https://doi.org/10.5061/dryad.vq83bk48r

Authors: Bonaventura Majolo, Federica Amici, Eliza Florence Dunn, Robin Watson

Date: 2026

Study description: A meta analysis assessing the cultural differences in pro- and antisocial behaviour across development

Licence:

CC0

Files:

Dataset.csv - The dataset used for the analysis

Analysis_script.Rmd - The R script used for any further processing of the data and the analysis.

prosocial_gamma_filtered.rds; prosocial_gamma_pubyear.rds; prosocial_gamma_yearcollected.rds; antisocial_gamma_pubyear.rds; antisocial_gamma_collected_year.rds; comparison_check.rds - R objects containing the fit rstan models for replicating the outputs provided in the manuscript. See the R notebook file for further details

Software:

R (version 4.6.0); packages: tidyverse, ggplot2, rstan, tidybayes, cowplot, patchwork

Data columns:

• Study_numeric - A numeric representation of the Study. Each Study has a unique identifying number.
• Study - The citation for where the data was extracted
• Dyad - The dyadic comparison for which the effect size refers
• absolute_G - Hedges' g calculated from the means, standard deviations, and N of the data points from the studies. Note that this raw data is not made publicly available based on the request of some of the original authors
• Gender - Whether the datapoint concerns Male, Female, or Mixed gendered participants
• Method - Which study method was used by the primary study to produce the data
• Average_age - Mean age of the participants
• pub.year - The year of publication of the paper
• data_collected - The stated year of data collection from the paper. In cases where this was not specified, this is assumed to match the publication year
• Behaviour - Whether the behaviour is prosocial or antisocial (see the manuscript for further details)
• total_N - The total sample size of the study
• Distance - The distance between the two countries/cultures in km

Data collection and pre-processing:

The data provided in the spreadsheet collates data from 69 publications extracted from the psychological literature. When a study contained data on more than two countries, we paired each country with all the other countries included in the study to maximise the number of comparisons we could analyse. For example, if a study contained data on country A, B, and C, we paired the data so that we could include the following pairwise comparisons: A-B, A-C, and B-C. All the studies include in our analyses provided the means and SD for each country, so that we did not have to run any data transformation (e.g., from SE to SD) and pre-processing. The outcome variable Hedges' g was calculated on a more detailed dataset which contained the means and standard deviations for each of the two countries/cultures listed in the study dyad. We are unable to make this detailed dataset publicly available based on requests from several authors of the primary studies to not openly share their data.

Purpose of the R code:

The R code is used to A) Calculate the descriptive statistics presented in the paper; B) fit several Bayesian models to the data to assess our RQ; C) Produce accompanying visualisations. Additionally, some further necessary data tidying and exclusions are done in R. In total, 100 lines of data are excluded based on either A) missing hedges' g values or B) missing information on the study method.

Human subjects data

The data in this submission is meta-analytical data extracted from published primary research. No human participants are identifiable in this dataset.