Assessment of anxiety in youth : a confirmatory factor analysis study of construct validity using multitrait-multimethod models

This dissertation examines the construct validity of separation anxiety disorder, social phobia, and generalized anxiety disorder (GAD) in a sample of clinically anxious children. Participants were 572 children, 6 to 17 years old (294 boys) who consented to participate in anxiety assessment and treatment. Data for 85% of the sample came from a federally-funded, multi-site RCT. The remaining 15% came from an intervention enhancement study conducted at a mental health clinic associated with a large public university. Clinician ratings of child anxiety were obtained from the Pediatric Anxiety Rating Scale (PARS) and the Anxiety Disorders Interview Schedule for Children and Parents (ADIS: C/P). Parent and child ratings of symptom severity were examined using the Screen for Child Anxiety and Related Disorders (SCARED) and the Multidimensional Anxiety Scale for Children (MASC). Discriminant and convergent validity were assessed using confirmatory factor analytic (CFA) techniques to test a multitrait-multimethod (MTMM) model. CFA analyses of the data supported the conceptualization of child anxiety as three separate traits. Separation anxiety disorder, social phobia, and GAD can be reliably differentiated in children. Discriminant validity of traits was supported because the three anxiety disorders showed meaningful divergence in the data by trait. Convergent validity of traits was also supported because the three-trait model fit the data considerably better than a no-trait model. Further, evidence for discriminant validity of methods showed that each informant provided unique information about anxiety symptoms. The best model fit was obtained for a subsample of 314 children younger than age 11. A review of factor-loadings for this model showed that separation anxiety and GAD were best measured by parent reports, and the clinician report had the strongest influence on social phobia. Specifically, each of three subscales of the SCARED-P and the ADIS: C/P for social phobia proved to be the most informative evaluation tools. As a result of this analysis, clinicians may be more confident that the current classification system of anxiety disorders is accurate and that discordant informant reports encountered in practice are more a reflection of unique perspective than of poor construct validity.