This study examines the validity of different measurement methods for assessing personality traits and their implications for studying political preferences. While self-reports are the predominant approach in the literature, they suffer from limitations, including measurement error and reverse causality. To address these issues, alternative strategies have been proposed, such as third-party assessments (e.g., parental reports) and lagged self-reports. Using longitudinal data from the Growing Up in Ireland dataset we compare four approaches: Self-reports, parental reports, lagged self-reports, and an instrumental variable strategy using lagged self-reports as instruments for contemporaneous ones to understand how these methods capture the Big Five personality traits and their association with political attitudes.
Our results reveal crucial differences between these methods. First, parental ratings of personality traits show minimal alignment with self-reported measures, both in terms of raw relationships (slopes and explained variation) and construct validity (stability over time). Parental ratings exhibit low autoregressive consistency explaining at most 5% of the variation in personality traits across waves, raising doubts about their reliability. Second, when examining the relationship between personality traits and political attitudes, self-reports and lagged self-reports provide consistent results, suggesting robustness to endogeneity concerns. However, the instrumental variable strategy significantly increases the estimated coefficients, highlighting pervasive measurement error in self-reported traits. Overall, our study highlights the limitations of third-party assessments such as parental reports and the need for researchers to carefully consider measurement error when studying personality traits and their political implications.