An Exploration of Item Level Responses on Popular Psychopathy Measures Using Exploratory Factor Analysis and Item Response Theory Methods

There is a growing literature suggesting there are notable sex differences in psychopathy, that females may not be assess validly by summed scores on some popular psychopathy measures. The current study examined possible sex differences in measurement in three popular self-report psychopathy measures—the Short Dark Triad 3 Psychopathy (SD3: psychopathy), the Levenson Self-Report Psychopathy Scale (LSRP), and the Triarchic Psychopathy Measure (TriPM)—in a non-incarcerated community sample in a with high levels of psychopathology, and a undergraduate student sample with low levels of psychopathology. Using exploratory factor analyses (EFA), graded response modeling (GRM), and differential item function (DIF) methods based on item response theory, the study searched for statistical and visual evidence of sex differences in structure, category responses, item, test, and differential item functioning. Relevant to the interpretation and expression of psychopathy, results of preliminary statistics provided evidence of sex differences in summed scores in favor of males in both samples. EFA results demonstrated similar factor structures related to the target construct, that generally replicated well across samples. Comparability coefficients across sex were quite high, providing no significant evidence of sex differences in structure. GRM methods suggested a potential lack of invariance due to different category and item endorsement rates on these scales by sex.

There was evidence that males and females with equivalent levels of psychopathy-related characteristics endorsed items differentially across response categories, and items were contributing to the total score of these measures differently by sex. There was evidence of significant trait level DIF that generally favored males; however, with the exception of 2 items displaying moderate impact of sexDIF, the impact of these findings was low. Results and implications for psychopathy research by sex are discussed.