Skip to content

Raw results for our survey on similarity of attitudes in argumentations

Notifications You must be signed in to change notification settings

hhucn/argumentation-similarity-survey-results

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 

Repository files navigation

Complete Results for our Survey on Human Assessment of Attitude Similarity in Argumentation

This repository contains the anonymized survey results we have collected for our research on metrics for calculating the distance between argumentation graphs.

The survey has been conducted with workers from Amazon Mechanical Turk (MTurk) from the US. Most questions has different scenarios which were randomly assigned to a participant during the survey. The results file only contains the answers of participants who answered at least 3 of 5 control questions correctly.

The file results.html contains each question which has been asked. The different scenarios (h3 level) or grouped together (h2 level). The h2 heading contains the exact question which has been asked, followed by the summary of all answers frequencies for this scenario, each scenario's text and answer frequencies. If multiple questions have been asked for the same scenario, its text is not repeated again. More complex scenarios include a graphic representing the argumentations; those graphics were not part of the questionnaire.

The answer we have expected from our hypotheses are marked bold. Significant answer frequencies are underlined in different styles depending on the significance level:

  • 10 %: dotted, medium blue dotted, medium blue
  • 5 %: dashed, orange dashed, orange
  • 1 %: dark red dark red

Note that only the p-values for our expected answers are valid, as no correction for multiple testing is applied. Each percentage is followed by the concrete ratio of given answers and Clopper–Pearson confidence intervals for α=0.05.

Demographic Information

counting only participants who answered at least 3 out of 5 control questions correctly

Age Group #
20-29 25
30-39 25
40-49 2
50-59 6
60-69 2
>69 0
Gender #
Male 53
Female 27
Rather not say 0

Mapping of Hypotheses to Survey Questions

Each question (or question group) has a unique identifier. The following table shows which questions have been used to support which hypothesis. For more complex scenarios in the questionnaire, a graphical visualization is included (cf. html for expected answers).

# Hypothesis Question Group Visualization of Questionnaire Scenario
1 Proportionally bigger overlap of opinions on position results in greater similarity than the absolute number of differences. D1s
2 Proportionally bigger overlap on arguments for/against a position results in greater similarity than the absolute number of differences. D1
3 A neutral opinion is between a positive and a negative opinion. D2nf, D2nF
4 Deviations in deeper parts have less contribution to dissimilarity than deviations in higher parts. D3
5 Weights of arguments have an influence even if they are the only difference. D5
6 Argumentation differences in a branch with lower importance contribute less to dissimilarity. D4i
7 No opinion is between a positive and a negative opinion. D2e, D2eF
8 An unknown opinion is between a positive and a negative opinion. D2i, D2iF
9 A statement for which no opinion is mentioned is like a statement for which we explicitly say the opinion is unknown. TD2
10 Not mentioning an argument and being against an argument have the same effect. D4e
11 Disagreeing on a position results in greater distance than having the same opinion on that position, but with contrary arguments. Mn
12 It is possible for a difference in arguments for/against positions to result in greater dissimilarity than a difference in opinions on those positions. Ap
13 Two argumentations with weak and contrary opinions on a statement can be closer than two argumentations with the same opinions, but with very different strength. D7p, D8
14 Two argumentations with weak arguments and contrary opinions on their premises can be closer than two argumentations with the same opinions, but with very different strength of arguments. D7a
15 When determining the attitude regarding a position, opinions (not) mentioned for a not-accepted argument have no influence. F1
16 Flipping the two important positions results in a bigger difference than flipping two less important positions. F2s1
17 Adding a new position can remove a previous dissimilarity. F2s2
18 Adding a new position as most important position can swap a previous similarity order. F2a1, F2a2
19 Agreeing with someone’s most important position is as important as having that person’s most important opinion matching mine. Pf
20 Adding another most important position results in greater dissimilarity than flipping the priorities of two positions. Wa
21 Having more similar priorities of opinions can result in greater similarity even with lower absolute number of same opinions. Wr1
22 Not mentioning a position results in greater dissimilarity than assigning lower priorities. Pp

Related publication

When using this dataset, please cite the following publication:

Brenneis, M., Mauve, M.: Do I Argue Like Them? A Human Baseline for Comparing Attitudes in Argumentations. In: Proceedings of the Workshop on Advances In Argumentation In Artificial Intelligence 2020. pp. 1–15. No. 2777 in CEUR Workshop Proceedings, Aachen (Nov 2020)

About

Raw results for our survey on similarity of attitudes in argumentations

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages