Abstract
The 64-item Hare Self-Report Psychopathy Scale (Hare SRP; CitationPaulhus, Neumann, & Hare, in press) is the most recent revision of the SRP, which has undergone numerous iterations. Little research has been conducted with this new edition; therefore, the goal of this study was to elucidate the factor structure as well as the criterion-related, convergent, and discriminant validity of the measure in a large sample of college students (N = 602). Confirmatory factor analyses revealed that the best fitting model was the original 4-factor model proposed by the authors of the Hare SRP (compared to a 1-factor, 2-factor, and 4-factor random model). The 4-factor model revealed superior fit for the data relative to the other alternative models. In addition, we elaborated on the psychometric properties of this 4-factor model in this sample. The Hare SRP total and factor scores evidenced good internal reliability as well as promising criterion-related, convergent, and discriminant validity in terms of predicting scores on conceptually relevant external criteria. Implications for theory and future research are discussed.
Notes
This measure has also been referred to in the literature as the SRP–III and SRP–IV. Because the commercially published version of the scale will be named the Hare Self-Report Psychopathy Scale (K. Williams, personal communication, March 15, 2011), we have decided to use this name.
The Cronbach's alpha and average interitem correlation values for all the scales in this sample are available from the authors on request.
The means and standard deviations for all the scales in this sample are available from the authors on request.
The means and standard deviations for all the scales in this sample are available from the authors on request.
One thoughtful reviewer questioned our use of CFA, as it has shown to demonstrate poor fit for several widely used personality measures with well-documented reliability and criterion-related validity (e.g., see Hopwood & Donnellan, Citation2010). In their analysis of the use of CFA and exploratory factor analyses (EFA) for inherently complex personality inventories, Hopwood and Donnellan (2010) concluded that EFAs might be more useful to evaluate model fit for these complex measures due to EFA's less stringent tests of model viability than CFA. As per their recommendation, we conducted an EFA on the 64-item Hare SRP with maximum likelihood estimation and oblique Geomin rotation in Mplus 5.21 (Muthén & Muthén, Citation2005). The EFA fit indexes for the fixed four-factor 64-item model were better than the CFA results for the 64-item model, but they were still not a good fit: Satorra-Bentler χ2(1766) = 3492.84, CFI = 0.79, RMSEA = 0.040 (90% CI = 0.038–0.042), SRMR = 0.04, AIC = 102192.93, and BIC = 103856.23. These EFA results suggest that incremental fit indexes have problems with item-level data even when optimal solutions with cross-loadings are considered.