Measurement and structural invariance of the antisocial process screening device

Despite increasing study of psychopathic traits in children and adolescents, evidence regarding the factor structure of these traits has been inconsistent across community, clinic-referred, and incarcerated samples. Empirical support exists for both 2-factor (Impulsivity-Conduct Problems and Callous-Unemotional) and 3-factor (Narcissism, Callous-Unemotional, and Impulsivity) models, and factorial invariance across various samples has not been either tested or supported in the extant literature. We conducted confirmatory factor analyses of the Antisocial Process Screening Device (APSD; Frick & Hare, 2001) in 838 nonreferred twin pairs (M = 10.57 years, SD = 3.19 years) and 251 clinic-referred children (M = 10.82 years, SD = 3.39 years). Factorial invariance was tested across zygosity (monozygotic vs. dizygotic twins), sex (males vs. females), and age (younger vs. older children, divided by median age of 10.37 years) in the community sample and across sample type in both the community and clinic-referred samples. Results suggested that the 3-factor model fit better than did the 2-factor model in both community and clinic-referred samples. Using the best fitting 3-factor model, full measurement and structural invariance were found across zygosity, sex, and age in the community sample. Full measurement and structural invariance were also found across sample type except for differences in factor means across samples, suggesting excellent psychometric properties of the APSD. These results strongly support the robustness of the 3-factor model of psychopathic traits in children as well as the generalizability of the APSD across samples.