[OSF; peer review] How does economic inequality relate to prosocial behavior? Existing theories and empirical studies from multiple disciplines have produced mixed results.
Here we conduct a systematic review and meta-analysis to systematically synthesize empirical studies.
Results: from 192 effect sizes and over 2.5 million observations in 100 studies show that:
the relationship varies from being negative to positive depending upon the study (95% prediction interval −0.450–0.343). However, on average, there is a small, negative relationship between economic inequality and prosocial behavior (r = −0.064, p = 0.004, 95% confidence interval −0.106 to −0.021).
There is generally no evidence that results depend upon characteristics of the studies, participants, the way prosocial behavior and inequality were assessed, and the publication discipline.
Given the prevalence of economic inequality and the importance of prosocial behavior, this systematic review and meta-analysis provides a timely study on the relationship between economic inequality and prosocial behavior.
…Publication bias: We deployed several approaches to test and remedy potential publication bias. The first was using the funnel plot and trim-and-fill approach78. As shown in Figure 3, the distribution in the funnel plot was roughly asymmetric. We used the trim-and-fill approach to assess the symmetry and adjust for any bias. However, results showed that the number of samples and the average effect size remained the same, indicating that there was not substantial publication bias. Next, the Egger’s test was used to check publication bias79. The Egger’s test results suggested a potential concern of publication bias (z = −3.32, p = 0.001). We also examined whether there was a statistically-significant difference in effect sizes between published and unpublished studies, but publication status was not a statistically-significant moderator, Q(1) = 0.24, p = 0.627. Hence, we argue that publication bias was not a serious issue in our meta-analysis. [With a funnel plot like that?]
…In addition, the Q-statistic indicated that there was much heterogeneity (Q(191) = 11,192.948, p < 0.001) and the I2-statistic showed that 99.82% of heterogeneity cannot be attributed to sample error. Thus, it was appropriate to use random-effects meta-analysis. The variance (sigma-between 0.201, sigma-within 0.072) suggested that between-study variance was higher than within-study variance.
…Conclusion: How does economic inequality relate to prosocial behavior? In this study, we conducted a meta-analysis to systematically synthesize empirical studies. We found the relationship varies from being negative to positive depending upon the study, but on average there is a small, negative relationship between economic inequality and prosocial behavior. Moderator tests demonstrated that the results were generally robust across study characteristics, participants, measures of prosocial behavior and inequality, and publication disciplines. The fact that economic inequality tends to be associated with less prosocial behavior should perhaps make us wonder if the societal costs of inequality are worth it.
[I don’t know how you get such wild differences in effect sizes with extraordinary heterogeneity, which are completely unmoderated by any of the variables (despite including moderators like study quality, publication-year, published vs non-published, experimental vs non-experimental, Africa versus America, public goods vs dictator game, quantified dozens of different ways, income vs wealth inequality, and units of analysis ranging from individual to country [!]), yielding a misshapen funnel plot, for an average effect extraordinarily close to zero and well within ‘crud factor’ systematic bias range, and conclude that there are any ‘societal costs’ at all!]