Examining the Factor Structure and Incremental Utility of the Contrast Avoidance Questionnaires via Bifactor Analysis

Abstract

The negative emotional contrast avoidance model posits that pathological worry is maintained by the avoidance of negative emotional shifts. The Contrast Avoidance Questionnaires (CAQ–Worry and CAQ–General Emotion) aim to assess contrast avoidance beliefs and behaviors. Questions remain around the factor structures of the CAQs, whether such structures replicate in nonundergraduate samples, and whether their domain-specific factors are valid for independent use. This study used bifactor analysis in a large community sample (N = 827) to address these gaps in the literature. Results supported bifactor models of both measures. Complementary analyses supported the multidimensionality of the CAQ–Worry, including its strong general factor and independent use of two domain-specific factors, rather than the original three domain-specific factors. The CAQ–General Emotion’s general factor was strong, but the merits of the Discomfort domain-specific factor require more exploration, and use of the Avoidance domain-specific factor is discouraged.

Notes

1 The hierarchical model of the CAQ–W was compared to the two-factor and three-factor bifactor models in terms of variance explained (R²) in PSWQ–10 and GAD–7 latent scores (all ps < .001). When predicting worry, both the two-factor and three-factor bifactor models explained 9% to 10% more variance in worry (R² = .53 and .54, respectively) than the CAQ–W hierarchical model (R² = .44) and 9% to 11% more variance in GAD symptoms (R² = .43 and .45, respectively) than the CAQ–W hierarchical model (R² = .34).

2 The correlated factors model of the CAQ–GE was compared to its bifactor counterpart in terms of variance explained (R²) in PSWQ–10 and GAD–7 latent scores (all ps < .001). The CAQ–GE bifactor model explained an equivalent amount of variance in worry (R² = .57) and GAD symptoms (R² = .47) compared to the CAQ–GE correlated factors model (R² = .56 and .48, respectively).