The relation between anxiety and depressive disorders in adults and children has received considerable attention in the past two decades. Clark and Watson (1991) proposed a tripartite model of anxiety and depression that explains the overlap between anxiety and depression. According to this model, anxiety and depression share a large component of negative affectivity (NA), which is a broad general factor of emotional distress. The second factor in the model, positive affectivity (PA), is relatively specific to depression and represents the degree of pleasurable engagement with the environment. The third factor in the model, physiological hyperarousal (PH), is relatively specific to anxiety and reflects somatic tension and arousal. Studies with adults have indicated that the tripartite model is able to address the issue of specificity in diagnosis and has implications for research and practice (e.g., Clark et al. 1994; Watson et al., 1988, 1995). Despite the strength of findings with adults, relatively few studies have examined the applicability of the tripartite model for children. Lonigan, Carey, and Finch (1994) found that anxious and depressesed children reported similarly high levels of NA; in contrast, only depressed children reported low levels of PA. Recently, Joiner, Catanzaro, & Laurent (1996) conducted an exploratory factor analysis on anxious and depressive symptoms in a group of 116 inpatient children and found evidence for the three factors, suggesting that the structure of the tripartite model could be applied to children.

The present study examined the applicability of the tripartite model of anxiety and depression in a sample of 280 children and adolescents who were admitted to an inpatient psychiatric facility because of problems associated with depression or anxiety. The children ranged in age from 6 to 17 years (M=11.02, SD=2.91) and had completed the Children's Depression Inventory (CDI) and the Revised Children's Manifest Anxiety Scale (RCMAS) as part of the standard assessment procedure. Using the items from the CDI and the RCMAS selected by Joiner et al. (1996) to serve as indicators of NA, PA, and PH, confirmatory factor analyses were conducted to test the fit of three-, two-, and one-factor models. These analyses revealed a significant difference between the three-factor model (CFI = ..80) and the two-factor model (CFI = .77; c2 difference = 22.7, df = 2, p<.001), and the one-factor model (CFI = .76, c2 difference = 30.6, df = 3, p<.001). Although the three-factor model had the best fit, the value of the comparative fit index suggested the model provided only a marginal fit to the data (i.e., CFIs > .90 indicate a good fit). Correlations between the latent factors indicated substantial factor overlap, particularly for NA-PA and NA-PH. According to the tripartite model, NA and PA are orthogonal factors. Therefore, a final analysis examined the effect of constraining the correlation between the NA and PA factor to zero. This constraint on the model resulted in a significant decrement in the fit of the model to the data (CFI = .78, c2 difference = 14.84, df =1, p<.001). These results provide partial support for the tripartite model in children. Although further refinement of measures of the tripartite model in children is required, the model provides a useful means of differentiating anxiety from depression and provides a means of identifying children potentially at risk for these disorders prior to the onset of significant symptomatology.