Some challenges in personality trait-health outcome research

Overwhelming evidence suggests that personality test scores are correlated with a range of health-related variables, both cross-sectionally and longitudinally. Although (explicitly) researchers often interpret these correlations cautiously, I argue that we cannot escape aiming at causal inferences and implicitly we often draw them. I will talk about two conditions pertinent to making causal inferences. First, for traits to be able to be causal, they have to reflect real and holistic attributes that exist independently of our operationalizations of them. This condition may be hard to verify, so it may often need to remain an assumption. Second, we should demonstrate that trait-outcome associations are independent of how the traits are operationalized. If the associations depend on particular items or facets that the trait-measures at hand happen to contain, then they likely pertain to these items or facets rather than the traits as such, and should be interpreted accordingly.