A taxometric investigation of the latent structure of eating disorders Bunmi O. Olatunji a, ⁎, Joshua J. Broman-Fulks b , Bethany G. Ciesielski a , Laci L. Zawilinski b , Shona Shewmaker c , David Wall c a Department of Psychology, Vanderbilt University, Nashville, TN, USA b Department of Psychology, Appalachian State University, Boone, NC, USA c Remuda Ranch Programs for Anxiety and Eating Disorders, Wickenburg, AZ, USA abstract article info Article history: Received 28 January 2011 Received in revised form 2 December 2011 Accepted 11 December 2011 Keywords: Eating disorders Dimensional Taxonic Taxometric The present study examined the latent structure of eating disorder symptoms in a large sample of patients with a diagnosis of anorexia nervosa restricting type, anorexia nervosa binge eating/purging type, and bulim- ia nervosa (n = 3747). Three taxometric procedures (MAXimum EIGenvalue (MAXEIG), Mean Above Minus Below A (MAMBAC), and Latent-Mode Factor Analysis (L-Mode)) were applied to self-reported symptoms of bulimia, drive for thinness, body dissatisfaction, as well as body mass index. Taxometric analysis among patients with the restricting and binge eating/purging subtype of anorexia and those with bulimia nervosa supported a dimensional latent structure of eating disorder symptoms. Taxometric analysis also revealed a dimensional latent structure of eating disorder symptoms among patients with the restricting and binge eat- ing/purging subtype of anorexia suggesting that the two anorexia subtypes may not represent discrete cate- gories. These findings suggest that the diagnosis and assessment of eating disorder symptoms should be conceptualized from a dimensional framework. © 2012 Elsevier Ireland Ltd. All rights reserved. 1. Introduction According to the Diagnostic and Statistical Manual of Mental Disor- ders (DSM-IV TR; American Psychiatric Association [APA], 2000), eat- ing disorders consist of distinct categories (e.g., anorexia nervosa) that are characterized by distinct symptoms (e.g., less than 85% of the normal weight). This view largely assumes that eating disorders are discrete entities that are qualitatively different from one another. Although some research has supported some of the distinctions drawn within the DSM-IV-TR among eating disorder subtypes (Keel et al., 2004), it has been observed that the assumption that eating dis- order diagnosis is entirely categorical may not have the level of con- struct validity that is often assumed, and therefore may not be the best system for advancing the field of eating disorders (Gordon et al., 2007). That is, it remains unclear if distinct types/subtypes are readily observed among those with eating disorders. For example, Eddy et al. (2008) found that the majority of women with anorexia nervosa experienced diagnostic crossover over a 7-year period: over half crossed between the restricting and binge eating/purging an- orexia nervosa subtypes; one-third crossed over to bulimia nervosa. The frequency of crossover between restricting-type and binge eat- ing/purging-type anorexia nervosa suggests that these two subtypes may not be distinct diagnostic categories. Delineating the extent to which eating disorder diagnosis is categor- ically distinct from each other may ultimately inform theoretical models of the diagnosis and classification of pathological eating behavior. It has been noted that taxometrics may help address this question in creating an eating disorder classification system with greater scientific validity and clinical utility (Wonderlich et al., 2007). Taxometrics is a set of sta- tistical procedures designed to discern the latent structure of phenom- ena by evaluating whether the numerical relations among various indicators of a conjectured taxon fit a pattern consistent with a naturally occurring class or a continuous dimension (Waller and Meehl, 1998). Accordingly, taxometrics may speak to whether eating disorder diagno- sis are dimensional or taxonic (representing nonarbitrary latent catego- ries whose members are qualitatively distinct from nonmembers). The procedures use multiple fallible, but non-redundant, indicators of a la- tent variable to distinguish between taxonic (i.e., categorical) and non- taxonic (i.e., dimensional) structures. Taxometric procedures involve graphing particular statistical summaries of the indicator interrelations and then identifying specific features in these graphs that indicate taxo- nic or dimensional latent structure. Quantitative features of these graphs are used to estimate critical parameters, such as the “base rate” or prevalence of any taxon that is obtained. Taxometric research examining the latent structure of eating dis- orders has begun to accumulate (Gleaves et al., 2000a, 2000b; Williamson et al., 2002). For example, Gleaves et al. (2000a, 2000b) found that although the subtypes of bulimia nervosa appeared differ- ent from one another only by degree, they were qualitatively differ- ent from normative eating behavior. Furthermore, the two subtypes Psychiatry Research 197 (2012) 97–102 ⁎ Corresponding author. Tel.: + 1 615 322 0060; fax: + 1 615 343 8449. E-mail address: olubunmi.o.olatunji@vanderbilt.edu (B.O. Olatunji). 0165-1781/$ – see front matter © 2012 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.psychres.2011.12.016 Contents lists available at SciVerse ScienceDirect Psychiatry Research journal homepage: www.elsevier.com/locate/psychres