Predicting More Behavior More of the Time: On the Behavioral Nature of Different Personality Trait Measures

Most research focuses on self-reported global trait ratings to approximate behavioral individual differences. However, several alternatives exist. Average self-reported states from experience sampling have been suggested to be closer to behavior, but little evidence supports this to date. Further, other-report should rely more strongly on observable behavior. Here, we predict behavior from different personality trait measures, highlighting type of aggregation (global vs. average state measures) and perspective (self- vs. other-report). We provide initial evidence with a multi-methodological study (N = 88) including six trait measures: global self-report, global informant-report, time-based self-reported states (fortnightly diaries), time-based peer-reported states, event-based self-reported states (after social interactions), and event-based peer-reported states. Measures were compared in the prediction of actual behavior in the laboratory one year later. We focused on four behaviors (warmth, expressiveness, self-confidence, nervousness) and corresponding global/state dimensions. Self-reported states were rarely incrementally predictive above global self-report and thus not as close to behavior as commonly believed. Generally, other-report was more consistently linked to behavior than self-report. Notably, other-reported states were substantially predictive beyond global informant-report and self-reported states, representing an underused method closer to behavior. We outline key future directions to go beyond self-report in behavioral personality science.