Stability of dyadic exchange: experimental evidence for the impact of shared group membership

This repository contains data and R code necessary to reproduce results reported in Karpiński et al. (2023). It consists of just three files:

• the files social-exchange-data.csv and social-identity-data.csv contain the data used in the analyses. More information about the variables is given below. In both files, columns (variables) are separated by semicolons;
• the file data-analysis.R is an R script file with R code used in the analyses of the data.

The data come form an experiment in which participants were divided into two categories, based on their preference for the paintings of Kandinsky over Klee, and then were given 20 opportunities to share valued resources with one another. After completing all 20 opportunities, the participants filled in a short questionnaire concerning their perceptions of their in-group (i.e., the category they had been assigned to) and their out-group. See the paper by Karpiński et al. (2023) for a complete description of the experimental setting, protocols, and instructions.

The data

Social exchange process

The data in social-exchange-data.csv come from the part of the experiment in which participants were sharing resources with other participants over 20 opportunities (or rounds). The data come in a long format, each observation being an instance of participant $i$ sharing their resources with participant $j$ in round $t$ of experimental session $k$. With 18 experimental sessions, 10 participants per session, 20 opportunities for sharing resources given to each participants, and 9 alters to share one’s resources with per round, there is a total of $18\times 20\times 10\times 9$ = 32,400 unique observations.

The following variables are included in the dataset:

• sessionid – session identifier;
• session_date – calendar date of the day on which the experiment took place;
• visible – a binary variable equal to “Yes” for experimental sessions in which participants’ social categories were made immediately visible to them, and “No” otherwise;
• split – a character string corresponding to the way in which participants were split between social categories. It assumes the following values: “5:5” (meaning that all 10 participants in a session were evenly split among the categories), “6:4” (meaning that 6 of the 10 participants in a session were assigned to one category, while the remaining 4 ended up in the other), and “7:3” (meaning that 7 of the 10 participants in a session were assigned to one category, while the remaining 3 ended up in the other);
• subjectid – subject’s identifier
• round – exchange opportunity, an integer ranging from 1 to 20
• subject – subject’s index number in the experimental session, an integer ranging from 1 to 10;
• alter – alter’s index number in the experimental session, an integer ranging from 1 to 10;
• gave – a binary variable specific to each subject-alter dyad. It is coded 1 if the subject chose to share their resources with the alter at a given exchange opportunity. Otherwise, it is coded as 0.
• received – a binary variable specific to each subject-alter dyad. It is coded 1 if the subject received resources from the alter at a given exchange opportunity. Otherwise, it is coded as 0.
• gave_last – a lagged version of gave: it is coded 1 if the subject shared their resources with the alter at the previous exchange opportunity. Otherwise, it is coded as 0. Note that this variable is not defined when round == 1.
• received_last – a lagged version of received: it is coded 1 if the subject received resources from the alter at the previous exchange opportunity. Otherwise, it is coded as 0. Note that this variable is not defined when round == 1.
• s_cat – subject’s social category, coded 1 for the Kandinsky group and 2 for the Klee group;
• a_cat – alter’s social category, coded 1 for the Kandinsky group and 2 for the Klee group;
• dyad_type – a binary variable distinguishing between “Intra-group” (where the subject and alter belong to the same social category) and “Inter-group” (where the subject and alter belong to different categories) dyads.

Social identity questionnaire

The dataset in social-identity-data.csv contains responses to a questionnaire that the participants filled in after having completed all 20 exchange opportunities. In the questionnaire, they were asked how they felt about their in-group and their out-group with respect to 4 dimensions (or domains): belongingness, commonality, closeness, and liking. The responses were coded on 7-point scale, with higher values indicating more positive evaluations. The questionnaire is based on the work by Yamagishi and Kiyonari (2000) and Aksoy (2015).

The data come in a long format, each observation corresponding to an evaluation along dimension $k$ by individual $i$ in experimental session $j$. With 18 session, 10 participants per session, and 4 dimensions of evaluation, the total number of observations is 720.

The dataset includes the following variables:

• sessionid – session identifier;
• session_date – calendar date of the day on which the experiment took place;
• visible – a binary variable equal to “Yes” for experimental sessions in which participants’ social categories were made immediately visible to them, and “No” otherwise;
• split – a character string corresponding to the way in which participants were split between social categories. It assumes the following values: “5:5” (meaning that all 10 participants in a session were evenly split among the categories), “6:4” (meaning that 6 of the 10 participants in a session were assigned to one category, while the remaining 4 ended up in the other), and “7:3” (meaning that 7 of the 10 participants in a session were assigned to one category, while the remaining 3 ended up in the other);
• subjectid – subject’s identifier
• subject – subject’s index number in the experimental session, an integer ranging from 1 to 10;
• s_cat – subject’s social category, coded 1 for the Kandinsky group and 2 for the Klee group;
• female – a dummy variable for being female, coded 1 for female and 0 for male;
• age – subject’s age (in years);
• domain – evaluation domain for the in-group and the out-group;
• eval_in – subjective evaluations of one’s in-group;
• eval_out – subjective evaluations of one’s out-group.

The variables sessionid, session_date, visible, split, subjectid, subject, s_cat are common to both datasets.

Acknowledgement

Karpiński et al. (2023) use data from an experiment designed to study homophily processes in a small-group social exchange setting. The experiment is a part of a larger research project on attraction to similar others and repulsion from dissimilar others as drivers of homophily in social relations. The research is supported by a grant from National Science Centre in Poland (grant number UMO-2017/25/B/HS6/02543).

References

Aksoy, Ozan. 2015. “Effects of Heterogeneity and Homophily on Cooperation.” Social Psychology Quarterly 78 (4): 324–44.

Karpiński, Zbigniew, Adam Kęska, Dariusz Przybysz, and John Skvoretz. 2023. “Stability of Dyadic Exchange: Experimental Evidence for the Impact of Shared Group Membership.” Social Science Research 116: 102940. https://doi.org/https://doi.org/10.1016/j.ssresearch.2023.102940.

Yamagishi, Toshio, and Toko Kiyonari. 2000. “The Group as the Container of Generalised Reciprocity.” Social Psychology Quarterly 63: 116–32.