We conducted two series of logistic regression analyses to examine the influence of personality traits on incident major and minor depression. First, we examined the independent influence of the five NEO-FFI domain scores and their interactions, controlling for covariates. These analyses informed our approach to the hypothesis-generating analyses. Second, we examined the effect of personality styles. As described in the Introduction, the hypothesis-testing analyses focused on the joint contributions of Neuroticism and Extraversion (N+E−) as well as Neuroticism and Conscientiousness (N+C−). To reduce the number of potential hypothesis-generating analyses we concentrated on those domains that were shown to confer risk in the first set of analyses. For example, if a high level of Neuroticism was shown to confer risk, we examined whether it, in combination with high or low levels of Openness and Agreeableness and low levels of Extraversion and Conscientiousness elevated risk (analyses of the combinations of high Neuroticism with low Extraversion or low Conscientiousness were construed as hypothesis-testing). In the first block, we entered all the covariates. In the second block, we entered four dummy-coded (0 or 1)
