Abstract
The structure of self-reported symptoms representative of the tripartite model was examined using data drawn from the Medical Outcomes Study (Tarlov et al., 1989). Participants were persons who had been diagnosed 48 months previously as suffering from either depression (N = 315) or hypertension (N = 403). Results of confirmatory factor analyses were broadly consistent with the tripartite model (L. A. Clark & Watson, 1991). Factors emerged corresponding to each of the 3 posited first-order dimensions of negative affect, positive affect, and physiologic arousal. Nonetheless, some discrepancies were found between the observed data and the hypothesized tripartite model. First, the obtained physiologic arousal factor was best viewed as reflecting nonspecific somatic distress rather than physiologic arousal. Finally, although differentiable in the strictest statistical sense, all three domains were significantly correlated (.36 to .86, absolute value). In particular, contrary to the tripartite model, positive and negative affect covaried markedly (-.81 to -.86). Findings raise issues concerning the utility of the tripartite model as a heuristic framework for enhancing understanding of individual differences in normal mood as well as mood disorders.