distribution cluster-mass statistics. output, printed, essentially output summary(df_timeclust) p-values (Probability column) null distribution (extremety empirical clusters context) can visualized plot() method:","code":"num_sub <- length(unique((response_window_clean$ParticipantName))) threshold_t <- qt(p = 1 - .05 / 2, df = num_sub - 1) threshold_t #> [1] 2.055529 df_timeclust <- make_time_cluster_data(response_time,   test = \"t.test\", paired = TRUE,   predictor_column = \"Target\",   threshold = threshold_t ) plot(df_timeclust) summary(df_timeclust) #> Test Type:    t.test  #> Predictor:    Target  #> Formula:  Pair(Prop[Target == \"Animate\"], Prop[Target == \"Inanimate\"]) ~      1  #> Summary of Clusters ====== #>   Cluster Direction SumStatistic StartTime EndTime #> 1       1  Positive    132.29900     16100   19300 #> 2       2  Positive     42.31067     19400   20800 system.time({   clust_analysis <- analyze_time_clusters( df_timeclust, within_subj = TRUE, paired = TRUE, samples = 150, quiet = TRUE   ) }) #>    user  system elapsed  #>   19.48    0.14   36.24 clust_analysis #> Test Type:    t.test  #> Predictor:    Target  #> Formula:  Pair(Prop[Target == \"Animate\"], Prop[Target == \"Inanimate\"]) ~      1  #> Null Distribution   ======  #>  Mean:        -1.4815  #>  2.5%:        -31.1268  #> 97.5%:        17.9841  #> Summary of Clusters ====== #>   Cluster Direction SumStatistic StartTime EndTime Probability #> 1       1  Positive    132.29900     16100   19300 0.000000000 #> 2       2  Positive     42.31067     19400   20800 0.006666667 plot(clust_analysis)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/eyetrackingR-comparison.html","id":"cpa-in-jlmerclusterperm","dir":"Articles","previous_headings":"","what":"CPA in {jlmerclusterperm}","title":"Comparison to eyetrackingR","text":"First, set jlmerclusterperm. comparability, use single thread: {jlmerclusterpm}, CPA can also conducted two steps: Make specification object make_jlmer_spec() first specify formula, data, grouping columns data. Instead paired t-test, specify linear model formula Prop ~ Target. output prepares data CPA subsetting applying contrast coding schemes, among things: Run permutation test specification using clusterpermute() kinds information returned: different pieces information available inspection using tidy(), returns dataframe underlying summary: contrast eyetrackingR, jlmerclusterperm provide custom plot() method. However, kinds plots can replicated lines ggplot code:  Note sign clusters (originally) flipped two approaches, trivial difference simply reflects default choice baseline t.text() vs. regression contrasts. can addressed setting contrasts prior running CPA (see related issue another article). Lastly, side--side comparison results:","code":"library(jlmerclusterperm) options(\"jlmerclusterperm.nthreads\" = 1) system.time({   # Note the overhead for initializing the Julia session   jlmerclusterperm_setup() }) #>    user  system elapsed  #>    0.04    0.01   26.13 spec <- make_jlmer_spec(   formula = Prop ~ Target,   data = response_time,   subject = \"ParticipantName\",   trial = \"Target\",   time = \"TimeBin\" ) #> ! Dropping 37 rows with missing values. spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: Prop ~ 1 + TargetInanimate #> Predictors: #>   Target: TargetInanimate #> Groupings: #>   Subject: ParticipantName #>   Trial: Target #>   Time: TimeBin #> Data: #> # A tibble: 2,933 × 5 #>    Prop TargetInanimate ParticipantName Target  TimeBin #>   <dbl>           <dbl> <fct>           <fct>     <dbl> #> 1     0               0 ANCAT18         Animate     161 #> 2     0               0 ANCAT18         Animate     162 #> 3     1               0 ANCAT18         Animate     163 #> # ℹ 2,930 more rows #> ──────────────────────────────────────────────────────────────────────────────── system.time({   cpa <- clusterpermute(spec, threshold = threshold_t, nsim = 150) }) #>    user  system elapsed  #>    0.05    0.00   16.29 cpa #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2.05552943864287) ─────────────────────── #> TargetInanimate (n = 150) #>   Mean (SD): -1.009 (11.03) #>   Coverage intervals: 95% [-25.856, 27.183] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2.05552943864287) ─────────── <empirical_clusters> ── #> TargetInanimate #>   [161, 191]: -130.192 (p=0.0066) #>   [194, 209]: -45.791 (p=0.0066) #> ──────────────────────────────────────────────────────────────────────────────── null_distribution <- tidy(cpa$null_cluster_dists) null_distribution #> # A tibble: 150 × 6 #>    predictor       start   end length sum_statistic sim   #>    <chr>           <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 TargetInanimate    NA    NA     NA          0    001   #>  2 TargetInanimate   208   209      2          5.09 002   #>  3 TargetInanimate   197   200      4         -8.94 003   #>  4 TargetInanimate   205   206      2         -4.32 004   #>  5 TargetInanimate    NA    NA     NA          0    005   #>  6 TargetInanimate   193   196      4        -10.8  006   #>  7 TargetInanimate   187   193      7         19.3  007   #>  8 TargetInanimate   177   180      4          9.34 008   #>  9 TargetInanimate   159   161      3         -7.57 009   #> 10 TargetInanimate    NA    NA     NA          0    010   #> # ℹ 140 more rows empirical_clusters <- tidy(cpa$empirical_clusters) empirical_clusters #> # A tibble: 2 × 7 #>   predictor       id    start   end length sum_statistic  pvalue #>   <chr>           <fct> <dbl> <dbl>  <dbl>         <dbl>   <dbl> #> 1 TargetInanimate 1       161   191     31        -130.  0.00662 #> 2 TargetInanimate 2       194   209     16         -45.8 0.00662 # We flip the sign of the statistic here for visual equivalence cpa_plot <- null_distribution %>%   ggplot(aes(-sum_statistic)) +   geom_histogram(     aes(y = after_stat(density)),     binwidth = 3   ) +   geom_density() +   geom_vline(     aes(xintercept = -sum_statistic, color = id),     data = empirical_clusters,     linetype = 2,     linewidth = 1   ) cpa_plot library(patchwork) plot(clust_analysis) + cpa_plot"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/eyetrackingR-comparison.html","id":"performance","dir":"Articles","previous_headings":"","what":"Performance","title":"Comparison to eyetrackingR","text":"jlmerclusterperm scales easily higher number simulations, especially fixed-effects models: eyetrackingR option parallelization, limited support platform dependent. contrast, jlmerclusterperm leverages Julia’s built-multi-threading support performant consistent: 10,000 simulations 7 threads:","code":"system.time({   clusterpermute(spec, threshold = threshold_t, nsim = 1000) }) #>    user  system elapsed  #>    0.00    0.04   12.34 options(\"jlmerclusterperm.nthreads\" = 7) system.time({   jlmerclusterperm_setup(restart = TRUE, verbose = FALSE) }) #>    user  system elapsed  #>    0.06    0.01   22.89 system.time({   clusterpermute(spec, threshold = threshold_t, nsim = 10000) }) #>    user  system elapsed  #>    0.10    0.16   52.57"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"background","dir":"Articles","previous_headings":"","what":"Background","title":"Garrison et al. 2020","text":"data comes eye-tracking study Garrison, Baudet, Breitfield, Aberman, & Bergelson (2020) “Familiarity plays small role noun comprehension 12-18 months”, way peekbankr corpus package. original study used growth curve model analyze children’s looks object familiar (“”) vs. unfamiliar (“”) different ages (median-split “younger” “older” groups). replicating specific analysis . Instead, follow Katie Von Holzen’s tutorial used dataset conduct simpler, cluster-based permutation analysis (CPA) differences two age groups proportion looks target. tutorial used eyetrackingR package data preparation CPA, vignette use dplyr jlmerclusterperm tasks, respectively. following set-code reads data prepares analysis: data ts_data_agg ’ll use replicate tutorial just 4 columns relevant cluster-permutation analysis: subject_id: Unique identifier subjects age: -participant factor variable (\"Younger\", \"Older\") time: continuous measure time 0-3975ms 25ms intervals prop: -subject proportion looks target object following replicates figure tutorial introduces cluster-permutation analysis:","code":"library(dplyr) remotes::install_github(\"langcog/peekbankr\", quiet = TRUE) library(peekbankr)  # Load data aoi_timepoints <- get_aoi_timepoints(dataset_name = \"garrison_bergelson_2020\") administrations <- get_administrations(dataset_name = \"garrison_bergelson_2020\")  # Pre-processing ## Filter for age groups of interest ps_data <- aoi_timepoints %>%   left_join(administrations, by = \"administration_id\") %>%   filter(!between(age, 14, 16)) %>%   mutate(age_binned = ifelse(age < 14, \"Younger\", \"Older\")) ## Filter for time window of interest ts_window <- ps_data %>%   filter(t_norm >= 0, t_norm < 4000) ## Identify trials to exclude (trackloss in >25% of samples) to_exclude <- ts_window %>%   group_by(subject_id, trial_id) %>%   summarize(prop = mean(aoi == \"missing\"), .groups = \"drop\") %>%   filter(prop > 0.25) ## Exclude disqualifying trials and keep information relevant for CPA ts_data <- ts_window %>%   anti_join(to_exclude, by = c(\"subject_id\", \"trial_id\")) %>%   mutate(     target = aoi == \"target\",     missing = aoi == \"missing\",     age = factor(age_binned, c(\"Younger\", \"Older\"))   ) %>%   select(subject_id, age, trial_id, time = t_norm, target, missing)  # Data of subject mean proportions of fixations to target ts_data_agg <- ts_data %>%   group_by(subject_id, age, time) %>%   summarize(prop = sum(target) / sum(!missing), .groups = \"drop\")  # Data of trial-level fixations to target (0s and 1s) ts_data_trials <- ts_data %>%   filter(!missing) %>%   mutate(target = as.integer(target)) %>%   select(subject_id, age, trial_id, time, target) ts_data_agg #> # A tibble: 2,560 × 4 #>    subject_id age    time  prop #>         <int> <fct> <dbl> <dbl> #>  1       1413 Older     0 0.7   #>  2       1413 Older    25 0.733 #>  3       1413 Older    50 0.7   #>  4       1413 Older    75 0.7   #>  5       1413 Older   100 0.677 #>  6       1413 Older   125 0.677 #>  7       1413 Older   150 0.645 #>  8       1413 Older   175 0.645 #>  9       1413 Older   200 0.581 #> 10       1413 Older   225 0.548 #> # ℹ 2,550 more rows library(ggplot2) tutorial_fig <- ggplot(ts_data_agg, aes(x = time, y = prop, color = age)) +   stat_smooth(method = \"loess\", span = 0.1, se = TRUE, aes(fill = age), alpha = 0.3) +   theme_bw() +   labs(y = \"Prop. of Target Looks\") +   geom_hline(yintercept = .5, color = \"black\") +   geom_vline(xintercept = 0, color = \"black\") +   coord_cartesian(ylim = c(0, 1)) +   theme(     legend.position = \"bottom\",     legend.title = element_blank()   ) tutorial_fig"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"outline","dir":"Articles","previous_headings":"","what":"Outline","title":"Garrison et al. 2020","text":"case study vignette showcases four features CPA jlmerclusterperm: Prepping data CPA using make_jlmer_spec() CPA one go clusterpermute() modular, step--step approach CPA Using logistic regression trial-level data 1s 0s Load package start Julia instance jlmerclusterperm_setup() proceeding.","code":"library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"a-prepping-a-specification-object","dir":"Articles","previous_headings":"","what":"A) Prepping a specification object","title":"Garrison et al. 2020","text":"Conducting cluster-based permutation analysis jlmerclusterm starts creating special specification object, compiles information necessary CPA. function make_jlmer_spec() returns specification object class <jlmer_spec> minimally requires two arguments: formula: R regression model formula data: data frame example, model prop using age predictor ts_data_agg data: printed display includes four pieces information specification object: Formula: model formula Julia syntax. , looks similar formula provided, contrasts “spelled ” (age becomes ageOlder) Predictors: list predictors form ( original : expanded ). case, one predictor age expanded ageOlder ’s treatment coded “Younger” reference level. Groupings: specified call make_jlmer_spec(). empty simple_spec provided formula data. Data: transformed dataframe whose columns terms expanded formula. Note default dummy coding applied discrete variable age - now ageOlder 0 “Younger” 1 “Older”. start bare specification object , lacks parts CPA, can use jlmerclusterperm’s simple interface Julia regression models. function jlmer() takes specification object input returns Julia regression model. model represents “global” model fitted entire data, collapsing across time. equivalent lm() model specifications: raise tangent jlmer() recommend sanity-checking design output global model prior using model specifications CPA. regression models fit Julia “hood”, want make sure output comparable expect R. global model can also tell upper bound model complexity. example, global model singular fit, time-wise models fitted subsets data likely . Functions CPA also take <jlmer_spec> object require information time (calculate time-wise statistics) subject/trial (bootstrapped permutation). ts_data_agg data trial-level information summarized subject, just leave . trial argument can omitted <jlmer_spec> object time sample/bin uniquely identified subject. Otherwise, trial column data uniquely identifies time values within subject (example, column items counterbalanced designed participant sees every item one variants).","code":"simple_spec <- make_jlmer_spec(formula = prop ~ 1 + age, data = ts_data_agg) simple_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: prop ~ 1 + ageOlder #> Predictors: #>   age: ageOlder #> Data: #> # A tibble: 2,560 × 2 #>    prop ageOlder #>   <dbl>    <dbl> #> 1 0.7          1 #> 2 0.733        1 #> 3 0.7          1 #> # ℹ 2,557 more rows #> ──────────────────────────────────────────────────────────────────────────────── global_jmod <- jlmer(simple_spec) tidy(global_jmod) #> # A tibble: 2 × 5 #>   term        estimate std.error statistic  p.value #>   <chr>          <dbl>     <dbl>     <dbl>    <dbl> #> 1 (Intercept)   0.523    0.00387     135.  0        #> 2 ageOlder      0.0984   0.00548      18.0 3.45e-72 library(broom) # for the `tidy()` method for `lm()` global_rmod <- lm(formula = prop ~ 1 + age, data = ts_data_agg) tidy(global_rmod) #> # A tibble: 2 × 5 #>   term        estimate std.error statistic  p.value #>   <chr>          <dbl>     <dbl>     <dbl>    <dbl> #> 1 (Intercept)   0.523    0.00387     135.  0        #> 2 ageOlder      0.0984   0.00548      18.0 4.45e-68 ts_data_agg_spec <- make_jlmer_spec(   formula = prop ~ 1 + age, data = ts_data_agg,   subject = \"subject_id\", time = \"time\",   trial = NULL # This is the default value ) ts_data_agg_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: prop ~ 1 + ageOlder #> Predictors: #>   age: ageOlder #> Groupings: #>   Subject: subject_id #>   Trial: #>   Time: time #> Data: #> # A tibble: 2,560 × 4 #>    prop ageOlder subject_id  time #>   <dbl>    <dbl>      <int> <dbl> #> 1 0.7          1       1413     0 #> 2 0.733        1       1413    25 #> 3 0.7          1       1413    50 #> # ℹ 2,557 more rows #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"b-cpa-with-clusterpermute","dir":"Articles","previous_headings":"","what":"B) CPA with clusterpermute()","title":"Garrison et al. 2020","text":"clusterpermute() function runs CPA one go, using information <jlmer_spec> object. addition specification object, must also supply (t-value) threshold number simulations run: result list two elements: <null_cluster_dists> object holds information distribution bootstrapped cluster-mass statistics random permutations data: <empirical_clusters> object holds per-predictor information clusters detected, p-values significance testing null: output explained next section, presented purposes simply note results similar reported Katie Von Holzen’s tutorial, copied . minor numerical differences due stochastic nature permutation test ’re comparing t-tests R regression models Julia.","code":"CPA_agg <- clusterpermute(   ts_data_agg_spec,   threshold = 2,   nsim = 100,   progress = FALSE # Suppress printing of progress for vignette ) CPA_agg #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> ageOlder (n = 100) #>   Mean (SD): -0.844 (29.71) #>   Coverage intervals: 95% [-46.664, 71.698] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 74.724 (p=0.0396) #>   [2675, 2775]: 11.017 (p=0.4356) #>   [3500, 3900]: 46.295 (p=0.0990) #> ──────────────────────────────────────────────────────────────────────────────── #> ## Summary of Clusters ====== #> ##   Cluster Direction SumStatistic StartTime EndTime Probability #> ## 1       1  Positive    74.723687       850    1550       0.051 #> ## 2       2  Positive     6.905622      2725    2800       0.405 #> ## 3       3  Positive    46.294644      3500    3925       0.111"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"c-a-step-by-step-approach","dir":"Articles","previous_headings":"","what":"C) A step-by-step approach","title":"Garrison et al. 2020","text":"clusterpermute() function showcased consists five parts, also standalone functions package: compute_timewise_statistics() extract_empirical_clusters() permute_timewise_statistics() extract_null_cluster_dists() calculate_clusters_pvalues() functions correspond algorithmic steps statistical test. clusterpermute() obviates need think individual components, knowing gives greater flexibility procedure. section walks CPA step--step using functions.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"empirical-clusters","dir":"Articles","previous_headings":"C) A step-by-step approach","what":"1) Empirical clusters","title":"Garrison et al. 2020","text":"compute_timewise_statistics() function takes <jlmer_spec> object returns timewise statistics form matrix row predictor column time point: example see 25ms, t-value age -0.127. consistent figure tutorial replicated show two lines mostly overlapping. Just demonstration purposes, use to_jlmer() quickly fit model using formula just data 25ms. see t-value age model exactly saw : construct empirical clusters timewise statistics. Clusters defined continuous span statistics sign whose absolute value passes certain threshold. extract_empirical_clusters() function takes timewise statistics threshold just : t-value threshold 2, detect three clusters data main effect age (ageOlder). numbers brackets show span cluster number right show sum t-values cluster (.k.. cluster-mass statistic). can collect <empirical_clusters> object data frame tidy(): add figure :  point know whether size clusters observe (un)likely emerge chance. calculate probability, need simulate null distribution.","code":"empirical_statistics <- compute_timewise_statistics(ts_data_agg_spec) dim(empirical_statistics) #> [1]   1 160 empirical_statistics[1, 1:5, drop = FALSE] # First 5 time points #>           Time #> Predictor          0         25         50         75        100 #>   ageOlder 0.1166025 -0.1273179 -0.2387778 -0.4022095 -0.2804397 to_jlmer(prop ~ 1 + age, data = filter(ts_data_agg, time == 25)) %>%   tidy() %>%   filter(term == \"ageOlder\") %>%   pull(statistic) #> [1] -0.1273179 empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 2) empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 74.724 #>   [2675, 2775]: 11.017 #>   [3500, 3900]: 46.295 #> ──────────────────────────────────────────────────────────────────────────────── empirical_clusters_df <- tidy(empirical_clusters) empirical_clusters_df #> # A tibble: 3 × 6 #>   predictor id    start   end length sum_statistic #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl> #> 1 ageOlder  1       850  1525     28          74.7 #> 2 ageOlder  2      2675  2775      5          11.0 #> 3 ageOlder  3      3500  3900     17          46.3 tutorial_fig +   geom_segment(     aes(x = start, xend = end, y = -Inf, yend = -Inf),     linewidth = 10, color = \"steelblue\",     inherit.aes = FALSE,     data = empirical_clusters_df   ) +   geom_text(     aes(       y = -Inf, x = start + (end - start) / 2,       label = paste(\"Cluster\", id)     ),     vjust = -2,     inherit.aes = FALSE,     data = empirical_clusters_df   )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"null-distribution","dir":"Articles","previous_headings":"C) A step-by-step approach","what":"2) Null distribution","title":"Garrison et al. 2020","text":"permute_timewise_statistics() function takes specification object number bootstrapped permutation data simulate (via nsim argument). returned 3-dimensional array like output compute_timewise_statistics() except additional dimension representing simulation: permutation algorithm preserves grouping structures data specified specification object. example, testing -subject predictor like age, age groups participants random swapped preserving temporal structure data. null distribution collection largest cluster-mass statistic simulated data. clusters detected simulation, contributes cluster-mass zero null. use extract_null_cluster_dists() construct null distribution, using threshold value 2 allow comparison empirical clusters: printed, returned <null_cluster_dists> object shows summary statistics. can use tidy() collect samples null data frame: null distribution can plotted like :  Now left test probability observing cluster-mass statistics extreme detected clusters empirical_clusters, using null_cluster_dists reference.","code":"null_statistics <- permute_timewise_statistics(ts_data_agg_spec, nsim = 100) # simulation by time by predictor dim(null_statistics) #> [1] 100 160   1 # First 5 time points of the first three simulations null_statistics[1:3, 1:5, 1, drop = FALSE] #> , , Predictor = ageOlder #>  #>      Time #> Sim            0        25        50        75        100 #>   001 -0.1037696 0.1387838 0.4583545 0.4337538 0.69136613 #>   002  1.4358330 1.2745381 1.2985888 0.9111359 1.16401311 #>   003  0.2545470 0.3885999 0.3374939 0.3200497 0.05603617 null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 2) null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> ageOlder (n = 100) #>   Mean (SD): -6.150 (40.29) #>   Coverage intervals: 95% [-114.564, 74.425] #> ──────────────────────────────────────────────────────────────────────────────── null_cluster_dists_df <- tidy(null_cluster_dists) null_cluster_dists_df #> # A tibble: 100 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 ageOlder     NA    NA     NA          0    001   #>  2 ageOlder     NA    NA     NA          0    002   #>  3 ageOlder   2175  2225      3         -6.37 003   #>  4 ageOlder     NA    NA     NA          0    004   #>  5 ageOlder   3425  3475      3         -6.67 005   #>  6 ageOlder     NA    NA     NA          0    006   #>  7 ageOlder     NA    NA     NA          0    007   #>  8 ageOlder     NA    NA     NA          0    008   #>  9 ageOlder     NA    NA     NA          0    009   #> 10 ageOlder   1525  2475     39       -114.   010   #> # ℹ 90 more rows plot(   density(null_cluster_dists_df$sum_statistic),   main = \"Null distribution of cluster-mass statistics\" )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"significance-test","dir":"Articles","previous_headings":"C) A step-by-step approach","what":"3) Significance test","title":"Garrison et al. 2020","text":"calculate_clusters_pvalues() function computes p-value empirical cluster returns augmented <empirical_clusters> object: , p-values slightly different test stochastic. separate vignette covers issues replicability, seed, rng. collected tidy(), p-values added column: Note reporting pvalues, ’s customary add 1 numerator denominator. observed data, however unlikely, one particular arrangement (permutation) data. controlled via add1 argument - false default calculate_clusters_pvalues() true clusterpermute(). now come full circle: pvalue-augmented <empirical_clusters> object, along <null_cluster_dists> object, also clusterpermute() function returned .","code":"empirical_clusters_tested <- calculate_clusters_pvalues(empirical_clusters, null_cluster_dists) empirical_clusters_tested #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 74.724 (p=0.1100) #>   [2675, 2775]: 11.017 (p=0.4000) #>   [3500, 3900]: 46.295 (p=0.1700) #> ──────────────────────────────────────────────────────────────────────────────── tidy(empirical_clusters_tested) #> # A tibble: 3 × 7 #>   predictor id    start   end length sum_statistic pvalue #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>  <dbl> #> 1 ageOlder  1       850  1525     28          74.7   0.11 #> 2 ageOlder  2      2675  2775      5          11.0   0.4  #> 3 ageOlder  3      3500  3900     17          46.3   0.17 calculate_clusters_pvalues(empirical_clusters, null_cluster_dists, add1 = TRUE) #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 74.724 (p=0.1188) #>   [2675, 2775]: 11.017 (p=0.4059) #>   [3500, 3900]: 46.295 (p=0.1782) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"aggregated-vs--trial-level-data","dir":"Articles","previous_headings":"D) As logistic regression","what":"Aggregated vs. trial-level data","title":"Garrison et al. 2020","text":"data using far aggregated data collapsed across trials within subject. words, exactly one measure proportion looks target per time point within subject: possible also fit logistic regression model trial-level measure “hits” (1s 0s) target area interest. un-aggregated, trial-level data comes data preparation code top vignette. demonstrate relationship two data sets: specification object fitting logistic regression models trial-level data requires two changes original: formula must predict binary target variable, instead prop trial grouping must specified uniquely identify time point within subject following appropriate <jlmer_spec> object CPA using logistic regression: , sanity check global model using family = \"binomial\": estimates similar global_jmod previous section, log-odds.","code":"nrow(ts_data_agg) #> [1] 2560 ts_data_agg %>%   distinct(subject_id, time) %>%   nrow() #> [1] 2560 ts_data_trials #> # A tibble: 62,466 × 5 #>    subject_id age   trial_id  time target #>         <int> <fct>    <int> <dbl>  <int> #>  1       1413 Older     2659     0      1 #>  2       1413 Older     2659    25      1 #>  3       1413 Older     2659    50      1 #>  4       1413 Older     2659    75      1 #>  5       1413 Older     2659   100      1 #>  6       1413 Older     2659   125      1 #>  7       1413 Older     2659   150      1 #>  8       1413 Older     2659   175      1 #>  9       1413 Older     2659   200      1 #> 10       1413 Older     2659   225      1 #> # ℹ 62,456 more rows identical(   ts_data_agg,   ts_data_trials %>%     group_by(subject_id, age, time) %>%     summarize(prop = mean(target), .groups = \"drop\") ) #> [1] TRUE ts_data_trial_spec <- make_jlmer_spec(   formula = target ~ 1 + age, data = ts_data_trials,   subject = \"subject_id\", trial = \"trial_id\", time = \"time\" ) ts_data_trial_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: target ~ 1 + ageOlder #> Predictors: #>   age: ageOlder #> Groupings: #>   Subject: subject_id #>   Trial: trial_id #>   Time: time #> Data: #> # A tibble: 62,466 × 5 #>   target ageOlder subject_id trial_id  time #>    <int>    <dbl>      <int>    <int> <dbl> #> 1      1        1       1413     2659     0 #> 2      1        1       1413     2659    25 #> 3      1        1       1413     2659    50 #> # ℹ 62,463 more rows #> ──────────────────────────────────────────────────────────────────────────────── global_jmod_binom <- jlmer(ts_data_trial_spec, family = \"binomial\") tidy(global_jmod_binom) #> # A tibble: 2 × 5 #>   term        estimate std.error statistic   p.value #>   <chr>          <dbl>     <dbl>     <dbl>     <dbl> #> 1 (Intercept)    0.113    0.0121      9.31 1.24e- 20 #> 2 ageOlder       0.392    0.0164     23.9  1.08e-126"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Garrison-et-al-2020.html","id":"repeating-the-cpa-with-logistic-regression","dir":"Articles","previous_headings":"D) As logistic regression","what":"Repeating the CPA with logistic regression","title":"Garrison et al. 2020","text":"repeat procedure use new specification object set family = \"binomial\": Instead 3 clusters, now detect 5. can annotate detected clusters time series data :  differences see aggregated vs. trial-level data completely expected follow straightforwardly following two facts: cluster-mass statistic derived size t-values time point, effect size. virtue unaggregated ts_data_trials data number observations, t-values overall inflated difference means (also see output global_jmod vs. global_jmod_binom well). Whereas aggregated data reflects complete pooling trial-level variation, binomial model trial-level data performs pooling treats trial independent. Thus, certain item experiment show outlier effect one time point, example, strongly influence mass cluster spanning given time point. first issue (inflated statistics) trivial just takes matter adjusting threshold. fact, good “stretching ” t-statistics, get higher resolution shape timewise statistics. However. second issue (pooling) troubling can warp fundamental shape timewise statistics. diagnose factors visualizing timewise statistics. also plot timewise statistics aggregated data comparison:  see trial-level binomial model overall inflated t-values, well pronounced “peaks” (specifically maximum 1300ms). middle-ground, partial pooling approaching using mixed-effects models CPA showcased case study vignettes. present purposes, raise threshold value (2.8) gets us close detecting least largest cluster empirical_statistics, spanning 850ms-1525ms: run cluster permutation test threshold just largest cluster: gives us pvalue cluster spanning 850ms 1525ms similar one reported aggregated data: appear converge higher nsim data set. also time 1000-simulation runs showcase speed jlmerclusterperm. Note use set_rng_state(), makes two CPAs generate permutations observed data.","code":"empirical_statistics_binom <- compute_timewise_statistics(ts_data_trial_spec, family = \"binomial\") empirical_clusters_binom <- extract_empirical_clusters(empirical_statistics_binom, threshold = 2) empirical_clusters_binom #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [650, 725]: 8.727 #>   [800, 1575]: 112.199 #>   [1775, 1875]: 12.764 #>   [2600, 2825]: 23.733 #>   [3475, 3975]: 62.714 #> ──────────────────────────────────────────────────────────────────────────────── empirical_clusters_binom_df <- tidy(empirical_clusters_binom) tutorial_fig +   geom_segment(     aes(x = start, xend = end, y = -Inf, yend = -Inf),     linewidth = 10, color = \"steelblue\",     inherit.aes = FALSE,     data = empirical_clusters_binom_df   ) +   geom_text(     aes(       y = -Inf, x = start + (end - start) / 2,       label = paste(\"Cluster\", id)     ),     vjust = -2,     inherit.aes = FALSE,     data = empirical_clusters_binom_df   ) timewise_statistics_fig <- ggplot(mapping = aes(time, statistic)) +   geom_hline(aes(yintercept = 0)) +   geom_line(     aes(color = \"aggregated\"),     linewidth = 1.5,     data = tidy(empirical_statistics)   ) +   geom_line(     aes(color = \"trial-level\"),     linewidth = 1.5,     data = tidy(empirical_statistics_binom)   ) timewise_statistics_fig +   geom_vline(     aes(xintercept = time),     data = tidy(empirical_statistics_binom) %>%       slice(which.max(abs(statistic)))   ) empirical_clusters_binom2 <- extract_empirical_clusters(empirical_statistics_binom, threshold = 2.8) empirical_clusters_binom2 #> ── Empirical clusters (t > 2.8) ──────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 102.450 #>   [3500, 3875]: 51.365 #> ──────────────────────────────────────────────────────────────────────────────── CPA_trial <- clusterpermute(   ts_data_trial_spec,   family = \"binomial\",   threshold = 2.8,   nsim = 100,   top_n = 1, # Just test the largest cluster   progress = FALSE # Suppress printing of progress for vignette ) CPA_trial$empirical_clusters #> ── Empirical clusters (t > 2.8) ──────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 102.450 (p=0.0297) #> ────────────────────────────────────────────────────────────────────────────────  CPA_agg$empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 74.724 (p=0.0396) #>   [2675, 2775]: 11.017 (p=0.4356) #>   [3500, 3900]: 46.295 (p=0.0990) #> ──────────────────────────────────────────────────────────────────────────────── system.time({   set_rng_state(123L)   CPA_agg_1000 <- clusterpermute(     ts_data_agg_spec,     threshold = 2,     nsim = 1000, top_n = 1, progress = FALSE   ) }) #>    user  system elapsed  #>    0.05    0.11   10.97 CPA_agg_1000 #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> ageOlder (n = 1000) #>   Mean (SD): 0.078 (40.29) #>   Coverage intervals: 95% [-76.748, 82.717] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 74.724 (p=0.0629) #> ──────────────────────────────────────────────────────────────────────────────── system.time({   set_rng_state(123L)   CPA_trial_1000 <- clusterpermute(     ts_data_trial_spec,     family = \"binomial\", threshold = 2.8,     nsim = 1000, top_n = 1, progress = FALSE   ) }) #>    user  system elapsed  #>    0.05    0.06   15.25 CPA_trial_1000 #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2.8) ──────────── <null_cluster_dists> ── #> ageOlder (n = 1000) #>   Mean (SD): -0.045 (55.09) #>   Coverage intervals: 95% [-119.239, 125.506] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2.8) ──────────────────────── <empirical_clusters> ── #> ageOlder #>   [850, 1525]: 102.450 (p=0.0689) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"background","dir":"Articles","previous_headings":"","what":"Background","title":"Geller et al. 2020","text":"data comes lexical decision task using pupilometry (specifically, measure pupil dilation) study differences processing difficulty print vs. cursive script. data available part gazer package Geller, Winn, Mahr, & Mirman 2020. follow Jason Geller’s tutorial used clusterperm package conduct cluster-based permutation analysis (CPA) differences pupil size print cursive conditions. data package (cursive_agg_data) looks slightly different used tutorial, ensure full reproducibility use exact data tutorial used: data comes prepared analysis box. following columns relevant analysis: subject: Unique identifier subjects script: within-participant factor variable (\"cursive\", \"print\") time: continuous measure time 0-2500ms 100ms intervals aggbaseline: response variable representing normalized (“baseline-corrected”) pupil size name cusrive_agg_data variable suggests, data aggregated within subject, collapsing across trials. following reproduces figure tutorial:  within-participant predictors like script, permutation algorithm randomly swap labels condition trials within subject. preserves temporal structure trial-level data (swapping trial-level grouping) well subject-level grouping structure (swapping trials across participants).","code":"library(dplyr) cursive_agg_data <- as_tibble(read.csv(\"https://raw.githubusercontent.com/jgeller112/drjasongeller/main/content/blog/2020-07-10-CBPT/blog_data.csv\")) cursive_agg_data #> # A tibble: 2,184 × 5 #>        X subject script  timebins aggbaseline #>    <int> <chr>   <chr>      <int>       <dbl> #>  1     1 10b.edf cursive        0      36.9   #>  2     2 10b.edf cursive      100      27.5   #>  3     3 10b.edf cursive      200      19.6   #>  4     4 10b.edf cursive      300      12.6   #>  5     5 10b.edf cursive      400       5.60  #>  6     6 10b.edf cursive      500       0.968 #>  7     7 10b.edf cursive      600      -5.98  #>  8     8 10b.edf cursive      700      -5.96  #>  9     9 10b.edf cursive      800       3.01  #> 10    10 10b.edf cursive      900      12.9   #> # ℹ 2,174 more rows library(ggplot2) script_fig <- ggplot(cursive_agg_data, aes(timebins, aggbaseline)) +   stat_summary(aes(linetype = script), geom = \"line\", linewidth = 1) script_fig"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"outline","dir":"Articles","previous_headings":"","what":"Outline","title":"Geller et al. 2020","text":"case study vignette showcases five features CPA jlmerclusterperm: Prepping data CPA using make_jlmer_spec() clusterpermute() default statistic = \"t\" comparison “t” “chisq” Specifying random effects Contrast coding Load package start Julia instance jlmerclusterperm_setup() proceeding.","code":"library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"a-prepping-a-specification-object","dir":"Articles","previous_headings":"","what":"A) Prepping a specification object","title":"Geller et al. 2020","text":"start simple specification object model aggbaseline using script predictor script_fig data: sanity check, fit global model data… …check ’s comparable expect lm(): full specification object CPA must also declare grouping structures present data. rule thumb every observation (row) data must uniquely identified combination columns subject, trial, time. CPA time series data time must always specified, subject must also always specified permutation algorithm respect subject-level grouping data. cursive_agg_data data collapses across trials within subject, column trial. However, leave trial argument unspecified observations uniquely identified subject time alone. instead 2 rows per subject-time combination, one script condition: Therefore need “dummy” column trial distinctly mark “cursive” vs. “script” trials. save new data cursive_agg: makes observations uniquely identified columns subject, trial, time: final specification object looks like following:","code":"simple_spec <- make_jlmer_spec(   formula = aggbaseline ~ 1 + script,   data = cursive_agg_data ) simple_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: aggbaseline ~ 1 + scriptprint #> Predictors: #>   script: scriptprint #> Data: #> # A tibble: 2,184 × 2 #>   aggbaseline scriptprint #>         <dbl>       <dbl> #> 1        36.9           0 #> 2        27.5           0 #> 3        19.6           0 #> # ℹ 2,181 more rows #> ──────────────────────────────────────────────────────────────────────────────── jlmer(simple_spec) #> <Julia object of type StatsModels.TableRegressionModel> #> ──────────────────────────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|)  Lower 95%  Upper 95% #> ──────────────────────────────────────────────────────────────────────── #> (Intercept)   46.7221     2.79622  16.71    <1e-61    41.2417    52.2026 #> scriptprint  -10.2582     3.95445  -2.59    0.0095   -18.0088    -2.5076 #> ──────────────────────────────────────────────────────────────────────── summary(lm(formula = aggbaseline ~ 1 + script, data = cursive_agg_data))$coefficients #>              Estimate Std. Error   t value     Pr(>|t|) #> (Intercept)  46.72214   2.796221 16.709029 4.512338e-59 #> scriptprint -10.25818   3.954454 -2.594083 9.547816e-03 cursive_agg_data %>%   count(subject, timebins) #> # A tibble: 1,092 × 3 #>    subject timebins     n #>    <chr>      <int> <int> #>  1 10b.edf        0     2 #>  2 10b.edf      100     2 #>  3 10b.edf      200     2 #>  4 10b.edf      300     2 #>  5 10b.edf      400     2 #>  6 10b.edf      500     2 #>  7 10b.edf      600     2 #>  8 10b.edf      700     2 #>  9 10b.edf      800     2 #> 10 10b.edf      900     2 #> # ℹ 1,082 more rows cursive_agg <- cursive_agg_data %>%   mutate(trial_type = paste0(script, \"_agg\")) cursive_agg #> # A tibble: 2,184 × 6 #>        X subject script  timebins aggbaseline trial_type  #>    <int> <chr>   <chr>      <int>       <dbl> <chr>       #>  1     1 10b.edf cursive        0      36.9   cursive_agg #>  2     2 10b.edf cursive      100      27.5   cursive_agg #>  3     3 10b.edf cursive      200      19.6   cursive_agg #>  4     4 10b.edf cursive      300      12.6   cursive_agg #>  5     5 10b.edf cursive      400       5.60  cursive_agg #>  6     6 10b.edf cursive      500       0.968 cursive_agg #>  7     7 10b.edf cursive      600      -5.98  cursive_agg #>  8     8 10b.edf cursive      700      -5.96  cursive_agg #>  9     9 10b.edf cursive      800       3.01  cursive_agg #> 10    10 10b.edf cursive      900      12.9   cursive_agg #> # ℹ 2,174 more rows cursive_agg %>%   count(subject, timebins, trial_type) %>%   distinct(n) #> # A tibble: 1 × 1 #>       n #>   <int> #> 1     1 cursive_agg_spec <- make_jlmer_spec(   formula = aggbaseline ~ 1 + script,   data = cursive_agg,   subject = \"subject\", trial = \"trial_type\", time = \"timebins\" ) cursive_agg_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: aggbaseline ~ 1 + scriptprint #> Predictors: #>   script: scriptprint #> Groupings: #>   Subject: subject #>   Trial: trial_type #>   Time: timebins #> Data: #> # A tibble: 2,184 × 5 #>   aggbaseline scriptprint subject trial_type  timebins #>         <dbl>       <dbl> <chr>   <chr>          <int> #> 1        36.9           0 10b.edf cursive_agg        0 #> 2        27.5           0 10b.edf cursive_agg      100 #> 3        19.6           0 10b.edf cursive_agg      200 #> # ℹ 2,181 more rows #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"b-cpa-with-default-statistic-t","dir":"Articles","previous_headings":"","what":"B) CPA with default statistic = \"t\"","title":"Geller et al. 2020","text":"CPA output original tutorial (using 100 simulations) copied comparison: Using threshold 2 default statistic = \"t\", clusterpermute() returns following: results look different original. due following: cluster-mass statistic (cms) lower. ran clusterpermute() default statistic = \"t\". Different clusters identified. simple model account subject-level variation, whereas original error term formula aggbaseline ~ script + Error(subject). sign cluster reversed. defaults aov() used original tutorial different default contrast coding regression model. now address issue turn, building CPA closely replicate results tutorial.","code":"#> #> effect   b0     b1  sign        cms           p #> #> script    0    100     1  10.121215   0.1584158 #> #> script  900   1000    -1  8.756152    0.1782178 #> #> script 1900   2500     1  82.198279   0.0099010 set_rng_state(123L) clusterpermute(   cursive_agg_spec,   statistic = \"t\", # Default value spelled out   threshold = 2,   nsim = 100,   progress = FALSE ) #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> scriptprint (n = 100) #>   Mean (SD): 0.000 (0.00) #>   Coverage intervals: 95% [0.000, 0.000] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> scriptprint #>   [2000, 2300]: -8.748 (p=0.0099) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"c-comparing-t-vs--chisq","dir":"Articles","previous_headings":"","what":"C) Comparing “t” vs. “chisq”","title":"Geller et al. 2020","text":"original tutorial used clusterperm::cluster_nhds() conduct CPA, fits ANOVA models time. , timewise statistic used chi-squared threshold determined p-value chi-squared statistic. Using chi-squared statistics p-value threshold also supported clusterpermute() using statistic = \"chisq\" (instead default \"t\"): returns cluster now numerically larger cluster-mass statistic, expected (chi-squared asymptotic t^2). , compare shape timewise statistics “t” threshold 2 “chisq” threshold p=0.05:  find clusters identified 2000ms-2300ms “t” “chisq”, peaks pronounced “chisq”. differences two inconsequential example, may produce different results cases. chi-squared statistic (jlmerclusterperm computes via likelihood ratio test) often preferred testing single parameters makes less assumptions tend robust (glmmFAQ). goodness--fit tests level predictor model formula, “chisq” less interpretable multi-level predictors (k-1 > 1). teasing apart contribution individual levels multi-level predictor, using statistic = \"t\" appropriate.","code":"set_rng_state(123L) clusterpermute(   cursive_agg_spec,   statistic = \"chisq\",   threshold = 0.05, # Threshold is now the p-value of the chi-squared statistic   nsim = 100,   progress = FALSE ) #> $null_cluster_dists #> ── Null cluster-mass distribution (chisq p < 0.05) ───── <null_cluster_dists> ── #> script (n = 100, df = 1) #>   Mean (SD): 0.000 (0.00) #>   Coverage intervals: 95% [0.000, 0.000] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (chisq p < 0.05) ───────────────── <empirical_clusters> ── #> script (df = 1) #>   [2000, 2300]: -19.085 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── timewise_ts <- compute_timewise_statistics(cursive_agg_spec, statistic = \"t\") timewise_chisqs <- compute_timewise_statistics(cursive_agg_spec, statistic = \"chisq\") library(ggplot2) timewise_fig <- ggplot(mapping = aes(x = time, y = statistic)) +   geom_line(     aes(color = \"fixed-t\"),     linewidth = 1.5,     data = tidy(timewise_ts)   ) +   geom_line(     aes(color = \"fixed-chisq\"),     linewidth = 1.5,     data = tidy(timewise_chisqs)   ) +   geom_hline(     aes(yintercept = c(-2, 2, qchisq(.95, df = 1), -qchisq(.95, df = 1))),     color = rep(c(\"#00BFC4\", \"#F8766D\"), each = 2), linetype = 2   ) +   scale_color_manual(     values = setNames(       c(\"#E69F00\", \"#56B4E9\", \"#009E73\", \"#F0E442\"),       c(\"fixed-t\", \"fixed-chisq\", \"re-intercept-chisq\", \"re-max-chisq\")     )   ) timewise_fig"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"d-specifying-random-effects","dir":"Articles","previous_headings":"","what":"D) Specifying random effects","title":"Geller et al. 2020","text":"biggest missing component point subject random effects, original tutorial captures via error term Error(subject). isn’t strict equivalent regression, specifying random intercepts subject (1 | subject) gets us close: repeat results original tutorial comparison: now plot timewise statistics random intercept models:  point may wonder whether results change much used maximal model random effects structure (1 + script | subject). first create another specification object maximal formula: compute timewise statistics: find chisq statistics maximal model virtually identical parsimonious, intercept-model (two lines overlap):  Parsimony incredibly important simulation - jlmerclusterperm fast, model complexity still major bottleneck. adding extra random effect terms (random slope correlation) negligible effects statistic, removing likely inconsequential CPA . show following CPA run, uses maximal model. Notice results identical intercept-re_CPA substantially slower: run maximal mixed model CPA purely demonstration purposes. random effects structure actually unidentifiable given aggregated data, point time 2 observations subject (mean condition). Lastly, back figure data, annotate clusters detected chisq p-value threshold 0.05 random intercept CPA:","code":"cursive_agg_spec_re <- make_jlmer_spec(   formula = aggbaseline ~ 1 + script + (1 | subject),   data = cursive_agg,   subject = \"subject\", time = \"timebins\", trial = \"trial_type\" ) set_rng_state(123L) system.time({   re_CPA <- clusterpermute(     cursive_agg_spec_re,     statistic = \"chisq\",     threshold = 0.05,     nsim = 100,     progress = FALSE   ) }) #>    user  system elapsed  #>    0.02    0.10    9.42 re_CPA #> $null_cluster_dists #> ── Null cluster-mass distribution (chisq p < 0.05) ───── <null_cluster_dists> ── #> script (n = 100, df = 1) #>   Mean (SD): 1.132 (12.58) #>   Coverage intervals: 95% [-21.081, 32.649] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (chisq p < 0.05) ───────────────── <empirical_clusters> ── #> script (df = 1) #>   [0, 100]: -9.770 (p=0.1881) #>   [900, 1000]: 8.521 (p=0.2475) #>   [1900, 2500]: -73.440 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── #> #> effect   b0     b1  sign        cms           p #> #> script    0    100     1  10.121215   0.1584158 #> #> script  900   1000    -1  8.756152    0.1782178 #> #> script 1900   2500     1  82.198279   0.0099010 timewise_chisqs_re <- compute_timewise_statistics(cursive_agg_spec_re, statistic = \"chisq\") timewise_fig_re <- timewise_fig +   geom_line(     aes(color = \"re-intercept-chisq\"),     linewidth = 1.5,     data = tidy(timewise_chisqs_re)   ) timewise_fig_re cursive_agg_spec_re_max <- make_jlmer_spec(   formula = aggbaseline ~ 1 + script + (1 + script | subject),   data = cursive_agg,   subject = \"subject\", time = \"timebins\", trial = \"trial_type\" ) timewise_chisqs_re_max <- compute_timewise_statistics(cursive_agg_spec_re_max, statistic = \"chisq\") #> ℹ 1 singular fit (3.85%). timewise_fig_re +   geom_line(     aes(color = \"re-max-chisq\"),     linewidth = 1.5,     data = tidy(timewise_chisqs_re_max)   ) set_rng_state(123L) system.time({   re_max_CPA <- clusterpermute(     cursive_agg_spec_re_max,     statistic = \"chisq\",     threshold = 0.05,     nsim = 100,     progress = FALSE   ) }) #>    user  system elapsed  #>    0.00    0.06   10.60 re_max_CPA #> $null_cluster_dists #> ── Null cluster-mass distribution (chisq p < 0.05) ───── <null_cluster_dists> ── #> script (n = 100, df = 1) #>   Mean (SD): 1.132 (12.58) #>   Coverage intervals: 95% [-21.081, 32.649] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (chisq p < 0.05) ───────────────── <empirical_clusters> ── #> script (df = 1) #>   [0, 100]: -9.770 (p=0.1881) #>   [900, 1000]: 8.521 (p=0.2475) #>   [1900, 2500]: -73.440 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── empirical_clusters_df <- tidy(extract_empirical_clusters(timewise_chisqs_re, threshold = 0.05)) script_fig +   geom_segment(     aes(       x = start, xend = end, y = -Inf, yend = -Inf,       color = factor(sign(sum_statistic))     ),     linewidth = 10,     inherit.aes = FALSE,     data = empirical_clusters_df   ) +   geom_text(     aes(       y = -Inf, x = start + (end - start) / 2,       label = paste(\"Cluster\", id)     ),     vjust = -2,     inherit.aes = FALSE,     data = empirical_clusters_df   ) +   labs(color = \"Sign of cluster\")"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/Geller-et-al-2020.html","id":"e-contrast-coding","dir":"Articles","previous_headings":"","what":"E) Contrast coding","title":"Geller et al. 2020","text":"last piece puzzle flipped sign effect. Whereas detect negative cluster line cursive line print, original tutorial reports positive cluster (output repeated ): Fixing trivial - just takes different choice contrast. example, can flip levels factor make “print” reference level: also sum-code set “print” -1 “cursive” 1: CPA operates domain timewise statistics (effect size), whether difference “print” “cursive” expressed unit 1 (treatment coding) 2 (contrast coding) bearing CPA. conclude, re-run time 1000-simulation CPA using intercept-model new treatment coding script “print” reference level. Finally, annotating just significant cluster figure:","code":"#> #> effect   b0     b1  sign        cms           p #> #> script    0    100     1  10.121215   0.1584158 #> #> script  900   1000    -1  8.756152    0.1782178 #> #> script 1900   2500     1  82.198279   0.0099010 rev_contrast_df <- cursive_agg rev_contrast_df$script <- factor(rev_contrast_df$script, levels = c(\"print\", \"cursive\")) contrasts(rev_contrast_df$script) #>         cursive #> print         0 #> cursive       1 reverse_contrast_spec <- make_jlmer_spec(   formula = aggbaseline ~ 1 + script + (1 | subject),   data = rev_contrast_df,   subject = \"subject\", time = \"timebins\", trial = \"trial_type\" ) set_rng_state(123L) clusterpermute(   reverse_contrast_spec,   statistic = \"chisq\",   threshold = 0.05,   nsim = 100,   progress = FALSE ) #> $null_cluster_dists #> ── Null cluster-mass distribution (chisq p < 0.05) ───── <null_cluster_dists> ── #> script (n = 100, df = 1) #>   Mean (SD): -1.132 (12.58) #>   Coverage intervals: 95% [-32.649, 21.081] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (chisq p < 0.05) ───────────────── <empirical_clusters> ── #> script (df = 1) #>   [0, 100]: 9.770 (p=0.1881) #>   [900, 1000]: -8.521 (p=0.2475) #>   [1900, 2500]: 73.440 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── sum_contrast_df <- cursive_agg sum_contrast_df$script <- factor(sum_contrast_df$script) contrasts(sum_contrast_df$script) <- contr.sum(2) contrasts(sum_contrast_df$script) #>         [,1] #> cursive    1 #> print     -1 sum_contrast_spec <- make_jlmer_spec(   formula = aggbaseline ~ 1 + script + (1 | subject),   data = sum_contrast_df,   subject = \"subject\", time = \"timebins\", trial = \"trial_type\" ) set_rng_state(123L) clusterpermute(   sum_contrast_spec,   statistic = \"chisq\",   threshold = 0.05,   nsim = 100,   progress = FALSE ) #> $null_cluster_dists #> ── Null cluster-mass distribution (chisq p < 0.05) ───── <null_cluster_dists> ── #> script (n = 100, df = 1) #>   Mean (SD): -1.132 (12.58) #>   Coverage intervals: 95% [-32.649, 21.081] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (chisq p < 0.05) ───────────────── <empirical_clusters> ── #> script (df = 1) #>   [0, 100]: 9.770 (p=0.1881) #>   [900, 1000]: -8.521 (p=0.2475) #>   [1900, 2500]: 73.440 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── set_rng_state(123L) system.time({   final_CPA <- clusterpermute(     reverse_contrast_spec,     statistic = \"chisq\",     threshold = 0.05,     nsim = 1000,     progress = FALSE   ) }) #>    user  system elapsed  #>    0.00    0.01   11.68 final_CPA #> $null_cluster_dists #> ── Null cluster-mass distribution (chisq p < 0.05) ───── <null_cluster_dists> ── #> script (n = 1000, df = 1) #>   Mean (SD): -0.107 (16.50) #>   Coverage intervals: 95% [-36.966, 41.530] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (chisq p < 0.05) ───────────────── <empirical_clusters> ── #> script (df = 1) #>   [0, 100]: 9.770 (p=0.2108) #>   [900, 1000]: -8.521 (p=0.2358) #>   [1900, 2500]: 73.440 (p=0.0090) #> ──────────────────────────────────────────────────────────────────────────────── signif_clusters_df <- tidy(final_CPA$empirical_clusters) %>%   filter(pvalue < 0.05) signif_clusters_df #> # A tibble: 1 × 7 #>   predictor id    start   end length sum_statistic  pvalue #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>   <dbl> #> 1 script    3      1900  2500      7          73.4 0.00899 script_fig +   geom_segment(     aes(x = start, xend = end, y = -Inf, yend = -Inf),     color = \"steelblue\", linewidth = 10,     inherit.aes = FALSE,     data = signif_clusters_df   ) +   geom_text(     aes(       y = -Inf, x = start + (end - start) / 2,       label = paste(\"p =\", signif(pvalue, 3))     ),     vjust = -2,     inherit.aes = FALSE,     data = signif_clusters_df   )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/jlmerclusterperm.html","id":"overview-of-cpa","dir":"Articles","previous_headings":"","what":"Overview of CPA","title":"Introduction to jlmerclusterperm","text":"Cluster-based permutation analysis (CPA) simulation-based, non-parametric statistical test difference groups time series. suitable analyzing densely-sampled time data (EEG eye-tracking research) research hypothesis often specified existence effect (e.g., predicted higher-order cognitive mechanisms) agnostic temporal details effect (precise moment emergence). CPA formalizes two intuitions means difference groups: countable unit difference (.e., cluster) contiguous, uninterrupted span sufficiently large differences time point (determined via threshold). degree extremity cluster (.e., cluster-mass statistic) measure sensitive magnitude difference, variability, sample size (e.g., t-statistic regression). CPA procedure identifies empirical clusters time series data tests significance observed cluster-mass statistics bootstrapped permutations data. bootstrapping procedure samples “null” (via random shuffling condition labels), yielding distribution cluster-mass statistics emerging chance. statistical significance cluster probability observing cluster-mass statistic extreme cluster’s simulated null distribution.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/jlmerclusterperm.html","id":"package-design","dir":"Articles","previous_headings":"","what":"Package design","title":"Introduction to jlmerclusterperm","text":"jlmerclusterperm provides wholesale piecemeal approach conducting CPA. main workhorse function package clusterpermute(), composed five smaller functions called internally succession. smaller functions representing algorithmic steps CPA also exported, allow control procedure (e.g., debugging diagnosing CPA run). See function documentation .","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/jlmerclusterperm.html","id":"organization-of-vignettes","dir":"Articles","previous_headings":"","what":"Organization of vignettes","title":"Introduction to jlmerclusterperm","text":"package vignettes roughly divided two groups: topics case studies. researcher data ready analysis, recommended go case studies first pluck relevant bits desired analysis order increasing complexity, case studies : Garrison et al. 2020, introduces package’s functions running CPA, demonstrating wholesale piecemeal approaches. also compares use linear vs. logistic regression calculation timewise statistics. Geller et al. 2020, demonstrates use random effects regression contrasts CPA specification. also compares use t vs. chisq timewise statistics. de Carvalho et al. 2021, showcases using custom regression contrasts test relationship multiple levels factor. also discusses issues surrounding complexity timewise regression model. topics cover general package features: Tidying output: tidy() ways collecting jlmerclusterperm objects tidy data. Julia interface: interacting Julia objects using JuliaConnectoR package. Reproducibility: using Julia RNG make CPA runs reproducible. Asynchronous CPA: running CPA background process using future package. Comparison eyetrackingR: Translating eyetrackingR code CPA jlmerclusterperm.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/jlmerclusterperm.html","id":"compared-to-other-implementations","dir":"Articles","previous_headings":"","what":"Compared to other implementations","title":"Introduction to jlmerclusterperm","text":"R packages cluster-based permutation analysis, clusterperm, eyetrackingR, permuco, permutes. Compared existing implementations, jlmerclusterperm package optimized CPAs based mixed-effects regression models, suitable typical experimental research data multiple, crossed grouping structures (e.g., subjects items). also package interface individual algorithmic steps CPA. Lastly, thanks Julia back-end, bootstrapping fast even mixed models.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/jlmerclusterperm.html","id":"further-readings","dir":"Articles","previous_headings":"","what":"Further readings","title":"Introduction to jlmerclusterperm","text":"original method paper CPA Maris & Oostenveld 2007 recent overview paper Meyer et al. 2021 (CPA EEG research, applies broadly) critical paper Sassenhagen & Draschkow 2019 DOs DON’Ts interpreting reporting CPA results. review paper eyetracking analysis methods Ito & Knoeferle 2022 compares CPA statistical techniques testing difference groups time series data. JOSS paper permuco package (Frossard & Renaud 2021) references therein. Benedikt V. Ehinger’s blog post, includes detailed graphical explainer CPA procedure.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/julia-interface.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Julia interface","text":"","code":"library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/julia-interface.html","id":"interacting-with-julia-objects","dir":"Articles","previous_headings":"","what":"Interacting with Julia objects","title":"Julia interface","text":"jlmerclusterperm functions collect Julia objects R objects, except jlmer() to_jlmer() return GLM.jl MixedModels.jl fitted model objects. can use functions JuliaConnectoR interact (pointers ) Julia objects: can also call functions Julia packages already installed. example, Effects.jl marginaleffects/emmeans-like features: Note jlmer() to_lmer() just conveniences sanity checking steps cluster-based permutation test, thus used general interface fitting regression models Julia serious analyses.","code":"jmod <- to_jlmer(Reaction ~ Days + (Days | Subject), lme4::sleepstudy) jmod #> <Julia object of type LinearMixedModel{Float64}> #> Variance components: #>             Column    Variance Std.Dev.   Corr. #> Subject  (Intercept)  565.51067 23.78047 #>          Days          32.68212  5.71683 +0.08 #> Residual              654.94145 25.59182 #> ────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|) #> ────────────────────────────────────────────────── #> (Intercept)  251.405      6.63226  37.91    <1e-99 #> Days          10.4673     1.50224   6.97    <1e-11 #> ────────────────────────────────────────────────── library(JuliaConnectoR) juliaLet(\"x.rePCA\", x = jmod) #> <Julia object of type NamedTuple{(:Subject,), Tuple{Vector{Float64}}}> #> (Subject = [0.5406660682947617, 1.0],) juliaCall(\"issingular\", jmod) #> [1] FALSE # juliaEval('using Pkg; Pkg.activate(); Pkg.add(\"Effects\")') juliaEval(\"using Effects\") juliaLet(\"effects(x.namedelements, y)\", x = list(Days = 2:5), y = jmod) #> <Julia object of type DataFrame> #> 4×5 DataFrame #>  Row │ Days   Reaction  err      lower    upper #>      │ Int64  Float64   Float64  Float64  Float64 #> ─────┼──────────────────────────────────────────── #>    1 │     2   272.34   7.09427  265.245  279.434 #>    2 │     3   282.807  7.70544  275.102  290.512 #>    3 │     4   293.274  8.55557  284.719  301.83 #>    4 │     5   303.742  9.58128  294.16   313.323"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/julia-interface.html","id":"julia-global-options","dir":"Articles","previous_headings":"","what":"Julia global options","title":"Julia interface","text":"functions involved cluster-based permutation analysis print progress bars update real time console. function julia_progress() controls whether show progress bar width printed progress bar.","code":"# Set Julia progress options and save old state old_progress_opts <- julia_progress(show = FALSE, width = 30) old_progress_opts #> $show #> [1] TRUE #> #> $width #> [1] 50 julia_progress() #> $show #> [1] FALSE #> #> $width #> [1] 30  # Restored old state julia_progress(old_progress_opts) identical(old_progress_opts, julia_progress()) #> [1] TRUE  # The syntax to reset progress options julia_progress(show = TRUE, width = \"auto\") julia_progress()"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/julia-interface.html","id":"the-julia-environment","dir":"Articles","previous_headings":"","what":"The Julia environment","title":"Julia interface","text":"following packages loaded Julia environment via jlmerclusterperm_setup():","code":"cat(readLines(system.file(\"julia/Project.toml\", package = \"jlmerclusterperm\")), sep = \"\\n\") #> [deps] #> DataFrames = \"a93c6f00-e57d-5684-b7b6-d8193f3e46c0\" #> Distributions = \"31c24e10-a181-5473-b8eb-7969acd0382f\" #> GLM = \"38e38edf-8417-5370-95a0-9cbb8c7f171a\" #> JlmerClusterPerm = \"2a59065a-9564-4903-82dd-0e42ce19d0e1\" #> MixedModels = \"ff71e718-51f3-5ec2-a782-8ffcbfa3c316\" #> ProgressMeter = \"92933f4c-e287-5a05-a399-4b506db050ca\" #> Random = \"9a3f8284-a2c9-5f02-9a11-845980a1fd5c\" #> Random123 = \"74087812-796a-5b5d-8853-05524746bad3\" #> StatsBase = \"2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91\" #> StatsModels = \"3eaba693-59b7-5ba5-a881-562e759f1c8d\" #> Suppressor = \"fd094767-a336-5f1f-9728-57cf17d0bbfb\" #>  #> # add DataFrames@1.3, GLM@1.8, MixedModels@4, ProgressMeter@1.7, Random123@1.6, StatsBase@0.33, StatsModels@0.7, Suppressor@0.2 #> # add Distributions, Random #> # dev JlmerClusterPerm #>  #> [compat] #> DataFrames = \"< 1.4\" #> GLM = \"1.8\" #> JlmerClusterPerm = \"=0.1.1\" #> MixedModels = \"4\" #> ProgressMeter = \"1.7\" #> Random123 = \"1.6\" #> StatsBase = \"0.33\" #> StatsModels = \"0.7\" #> Suppressor = \"0.2\" #> julia = \"1.8\""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/julia-interface.html","id":"optional-configurations","dir":"Articles","previous_headings":"","what":"Optional configurations","title":"Julia interface","text":"speed-ups model fitting, install MKL.jl (Intel core) AppleAccelerate.jl (Mac) load start modifying startup.jl file. example, Windows machine startup file located : load MKL.jl:","code":"JuliaConnectoR::juliaCall(\"Base._local_julia_startup_file\") #> \"C:/Users/jchoe/.julia/config/startup.jl\" if contains(first(Sys.cpu_info()).model, \"Intel\")   using MKL end"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/reproducibility.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Reproducibility","text":"Minimal example data:  jlmer specification object used showcase:","code":"library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE) library(dplyr, warn.conflicts = FALSE) library(ggplot2) chickweights <- as_tibble(ChickWeight) %>%   mutate(diet = forcats::fct_collapse(Diet, A = c(1, 3), B = c(2, 4))) %>%   filter(Time %in% 0:20) ggplot(chickweights, aes(Time, weight, color = diet)) +   stat_summary(fun.args = list(mult = 1.96)) +   stat_summary(geom = \"line\") +   theme_minimal(base_size = 18) jlmer_spec <- make_jlmer_spec(   formula = weight ~ diet + (1 | Chick), data = chickweights,   subject = \"Chick\", time = \"Time\" ) jlmer_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + dietB + (1 | Chick) #> Predictors: #>   diet: dietB #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 4 #>   weight dietB Chick  Time #>    <dbl> <dbl> <chr> <dbl> #> 1     42     0 1         0 #> 2     51     0 1         2 #> 3     59     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/reproducibility.html","id":"rng-seed-vs--counter","dir":"Articles","previous_headings":"","what":"RNG seed vs. counter","title":"Reproducibility","text":"CPA powered bootstrapping, inherently stochastic procedure. Reproducibility guaranteed clusterpermute() calls ran seed counter values. properties Julia random number generator used jlmerclusterperm refer two slightly different concepts: seed conceptually equivalent “seed” talked R, set.seed(). seed used RNG’s initialization, locking specific state persists session. jlmerclusterperm, default seed 1 primarily controlled via options(\"jlmerclusterperm.seed\") must set calling jlmerclusterperm_setup(). counter initializes 0 increments value every time RNG consumed produce random number. equivalent concept base R, though ’s handy intuitive way interfacing RNG’s state interactive session. counter can manipulated via functions reset_rng_state(), set_rng_state(), get_rng_state(). rest vignette showcase interacting counter state, useful practice.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/reproducibility.html","id":"behavior-of-rng-counter","dir":"Articles","previous_headings":"","what":"Behavior of RNG counter","title":"Reproducibility","text":"start fresh session, counter initialized zero. specify seed, ’s one: start simple clusterpermute() 100 simulations. reproducibility profile run seed 1 counter 0: run, counter now incremented 3712: call clusterpermute() now runs new counter value, produces different result. , expected given stochastic nature CPA (thus need run tens thousands simulations reporting results): second run also increments counter state, note ’s exactly amount (possible predict): difference two runs specifically null distribution cluster-mass statistics. 100-simulation samples drawn share underlying (null) distribution, though identical one another (rarely ).","code":"get_rng_state() #> [1] 0 get_rng_seed() #> [1] 1 CPA_1_0 <- clusterpermute(jlmer_spec, threshold = 1, nsim = 100, progress = FALSE) CPA_1_0 #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 100) #>   Mean (SD): 0.724 (5.54) #>   Coverage intervals: 95% [-10.847, 12.592] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 (p=0.0198) #> ──────────────────────────────────────────────────────────────────────────────── get_rng_state() #> [1] 3712 CPA_1_3712 <- clusterpermute(jlmer_spec, threshold = 1, nsim = 100, progress = FALSE) CPA_1_3712 #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 100) #>   Mean (SD): -0.618 (8.51) #>   Coverage intervals: 95% [-18.328, 19.076] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 (p=0.0792) #> ──────────────────────────────────────────────────────────────────────────────── get_rng_state() #> [1] 7427 ggplot(mapping = aes(sum_statistic)) +   geom_density(     aes(fill = \"1\"),     alpha = .5,     data = tidy(CPA_1_0$null_cluster_dists)   ) +   geom_density(     aes(fill = \"3712\"),     alpha = .5,     data = tidy(CPA_1_3712$null_cluster_dists)   ) +   guides(fill = guide_legend(\"Counter\")) +   theme_classic()"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/reproducibility.html","id":"manipulating-rng-counter","dir":"Articles","previous_headings":"","what":"Manipulating RNG counter","title":"Reproducibility","text":"Using set_rng_counter(), possible reproduce prior run. example, reproduce second clusterpermute() call, set counter back 3712 set_rng_state(): results subsequent run counter state identical previous run: reproduce first CPA ran counter 0, can use set_rng_state() simply use reset_rng_state() alias set_rng_state(0): behavior can observed specific step simulating null distribution well. function corresponding step permute_timewise_statistics() run 100 simulations step counter zero: Running cluster detection algorithm result using extract_null_cluster_dists() returns null cluster-mass distribution identical CPA_1_0:","code":"set_rng_state(3712) get_rng_state() #> [1] 3712 identical(   CPA_1_3712,   clusterpermute(jlmer_spec, threshold = 1, nsim = 100, progress = FALSE) ) #> [1] TRUE reset_rng_state() get_rng_state() #> [1] 0 identical(   CPA_1_0,   clusterpermute(jlmer_spec, threshold = 1, nsim = 100, progress = FALSE) ) #> [1] TRUE reset_rng_state() get_rng_state() #> [1] 0 null_statistics <- permute_timewise_statistics(jlmer_spec, nsim = 100) get_rng_state() #> [1] 3712 null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 1) null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 100) #>   Mean (SD): 0.724 (5.54) #>   Coverage intervals: 95% [-10.847, 12.592] #> ──────────────────────────────────────────────────────────────────────────────── identical(null_cluster_dists, CPA_1_0$null_cluster_dists) #> [1] TRUE"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/reproducibility.html","id":"technical-details-and-notes","dir":"Articles","previous_headings":"","what":"Technical details and notes","title":"Reproducibility","text":"Julia back-end uses thread-safe, counter-based RNG called Threefry (specifically, Random123.Threefry2x) comes Random123.jl Julia library adaptation original algorithm developed Salmon, Moraes, Dror, Shaw (2011). RNG initialized “rounds” value 20, also default. Threefry used speed stability across Julia versions. Additionally, noted earlier, RNG initializes explicit seed 1. behavior different case R, random seed chosen user every new session. consequence absence explicit seeding via options(\"jlmerclusterperm.seed\"), first run every new session (.e., runs counter value zero) identical. Given interface RNG counter, many unique advantages using different seed manipulating RNG counter. interface RNG seed also provided via set_rng_seed() get_rng_seed(). Lastly, please aware changing seed R session (e.g., via set.seed()) effect jlmerclusterperm functions powered Julia back-end.","code":"# Restart Julia session jlmerclusterperm_setup(restart = TRUE, verbose = FALSE)  # RNG seed is again 1 get_rng_seed() #> [1] 1  # So results of the first CPA of the session (with counter = 0) is the same get_rng_state() #> [1] 0 identical(   CPA_1_0,   clusterpermute(jlmer_spec, threshold = 1, nsim = 100, progress = FALSE) ) #> [1] TRUE"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/reproducibility.html","id":"miscellaneous","dir":"Articles","previous_headings":"","what":"Miscellaneous","title":"Reproducibility","text":"may wondering whether two CPAs overlap range counter values also share overlapping samples null. actually case practice shuffling CPA implemented:  two colored lines represent two 100-simulation bootstrap permutations differ initial RNG counter value 1. point shape two lines look different - one just shifted form another. formal investigations (e.g., simulating power CPA), ’s course best practice generate random seeds. counter convenience balances interactivity reproducibility.","code":"# 100 simulations of timewise statistics ## First CPA using counter at 1 set_rng_state(1) x <- permute_timewise_statistics(jlmer_spec, nsim = 100) ## Second CPA using counter at 2 set_rng_state(2) y <- permute_timewise_statistics(jlmer_spec, nsim = 100)  # The null cluster-mass statistics x_null <- extract_null_cluster_dists(x, threshold = 1) x_null_stats <- tidy(x_null)$sum_statistic y_null <- extract_null_cluster_dists(y, threshold = 1) y_null_stats <- tidy(y_null)$sum_statistic  # Relationship between two CPAs with similar initial counters matplot(cbind(x_null_stats, y_null_stats), type = \"b\", pch = 1)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Tidying output","text":"Minimal example data:  jlmer specification object used showcase:","code":"library(jlmerclusterperm) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) library(dplyr, warn.conflicts = FALSE) library(ggplot2) chickweights <- as_tibble(ChickWeight) %>%   mutate(diet = forcats::fct_collapse(Diet, A = c(1, 3), B = c(2, 4))) %>%   filter(Time %in% 0:20) ggplot(chickweights, aes(Time, weight, color = diet)) +   stat_summary(fun.args = list(mult = 1.96)) +   stat_summary(geom = \"line\") +   theme_minimal(base_size = 18) jlmer_spec <- make_jlmer_spec(   formula = weight ~ diet + (1 | Chick), data = chickweights,   subject = \"Chick\", time = \"Time\" ) jlmer_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + dietB + (1 | Chick) #> Predictors: #>   diet: dietB #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 4 #>   weight dietB Chick  Time #>    <dbl> <dbl> <chr> <dbl> #> 1     42     0 1         0 #> 2     51     0 1         2 #> 3     59     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"a-tour-of-tidy-methods","dir":"Articles","previous_headings":"","what":"A tour of tidy() methods","title":"Tidying output","text":"non-<data.frame> objects returned jlmerclusterperm functions tidy() method return data frame representation object.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"timewise_statistics-objects","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"<timewise_statistics> objects","title":"Tidying output","text":"Functions compute_timewise_statistics() permute_timewise_statistics() return object class <timewise_statistics>, array statistics. Note isn’t stylized print() method class. Instead, recommended use tidy() inspect results data frame statistic represented row. case compute_timewise_statistics(), output matrix (2D array) predictor time. dimension becomes column tidy() data: case compute_timewise_statistics(), output 3D array simulation, time, predictor. , dimension becomes column tidy() data:","code":"empirical_statistics <- compute_timewise_statistics(jlmer_spec) class(empirical_statistics) #> [1] \"timewise_statistics\" dim(empirical_statistics) # predictor x time #> [1]  1 11 tidy(empirical_statistics) #> # A tibble: 11 × 3 #>    predictor  time statistic #>    <chr>     <dbl>     <dbl> #>  1 dietB         0    -1.09  #>  2 dietB         2     2.29  #>  3 dietB         4     3.14  #>  4 dietB         6     4.01  #>  5 dietB         8     2.90  #>  6 dietB        10     2.44  #>  7 dietB        12     2.18  #>  8 dietB        14     1.26  #>  9 dietB        16     0.673 #> 10 dietB        18     0.532 #> 11 dietB        20     0.809 null_statistics <- permute_timewise_statistics(jlmer_spec, nsim = 1000) class(null_statistics) #> [1] \"timewise_statistics\" dim(null_statistics) # simulation x time x predictor #> [1] 1000   11    1 tidy(null_statistics) #> # A tibble: 11,000 × 4 #>    predictor  time statistic sim   #>    <chr>     <dbl>     <dbl> <fct> #>  1 dietB         0   -0.309  0001  #>  2 dietB         0    2.10   0002  #>  3 dietB         0   -2.83   0003  #>  4 dietB         0   -0.568  0004  #>  5 dietB         0    2.10   0005  #>  6 dietB         0   -1.09   0006  #>  7 dietB         0   -0.830  0007  #>  8 dietB         0   -0.0515 0008  #>  9 dietB         0   -0.0515 0009  #> 10 dietB         0    0.206  0010  #> # ℹ 10,990 more rows"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"empirical_clusters-objects","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"<empirical_clusters> objects","title":"Tidying output","text":"<empirical_clusters> object returned extract_empirical_clusters() print method defined, makes easier eye: tidy() data frame representation also contains information like length, number time points cluster spans. id variable uniquely identifies cluster within predictor.","code":"empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 1) class(empirical_clusters) #> [1] \"empirical_clusters\" empirical_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 #> ──────────────────────────────────────────────────────────────────────────────── tidy(empirical_clusters) #> # A tibble: 1 × 6 #>   predictor id    start   end length sum_statistic #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl> #> 1 dietB     1         2    14      7          18.2"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"null_cluster_dists-objects","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"<null_cluster_dists> objects","title":"Tidying output","text":"<null_cluster_dists> object returned extract_null_cluster_dists() also print method defined, displays summary statistics: null distribution constructed largest cluster-mass statistic observed bootstrapped permutations original data. words, simulation contributes one sample null. tidy() representation, simulation predictor represents row. sum_statistic column represents size largest cluster simulation. clusters found, column zero values start, end, length NA. output tidy(null_statistics) tidy(null_cluster_dists) form relational data. example, can use left_join() find time wise statistics permuted data extreme cluster-mass statistic: Note sum_statistic simply sum statistic values span largest cluster:","code":"null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 1) class(null_cluster_dists) #> [1] \"null_cluster_dists\" null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 1000) #>   Mean (SD): 0.014 (7.11) #>   Coverage intervals: 95% [-15.360, 15.902] #> ──────────────────────────────────────────────────────────────────────────────── tidy(null_cluster_dists) #> # A tibble: 1,000 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 dietB         8    20      7         11.1  0001  #>  2 dietB         6    10      3          3.75 0002  #>  3 dietB        18    20      2          2.07 0003  #>  4 dietB         2    10      5          6.00 0004  #>  5 dietB         4    10      4          5.91 0005  #>  6 dietB        NA    NA     NA          0    0006  #>  7 dietB         8    20      7         12.2  0007  #>  8 dietB         2     8      4          5.29 0008  #>  9 dietB        NA    NA     NA          0    0009  #> 10 dietB        NA    NA     NA          0    0010  #> # ℹ 990 more rows largest_permuted_cluster <- tidy(null_cluster_dists) %>%   slice(which.max(abs(sum_statistic))) %>%   left_join(tidy(null_statistics), by = c(\"predictor\", \"sim\")) largest_permuted_cluster #> # A tibble: 11 × 8 #>    predictor start   end length sum_statistic sim    time statistic #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> <dbl>     <dbl> #>  1 dietB         2    20     10          29.4 0319      0     -1.09 #>  2 dietB         2    20     10          29.4 0319      2      1.67 #>  3 dietB         2    20     10          29.4 0319      4      2.56 #>  4 dietB         2    20     10          29.4 0319      6      2.97 #>  5 dietB         2    20     10          29.4 0319      8      3.47 #>  6 dietB         2    20     10          29.4 0319     10      3.54 #>  7 dietB         2    20     10          29.4 0319     12      3.49 #>  8 dietB         2    20     10          29.4 0319     14      3.54 #>  9 dietB         2    20     10          29.4 0319     16      3.05 #> 10 dietB         2    20     10          29.4 0319     18      2.65 #> 11 dietB         2    20     10          29.4 0319     20      2.48 largest_permuted_cluster %>%   filter(between(time, start, end)) %>%   summarize(sum_statistic = unique(sum_statistic), sum(statistic)) #> # A tibble: 1 × 2 #>   sum_statistic `sum(statistic)` #>           <dbl>            <dbl> #> 1          29.4             29.4"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"in-other-contexts","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"In other contexts","title":"Tidying output","text":"output calculate_clusters_pvalues() just <empirical_clusters> object augmented p-values: available, p-values returned column tidy() data: Note p-value cluster 0.027 27 samples null extreme observed: case one-fell-swoop function clusterpermute(), <null_cluster_dists> object <empirical_clusters> object returned list: can tidy() elements list lapply() purrr::map():","code":"tested_clusters <- calculate_clusters_pvalues(empirical_clusters, null_cluster_dists) tested_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 (p=0.0270) #> ──────────────────────────────────────────────────────────────────────────────── tidy(tested_clusters) #> # A tibble: 1 × 7 #>   predictor id    start   end length sum_statistic pvalue #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>  <dbl> #> 1 dietB     1         2    14      7          18.2  0.027 tidy(null_cluster_dists) %>%   filter(abs(sum_statistic) >= abs(tidy(tested_clusters)$sum_statistic)) #> # A tibble: 27 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 dietB         0    20     11         -18.7 0072  #>  2 dietB         0    20     11         -19.4 0114  #>  3 dietB         6    20      8          19.0 0118  #>  4 dietB         0    20     11         -23.6 0137  #>  5 dietB         4    20      9         -21.7 0142  #>  6 dietB         2    20     10          24.7 0268  #>  7 dietB         4    20      9          22.7 0289  #>  8 dietB         2    20     10          29.4 0319  #>  9 dietB         2    20     10          26.4 0325  #> 10 dietB         4    20      9          26.5 0384  #> # ℹ 17 more rows # Timing the entire process this time system.time({   full_test <- clusterpermute(jlmer_spec, threshold = 1, nsim = 1000, progress = FALSE) }) #>    user  system elapsed  #>    0.02    0.00    2.14 full_test #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 1000) #>   Mean (SD): 0.066 (7.61) #>   Coverage intervals: 95% [-16.959, 17.216] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 (p=0.0420) #> ──────────────────────────────────────────────────────────────────────────────── lapply(full_test, tidy) #> $null_cluster_dists #> # A tibble: 1,000 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 dietB        14    20      4         -5.26 0001  #>  2 dietB        14    20      4          9.22 0002  #>  3 dietB         6    20      8         11.8  0003  #>  4 dietB        NA    NA     NA          0    0004  #>  5 dietB        NA    NA     NA          0    0005  #>  6 dietB        16    20      3         -3.69 0006  #>  7 dietB        14    20      4          6.23 0007  #>  8 dietB        NA    NA     NA          0    0008  #>  9 dietB        12    20      5          5.38 0009  #> 10 dietB        NA    NA     NA          0    0010  #> # ℹ 990 more rows #>  #> $empirical_clusters #> # A tibble: 1 × 7 #>   predictor id    start   end length sum_statistic pvalue #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>  <dbl> #> 1 dietB     1         2    14      7          18.2 0.0420"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying-output.html","id":"tidying-julia-model-objects","dir":"Articles","previous_headings":"","what":"Tidying Julia model objects","title":"Tidying output","text":"Functions jlmer() to_jlmer() return object class <jlmer_mod>, JuliaConnectoR pointer Julia model object: can use tidy() glance() return information Julia model data frame: mimic behavior methods broom/broom.mixed defined (g)lm/(g)lmer models: Note Julia MixedModels.jl defaults REML = FALSE, whereas TRUE default lme4.","code":"jlmer_mod <- to_jlmer(weight ~ diet + (1 | Chick), chickweights) class(jlmer_mod) #> [1] \"jlmer_mod\"        \"JuliaStructProxy\" \"JuliaProxy\" jlmer_mod #> <Julia object of type LinearMixedModel> #> Variance components: #>             Column    Variance Std.Dev. #> Chick    (Intercept)   305.1809 17.4694 #> Residual              3854.0809 62.0812 #>  ────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|) #> ────────────────────────────────────────────────── #> (Intercept)  108.16       4.76331  22.71    <1e-99 #> dietB         12.2644     7.45319   1.65    0.0999 #> ────────────────────────────────────────────────── tidy(jlmer_mod) #> # A tibble: 4 × 7 #>   effect   group    term            estimate std.error statistic    p.value #>   <chr>    <chr>    <chr>              <dbl>     <dbl>     <dbl>      <dbl> #> 1 fixed    NA       (Intercept)        108.       4.76     22.7   3.83e-114 #> 2 fixed    NA       dietB               12.3      7.45      1.65  9.99e-  2 #> 3 ran_pars Chick    sd__(Intercept)     17.5     NA        NA    NA         #> 4 ran_pars Residual sd__Observation     62.1     NA        NA    NA     glance(jlmer_mod) #> # A tibble: 1 × 8 #>    nobs    df sigma logLik   AIC   BIC deviance df.residual #>   <int> <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <int> #> 1   533     4  62.1 -2972. 5952. 5969.    5944.         529 library(lme4) library(broom.mixed) lmer_mod <- lmer(weight ~ diet + (1 | Chick), chickweights, REML = FALSE) tidy(lmer_mod) #> # A tibble: 4 × 6 #>   effect   group    term            estimate std.error statistic #>   <chr>    <chr>    <chr>              <dbl>     <dbl>     <dbl> #> 1 fixed    NA       (Intercept)        108.       4.76     22.7  #> 2 fixed    NA       dietB               12.3      7.45      1.65 #> 3 ran_pars Chick    sd__(Intercept)     17.5     NA        NA    #> 4 ran_pars Residual sd__Observation     62.1     NA        NA glance(lmer_mod) #> # A tibble: 1 × 7 #>    nobs sigma logLik   AIC   BIC deviance df.residual #>   <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <int> #> 1   533  62.1 -2972. 5952. 5969.    5944.         529"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Tidying output","text":"Minimal example data:  jlmer specification object used vignette:","code":"library(jlmerclusterperm) system.time(jlmerclusterperm_setup(verbose = FALSE)) #>    user  system elapsed  #>    0.03    0.03   24.11 julia_progress(show = FALSE) library(dplyr, warn.conflicts = FALSE) library(ggplot2) chickweights <- as_tibble(ChickWeight) %>%    mutate(diet = forcats::fct_collapse(Diet, A = c(1, 3), B = c(2, 4))) %>%    filter(Time %in% 0:20) ggplot(chickweights, aes(Time, weight, color = diet)) +   stat_summary(fun.args = list(mult = 1.96)) +   stat_summary(geom = \"line\") +   theme_minimal(base_size = 18) jlmer_spec <- make_jlmer_spec(   formula = weight ~ diet + (1 | Chick), data = chickweights,   subject = \"Chick\", time = \"Time\" ) jlmer_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + dietB + (1 | Chick) #> Predictors: #>   diet: dietB #> Specials: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 4 #>   weight dietB Chick  Time #>    <dbl> <dbl> <chr> <dbl> #> 1     42     0 1         0 #> 2     51     0 1         2 #> 3     59     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"a-tour-of-tidy-methods","dir":"Articles","previous_headings":"","what":"A tour of tidy() methods","title":"Tidying output","text":"non-<data.frame> objects returned jlmerclusterperm functions tidy() method return data frame representation object.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"timewise_statistics-objects","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"<timewise_statistics> objects","title":"Tidying output","text":"Functions compute_timewise_statistics() permute_timewise_statistics() return object class <timewise_statistics>, array statistics. Note isn’t stylized print() method class. Instead, recommended use tidy() inspect results data frame statistic represented row. case compute_timewise_statistics(), output matrix (2D array) predictor time. dimension becomes column tidy() data: case compute_timewise_statistics(), output 3D array simulation, time, predictor. , dimension becomes column tidy() data:","code":"empirical_statistics <- compute_timewise_statistics(jlmer_spec) class(empirical_statistics) #> [1] \"timewise_statistics\" dim(empirical_statistics) # predictor x time #> [1]  1 11 tidy(empirical_statistics) #> # A tibble: 11 × 3 #>    predictor  time statistic #>    <chr>     <dbl>     <dbl> #>  1 dietB         0    -1.09  #>  2 dietB         2     2.29  #>  3 dietB         4     3.14  #>  4 dietB         6     4.01  #>  5 dietB         8     2.90  #>  6 dietB        10     2.44  #>  7 dietB        12     2.18  #>  8 dietB        14     1.26  #>  9 dietB        16     0.673 #> 10 dietB        18     0.532 #> 11 dietB        20     0.809 null_statistics <- permute_timewise_statistics(jlmer_spec, nsim = 1000) class(null_statistics) #> [1] \"timewise_statistics\" dim(null_statistics) # simulation x time x predictor #> [1] 1000   11    1 tidy(null_statistics) #> # A tibble: 11,000 × 4 #>    predictor  time statistic sim   #>    <chr>     <dbl>     <dbl> <fct> #>  1 dietB         0   -0.309  0001  #>  2 dietB         0    2.10   0002  #>  3 dietB         0   -2.83   0003  #>  4 dietB         0   -0.568  0004  #>  5 dietB         0    2.10   0005  #>  6 dietB         0   -1.09   0006  #>  7 dietB         0   -0.830  0007  #>  8 dietB         0   -0.0515 0008  #>  9 dietB         0   -0.0515 0009  #> 10 dietB         0    0.206  0010  #> # ℹ 10,990 more rows"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"empirical_clusters-objects","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"<empirical_clusters> objects","title":"Tidying output","text":"<empirical_clusters> object returned extract_empirical_clusters() print method defined, makes easier eye: tidy() data frame representation also contains information like length, number time points cluster spans. id variable uniquely identifies cluster within predictor.","code":"empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 1) class(empirical_clusters) #> [1] \"empirical_clusters\" empirical_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 #> ──────────────────────────────────────────────────────────────────────────────── tidy(empirical_clusters) #> # A tibble: 1 × 6 #>   predictor id    start   end length sum_statistic #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl> #> 1 dietB     1         2    14      7          18.2"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"null_cluster_dists-objects","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"<null_cluster_dists> objects","title":"Tidying output","text":"<null_cluster_dists> object returned extract_null_cluster_dists() also print method defined, displays summary statistics: null distribution constructed largest cluster-mass statistic observed bootstrapped permutations original data. words, simulation contributes one sample null. tidy() representation, simulation predictor represents row. sum_statistic column represents size largest cluster simulation. clusters found, column zero values start, end, length NA. output tidy(null_statistics) tidy(null_cluster_dists) form relational data. example, can use left_join() find time wise statistics permuted data extreme cluster-mass statistic: Note sum_statistic simply sum statistic values span largest cluster:","code":"null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 1) class(null_cluster_dists) #> [1] \"null_cluster_dists\" null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 1000) #>   Mean (SD): 0.014 (7.11) #>   Coverage intervals: 95% [-15.360, 15.902] #> ──────────────────────────────────────────────────────────────────────────────── tidy(null_cluster_dists) #> # A tibble: 1,000 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 dietB         8    20      7         11.1  0001  #>  2 dietB         6    10      3          3.75 0002  #>  3 dietB        18    20      2          2.07 0003  #>  4 dietB         2    10      5          6.00 0004  #>  5 dietB         4    10      4          5.91 0005  #>  6 dietB        NA    NA     NA          0    0006  #>  7 dietB         8    20      7         12.2  0007  #>  8 dietB         2     8      4          5.29 0008  #>  9 dietB        NA    NA     NA          0    0009  #> 10 dietB        NA    NA     NA          0    0010  #> # ℹ 990 more rows largest_permuted_cluster <- tidy(null_cluster_dists) %>%    slice(which.max(abs(sum_statistic))) %>%    left_join(tidy(null_statistics), by = c(\"predictor\", \"sim\")) largest_permuted_cluster #> # A tibble: 11 × 8 #>    predictor start   end length sum_statistic sim    time statistic #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> <dbl>     <dbl> #>  1 dietB         2    20     10          29.4 0319      0     -1.09 #>  2 dietB         2    20     10          29.4 0319      2      1.67 #>  3 dietB         2    20     10          29.4 0319      4      2.56 #>  4 dietB         2    20     10          29.4 0319      6      2.97 #>  5 dietB         2    20     10          29.4 0319      8      3.47 #>  6 dietB         2    20     10          29.4 0319     10      3.54 #>  7 dietB         2    20     10          29.4 0319     12      3.49 #>  8 dietB         2    20     10          29.4 0319     14      3.54 #>  9 dietB         2    20     10          29.4 0319     16      3.05 #> 10 dietB         2    20     10          29.4 0319     18      2.65 #> 11 dietB         2    20     10          29.4 0319     20      2.48 largest_permuted_cluster %>%    filter(between(time, start, end)) %>%    summarize(sum_statistic = unique(sum_statistic), sum(statistic)) #> # A tibble: 1 × 2 #>   sum_statistic `sum(statistic)` #>           <dbl>            <dbl> #> 1          29.4             29.4"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"in-other-contexts","dir":"Articles","previous_headings":"A tour of tidy() methods","what":"In other contexts","title":"Tidying output","text":"output calculate_clusters_pvalues() just <empirical_clusters> object augmented p-values: available, p-values returned column tidy() data: Note p-value cluster 0.027 27 samples null extreme observed: case one-fell-swoop function clusterpermute(), <null_cluster_dists> object <empirical_clusters> object returned list: can tidy() elements list lapply() purrr::map():","code":"tested_clusters <- calculate_clusters_pvalues(empirical_clusters, null_cluster_dists) tested_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 (p=0.0270) #> ──────────────────────────────────────────────────────────────────────────────── tidy(tested_clusters) #> # A tibble: 1 × 7 #>   predictor id    start   end length sum_statistic pvalue #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>  <dbl> #> 1 dietB     1         2    14      7          18.2  0.027 tidy(null_cluster_dists) %>%    filter(abs(sum_statistic) >= abs(tidy(tested_clusters)$sum_statistic)) #> # A tibble: 27 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 dietB         0    20     11         -18.7 0072  #>  2 dietB         0    20     11         -19.4 0114  #>  3 dietB         6    20      8          19.0 0118  #>  4 dietB         0    20     11         -23.6 0137  #>  5 dietB         4    20      9         -21.7 0142  #>  6 dietB         2    20     10          24.7 0268  #>  7 dietB         4    20      9          22.7 0289  #>  8 dietB         2    20     10          29.4 0319  #>  9 dietB         2    20     10          26.4 0325  #> 10 dietB         4    20      9          26.5 0384  #> # ℹ 17 more rows # Timing it for fun this time system.time({   full_test <- clusterpermute(jlmer_spec, threshold = 1, nsim = 1000, progress = FALSE) }) #>    user  system elapsed  #>    0.00    0.01    6.53 full_test$null_cluster_dists #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> dietB (n = 1000) #>   Mean (SD): 0.066 (7.61) #>   Coverage intervals: 95% [-16.959, 17.216] #> ──────────────────────────────────────────────────────────────────────────────── full_test$empirical_clusters #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> dietB #>   [2, 14]: 18.231 (p=0.0420) #> ──────────────────────────────────────────────────────────────────────────────── lapply(full_test, tidy) #> $null_cluster_dists #> # A tibble: 1,000 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 dietB        14    20      4         -5.26 0001  #>  2 dietB        14    20      4          9.22 0002  #>  3 dietB         6    20      8         11.8  0003  #>  4 dietB        NA    NA     NA          0    0004  #>  5 dietB        NA    NA     NA          0    0005  #>  6 dietB        16    20      3         -3.69 0006  #>  7 dietB        14    20      4          6.23 0007  #>  8 dietB        NA    NA     NA          0    0008  #>  9 dietB        12    20      5          5.38 0009  #> 10 dietB        NA    NA     NA          0    0010  #> # ℹ 990 more rows #>  #> $empirical_clusters #> # A tibble: 1 × 7 #>   predictor id    start   end length sum_statistic pvalue #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>  <dbl> #> 1 dietB     1         2    14      7          18.2 0.0420"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/articles/tidying_output.html","id":"tidying-julia-model-objects","dir":"Articles","previous_headings":"","what":"Tidying Julia model objects","title":"Tidying output","text":"Functions jlmer() to_jlmer() return object class <jlmer_mod>, JuliaConnectoR pointer Julia model object: can use tidy() glance() return information Julia model data frame: mimic behavior methods broom/broom.mixed defined (g)lm/(g)lmer models: Note Julia MixedModels.jl defaults REML = FALSE, whereas TRUE default lme4.","code":"jlmer_mod <- to_jlmer(weight ~ diet + (1 | Chick), chickweights) class(jlmer_mod) #> [1] \"jlmer_mod\"        \"JuliaStructProxy\" \"JuliaProxy\" jlmer_mod #> <Julia object of type LinearMixedModel{Float64}> #> Variance components: #>             Column    Variance Std.Dev. #> Chick    (Intercept)   305.1809 17.4694 #> Residual              3854.0809 62.0812 #>  ────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|) #> ────────────────────────────────────────────────── #> (Intercept)  108.16       4.76331  22.71    <1e-99 #> dietB         12.2644     7.45319   1.65    0.0999 #> ────────────────────────────────────────────────── tidy(jlmer_mod) #> # A tibble: 4 × 7 #>   effect   group    term            estimate std.error statistic    p.value #>   <chr>    <chr>    <chr>              <dbl>     <dbl>     <dbl>      <dbl> #> 1 fixed    NA       (Intercept)        108.       4.76     22.7   3.83e-114 #> 2 fixed    NA       dietB               12.3      7.45      1.65  9.99e-  2 #> 3 ran_pars Chick    sd__(Intercept)     17.5     NA        NA    NA         #> 4 ran_pars Residual sd__Observation     62.1     NA        NA    NA     glance(jlmer_mod) #> # A tibble: 1 × 7 #>    nobs sigma logLik   AIC   BIC deviance df.residual #>   <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <int> #> 1   533  62.1 -2972. 5952. 5969.    5944.         529 library(lme4) library(broom.mixed) lmer_mod <- lmer(weight ~ diet + (1 | Chick), chickweights, REML = FALSE) tidy(lmer_mod) #> # A tibble: 4 × 6 #>   effect   group    term            estimate std.error statistic #>   <chr>    <chr>    <chr>              <dbl>     <dbl>     <dbl> #> 1 fixed    NA       (Intercept)        108.       4.76     22.7  #> 2 fixed    NA       dietB               12.3      7.45      1.65 #> 3 ran_pars Chick    sd__(Intercept)     17.5     NA        NA    #> 4 ran_pars Residual sd__Observation     62.1     NA        NA glance(lmer_mod) #> # A tibble: 1 × 7 #>    nobs sigma logLik   AIC   BIC deviance df.residual #>   <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <int> #> 1   533  62.1 -2972. 5952. 5969.    5944.         529"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/authors.html","id":null,"dir":"","previous_headings":"","what":"Authors","title":"Authors and Citation","text":"June Choe. Author, maintainer, copyright holder.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/authors.html","id":"citation","dir":"","previous_headings":"","what":"Citation","title":"Authors and Citation","text":"Choe J (2024). jlmerclusterperm: Cluster-Based Permutation Analysis Densely Sampled Time Data. https://github.com/yjunechoe/jlmerclusterperm, https://yjunechoe.github.io/jlmerclusterperm/.","code":"@Manual{,   title = {jlmerclusterperm: Cluster-Based Permutation Analysis for Densely Sampled Time Data},   author = {June Choe},   year = {2024},   note = {https://github.com/yjunechoe/jlmerclusterperm, https://yjunechoe.github.io/jlmerclusterperm/}, }"},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"our-pledge","dir":"","previous_headings":"","what":"Our Pledge","title":"Contributor Covenant Code of Conduct","text":"members, contributors, leaders pledge make participation community harassment-free experience everyone, regardless age, body size, visible invisible disability, ethnicity, sex characteristics, gender identity expression, level experience, education, socio-economic status, nationality, personal appearance, race, caste, color, religion, sexual identity orientation. pledge act interact ways contribute open, welcoming, diverse, inclusive, healthy community.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"our-standards","dir":"","previous_headings":"","what":"Our Standards","title":"Contributor Covenant Code of Conduct","text":"Examples behavior contributes positive environment community include: Demonstrating empathy kindness toward people respectful differing opinions, viewpoints, experiences Giving gracefully accepting constructive feedback Accepting responsibility apologizing affected mistakes, learning experience Focusing best just us individuals, overall community Examples unacceptable behavior include: use sexualized language imagery, sexual attention advances kind Trolling, insulting derogatory comments, personal political attacks Public private harassment Publishing others’ private information, physical email address, without explicit permission conduct reasonably considered inappropriate professional setting","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"enforcement-responsibilities","dir":"","previous_headings":"","what":"Enforcement Responsibilities","title":"Contributor Covenant Code of Conduct","text":"Community leaders responsible clarifying enforcing standards acceptable behavior take appropriate fair corrective action response behavior deem inappropriate, threatening, offensive, harmful. Community leaders right responsibility remove, edit, reject comments, commits, code, wiki edits, issues, contributions aligned Code Conduct, communicate reasons moderation decisions appropriate.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"scope","dir":"","previous_headings":"","what":"Scope","title":"Contributor Covenant Code of Conduct","text":"Code Conduct applies within community spaces, also applies individual officially representing community public spaces. Examples representing community include using official e-mail address, posting via official social media account, acting appointed representative online offline event.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"enforcement","dir":"","previous_headings":"","what":"Enforcement","title":"Contributor Covenant Code of Conduct","text":"Instances abusive, harassing, otherwise unacceptable behavior may reported community leaders responsible enforcement codeofconduct@rstudio.com. complaints reviewed investigated promptly fairly. community leaders obligated respect privacy security reporter incident.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"enforcement-guidelines","dir":"","previous_headings":"","what":"Enforcement Guidelines","title":"Contributor Covenant Code of Conduct","text":"Community leaders follow Community Impact Guidelines determining consequences action deem violation Code Conduct:","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"id_1-correction","dir":"","previous_headings":"Enforcement Guidelines","what":"1. Correction","title":"Contributor Covenant Code of Conduct","text":"Community Impact: Use inappropriate language behavior deemed unprofessional unwelcome community. Consequence: private, written warning community leaders, providing clarity around nature violation explanation behavior inappropriate. public apology may requested.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"id_2-warning","dir":"","previous_headings":"Enforcement Guidelines","what":"2. Warning","title":"Contributor Covenant Code of Conduct","text":"Community Impact: violation single incident series actions. Consequence: warning consequences continued behavior. interaction people involved, including unsolicited interaction enforcing Code Conduct, specified period time. includes avoiding interactions community spaces well external channels like social media. Violating terms may lead temporary permanent ban.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"id_3-temporary-ban","dir":"","previous_headings":"Enforcement Guidelines","what":"3. Temporary Ban","title":"Contributor Covenant Code of Conduct","text":"Community Impact: serious violation community standards, including sustained inappropriate behavior. Consequence: temporary ban sort interaction public communication community specified period time. public private interaction people involved, including unsolicited interaction enforcing Code Conduct, allowed period. Violating terms may lead permanent ban.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"id_4-permanent-ban","dir":"","previous_headings":"Enforcement Guidelines","what":"4. Permanent Ban","title":"Contributor Covenant Code of Conduct","text":"Community Impact: Demonstrating pattern violation community standards, including sustained inappropriate behavior, harassment individual, aggression toward disparagement classes individuals. Consequence: permanent ban sort public interaction within community.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CODE_OF_CONDUCT.html","id":"attribution","dir":"","previous_headings":"","what":"Attribution","title":"Contributor Covenant Code of Conduct","text":"Code Conduct adapted Contributor Covenant, version 2.1, available https://www.contributor-covenant.org/version/2/1/code_of_conduct.html. Community Impact Guidelines inspired [Mozilla’s code conduct enforcement ladder][https://github.com/mozilla/inclusion]. answers common questions code conduct, see FAQ https://www.contributor-covenant.org/faq. Translations available https://www.contributor-covenant.org/translations.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CONTRIBUTING.html","id":null,"dir":"","previous_headings":"","what":"Contributing to jlmerclusterperm","title":"Contributing to jlmerclusterperm","text":"outlines propose change jlmerclusterperm.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CONTRIBUTING.html","id":"fixing-typos","dir":"","previous_headings":"","what":"Fixing typos","title":"Contributing to jlmerclusterperm","text":"can fix typos, spelling mistakes, grammatical errors documentation directly using GitHub web interface, long changes made source file. generally means ’ll need edit roxygen2 comments .R, .Rd file. can find .R file generates .Rd reading comment first line.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CONTRIBUTING.html","id":"bigger-changes","dir":"","previous_headings":"","what":"Bigger changes","title":"Contributing to jlmerclusterperm","text":"want make bigger change, ’s good idea first file issue make sure someone team agrees ’s needed. ’ve found bug, please file issue illustrates bug minimal reprex (also help write unit test, needed).","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CONTRIBUTING.html","id":"pull-request-process","dir":"","previous_headings":"Bigger changes","what":"Pull request process","title":"Contributing to jlmerclusterperm","text":"Fork package clone onto computer. haven’t done , recommend using usethis::create_from_github(\"yjunechoe/jlmerclusterperm\", fork = TRUE). Install development dependencies devtools::install_dev_deps(), make sure package passes R CMD check running devtools::check(). R CMD check doesn’t pass cleanly, ’s good idea ask help continuing. Create Git branch pull request (PR). recommend using usethis::pr_init(\"brief-description--change\"). Make changes, commit git, create PR running usethis::pr_push(), following prompts browser. title PR briefly describe change. body PR contain Fixes #issue-number. user-facing changes, add bullet top NEWS.md (.e. just first header). Follow style described https://style.tidyverse.org/news.html.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CONTRIBUTING.html","id":"code-style","dir":"","previous_headings":"Bigger changes","what":"Code style","title":"Contributing to jlmerclusterperm","text":"New code follow tidyverse style guide. can use styler package apply styles, please don’t restyle code nothing PR. use roxygen2, Markdown syntax, documentation. use testthat unit tests. Contributions test cases included easier accept.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/CONTRIBUTING.html","id":"code-of-conduct","dir":"","previous_headings":"","what":"Code of Conduct","title":"Contributing to jlmerclusterperm","text":"Please note jlmerclusterperm project released Contributor Code Conduct. contributing project agree abide terms.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/index.html","id":"jlmerclusterperm-","dir":"","previous_headings":"","what":"jlmerclusterperm","title":"jlmerclusterperm","text":"Julia GLM.jl MixedModels.jl based implementation cluster-based permutation test time series data, powered JuliaConnectoR.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/index.html","id":"installation-and-usage","dir":"","previous_headings":"","what":"Installation and usage","title":"jlmerclusterperm","text":"Install released version jlmerclusterperm CRAN: install development version GitHub : Using jlmerclusterperm requires prior installation Julia programming language, can downloaded either official website using command line utility juliaup. Julia version >=1.8 required 1.9 higher preferred substantial speed improvements. using functions jlmerclusterperm, initial setup required via calling jlmerclusterperm_setup(). first call system install necessary dependencies (happens takes around 10-15 minutes). Subsequent calls jlmerclusterperm_setup() incur small overhead around 30 seconds, plus slight delays first-time function calls. pay front start-warm-costs get blazingly-fast functions package. See Get Started page package website background tutorials.","code":"install.packages(\"jlmerclusterperm\") # install.packages(\"remotes\") remotes::install_github(\"yjunechoe/jlmerclusterperm\") # Both lines must be run at the start of each new session library(jlmerclusterperm) jlmerclusterperm_setup()"},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/index.html","id":"wholesale-cpa-with-clusterpermute","dir":"","previous_headings":"Quick tour of package functionalities","what":"Wholesale CPA with clusterpermute()","title":"jlmerclusterperm","text":"time series data:  Preparing specification object make_jlmer_spec(): Cluster-based permutation test clusterpermute(): Including random effects:","code":"chickweights <- ChickWeight chickweights$Time <- as.integer(factor(chickweights$Time)) matplot(   tapply(chickweights$weight, chickweights[c(\"Time\", \"Diet\")], mean),   type = \"b\", lwd = 3, ylab = \"Weight\", xlab = \"Time\" ) chickweights_spec <- make_jlmer_spec(   formula = weight ~ 1 + Diet,   data = chickweights,   subject = \"Chick\", time = \"Time\" ) chickweights_spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #>   weight Diet2 Diet3 Diet4 Chick Time #> 1     42     0     0     0     1    1 #> 2     51     0     0     0     1    2 #> 3     59     0     0     0     1    3 #>  [ reached 'max' / getOption(\"max.print\") -- omitted 575 rows ] #> ──────────────────────────────────────────────────────────────────────────────── set_rng_state(123L) clusterpermute(   chickweights_spec,   threshold = 2.5,   nsim = 100 ) #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2.5) ──────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): -0.039 (1.89) #>   Coverage intervals: 95% [-2.862, 0.000] #> Diet3 (n = 100) #>   Mean (SD): -0.129 (2.02) #>   Coverage intervals: 95% [0.000, 0.000] #> Diet4 (n = 100) #>   Mean (SD): 0.296 (3.21) #>   Coverage intervals: 95% [0.000, 5.797] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2.5) ──────────────────────── <empirical_clusters> ── #> Diet2 #>   [3, 4]: 6.121 (p=0.0495) #> Diet3 #>   [3, 12]: 35.769 (p=0.0099) #> Diet4 #>   [2, 8]: 32.442 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── chickweights_re_spec <- make_jlmer_spec(   formula = weight ~ 1 + Diet + (1 | Chick),   data = chickweights,   subject = \"Chick\", time = \"Time\" ) set_rng_state(123L) clusterpermute(   chickweights_re_spec,   threshold = 2.5,   nsim = 100 )$empirical_clusters #> ── Empirical clusters (t > 2.5) ──────────────────────── <empirical_clusters> ── #> Diet2 #>   [3, 4]: 6.387 (p=0.0594) #> Diet3 #>   [2, 12]: 39.919 (p=0.0099) #> Diet4 #>   [2, 8]: 33.853 (p=0.0099) #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/index.html","id":"piecemeal-approach-to-cpa","dir":"","previous_headings":"Quick tour of package functionalities","what":"Piecemeal approach to CPA","title":"jlmerclusterperm","text":"Computing time-wise statistics observed data:  Identifying empirical clusters: Simulating null distribution: Significance testing cluster-mass statistic: Iterating range threshold values:","code":"empirical_statistics <- compute_timewise_statistics(chickweights_spec) matplot(t(empirical_statistics), type = \"b\", pch = 1, lwd = 3, ylab = \"t-statistic\") abline(h = 2.5, lty = 3) empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 2.5) empirical_clusters #> ── Empirical clusters (t > 2.5) ──────────────────────── <empirical_clusters> ── #> Diet2 #>   [3, 4]: 6.121 #> Diet3 #>   [3, 12]: 35.769 #> Diet4 #>   [2, 8]: 32.442 #> ──────────────────────────────────────────────────────────────────────────────── set_rng_state(123L) null_statistics <- permute_timewise_statistics(chickweights_spec, nsim = 100) null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 2.5) null_cluster_dists #> ── Null cluster-mass distribution (t > 2.5) ──────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): -0.039 (1.89) #>   Coverage intervals: 95% [-2.862, 0.000] #> Diet3 (n = 100) #>   Mean (SD): -0.129 (2.02) #>   Coverage intervals: 95% [0.000, 0.000] #> Diet4 (n = 100) #>   Mean (SD): 0.296 (3.21) #>   Coverage intervals: 95% [0.000, 5.797] #> ──────────────────────────────────────────────────────────────────────────────── calculate_clusters_pvalues(empirical_clusters, null_cluster_dists, add1 = TRUE) #> ── Empirical clusters (t > 2.5) ──────────────────────── <empirical_clusters> ── #> Diet2 #>   [3, 4]: 6.121 (p=0.0495) #> Diet3 #>   [3, 12]: 35.769 (p=0.0099) #> Diet4 #>   [2, 8]: 32.442 (p=0.0099) #> ──────────────────────────────────────────────────────────────────────────────── walk_threshold_steps(empirical_statistics, null_statistics, steps = c(2, 2.5, 3)) #>    threshold predictor id start end length sum_statistic     pvalue #> 1        2.0     Diet2  1     3   5      3      8.496376 0.07920792 #> 2        2.0     Diet3  1     2  12     11     38.216035 0.00990099 #> 3        2.0     Diet4  1     2  12     11     41.651468 0.00990099 #> 4        2.5     Diet2  1     3   4      2      6.121141 0.04950495 #> 5        2.5     Diet3  1     3  12     10     35.768957 0.00990099 #> 6        2.5     Diet4  1     2   8      7     32.442352 0.00990099 #> 31       3.0     Diet3  1     3   5      3     12.719231 0.00990099 #> 21       3.0     Diet3  2     9  12      4     14.037622 0.00990099 #> 41       3.0     Diet4  1     2   7      6     29.659402 0.00990099"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/index.html","id":"acknowledgments","dir":"","previous_headings":"","what":"Acknowledgments","title":"jlmerclusterperm","text":"paper Maris & Oostenveld (2007) originally proposed cluster-based permutation analysis. JuliaConnectoR package powering R interface Julia. Julia packages GLM.jl MixedModels.jl fast implementations (mixed effects) regression models. Existing implementations CPA R (permuco, permutes, etc.) whose designs inspired CPA interface jlmerclusterperm.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/index.html","id":"citations","dir":"","previous_headings":"","what":"Citations","title":"jlmerclusterperm","text":"use jlmerclusterperm cluster-based permutation test mixed-effects models research, please cite one () following see fit. cite jlmerclusterperm: Choe, J. (2023). jlmerclusterperm: Cluster-Based Permutation Analysis Densely Sampled Time Data. R package version 1.1.1. https://cran.r-project.org/package=jlmerclusterperm. cite cluster-based permutation test: Maris, E., & Oostenveld, R. (2007). Nonparametric statistical testing EEG- MEG-data. Journal Neuroscience Methods, 164, 177–190. doi: 10.1016/j.jneumeth.2007.03.024. cite Julia programming language: Bezanson, J., Edelman, ., Karpinski, S., & Shah, V. B. (2017). Julia: Fresh Approach Numerical Computing. SIAM Review, 59(1), 65–98. doi: 10.1137/141000671. cite GLM.jl MixedModels.jl Julia libraries, consult Zenodo pages: GLM: https://doi.org/10.5281/zenodo.3376013 MixedModels: https://zenodo.org/badge/latestdoi/9106942","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/LICENSE.html","id":null,"dir":"","previous_headings":"","what":"MIT License","title":"MIT License","text":"Copyright (c) 2023 June Choe Permission hereby granted, free charge, person obtaining copy software associated documentation files (“Software”), deal Software without restriction, including without limitation rights use, copy, modify, merge, publish, distribute, sublicense, /sell copies Software, permit persons Software furnished , subject following conditions: copyright notice permission notice shall included copies substantial portions Software. SOFTWARE PROVIDED “”, WITHOUT WARRANTY KIND, EXPRESS IMPLIED, INCLUDING LIMITED WARRANTIES MERCHANTABILITY, FITNESS PARTICULAR PURPOSE NONINFRINGEMENT. EVENT SHALL AUTHORS COPYRIGHT HOLDERS LIABLE CLAIM, DAMAGES LIABILITY, WHETHER ACTION CONTRACT, TORT OTHERWISE, ARISING , CONNECTION SOFTWARE USE DEALINGS SOFTWARE.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/calculate_clusters_pvalues.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate bootstrapped p-values of cluster-mass statistics — calculate_clusters_pvalues","title":"Calculate bootstrapped p-values of cluster-mass statistics — calculate_clusters_pvalues","text":"Calculate bootstrapped p-values cluster-mass statistics","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/calculate_clusters_pvalues.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate bootstrapped p-values of cluster-mass statistics — calculate_clusters_pvalues","text":"","code":"calculate_clusters_pvalues(   empirical_clusters,   null_cluster_dists,   add1 = FALSE )  clusters_are_comparable(empirical_clusters, null_cluster_dists, error = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/calculate_clusters_pvalues.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate bootstrapped p-values of cluster-mass statistics — calculate_clusters_pvalues","text":"empirical_clusters empirical_clusters object null_cluster_dists null_cluster_dists object add1 Whether add 1 numerator denominator calculating p-value. Use TRUE effectively count observed statistic part permuted null distribution (recommended larger nsim prior publishing results). error Whether throw error incompatible","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/calculate_clusters_pvalues.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate bootstrapped p-values of cluster-mass statistics — calculate_clusters_pvalues","text":"empirical_clusters object augmented p-values.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/calculate_clusters_pvalues.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculate bootstrapped p-values of cluster-mass statistics — calculate_clusters_pvalues","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Make empirical clusters empirical_statistics <- compute_timewise_statistics(spec) empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 2) empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 #> Diet3 #>   [2, 20]: 34.628 #> Diet4 #>   [2, 20]: 39.371 #> ────────────────────────────────────────────────────────────────────────────────  # Make null cluster-mass distribution reset_rng_state() null_statistics <- permute_timewise_statistics(spec, nsim = 100) null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 2)  # Significance test the empirical cluster(s) from each predictor against the simulated null calculate_clusters_pvalues(empirical_clusters, null_cluster_dists) #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 (p=0.0200) #> Diet3 #>   [2, 20]: 34.628 (p=0.0000) #> Diet4 #>   [2, 20]: 39.371 (p=0.0000) #> ────────────────────────────────────────────────────────────────────────────────  # Set `add1 = TRUE` to normalize by adding 1 to numerator and denominator calculate_clusters_pvalues(empirical_clusters, null_cluster_dists, add1 = TRUE) #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 (p=0.0297) #> Diet3 #>   [2, 20]: 34.628 (p=0.0099) #> Diet4 #>   [2, 20]: 39.371 (p=0.0099) #> ────────────────────────────────────────────────────────────────────────────────  # This sequence of procedures is equivalent to `clusterpermute()` reset_rng_state() clusterpermute(spec, threshold = 2, nsim = 100, progress = FALSE) #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): 0.266 (2.13) #>   Coverage intervals: 95% [0.000, 6.579] #> Diet3 (n = 100) #>   Mean (SD): -0.124 (2.75) #>   Coverage intervals: 95% [-4.777, 5.707] #> Diet4 (n = 100) #>   Mean (SD): 0.007 (2.50) #>   Coverage intervals: 95% [0.000, 0.000] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 (p=0.0297) #> Diet3 #>   [2, 20]: 34.628 (p=0.0099) #> Diet4 #>   [2, 20]: 39.371 (p=0.0099) #> ────────────────────────────────────────────────────────────────────────────────  # The empirical clusters and the null cluster-mass distribution must be comparable empirical_clusters2 <- extract_empirical_clusters(empirical_statistics, threshold = 3) # For example, below code errors because thresholds are different (2 vs. 3) try( calculate_clusters_pvalues(empirical_clusters2, null_cluster_dists) ) #> Error in clusters_are_comparable(empirical_clusters, null_cluster_dists,  :  #>   Cluster-mass statistics between empirical and null are not comparable. #> ℹ `empirical_clusters` and `null_cluster_dists` must share the same `statistic` #>   and `threshold`. #> ✖ threshold: empirical uses 3 but null uses 2.  # Check for compatibility with `clusters_are_comparable()` clusters_are_comparable(empirical_clusters, null_cluster_dists) #> [1] TRUE clusters_are_comparable(empirical_clusters2, null_cluster_dists) #> [1] FALSE  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/clusterpermute.html","id":null,"dir":"Reference","previous_headings":"","what":"Conduct a cluster-based permutation test — clusterpermute","title":"Conduct a cluster-based permutation test — clusterpermute","text":"Conduct cluster-based permutation test","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/clusterpermute.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Conduct a cluster-based permutation test — clusterpermute","text":"","code":"clusterpermute(   jlmer_spec,   family = c(\"gaussian\", \"binomial\"),   statistic = c(\"t\", \"chisq\"),   threshold,   nsim = 100L,   predictors = NULL,   binned = FALSE,   top_n = Inf,   add1 = TRUE,   ...,   progress = TRUE )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/clusterpermute.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Conduct a cluster-based permutation test — clusterpermute","text":"jlmer_spec Data prepped jlmer make_jlmer_spec() family GLM family. Currently supports \"gaussian\" \"binomial\". statistic Test statistic calculating cluster mass. Can one \"t\" (default) regression model output \"chisq\" likelihood ratio test (takes twice long calculate). threshold threshold value statistic must pass contribute cluster mass. Interpretation differs choice statistic (): statistic = \"t\", threshold t-value (beta/std.err) regression model. statistic = \"chisq\", threshold p-value chi-squared statistics likelihood ratio tests. nsim Number simulations description predictors (Optional) subset predictors test. Defaults NULL tests predictors. binned Whether data aggregated/collapsed time bins. Defaults FALSE, requires cluster span least two time points. TRUE, allows length-1 clusters exist. top_n many clusters return, order size cluster-mass statistic. Defaults Inf return detected clusters. add1 Whether add 1 numerator denominator calculating p-value. Use TRUE effectively count observed statistic part permuted null distribution (recommended larger nsim prior publishing results). ... Optional arguments passed Julia model fitting. Defaults fast = TRUE (family = \"binomial\") progress = FALSE. progress Defaults TRUE, prints progress step cluster permutation test.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/clusterpermute.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Conduct a cluster-based permutation test — clusterpermute","text":"list null_cluster_dists empirical_clusters p-values","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/clusterpermute.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Conduct a cluster-based permutation test — clusterpermute","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Should minimally provide `threshold` and `nsim`, in addition to the spec object reset_rng_state() CPA <- clusterpermute(spec, threshold = 2, nsim = 100, progress = FALSE) CPA #> $null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): 0.266 (2.13) #>   Coverage intervals: 95% [0.000, 6.579] #> Diet3 (n = 100) #>   Mean (SD): -0.124 (2.75) #>   Coverage intervals: 95% [-4.777, 5.707] #> Diet4 (n = 100) #>   Mean (SD): 0.007 (2.50) #>   Coverage intervals: 95% [0.000, 0.000] #> ──────────────────────────────────────────────────────────────────────────────── #>  #> $empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 (p=0.0297) #> Diet3 #>   [2, 20]: 34.628 (p=0.0099) #> Diet4 #>   [2, 20]: 39.371 (p=0.0099) #> ────────────────────────────────────────────────────────────────────────────────  # CPA is a list of `<null_cluster_dists>` and `<empirical_clusters>` objects sapply(CPA, class) #>   null_cluster_dists   empirical_clusters  #> \"null_cluster_dists\" \"empirical_clusters\"   # You can extract the individual components for further inspection CPA$null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): 0.266 (2.13) #>   Coverage intervals: 95% [0.000, 6.579] #> Diet3 (n = 100) #>   Mean (SD): -0.124 (2.75) #>   Coverage intervals: 95% [-4.777, 5.707] #> Diet4 (n = 100) #>   Mean (SD): 0.007 (2.50) #>   Coverage intervals: 95% [0.000, 0.000] #> ──────────────────────────────────────────────────────────────────────────────── CPA$empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 (p=0.0297) #> Diet3 #>   [2, 20]: 34.628 (p=0.0099) #> Diet4 #>   [2, 20]: 39.371 (p=0.0099) #> ────────────────────────────────────────────────────────────────────────────────  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/cluster_permutation_tidiers.html","id":null,"dir":"Reference","previous_headings":"","what":"Tidiers for cluster permutation test objects — cluster_permutation_tidiers","title":"Tidiers for cluster permutation test objects — cluster_permutation_tidiers","text":"Tidiers cluster permutation test objects","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/cluster_permutation_tidiers.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tidiers for cluster permutation test objects — cluster_permutation_tidiers","text":"","code":"# S3 method for timewise_statistics tidy(x, ...)  # S3 method for empirical_clusters tidy(x, ...)  # S3 method for null_cluster_dists tidy(x, ...)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/cluster_permutation_tidiers.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tidiers for cluster permutation test objects — cluster_permutation_tidiers","text":"x object class <timewise_statistics>, <empirical_clusters>, <null_cluster_dists> ... Unused","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/cluster_permutation_tidiers.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Tidiers for cluster permutation test objects — cluster_permutation_tidiers","text":"data frame","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/cluster_permutation_tidiers.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tidiers for cluster permutation test objects — cluster_permutation_tidiers","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Method for `<timewise_statistics>` empirical_statistics <- compute_timewise_statistics(spec) class(empirical_statistics) #> [1] \"timewise_statistics\" tidy(empirical_statistics) #> # A tibble: 33 × 3 #>    predictor  time statistic #>    <chr>     <dbl>     <dbl> #>  1 Diet2         0    -1.60  #>  2 Diet3         0    -1.37  #>  3 Diet4         0    -0.916 #>  4 Diet2         2     1.67  #>  5 Diet3         2     2.45  #>  6 Diet4         2     3.53  #>  7 Diet2         4     2.60  #>  8 Diet3         4     4.48  #>  9 Diet4         4     6.27  #> 10 Diet2         6     3.52  #> # ℹ 23 more rows  reset_rng_state() null_statistics <- permute_timewise_statistics(spec, nsim = 100) class(null_statistics) #> [1] \"timewise_statistics\" tidy(null_statistics) #> # A tibble: 3,300 × 4 #>    predictor  time statistic sim   #>    <chr>     <dbl>     <dbl> <fct> #>  1 Diet2         0    -0.898 001   #>  2 Diet2         0     1.40  002   #>  3 Diet2         0    -1.44  003   #>  4 Diet2         0    -0.449 004   #>  5 Diet2         0     1.38  005   #>  6 Diet2         0    -0.692 006   #>  7 Diet2         0     0.113 007   #>  8 Diet2         0     0.334 008   #>  9 Diet2         0     0.558 009   #> 10 Diet2         0    -0.449 010   #> # ℹ 3,290 more rows  # Method for `<empirical_clusters>` empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 2) class(empirical_clusters) #> [1] \"empirical_clusters\" tidy(empirical_clusters) #> # A tibble: 3 × 6 #>   predictor id    start   end length sum_statistic #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl> #> 1 Diet2     1         4     8      3          8.50 #> 2 Diet3     1         2    20     10         34.6  #> 3 Diet4     1         2    20     10         39.4   # Method for `<null_cluster_dists>` null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 2) class(null_cluster_dists) #> [1] \"null_cluster_dists\" tidy(null_cluster_dists) #> # A tibble: 300 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 Diet2        10    20      6          14.8 001   #>  2 Diet2        NA    NA     NA           0   002   #>  3 Diet2        NA    NA     NA           0   003   #>  4 Diet2        NA    NA     NA           0   004   #>  5 Diet2        NA    NA     NA           0   005   #>  6 Diet2        NA    NA     NA           0   006   #>  7 Diet2        NA    NA     NA           0   007   #>  8 Diet2        NA    NA     NA           0   008   #>  9 Diet2        NA    NA     NA           0   009   #> 10 Diet2        NA    NA     NA           0   010   #> # ℹ 290 more rows  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/compute_timewise_statistics.html","id":null,"dir":"Reference","previous_headings":"","what":"Fit Julia regression models to each time point of a time series data — compute_timewise_statistics","title":"Fit Julia regression models to each time point of a time series data — compute_timewise_statistics","text":"Fit Julia regression models time point time series data","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/compute_timewise_statistics.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fit Julia regression models to each time point of a time series data — compute_timewise_statistics","text":"","code":"compute_timewise_statistics(   jlmer_spec,   family = c(\"gaussian\", \"binomial\"),   statistic = c(\"t\", \"chisq\"),   ... )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/compute_timewise_statistics.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fit Julia regression models to each time point of a time series data — compute_timewise_statistics","text":"jlmer_spec Data prepped jlmer make_jlmer_spec() family GLM family. Currently supports \"gaussian\" \"binomial\". statistic Test statistic calculating cluster mass. Can one \"t\" (default) regression model output \"chisq\" likelihood ratio test (takes twice long calculate). ... Optional arguments passed Julia model fitting. Defaults fast = TRUE (family = \"binomial\") progress = FALSE.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/compute_timewise_statistics.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fit Julia regression models to each time point of a time series data — compute_timewise_statistics","text":"predictor--time matrix cluster statistics, class timewise_statistics.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/compute_timewise_statistics.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Fit Julia regression models to each time point of a time series data — compute_timewise_statistics","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Predictor x Time matrix of t-statistics from regression output empirical_statistics <- compute_timewise_statistics(spec) round(empirical_statistics, 2) #>          Time #> Predictor     0    2    4    6    8   10   12   14   16   18   20 #>     Diet2 -1.60 1.67 2.60 3.52 2.38 1.91 1.94 1.34 1.16 1.40 1.51 #>     Diet3 -1.37 2.45 4.48 4.54 3.70 2.97 3.06 2.98 3.05 3.60 3.80 #>     Diet4 -0.92 3.53 6.27 7.00 5.12 4.07 3.66 2.78 2.16 2.13 2.64  # Collect as dataframe with `tidy()` empirical_statistics_df <- tidy(empirical_statistics) empirical_statistics_df #> # A tibble: 33 × 3 #>    predictor  time statistic #>    <chr>     <dbl>     <dbl> #>  1 Diet2         0    -1.60  #>  2 Diet3         0    -1.37  #>  3 Diet4         0    -0.916 #>  4 Diet2         2     1.67  #>  5 Diet3         2     2.45  #>  6 Diet4         2     3.53  #>  7 Diet2         4     2.60  #>  8 Diet3         4     4.48  #>  9 Diet4         4     6.27  #> 10 Diet2         6     3.52  #> # ℹ 23 more rows  # Timewise statistics are from regression models fitted to each time point # - Note the identical statistics at `Time == 0` empirical_statistics_df %>%   filter(time == 0) #> # A tibble: 3 × 3 #>   predictor  time statistic #>   <chr>     <dbl>     <dbl> #> 1 Diet2         0    -1.60  #> 2 Diet3         0    -1.37  #> 3 Diet4         0    -0.916 to_jlmer(weight ~ 1 + Diet, filter(ChickWeight, Time == 0)) %>%   tidy() %>%   select(term, statistic) #> # A tibble: 4 × 2 #>   term        statistic #>   <chr>           <dbl> #> 1 (Intercept)   164.    #> 2 Diet2          -1.60  #> 3 Diet3          -1.37  #> 4 Diet4          -0.916  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_empirical_clusters.html","id":null,"dir":"Reference","previous_headings":"","what":"Detect largest clusters from a time sequence of predictor statistics — extract_empirical_clusters","title":"Detect largest clusters from a time sequence of predictor statistics — extract_empirical_clusters","text":"Detect largest clusters time sequence predictor statistics","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_empirical_clusters.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Detect largest clusters from a time sequence of predictor statistics — extract_empirical_clusters","text":"","code":"extract_empirical_clusters(   empirical_statistics,   threshold,   binned = FALSE,   top_n = Inf )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_empirical_clusters.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Detect largest clusters from a time sequence of predictor statistics — extract_empirical_clusters","text":"empirical_statistics predictor--time matrix empirical timewise statistics. threshold threshold value statistic must pass contribute cluster mass. Interpretation differs choice statistic (): statistic = \"t\", threshold t-value (beta/std.err) regression model. statistic = \"chisq\", threshold p-value chi-squared statistics likelihood ratio tests. binned Whether data aggregated/collapsed time bins. Defaults FALSE, requires cluster span least two time points. TRUE, allows length-1 clusters exist. top_n many clusters return, order size cluster-mass statistic. Defaults Inf return detected clusters.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_empirical_clusters.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Detect largest clusters from a time sequence of predictor statistics — extract_empirical_clusters","text":"empirical_clusters object.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_empirical_clusters.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Detect largest clusters from a time sequence of predictor statistics — extract_empirical_clusters","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Empirical clusters are derived from the timewise statistics empirical_statistics <- compute_timewise_statistics(spec) empirical_clusters <- extract_empirical_clusters(empirical_statistics, threshold = 2) empirical_clusters #> ── Empirical clusters (t > 2) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [4, 8]: 8.496 #> Diet3 #>   [2, 20]: 34.628 #> Diet4 #>   [2, 20]: 39.371 #> ────────────────────────────────────────────────────────────────────────────────  # Collect as dataframe with `tidy()` empirical_clusters_df <- tidy(empirical_clusters) empirical_clusters_df #> # A tibble: 3 × 6 #>   predictor id    start   end length sum_statistic #>   <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl> #> 1 Diet2     1         4     8      3          8.50 #> 2 Diet3     1         2    20     10         34.6  #> 3 Diet4     1         2    20     10         39.4   # Changing the `threshold` value identifies different clusters extract_empirical_clusters(empirical_statistics, threshold = 1) #> ── Empirical clusters (t > 1) ────────────────────────── <empirical_clusters> ── #> Diet2 #>   [2, 20]: 19.427 #> Diet3 #>   [2, 20]: 34.628 #> Diet4 #>   [2, 20]: 39.371 #> ────────────────────────────────────────────────────────────────────────────────  # A predictor can have zero or multiple clusters associated with it extract_empirical_clusters(empirical_statistics, threshold = 3) #> ── Empirical clusters (t > 3) ────────────────────────── <empirical_clusters> ── #> Diet3 #>   [4, 8]: 12.719 #>   [16, 20]: 10.449 #> Diet4 #>   [2, 12]: 29.659 #> ──────────────────────────────────────────────────────────────────────────────── #> ! No clusters found for Diet2  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_null_cluster_dists.html","id":null,"dir":"Reference","previous_headings":"","what":"Construct a null distribution of cluster-mass statistics — extract_null_cluster_dists","title":"Construct a null distribution of cluster-mass statistics — extract_null_cluster_dists","text":"Construct null distribution cluster-mass statistics","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_null_cluster_dists.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Construct a null distribution of cluster-mass statistics — extract_null_cluster_dists","text":"","code":"extract_null_cluster_dists(null_statistics, threshold, binned = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_null_cluster_dists.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Construct a null distribution of cluster-mass statistics — extract_null_cluster_dists","text":"null_statistics simulation--time--predictor 3D array null (permuted) timewise statistics. threshold threshold value statistic must pass contribute cluster mass. Interpretation differs choice statistic (): statistic = \"t\", threshold t-value (beta/std.err) regression model. statistic = \"chisq\", threshold p-value chi-squared statistics likelihood ratio tests. binned Whether data aggregated/collapsed time bins. Defaults FALSE, requires cluster span least two time points. TRUE, allows length-1 clusters exist.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_null_cluster_dists.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Construct a null distribution of cluster-mass statistics — extract_null_cluster_dists","text":"null_cluster_dists object.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/extract_null_cluster_dists.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Construct a null distribution of cluster-mass statistics — extract_null_cluster_dists","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Null cluster-mass distributions are derived from the permuted timewise statistics reset_rng_state() null_statistics <- permute_timewise_statistics(spec, nsim = 100) null_cluster_dists <- extract_null_cluster_dists(null_statistics, threshold = 2) null_cluster_dists #> ── Null cluster-mass distribution (t > 2) ────────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): 0.266 (2.13) #>   Coverage intervals: 95% [0.000, 6.579] #> Diet3 (n = 100) #>   Mean (SD): -0.124 (2.75) #>   Coverage intervals: 95% [-4.777, 5.707] #> Diet4 (n = 100) #>   Mean (SD): 0.007 (2.50) #>   Coverage intervals: 95% [0.000, 0.000] #> ────────────────────────────────────────────────────────────────────────────────  # Collect as dataframe with `tidy()` # - Each simulation contributes one (largest) cluster-mass statistic to the null # - When no clusters are found, the `sum_statistic` value is zero null_cluster_dists_df <- tidy(null_cluster_dists) null_cluster_dists_df #> # A tibble: 300 × 6 #>    predictor start   end length sum_statistic sim   #>    <chr>     <dbl> <dbl>  <dbl>         <dbl> <fct> #>  1 Diet2        10    20      6          14.8 001   #>  2 Diet2        NA    NA     NA           0   002   #>  3 Diet2        NA    NA     NA           0   003   #>  4 Diet2        NA    NA     NA           0   004   #>  5 Diet2        NA    NA     NA           0   005   #>  6 Diet2        NA    NA     NA           0   006   #>  7 Diet2        NA    NA     NA           0   007   #>  8 Diet2        NA    NA     NA           0   008   #>  9 Diet2        NA    NA     NA           0   009   #> 10 Diet2        NA    NA     NA           0   010   #> # ℹ 290 more rows  # Changing the `threshold` value changes the shape of the null extract_null_cluster_dists(null_statistics, threshold = 1) #> ── Null cluster-mass distribution (t > 1) ────────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): 0.565 (5.52) #>   Coverage intervals: 95% [-9.681, 12.061] #> Diet3 (n = 100) #>   Mean (SD): 0.089 (6.27) #>   Coverage intervals: 95% [-15.847, 12.185] #> Diet4 (n = 100) #>   Mean (SD): 0.715 (5.23) #>   Coverage intervals: 95% [-7.843, 11.728] #> ──────────────────────────────────────────────────────────────────────────────── extract_null_cluster_dists(null_statistics, threshold = 3) #> ── Null cluster-mass distribution (t > 3) ────────────── <null_cluster_dists> ── #> Diet2 (n = 100) #>   Mean (SD): 0.000 (0.00) #>   Coverage intervals: 95% [0.000, 0.000] #> Diet3 (n = 100) #>   Mean (SD): 0.000 (0.00) #>   Coverage intervals: 95% [0.000, 0.000] #> Diet4 (n = 100) #>   Mean (SD): 0.000 (0.00) #>   Coverage intervals: 95% [0.000, 0.000] #> ────────────────────────────────────────────────────────────────────────────────  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmer.html","id":null,"dir":"Reference","previous_headings":"","what":"Fit a Julia regression model using jlmer specifications — jlmer","title":"Fit a Julia regression model using jlmer specifications — jlmer","text":"Fit Julia regression model using jlmer specifications","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fit a Julia regression model using jlmer specifications — jlmer","text":"","code":"jlmer(jlmer_spec, family = c(\"gaussian\", \"binomial\"), ..., progress = FALSE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fit a Julia regression model using jlmer specifications — jlmer","text":"jlmer_spec Data prepped jlmer make_jlmer_spec() family GLM family. Currently supports \"gaussian\" \"binomial\". ... Optional arguments passed Julia model fitting. progress TRUE, prints timing iterations.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fit a Julia regression model using jlmer specifications — jlmer","text":"jlmer_mod object.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmer.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Fit a Julia regression model using jlmer specifications — jlmer","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  # Fitting a regression model with a specification object spec <- make_jlmer_spec(weight ~ 1 + Diet, ChickWeight) jlmer(spec) #> <Julia object of type StatsModels.TableRegressionModel> #> ──────────────────────────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|)  Lower 95%  Upper 95% #> ──────────────────────────────────────────────────────────────────────── #> (Intercept)  102.645      4.67395  21.96    <1e-99   93.4847    111.806 #> Diet2         19.9712     7.86744   2.54    0.0111    4.55132    35.3911 #> Diet3         40.3045     7.86744   5.12    <1e-06   24.8847     55.7244 #> Diet4         32.6173     7.91046   4.12    <1e-04   17.113      48.1215 #> ────────────────────────────────────────────────────────────────────────  # `lm()` equivalent summary(lm(weight ~ 1 + Diet, ChickWeight))$coef #>              Estimate Std. Error   t value     Pr(>|t|) #> (Intercept) 102.64545   4.673954 21.961161 4.712762e-78 #> Diet2        19.97121   7.867437  2.538465 1.139708e-02 #> Diet3        40.30455   7.867437  5.122958 4.113938e-07 #> Diet4        32.61726   7.910461  4.123307 4.286352e-05  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm-package.html","id":null,"dir":"Reference","previous_headings":"","what":"jlmerclusterperm: Cluster-Based Permutation Analysis for Densely Sampled Time Data — jlmerclusterperm-package","title":"jlmerclusterperm: Cluster-Based Permutation Analysis for Densely Sampled Time Data — jlmerclusterperm-package","text":"implementation fast cluster-based permutation analysis (CPA) densely-sampled time data developed Maris & Oostenveld, 2007 doi:10.1016/j.jneumeth.2007.03.024 . Supports (generalized, mixed-effects) regression models calculation timewise statistics. Provides wholesale piecemeal interface CPA procedure emphasis interpretability diagnostics. Integrates 'Julia' libraries 'MixedModels.jl' 'GLM.jl' performance improvements, additional functionalities interfacing 'Julia' 'R' powered 'JuliaConnectoR' package.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm-package.html","id":"author","dir":"Reference","previous_headings":"","what":"Author","title":"jlmerclusterperm: Cluster-Based Permutation Analysis for Densely Sampled Time Data — jlmerclusterperm-package","text":"Maintainer: June Choe jchoe001@gmail.com (ORCID) [copyright holder]","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm_setup.html","id":null,"dir":"Reference","previous_headings":"","what":"Initial setup for the jlmerclusterperm package — jlmerclusterperm_setup","title":"Initial setup for the jlmerclusterperm package — jlmerclusterperm_setup","text":"Initial setup jlmerclusterperm package","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm_setup.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initial setup for the jlmerclusterperm package — jlmerclusterperm_setup","text":"","code":"jlmerclusterperm_setup(..., restart = TRUE, verbose = TRUE)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm_setup.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initial setup for the jlmerclusterperm package — jlmerclusterperm_setup","text":"... Ignored. restart Whether set fresh Julia session, given one already running. FALSE jlmerclusterperm_setup() already called, nothing happens. verbose Print progress messages Julia console","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm_setup.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initial setup for the jlmerclusterperm package — jlmerclusterperm_setup","text":"TRUE","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/jlmerclusterperm_setup.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initial setup for the jlmerclusterperm package — jlmerclusterperm_setup","text":"","code":"# \\donttest{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE)  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_model_tidiers.html","id":null,"dir":"Reference","previous_headings":"","what":"Tidier methods for Julia regression models — julia_model_tidiers","title":"Tidier methods for Julia regression models — julia_model_tidiers","text":"Tidier methods Julia regression models","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_model_tidiers.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tidier methods for Julia regression models — julia_model_tidiers","text":"","code":"# S3 method for jlmer_mod tidy(x, effects = c(\"var_model\", \"ran_pars\", \"fixed\"), ...)  # S3 method for jlmer_mod glance(x, ...)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_model_tidiers.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tidier methods for Julia regression models — julia_model_tidiers","text":"x object class jlmer_mod effects One \"var_model\", \"ran_pars\", \"fixed\" ... Unused","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_model_tidiers.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Tidier methods for Julia regression models — julia_model_tidiers","text":"data frame","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_model_tidiers.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tidier methods for Julia regression models — julia_model_tidiers","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  # Fixed-effects only model mod1 <- to_jlmer(weight ~ 1 + Diet, ChickWeight) tidy(mod1) #> # A tibble: 4 × 5 #>   term        estimate std.error statistic   p.value #>   <chr>          <dbl>     <dbl>     <dbl>     <dbl> #> 1 (Intercept)    103.       4.67     22.0  6.77e-107 #> 2 Diet2           20.0      7.87      2.54 1.11e-  2 #> 3 Diet3           40.3      7.87      5.12 3.01e-  7 #> 4 Diet4           32.6      7.91      4.12 3.73e-  5 glance(mod1) #> # A tibble: 1 × 8 #>    nobs    df sigma logLik   AIC   BIC deviance df.residual #>   <dbl> <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <dbl> #> 1   578     5  69.3 -3268. 6546. 6568. 2758693.         573  # Mixed model mod2 <- to_jlmer(weight ~ 1 + Diet + (1 | Chick), ChickWeight) tidy(mod2) #> # A tibble: 6 × 7 #>   effect   group    term            estimate std.error statistic   p.value #>   <chr>    <chr>    <chr>              <dbl>     <dbl>     <dbl>     <dbl> #> 1 fixed    NA       (Intercept)        102.       5.78     17.6   2.74e-69 #> 2 fixed    NA       Diet2               20.8      9.79      2.13  3.34e- 2 #> 3 fixed    NA       Diet3               41.2      9.79      4.20  2.63e- 5 #> 4 fixed    NA       Diet4               33.3      9.83      3.39  7.08e- 4 #> 5 ran_pars Chick    sd__(Intercept)     15.7     NA        NA    NA        #> 6 ran_pars Residual sd__Observation     67.3     NA        NA    NA        glance(mod2) #> # A tibble: 1 × 8 #>    nobs    df sigma logLik   AIC   BIC deviance df.residual #>   <int> <int> <dbl>  <dbl> <dbl> <dbl>    <dbl>       <int> #> 1   578     6  67.3 -3265. 6542. 6568.    6530.         572  # Select which of fixed/random effects to return tidy(mod2, effects = \"fixed\") #> # A tibble: 4 × 5 #>   term        estimate std.error statistic  p.value #>   <chr>          <dbl>     <dbl>     <dbl>    <dbl> #> 1 (Intercept)    102.       5.78     17.6  2.74e-69 #> 2 Diet2           20.8      9.79      2.13 3.34e- 2 #> 3 Diet3           41.2      9.79      4.20 2.63e- 5 #> 4 Diet4           33.3      9.83      3.39 7.08e- 4 tidy(mod2, effects = \"ran_pars\") #> # A tibble: 2 × 3 #>   group    term            estimate #>   <chr>    <chr>              <dbl> #> 1 Chick    sd__(Intercept)     15.7 #> 2 Residual sd__Observation     67.3  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_progress.html","id":null,"dir":"Reference","previous_headings":"","what":"Set/get options for Julia progress bar — julia_progress","title":"Set/get options for Julia progress bar — julia_progress","text":"Set/get options Julia progress bar","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_progress.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set/get options for Julia progress bar — julia_progress","text":"","code":"julia_progress(show, width)"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_progress.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set/get options for Julia progress bar — julia_progress","text":"show Whether show progress bar. may also pass list \"show\" \"width\". width Width progress bar. \"auto\", adjusts progress bar width fit console.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_progress.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set/get options for Julia progress bar — julia_progress","text":"Previous values show width","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_progress.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set/get options for Julia progress bar — julia_progress","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) # }  # Show current progress options julia_progress() #> $show #> [1] TRUE #>  #> $width #> [1] 36 #>   # Set options and save previous options old_progress_opts <- julia_progress(show = FALSE, width = 100) julia_progress() #> $show #> [1] FALSE #>  #> $width #> [1] 100 #>   # Restoring progress settings by passing a list of old options old_progress_opts #> $show #> [1] TRUE #>  #> $width #> [1] 36 #>  julia_progress(old_progress_opts) identical(julia_progress(), old_progress_opts) #> [1] TRUE  # Alternatively, reset to default settings using this syntax: julia_progress(show = TRUE, width = \"auto\")  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_rng.html","id":null,"dir":"Reference","previous_headings":"","what":"Interface to the Julia RNG — julia_rng","title":"Interface to the Julia RNG — julia_rng","text":"Interface Julia RNG","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_rng.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Interface to the Julia RNG — julia_rng","text":"","code":"set_rng_state(i)  reset_rng_state()  get_rng_state()  set_rng_seed(seed)  get_rng_seed()"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_rng.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Interface to the Julia RNG — julia_rng","text":"Counter number seed Seed","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_rng.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Interface to the Julia RNG — julia_rng","text":"current seed counter","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_rng.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Interface to the Julia RNG — julia_rng","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  # RNG initializes to seed=1 counter=0 get_rng_seed() #> [1] 1 get_rng_state() #> [1] 0  # setter/getter for RNG counter set_rng_state(123) get_rng_state() #> [1] 123  # setter/getter for RNG seed set_rng_seed(2) get_rng_seed() #> [1] 2  # restore to initial setting (seed=1, counter=0) set_rng_seed(1) set_rng_state(0)  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_setup_ok.html","id":null,"dir":"Reference","previous_headings":"","what":"Check Julia setup requirements for jlmerclusterperm — julia_setup_ok","title":"Check Julia setup requirements for jlmerclusterperm — julia_setup_ok","text":"Check Julia setup requirements jlmerclusterperm","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_setup_ok.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check Julia setup requirements for jlmerclusterperm — julia_setup_ok","text":"","code":"julia_setup_ok()"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_setup_ok.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check Julia setup requirements for jlmerclusterperm — julia_setup_ok","text":"Boolean","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/julia_setup_ok.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Check Julia setup requirements for jlmerclusterperm — julia_setup_ok","text":"","code":"julia_setup_ok() #> [1] TRUE"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/make_jlmer_spec.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a specifications object for fitting regression models in Julia — make_jlmer_spec","title":"Create a specifications object for fitting regression models in Julia — make_jlmer_spec","text":"Create specifications object fitting regression models Julia","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/make_jlmer_spec.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a specifications object for fitting regression models in Julia — make_jlmer_spec","text":"","code":"make_jlmer_spec(   formula,   data,   subject = NULL,   trial = NULL,   time = NULL,   drop_terms = NULL,   ... )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/make_jlmer_spec.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a specifications object for fitting regression models in Julia — make_jlmer_spec","text":"formula Model formula R syntax data data frame subject Column subjects data. trial Column trials data. Must uniquely identify time series within subject (example, column items counterbalanced design subject sees exactly one item). time Column time data. drop_terms (Optional) terms drop reconstructed model formula ... Unused, extensibility.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/make_jlmer_spec.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a specifications object for fitting regression models in Julia — make_jlmer_spec","text":"object class jlmer_spec.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/make_jlmer_spec.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a specifications object for fitting regression models in Julia — make_jlmer_spec","text":"","code":"# Bare specification object (minimal spec for fitting a global model) spec <- make_jlmer_spec(weight ~ 1 + Diet, ChickWeight) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Data: #> # A tibble: 578 × 4 #>   weight Diet2 Diet3 Diet4 #>    <dbl> <dbl> <dbl> <dbl> #> 1     42     0     0     0 #> 2     51     0     0     0 #> 3     59     0     0     0 #> # ℹ 575 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Constraints on specification for CPA: # 1) The combination of `subject`, `trial`, and `time` must uniquely identify rows in the data # 2) `time` must have constant sampling rate (i.e., evenly spaced values) spec_wrong <- make_jlmer_spec(   weight ~ 1 + Diet, ChickWeight,   time = \"Time\" ) #> ! Grouping column \"Time\" does not uniquely identify rows in the data. #> ! Sampling rate for the `time` column \"Time\" is not constant - may affect interpretability of results. unique(ChickWeight$Time) #>  [1]  0  2  4  6  8 10 12 14 16 18 20 21  # Corrected specification for the above spec_correct <- make_jlmer_spec(   weight ~ 1 + Diet, subset(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec_correct #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_by_predictor.html","id":null,"dir":"Reference","previous_headings":"","what":"Permute data while respecting grouping structure(s) of observations — permute_by_predictor","title":"Permute data while respecting grouping structure(s) of observations — permute_by_predictor","text":"Permute data respecting grouping structure(s) observations","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_by_predictor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Permute data while respecting grouping structure(s) of observations — permute_by_predictor","text":"","code":"permute_by_predictor(   jlmer_spec,   predictors,   predictor_type = c(\"guess\", \"between_participant\", \"within_participant\"),   n = 1L )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_by_predictor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Permute data while respecting grouping structure(s) of observations — permute_by_predictor","text":"jlmer_spec Data prepped jlmer make_jlmer_spec() predictors vector terms model. multiple, must form levels one predictor. predictor_type Whether predictor \"between_participant\" \"within_participant\". Defaults \"guess\". n Number permuted samples data generate. Defaults 1L.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_by_predictor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Permute data while respecting grouping structure(s) of observations — permute_by_predictor","text":"long dataframe permuted re-samples .id column representing replication IDs.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_by_predictor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Permute data while respecting grouping structure(s) of observations — permute_by_predictor","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  # Example data setup chickweights_df <- ChickWeight chickweights_df <- chickweights_df[chickweights_df$Time <= 20, ] chickweights_df$DietInt <- as.integer(chickweights_df$Diet) head(chickweights_df) #> Grouped Data: weight ~ Time | Chick #>   weight Time Chick Diet DietInt #> 1     42    0     1    1       1 #> 2     51    2     1    1       1 #> 3     59    4     1    1       1 #> 4     64    6     1    1       1 #> 5     76    8     1    1       1 #> 6     93   10     1    1       1  # Example 1: Spec object using the continuous `DietInt` predictor chickweights_spec1 <- make_jlmer_spec(   formula = weight ~ 1 + DietInt,   data = chickweights_df,   subject = \"Chick\", time = \"Time\" ) chickweights_spec1 #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + DietInt #> Predictors: #>   DietInt: DietInt #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 4 #>   weight DietInt Chick  Time #>    <dbl>   <dbl> <ord> <dbl> #> 1     42       1 1         0 #> 2     51       1 1         2 #> 3     59       1 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Shuffling `DietInt` values guesses `predictor_type = \"between_participant\"` reset_rng_state() spec1_perm1 <- permute_by_predictor(chickweights_spec1, predictors = \"DietInt\") #> ℹ Guessed `predictor_type` to be \"between_participant\" # This calls the same shuffling algorithm for CPA in Julia, so counter is incremented get_rng_state() #> [1] 38  # Shuffling under shared RNG state reproduces the same permutation of the data reset_rng_state() spec1_perm2 <- permute_by_predictor(chickweights_spec1, predictors = \"DietInt\") #> ℹ Guessed `predictor_type` to be \"between_participant\" identical(spec1_perm1, spec1_perm2) #> [1] TRUE  # Example 2: Spec object using the multilevel `Diet` predictor chickweights_spec2 <- make_jlmer_spec(   formula = weight ~ 1 + Diet,   data = chickweights_df,   subject = \"Chick\", time = \"Time\" ) chickweights_spec2 #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Levels of a category are automatically shuffled together reset_rng_state() spec2_perm1 <- permute_by_predictor(chickweights_spec2, predictors = \"Diet2\") #> ℹ Guessed `predictor_type` to be \"between_participant\" #> ℹ Shuffling all levels of the factor together (\"Diet2\", \"Diet3\", and \"Diet4\") reset_rng_state() spec2_perm2 <- permute_by_predictor(chickweights_spec2, predictors = c(\"Diet2\", \"Diet3\", \"Diet4\")) #> ℹ Guessed `predictor_type` to be \"between_participant\" identical(spec2_perm1, spec2_perm2) #> [1] TRUE  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_timewise_statistics.html","id":null,"dir":"Reference","previous_headings":"","what":"Simulate cluster-mass statistics via bootstrapped permutations — permute_timewise_statistics","title":"Simulate cluster-mass statistics via bootstrapped permutations — permute_timewise_statistics","text":"Simulate cluster-mass statistics via bootstrapped permutations","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_timewise_statistics.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Simulate cluster-mass statistics via bootstrapped permutations — permute_timewise_statistics","text":"","code":"permute_timewise_statistics(   jlmer_spec,   family = c(\"gaussian\", \"binomial\"),   statistic = c(\"t\", \"chisq\"),   nsim = 100L,   predictors = NULL,   ... )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_timewise_statistics.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Simulate cluster-mass statistics via bootstrapped permutations — permute_timewise_statistics","text":"jlmer_spec Data prepped jlmer make_jlmer_spec() family GLM family. Currently supports \"gaussian\" \"binomial\". statistic Test statistic calculating cluster mass. Can one \"t\" (default) regression model output \"chisq\" likelihood ratio test (takes twice long calculate). nsim Number simulations description predictors (Optional) subset predictors test. Defaults NULL tests predictors. ... Optional arguments passed Julia model fitting. Defaults fast = TRUE (family = \"binomial\") progress = FALSE.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_timewise_statistics.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Simulate cluster-mass statistics via bootstrapped permutations — permute_timewise_statistics","text":"simulation--time--predictor 3D array cluster statistics, class timewise_statistics.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/permute_timewise_statistics.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Simulate cluster-mass statistics via bootstrapped permutations — permute_timewise_statistics","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  library(dplyr, warn.conflicts = FALSE)  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, filter(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Simulation x Time x Predictor array of t-statistics from regression output reset_rng_state() null_statistics <- permute_timewise_statistics(spec, nsim = 3) round(null_statistics, 2) #> , , Predictor = Diet2 #>  #>    Time #> Sim     0    2    4    6    8   10   12   14   16   18   20 #>   1 -0.90 0.41 0.39 1.12 1.58 2.13 2.34 2.58 2.47 2.67 2.61 #>   2  1.40 0.90 0.96 1.44 1.61 1.45 1.23 0.90 0.80 0.78 0.76 #>   3 -1.44 1.23 0.73 0.88 0.93 0.69 0.58 0.61 0.77 1.04 1.43 #>  #> , , Predictor = Diet3 #>  #>    Time #> Sim    0     2     4     6     8    10    12    14    16    18    20 #>   1 0.00 -0.21  0.11 -0.68 -0.66 -0.38 -0.01 -0.02 -0.17 -0.20 -0.12 #>   2 0.70  0.56 -0.10  0.17  0.37  0.37  0.55  0.21  0.46  0.44  0.02 #>   3 0.72  1.90  0.85  0.20  0.19 -0.09 -0.34 -0.24  0.34  0.44  0.83 #>  #> , , Predictor = Diet4 #>  #>    Time #> Sim     0     2     4     6     8    10    12   14    16    18    20 #>   1  0.45 -0.82 -0.33 -0.68 -0.36 -0.35 -0.41 0.19 -0.04 -0.16 -0.16 #>   2  2.09 -0.69 -0.10  0.34  0.56  0.63  0.35 0.05 -0.05 -0.26 -0.43 #>   3 -2.16 -1.38 -1.35 -0.85 -0.02 -0.12  0.01 0.59  0.71  0.67  0.53 #>   # Collect as dataframe with `tidy()` permuted_timewise_stats_df <- tidy(null_statistics) permuted_timewise_stats_df #> # A tibble: 99 × 4 #>    predictor  time statistic sim   #>    <chr>     <dbl>     <dbl> <fct> #>  1 Diet2         0 -8.98e- 1 1     #>  2 Diet2         0  1.40e+ 0 2     #>  3 Diet2         0 -1.44e+ 0 3     #>  4 Diet3         0  2.34e-15 1     #>  5 Diet3         0  6.98e- 1 2     #>  6 Diet3         0  7.21e- 1 3     #>  7 Diet4         0  4.49e- 1 1     #>  8 Diet4         0  2.09e+ 0 2     #>  9 Diet4         0 -2.16e+ 0 3     #> 10 Diet2         2  4.12e- 1 1     #> # ℹ 89 more rows  # Permutations ran under the same RNG state are identical reset_rng_state() null_statistics2 <- permute_timewise_statistics(spec, nsim = 3) identical(null_statistics, null_statistics2) #> [1] TRUE  get_rng_state() #> [1] 111 null_statistics3 <- permute_timewise_statistics(spec, nsim = 3) identical(null_statistics, null_statistics3) #> [1] FALSE  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/reexports.html","id":null,"dir":"Reference","previous_headings":"","what":"Objects exported from other packages — reexports","title":"Objects exported from other packages — reexports","text":"objects imported packages. Follow links see documentation. generics glance, tidy JuliaConnectoR stopJulia","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/to_jlmer.html","id":null,"dir":"Reference","previous_headings":"","what":"Fit a Julia regression model using lme4 syntax — to_jlmer","title":"Fit a Julia regression model using lme4 syntax — to_jlmer","text":"Fit Julia regression model using lme4 syntax","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/to_jlmer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fit a Julia regression model using lme4 syntax — to_jlmer","text":"","code":"to_jlmer(   formula,   data,   family = c(\"gaussian\", \"binomial\"),   jlmer_spec_opts = list(),   ...,   progress = FALSE )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/to_jlmer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fit a Julia regression model using lme4 syntax — to_jlmer","text":"formula Model formula R syntax data data frame family GLM family. Currently supports \"gaussian\" \"binomial\". jlmer_spec_opts List options passed make_jlmer_spec() ... Optional arguments passed Julia model fitting. progress TRUE, prints timing iterations.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/to_jlmer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fit a Julia regression model using lme4 syntax — to_jlmer","text":"jlmer_mod object.","code":""},{"path":[]},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/to_jlmer.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Fit a Julia regression model using lme4 syntax — to_jlmer","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  # Fitting a regression model with R formula syntax to_jlmer(weight ~ 1 + Diet, ChickWeight) #> <Julia object of type StatsModels.TableRegressionModel> #> ──────────────────────────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|)  Lower 95%  Upper 95% #> ──────────────────────────────────────────────────────────────────────── #> (Intercept)  102.645      4.67395  21.96    <1e-99   93.4847    111.806 #> Diet2         19.9712     7.86744   2.54    0.0111    4.55132    35.3911 #> Diet3         40.3045     7.86744   5.12    <1e-06   24.8847     55.7244 #> Diet4         32.6173     7.91046   4.12    <1e-04   17.113      48.1215 #> ────────────────────────────────────────────────────────────────────────  # `lm()` equivalent summary(lm(weight ~ 1 + Diet, ChickWeight))$coef #>              Estimate Std. Error   t value     Pr(>|t|) #> (Intercept) 102.64545   4.673954 21.961161 4.712762e-78 #> Diet2        19.97121   7.867437  2.538465 1.139708e-02 #> Diet3        40.30455   7.867437  5.122958 4.113938e-07 #> Diet4        32.61726   7.910461  4.123307 4.286352e-05  # Fitting a mixed model with {lme4} syntax to_jlmer(weight ~ 1 + Diet + (1 | Chick), ChickWeight) #> <Julia object of type LinearMixedModel> #> Variance components: #>             Column    Variance Std.Dev. #> Chick    (Intercept)   247.2559 15.7244 #> Residual              4527.9661 67.2902 #>  ────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|) #> ────────────────────────────────────────────────── #> (Intercept)  101.781      5.78499  17.59    <1e-68 #> Diet2         20.8353     9.79412   2.13    0.0334 #> Diet3         41.1687     9.79412   4.20    <1e-04 #> Diet4         33.284      9.82881   3.39    0.0007 #> ──────────────────────────────────────────────────  # Passing MixedModels.jl fit options to the `...` to_jlmer(weight ~ 1 + Diet + (1 | Chick), ChickWeight, REML = TRUE) #> <Julia object of type LinearMixedModel> #> Variance components: #>             Column    Variance Std.Dev. #> Chick    (Intercept)   305.7755 17.4864 #> Residual              4526.5647 67.2797 #>  ────────────────────────────────────────────────── #>                 Coef.  Std. Error      z  Pr(>|z|) #> ────────────────────────────────────────────────── #> (Intercept)  101.634      6.04044  16.83    <1e-62 #> Diet2         20.9832    10.2365    2.05    0.0404 #> Diet3         41.3165    10.2365    4.04    <1e-04 #> Diet4         33.4041    10.27      3.25    0.0011 #> ──────────────────────────────────────────────────  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/walk_threshold_steps.html","id":null,"dir":"Reference","previous_headings":"","what":"Test the probability of cluster-mass statistics over a range of threshold values — walk_threshold_steps","title":"Test the probability of cluster-mass statistics over a range of threshold values — walk_threshold_steps","text":"Test probability cluster-mass statistics range threshold values","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/walk_threshold_steps.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test the probability of cluster-mass statistics over a range of threshold values — walk_threshold_steps","text":"","code":"walk_threshold_steps(   empirical_statistics,   null_statistics,   steps,   top_n = Inf,   binned = FALSE,   add1 = TRUE,   progress = TRUE )"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/walk_threshold_steps.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test the probability of cluster-mass statistics over a range of threshold values — walk_threshold_steps","text":"empirical_statistics predictor--time matrix empirical timewise statistics. null_statistics simulation--time--predictor 3D array null (permuted) timewise statistics. steps vector threshold values test top_n many clusters return, order size cluster-mass statistic. Defaults Inf return detected clusters. binned Whether data aggregated/collapsed time bins. Defaults FALSE, requires cluster span least two time points. TRUE, allows length-1 clusters exist. add1 Whether add 1 numerator denominator calculating p-value. Use TRUE effectively count observed statistic part permuted null distribution (recommended larger nsim prior publishing results). progress Whether display progress bar","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/walk_threshold_steps.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test the probability of cluster-mass statistics over a range of threshold values — walk_threshold_steps","text":"data frame predictor clusters-mass statistics threshold.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/reference/walk_threshold_steps.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Test the probability of cluster-mass statistics over a range of threshold values — walk_threshold_steps","text":"","code":"# \\donttest{ # \\dontshow{ options(\"jlmerclusterperm.nthreads\" = 2) jlmerclusterperm_setup(verbose = FALSE) julia_progress(show = FALSE) # }  # Specification object spec <- make_jlmer_spec(   weight ~ 1 + Diet, subset(ChickWeight, Time <= 20),   subject = \"Chick\", time = \"Time\" ) spec #> ── jlmer specification ───────────────────────────────────────── <jlmer_spec> ── #> Formula: weight ~ 1 + Diet2 + Diet3 + Diet4 #> Predictors: #>   Diet: Diet2, Diet3, Diet4 #> Groupings: #>   Subject: Chick #>   Trial: #>   Time: Time #> Data: #> # A tibble: 533 × 6 #>   weight Diet2 Diet3 Diet4 Chick  Time #>    <dbl> <dbl> <dbl> <dbl> <ord> <dbl> #> 1     42     0     0     0 1         0 #> 2     51     0     0     0 1         2 #> 3     59     0     0     0 1         4 #> # ℹ 530 more rows #> ────────────────────────────────────────────────────────────────────────────────  # Compute timewise statistics for the observed and permuted data empirical_statistics <- compute_timewise_statistics(spec) null_statistics <- permute_timewise_statistics(spec, nsim = 100)  # Test cluster mass/probability under different threshold values walk_threshold_steps(empirical_statistics, null_statistics, steps = 1:3,                      progress = FALSE) #> # A tibble: 9 × 8 #>   threshold predictor id    start   end length sum_statistic  pvalue #>       <int> <chr>     <fct> <dbl> <dbl>  <dbl>         <dbl>   <dbl> #> 1         1 Diet2     1         2    20     10         19.4  0.00990 #> 2         1 Diet3     1         2    20     10         34.6  0.00990 #> 3         1 Diet4     1         2    20     10         39.4  0.00990 #> 4         2 Diet2     1         4     8      3          8.50 0.0297  #> 5         2 Diet3     1         2    20     10         34.6  0.00990 #> 6         2 Diet4     1         2    20     10         39.4  0.00990 #> 7         3 Diet3     1         4     8      3         12.7  0.00990 #> 8         3 Diet3     2        16    20      3         10.4  0.00990 #> 9         3 Diet4     1         2    12      6         29.7  0.00990  # \\dontshow{ JuliaConnectoR::stopJulia() # } # }"},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-111","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.1.1","title":"jlmerclusterperm 1.1.1","text":"CRAN release: 2024-02-08","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-111-1","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.1.1","title":"jlmerclusterperm 1.1.1","text":"CRAN release: 2024-02-08 informative warnings constructing spec object Fixed minor bugs interfacing Julia Fixed bug reporting convergence warnings one encountered","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-110","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.1.0","title":"jlmerclusterperm 1.1.0","text":"CRAN release: 2023-10-07 Minor breaking change: Slightly loosened default convergence criterion permute_timewise_statistics() Fixes CRAN: Specified importing R_user_dir() backports missing previously","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-106","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.6","title":"jlmerclusterperm 1.0.6","text":"CRAN release: 2023-09-19 Fixed bug add1 argument walk_threshold_steps() passed properly R < 4.0 falls back using backports R_user_dir() Fixes CRAN: Downgrade DataFrames.jl dependency 1.3 avoid mysterious pre-compilation failure CRAN windows checks","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-105","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.5","title":"jlmerclusterperm 1.0.5","text":"CRAN release: 2023-09-10 user-visible changes Fixes CRAN: Ensure minimum version requirements Tables.jl DataAPI.jl met","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-104","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.4","title":"jlmerclusterperm 1.0.4","text":"CRAN release: 2023-09-06 user-visible changes Restructured internal Julia code module (JlmerClusterPerm.jl) Upon activating project, Manifest.toml cached tools::R_user_dir() speed pre-compilation.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-103","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.3","title":"jlmerclusterperm 1.0.3","text":"CRAN release: 2023-07-19 Fixes CRAN: Use string package version comparison","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-102","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.2","title":"jlmerclusterperm 1.0.2","text":"CRAN release: 2023-07-15","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"bug-fixes-1-0-2","dir":"Changelog","previous_headings":"","what":"Bug fixes","title":"jlmerclusterperm 1.0.2","text":"jlmerclusterperm_setup() now works rc alpha builds Julia Fixes CRAN: exclusion Manifest.toml now specified .Rbuildignore - Project.toml bundled intended.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-101","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.1","title":"jlmerclusterperm 1.0.1","text":"CRAN release: 2023-07-02 jlmerclusterperm_setup() now exits early Julia version requirement (>=1.8) met. Fixes CRAN: Check Julia version requirement proceeding examples tests","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-100","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 1.0.0","title":"jlmerclusterperm 1.0.0","text":"CRAN release: 2023-06-27","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"breaking-changes-1-0-0","dir":"Changelog","previous_headings":"","what":"Breaking changes","title":"jlmerclusterperm 1.0.0","text":"threshold_steps argument walk_threshold_steps() renamed steps.","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"new-features-1-0-0","dir":"Changelog","previous_headings":"","what":"New features","title":"jlmerclusterperm 1.0.0","text":"New functions interface Julia RNG seed: get_rng_seed() set_rng_seed()","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"other-improvements-1-0-0","dir":"Changelog","previous_headings":"","what":"Other improvements","title":"jlmerclusterperm 1.0.0","text":"jlmerclusterperm_setup() now echos Pkg.instantiate() print precompilation information upon first setup call","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-020","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 0.2.0","title":"jlmerclusterperm 0.2.0","text":"Added vignettes. Significant usability improvements","code":""},{"path":"https://yjunechoe.github.io/jlmerclusterperm/news/index.html","id":"jlmerclusterperm-010","dir":"Changelog","previous_headings":"","what":"jlmerclusterperm 0.1.0","title":"jlmerclusterperm 0.1.0","text":"Initial release","code":""}]
diff --git a/docs/sitemap.xml b/docs/sitemap.xml
index 516cadf..acc3ac1 100644
--- a/docs/sitemap.xml
+++ b/docs/sitemap.xml
@@ -9,6 +9,9 @@
   <url>
     <loc>https://yjunechoe.github.io/jlmerclusterperm/articles/articles/deCarvalho-et-al-2021.html</loc>
   </url>
+  <url>
+    <loc>https://yjunechoe.github.io/jlmerclusterperm/articles/articles/eyetrackingR-comparison.html</loc>
+  </url>
   <url>
     <loc>https://yjunechoe.github.io/jlmerclusterperm/articles/articles/Garrison-et-al-2020.html</loc>
   </url>