Cross-Measure Equivalence and Communicability in the Assessment of Depression

Abstract

All measures of depression yield a global summary scale indicating the severity of depressive symptoms, implicitly conceptualized as a homogeneous construct. However, depression is a heterogeneous construct, with different presentations, subtypes, correlates, and responses to interventions. In response, the National Institute of Mental Health (NIMH) has suggested changes in the way depression is assessed, moving the focus to specific factors, such as cognitive, somatic, or affective symptoms. Still, there is little factor overlap between measures, and shared factors are weighted differently. To help fulfill NIMH’s strategic plan, this study used canonical correlation analysis (CCA) to explore shared latent variables and redundancy across the measures. It also analyzed the psychometric properties of factor-based subscales in the Beck Depression Inventory–2nd edition (BDI-II), Center for Epidemiologic Studies Depression scale (CES-D), Inventory for Depression and Anxiety Symptoms (IDAS), and Inventory of Depressive Symptomatology (IDS). Using a diverse sample of 218 students who reported at least mild depressive symptoms, this study found that the IDAS was best aligned with NIMH’s strategic plan; it has complete DSM-IV/DSM-5 symptom coverage and content-valid, psychometrically sound subscales. The BDI-II, CES-D, and IDS did not have consistent subscales, nor had incomplete or incongruent coverage of DSM criteria. Furthermore, CCA revealed low redundancy across measures (23% to 41% shared variance). These results suggest that different measures of depression do not measure the same construct. As a partial solution, empirical conversion tables were provided for researchers and clinicians to empirically compare total scores from different measures.