Reducing Prejudice: A Meta-Analysis and Experiment on the Contact Hypothesis

For decades, intergroup contact has been considered one of the most important tools for reducing prejudice and improving intergroup relations. This paper meta-analyses the experimental literature on the contact hypothesis. The analysis is based on 209 outcomes from 61 papers, involving more than 30,000 people. The main finding is that contact interventions are effective in reducing prejudice, with an average effect of approximately 0.3 standard deviation and few instances of backlash. However, I identify several worrying signs of publication bias and p-hacking, although these do not impact the significance of the main result. I also find substantial heterogeneity in contact effects, but the characteristics of contact interventions considered in the literature have limited power to explain this heterogeneity. In particular, the common-goals condition, often emphasized as beneficial, is negatively correlated with effect size, although this relationship is descriptive. An analysis using external participants and Large Language Models also reveals considerable diversity in how contact interventions satisfy the conditions emphasized in the literature. Overall, the findings suggest that while contact interventions can reduce prejudice, the evidence currently provides limited guidance on which types of contact are most effective, highlighting the need for experiments that directly manipulate contact characteristics and for a better theoretical understanding of the mechanisms underlying contact effects.