The structure of personality pathology: Both general ('g') and specific ('s') factors?

Recent editions of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5; American Psychiatric Association, 2013) conceptualize personality disorders (PDs) as categorical constructs, but high PD co-occurrence suggests underlying latent dimensions. Moreover, several borderline PD criteria resemble Criterion A of the new DSM-5 Section III general criteria for personality pathology (i.e., self and interpersonal dysfunction). We evaluated a bifactor model of PD pathology in which a general factor and several specific factors of personality pathology (PD 'g' and 's' factors, respectively) account for the covariance among PD criteria. In particular, we examined the extent to which the borderline PD criteria would load exclusively onto the g-factor versus on both the g- and one or more s-factors. A large (N = 966) sample of inpatients were interviewed for six DSM-IV (American Psychiatric Association, 1994) PDs using the (Structured Clinical Interview for Personality Disorders (SCID-II; First, Spitzer, Gibbon, Williams, & Benjamin, 1994) with no skip-outs. We ran a series of confirmatory, exploratory, and bifactor exploratory factor analyses on the rated PD criteria. The confirmatory analysis largely replicated the DSM PDs, but with high factor correlations. The "standard" exploratory analysis replicated four of the DSM PDs fairly well, but nearly half the criteria cross-loaded. In the bifactor analysis, borderline PD criteria loaded only on the general factor; the remaining PDs loaded either on both the general and a specific factor or largely only on a specific factor. Results are interpreted in the context of several possibilities to define the nature of the general factor.