BRIEF REPORTS Pathological Gambling Subtypes David D. Vachon Purdue University R. Michael Bagby Centre for Addiction and Mental Health, University of Toronto Although pathological gambling (PG) is regarded in the 4th edition of the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 1994) as a unitary diagnostic construct, it is likely composed of distinct subtypes. In the current report, the authors used cluster analyses of personality traits with a non-treatment-seeking community sample of gamblers and identified 3 PG subtypes. Gamblers partitioned into a simple PG cluster, characterized by low rates of comorbid psychopathology and trait scores near the normative mean; a hedonic PG cluster, characterized by moderate rates of comorbid psychopathology and a proclivity for excitement seeking and positive affect; and a demoralized PG cluster, characterized by high rates of comorbid psychopathology and a propensity toward negative affect, low positive emotionality, and disinhibition. The findings provide preliminary support for an empirically based typology of gamblers, distinguishable in terms of personality structure, which may reflect different etiologies. Keywords: pathological gambling, five-factor model of personality, personality disorders, gambling subtypes, cluster analysis Pathological gambling (PG) places an estimated $5 billion bur- den on American society and is associated with diminished quality of life and increased financial, domestic, and legal difficulties (National Gambling Impact Study Commission, 1999). Currently, treatment for PG is influenced by an array of perspectives, each of which acknowledges the interaction of biopsychosocial variables in the etiology of PG but stresses a different set of critical pro- cesses. Although these models typically conceptualize pathologi- cal gamblers (PGs) as a single, homogenous population, converg- ing lines of investigation suggest that PGs may form a heterogeneous group composed of qualitatively unique subtypes (Blaszczynski & Nower, 2002; Ledgerwood & Petry, 2006; Steel & Blaszczynski, 1998). An empirically derived taxonomy of PG may enhance therapeutic outcomes by providing specific guide- lines on how to tailor treatment for different types of PGs. PG is typically linked to impulsive, externalizing psychopathology, such as antisocial personality disorder (Petry, Stinson, & Grant, 2005) and drug and alcohol disorders (Slutske, Caspi, Moffitt, & Poulton, 2005). Genetic data from more than 8,000 twins included in the Vietnam Era Twin Registry suggest a hereditary overlap between PG and antisocial behaviors (Slutske et al., 2001), as well as alcohol dependence (Slutske et al., 2000). A second literature links PG with internalizing psychopathol- ogy; for example, the adjusted odds ratios for internalizing disorders in PGs range from 3.9 to 4.4 for anxiety and mood disorders, respectively (Petry et al., 2005), and PG has more substantial genetic correlations with depression than with any other psychiatric disorder (Potenza, Xian, Shah, Scherrer, & Eisen, 2005). Finally, PG has also been linked to maladaptive cognitions, “normal” conditioning processes, and ecological factors such as accessibility and availability. Blaszczynski and Nower (2002) proposed three subtypes of problem gamblers, each described by a separate pathway culmi- nating in PG. This qualitative taxonomy of gambling subtypes remains empirically untested and influenced largely by investiga- tions of nonrepresentative, treatment-seeking PGs (which embody only 7%–12% of all PGs; Slutske, 2006). In the current investi- gation, we sought to identify empirically derived PG subtypes in a community sample of non-treatment-seeking PGs by using a widely accepted model of personality—the five-factor model (FFM). The FFM, as measured by the Revised NEO Personality Inventory (NEO-PI-R; Costa & McCrae, 1992), offers an appeal- ing framework for deriving PG subtypes; for example, it includes four lower order facet traits that predict 66% of the variance in the most commonly used measures of impulsivity (Whiteside & Lynam, 2001); as PG is defined as an impulse-control disorder, this predictive capacity is particularly important. Also, because the FFM was derived from basic research on personality in an attempt to organize the trait terms of the English language rather than to predict certain criteria or pathology, one begins with simple traits that were developed independently of PG criteria and contain little representation of explicit gambling behaviors. Finally, substantial evidence suggests that FFM traits have high reliability, stability, David D. Vachon, Department of Psychological Sciences, Purdue Univer- sity; R. Michael Bagby, Centre for Addiction and Mental Health, Department of Psychiatry, University of Toronto, Toronto, Ontario, Canada. This study was supported by an operating grant from the Ontario Problem Gambling Research Centre to R. Michael Bagby (principal inves- tigator). The authors acknowledge the earlier contributions of Peter Far- volden, Jon Oakman, and Tony Toneatto and thank Lena Quilty, Martin Sellbom, and Shauna Kushner for their helpful comments. Correspondence concerning this article should be addressed to R. Michael Bagby, Centre for Addiction and Mental Health, 250 College Street, Suite 647A, Toronto, Ontario M5T 1R8, Canada. E-mail: michael_bagby@ camh.net Psychological Assessment © 2009 American Psychological Association 2009, Vol. 21, No. 4, 608 – 615 1040-3590/09/$12.00 DOI: 10.1037/a0016846 608