Taxometric Analyses of Unipolar Depression

Solomon, Ari

Abstract

The proposed research is designed to resolve two crucial questions about unipolar clinical depression: First, does clinical depression differ categorically (i.e., qualitatively) from limited depressive symptoms, or do these phenomena differ only in degree? Second, does vulnerability to clinical depression occur along a continuum, or categorically? Resolution of these twin """"""""continuity questions of depression"""""""" would have important implications for future research on depression's etiology, prevention, and treatment. To our knowledge, all existing efforts to address these questions have relied on analytic strategies which cannot detect latent classes (such as discriminant function or multiple regression analysis), strategies which generate latent classes even when none exists in nature (such as cluster analysis and latent class analysis), inappropriate samples, and/or inappropriate class indicator variables. The proposed studies will eliminate these limitations by applying Meehl's taxometric method to large, longitudinal data sets of rigorously assessed community samples. The taxometric method's multiple consistency tests offer unrivalled power and specificity for the detection of latent taxa. Indicator variables will be derived initially from the Oregon Adolescent Depression Project (OADP). The OADP is an ongoing study of onset and recurrence of clinical depression and risk factors for internalizing psychopathology in a large, locally representative adolescent sample. The extended longitudinal nature of the OADP data set (the current wave of follow-up data is reassessing subjects at approximately age 30), its over sampling of high-risk adolescents, low attrition, and inclusion of sophisticated markers of depression risk afford an unique opportunity to adequately investigate this issue. If any discontinuities in the taxonic structure of depressive symptomatology and vulnerability are detected with the OADP data set, the findings will then be extended to adults and older adults by applying the taxometric method to large cross-sectional community samples.