Temperament, Personality, and Psychopathology in Youth: A Preregistered Multilevel Meta-Analysis and Preregistered Large-Scale Replication and Extension

The current pair of highly-powered, preregistered studies estimate the associations between personality/temperament and youth psychopathology, and evaluate the generalizability of these associations across diverse populations and research designs. In Study 1, a multilevel meta-analysis of 147 studies (N = 46,369) quantified the associations between Five-Factor Model personality traits and internalizing/externalizing psychopathology dimensions among subjects aged 1.5 to 18. All traits but openness evidenced small-to-medium bivariate associations with both psychopathology dimensions (z = |.18 - .46|). After adjusting for the covariance between internalizing and externalizing, each dimension showed a distinct personality profile. Internalizing was associated with neuroticism (β = .42) and negative-extraversion (β = -.37). Externalizing was associated with extraversion (β = .30), negative-conscientiousness (β = -.46), and negative-agreeableness (β = -.50). Most variance was driven by informant effects, particularly within the externalizing dimension and when personality and psychopathology were informed by the same person (i.e., mono-informant). Study 2 was a preregistered replication and extension of Study 1 in a large cohort of youth (N = 10,414) using individual-participant, item-level data. Most of Study 1’s associations were replicated for higher-order scales (e.g., Effortful Control) and for their lower-order scales (e.g., Activation Control). Again, mono-informant effects were more pronounced than cross-informant effects. Taken together, these studies provide the first quantitative synthesis of associations between personality/temperament and psychopathology in youth. These preregistered quantitative findings are largely consistent with decades of theory (e.g., Tackett, 2006), evidence a robustness of effects across populations, and highlight the impact of mono-informant effects.