Journal of Personality Disorders, 24(3), 327–343, 2010  2010 The Guilford Press MULTIDIMENSIONAL FACTOR STRUCTURE OF POSITIVE SCHIZOTYPY David C. Cicero, PhD, and John G. Kerns, PhD Schizotypy refers to traits similar to schizophrenia symptoms and is related to cluster A personality disorders. Previous factor analytic stud- ies have found a positive schizotypy factor distinct from a negative fac- tor. However, some evidence suggests that the positive factor may itself be multidimensional, but the factor structure of positive schizotypy is still unclear. The current study provided converging evidence through four different analyses that positive schizotypy is multidimensional. First, a factor model with three positive schizotypy factors (paranoia, referential thinking, and cognitive-perceptual) fit the data better than models with fewer than three factors. Second, a factor model with a second-order (i.e., higher-order) positive schizotypy factor fit the data significantly worse than a factor model without a second-order factor in which first-order factors were allowed to correlate freely, suggesting that the second-order factor does not completely account for relations among the first-order factors. Third, a Schmid-Leiman transformation found that even after accounting for the second-order factor that mean- ingful variance was attributed to the first-order factors. Finally, the three positive schizotypy factors displayed differential relations with five-factor model personality traits. Overall, results suggest that posi- tive schizotypy is composed of correlated but distinct factors. Schizotypy refers to traits or symptoms that are similar to symptoms of schizophrenia but in a diminished form, and research on schizotypy may provide insight into liability for schizophrenia while removing confounds associated with schizophrenia research (e.g., Chapman, Chapman, Rau- lin, & Edell, 1978; Neale & Oltmanns, 1980). Schizotypy traits are also associated with several personality disorders including schizotypal, schiz- oid, and paranoid personality disorders, which have overlapping diagnos- tic criteria and are often comorbid (Lenzenweger, Lane, Loranger, & Kessler, 2007). Dimensional models of personality disorders (e.g., the five-factor model; Widiger & Trull, 2007) have been unclear as to how well they can From University of Missouri. Work on this article was supported by National Institute of Mental Health Grant MH072706, National Institute on Drug Abuse Grant DA022405, and a MU Research Board grant. Address correspondence to John G. Kerns, University of Missouri, Department of Psychologi- cal Sciences, 214 McAlester Hall, Columbia, MO, 65211; E-mail: KernsJ@missouri.edu 327