Latent structure of self-reported depression in undergraduates: Using taxometric procedures and information-theoretic latent variable modeling Yasuyuki Okumura a, * , Shinji Sakamoto b , Atsuko Tomoda c , Nobuhiko Kijima d a Graduate School of Literature and Social Sciences, Nihon University, 3-25-40 Sakura-josui, Setagaya-ku, Tokyo 156-8550, Japan b Department of Psychology, College of Humanities and Sciences, Nihon University, Japan c Faculty of Human and Social Studies, Saitama Institute of Technology, Japan d Faculty of Business and Commerce, Keio University, Japan article info Article history: Received 6 April 2008 Received in revised form 14 September 2008 Accepted 26 September 2008 Available online 12 November 2008 Keywords: Depression Taxometric Continuity issue Analogue study Validity abstract Many depression studies dichotomize the participants as depressed or nondepressed on the basis of a self-reported depression scale score. However, there is little statistical rationale for such a procedure. The purpose of this study was to clarify whether depression assessed by the Zung Self-rating Depression Scale (SDS) is categorical or dimensional by using taxometric procedures and information-theoretic latent variable modeling. The present sample comprises 2187 Japanese undergraduates. The latent struc- ture of depression was evaluated by mean above minus below a cut (MAMBAC) and maximum eigen- value (MAXEIG). In addition, in order to test convergent validity, we performed information-theoretic latent variable modeling. Both taxometric procedures (MAMBAC and MAXEIG) and information-theoretic latent variable modeling, using indicators drawn from the SDS, were consistent with the dimen- sional latent structure of depression. The latent structure of depression assessed by the SDS is dimensional in undergraduates. For the assessment of depression, researchers should not stratify the participants on the basis of a conventional cutoff score. In addition, it is recommended that researchers use correlational designs, which include individuals with a wider range of severity with regard to depressive symptoms, rather than group comparative research designs. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction Many depression studies in personality and social psychology have used undergraduates as participants. Typically, in these stud- ies, the Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) is administered to a sample of undergrad- uates. Participants are then categorized as depressed or nonde- pressed on the basis of their BDI scores (Vredenburg, Flett, & Krames, 1993). Such a procedure is referred to as dichotomization (Cohen, 1983). Okumura, Kameyama, Katsuya, and Sakamoto (2008) reviewed 974 depression studies in 18 Japanese journals published between 1990 and 2006, and found that 26% of studies used dichotomization. Dichotomizing a continuous variable implicitly assumes that there actually exist distinct groups (i.e., de- pressed or nondepressed) of individuals on the variable in question (MacCallum, Zhang, Preacher, & Rucker, 2002). One of the central debates in depression literature is whether depression is categorical (i.e., misery or distress might be qualita- tively different from diagnosable depression) or dimensional (i.e., depression exists along continua with normality) in nature. There are staunch advocates of both positions (e.g., Coyne, 1994; Vredenburg et al., 1993). The proponents of the categorical model tend to maintain that milder forms of depression are limited to a dysphoric mood state, whereas clinical depression is a distinct en- tity. In contrast, the advocates of the dimensional model tend to maintain that mild depressive states in persons drawn from non- clinical populations represent the low end of a continuum of sever- ity, with clinical depression at the opposite pole. Flett, Vredenburg, and Krames (1997) described four ap- proaches that have been applied to investigate whether depression is categorical or dimensional, namely, the phenomenological, typo- logical, etiological, and psychometric approaches. Recognizing the methodological and statistical limitation of existing evidence, Flett et al. (1997) called for studies utilizing sophisticated statistical techniques, called taxometric procedures. Taxometric procedures provide an empirical means of determining whether a latent vari- able (e.g., depression) is typologically distinct from normal behav- ioral functioning (Ruscio, Haslam, & Ruscio, 2006; Waller & Meehl, 1998). Taxometric procedures have been performed to investigate the latent structure underlying various mental disorders (Haslam & Kim, 2002). To our knowledge, three taxometric studies have explored the latent structure of depression, using a sample of undergraduates. 0191-8869/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.paid.2008.09.021 * Corresponding author. Tel.: +81 3 5317 9720; fax: +81 3 5317 9427. E-mail address: yokumura@blue.zero.jp (Y. Okumura). Personality and Individual Differences 46 (2009) 166–171 Contents lists available at ScienceDirect Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid