Common Versus Unique Variance Across Measures of Worry and Rumination: Predictive Utility and Mediational Models for Anxiety and Depression

Repetitive negative thinking (RNT) has been identified as a transdiagnostic construct. However, diagnosis-specific questionnaires have traditionally been used to measure RNT across emotional disorders, and thus the degree to which they assess shared versus unique aspects of RNT is unclear. Furthermore, the degree to which shared versus unique variance across these measures contributes to the prediction of anxiety and depression symptoms is yet to be fully understood. This study had three aims. First, confirmatory factor analysis was used to test the degree to which two common, diagnosis-specific questionnaires assess common versus unique variance in RNT. One questionnaire measured worry whereas the other measured two aspects of rumination (brooding, reflection). Second, the contribution of the shared and unique variance in predicting symptoms of anxiety and depression was determined. Third, the role of shared and unique variance in mediating the relationships between the vulnerability factor of negative affectivity and symptoms of anxiety and depression was assessed. Participants (N = 450) presented for treatment at a specialist clinic with anxiety and affective disorders (54% women). A nested four-factor model (RNT, worry, reflection, brooding) provided a superior fit to two-factor (worry, rumination) and three-factor (worry, reflection, brooding) oblique models. Structural equation modeling showed that RNT, brooding, and worry uniquely predicted anxiety and depression, and reflection also predicted depression. RNT partially mediated the relationship between negative affectivity and anxiety, and brooding partially mediated this relationship for depression and anxiety. Our findings suggest that both RNT and brooding may be separable and transdiagnostic constructs.