Direkt zum Inhalt

Using Within-Person Change in Three Large Panel Studies to Estimate Personality Age Trajectories

SOEPpapers 1191, 86 S.

Ingo S. Seifert, Julia M. Rohrer, Stefan C. Schmukle


get_appDownload (PDF  1.36 MB)


How does personality change when people get older? Numerous studies have investigated this question, overall supporting the idea of so-called personality maturation. However, heterogeneous findings have left open questions, such as whether maturation continues in old age and how large the effects are. We suggest that the heterogeneity is partly rooted in methodological issues. First, studies may have failed to recover age effects, as they did not stringently separate within-person changes from confounding between-person differences. Second, items supposedly belonging to the same trait may show different individual trajectories, thus rendering results sensitive to the specific set of items used. We analyzed panel data from Australia (N = 15,268; Study 1), Germany (N = 22,833; Study 2), and the Netherlands (N = 10,163; Study 3) to investigate age trends in the Big Five on the levels of both scores and items. We applied a fixed effects approach that incorporates only within-person changes over time. Developmental trends in the Big Five scores were generally moderate to large and broadly confirmed personality maturation at younger ages. At older ages, maturation consistently continued for Neuroticism, whereas we found mixed evidence for such changes in Conscientiousness and Agreeableness. Furthermore, in each study, individual items showed age trends that diverged from the rest of the corresponding trait; and these differential patterns could be partly replicated across the three studies. Our results highlight the importance of items in the study of personality development and provide an explanation for previously unaccounted for variability in age trends.

Topics: Personality, Family

Keywords: Personality development, mean-level change, Big Five, panel studies, fixed effects modeling
Frei zugängliche Version: (econstor)