Why are depression and anxiety correlated? A test of the tripartite model

L. A. Clark and D. Watson's (1991) tripartite model groups the symptoms of depression and anxiety into 3 components: nonspecific symptoms of general distress, which do not distinguish depression and anxiety; physiologic arousal, which is relatively unique to anxiety; and anhedonia (or low positive affect), which is unique to depression. Structural equation modeling was used to test this model with self-report data from 3 different samples: outpatients seeking treatment for mood disorder, anxiety disorder, or both (n = 483) outpatients seeking treatment for substance abuse (n = 453), and college students (n = 516). The tripartite model did not fit any of the sample covariance matrices, because the nonspecific symptoms of depression and anxiety could not be adequately represented by a single General Distress factor. An alternative model, in which the Anhedonia and Nonspecific Depression factors loaded on a second-order Depression factor, while the Somatic Arousal and Nonspecific Anxiety factors loaded on a second-order Anxiety factor, produced an excellent fit in all the groups. The Nonspecific Depression and Nonspecific Anxiety factors were the most valid and specific indicators of depression and anxiety, respectively. Anhedonia and Somatic Arousal were significantly less valid measures of depression and anxiety. The implications of these findings are discussed.