A replication attempt of personality change through randomized peer groups

Personality development theories and descriptive evidence highlight young adulthood as a period conducive to personality change. Young adults experience important transitions, such as starting university education, and establish new relationships. However, few empirical tests of the direct influence of university student peer groups on individual change in personality traits exist. Here, we conceptually replicated an earlier study and randomly allocated students (N = 351) to study groups of 3 to 6 peers that met over the course of the first semester. This randomization approach allowed the estimation of peer group socialization effects unconfounded by selection effects. We conducted a series of robustness checks to determine whether regression to the mean constitutes an alternative explanation for the results. The preregistered, confirmatory models seemed to support the hypothesis of peer group influence for all Big Five traits, showing patterns of assimilation where an individual’s personality became more similar to the personality of the remaining group members over time. Robustness checks, however, demonstrated convincingly that these patterns were due to the statistical artifact of regression to the mean and should not be interpreted as substantive evidence for peer group influence. We discuss theoretical and methodological implications for studying personality development through social relationships.