We examined the association between personality and IL-6 in several ways. First, we examined the partial correlations between personality and IL-6, controlling for sex and age and the other covariates. In addition to the Pearson correlations, to ensure that the correlations are not distorted by outliers, we also present the non-parametric Spearman rank correlations on age and sex residualized scores. Second, we tested whether these associations were moderated by either sex or age, using Aiken and West's (1991) methodology for testing interactions. We used age both as a continuous variable and, because of a nonlinear increase in IL-6 in old age, a dichotomous variable split at age 65. Third, to follow-up on the broad domain-level findings, we examined the association between IL-6 and the more specific facets of personality. Fourth, we tested whether the findings held when individuals with acute inflammation were excluded from the analyses and if the findings were similar using a related marker of systemic inflammation, C-reactive protein. Fifth, we tested whether the covariates mediated the association between personality and IL-6. Finally, we used logistic regression to test whether personality predicts which participants would exceed clinical thresholds for increased risk of disability and mortality.
