Partialing alters interpersonal correlates of negative affective symptoms and traits: A circumplex illustration.

Abstract

Negative affective symptoms (e.g., anxiety, depression, and anger) are correlated and have parallel associations with outcomes, as do related personality traits (i.e., facets of neuroticism), often prompting statistical control (i.e., partialing) to determine independent effects. However, such adjustments among predictor variables can alter their construct validity. In three studies, the interpersonal circumplex (IPC) and a related analytic approach (i.e., Structural Summary Method) were used to evaluate changes in interpersonal correlates of negative affective characteristics resulting from partialing. Samples of undergraduates (Sample 1 n=3283; Sample 2=688) and married couples (n=300 couples) completed self-report (three samples) and partner rating (sample 3) measures of anxiety, depression and anger, and IPC measures of interpersonal style. Anxiety, depression, and anger had expected interpersonal correlates across samples. Partialing depression eliminated interpersonal correlates of anxiety. When anxiety was controlled, depression measures were more strongly associated with submissiveness and less closely associated with low warmth. Adjustments involving anger magnified differences in dominance versus submissiveness associated with the negative affects. Removal of overlap among negative affective measures via partialing alters their interpersonal correlates, potentially complicating interpretation of adjusted associations.