Associations between self-rated personality traits and automobile fatality rates across small geographic areas

We explore how variation in fatal automobile crashes across small geographic areas is associated with geographic variation in self-rated psychological and behavioral characteristics. Specifically, estimates of ZIP-code-level automobile fatality rates were linked to a separate dataset comprising 2.8 million responses to a widely used self-report personality questionnaire. To control for an area's wealth and population density, associations were estimated at the slightly larger level of US Census ZIP Code Tabulation Areas (ZCTAs). Relationships were estimated using an intraclass adjustment method, where the estimated correlation between individual scores and ZCTA-level fatality rates was divided by the square root of the estimated within-ZCTA intraclass correlation. Further, the ZCTA-level correlation was estimated separately for each US state to examine the consistency of associations. Rates of fatal crash involvement tended to be higher in areas where respondents were more likely to describe themselves as being depressed, moody, and quarrelsome. Some relationships were less intuitive; for instance, rates of fatal crash involvement tended to be higher in areas where respondents described themselves as more helpful and as less easily distracted. However, many small-area-level associations between self-rated personality and fatality rates were found to be in the same direction across more states than expected by chance. This pattern indicates that even many of the less intuitive associations were often highly reliable and general, which has implications for identifying geographic areas with heightened automobile fatality risk.