Abstract
The Inventory of Callous-Unemotional traits (ICU) is a widely used measure of the affective aspects of psychopathy in children and adolescents. Although a 3-bifactor model of the ICU has often been supported, method-variance effects and mixed results for the Unemotional subscale raise concerns. This study applied a multi-informant design to investigate the structural and concurrent validity of the self-, parent- and teacher-versions of the ICU in a Norwegian at-risk adolescent sample (N = 160, female = 46.3%, mean age = 14.73 years, SD = 1.47). Confirmatory Factor Analysis demonstrated method-variance effects of the standard- and reverse-scored items. The best fitting model was the 12-item 2-factor ICU, comprising a Callousness and an Uncaring factor. The scale reliabilities were acceptable with Cronbach alphas ranging from .810 to .906 across respondent groups. Cross-rater reliability was poor, with Intra Class Correlations ranging from .170 to .226. The concurrent validity of the ICU12 was supported by within-rater associations to externalizing problems, aggression, and problematic alcohol use. Cross-rater associations of ICU12-scores to self-reported delinquency and lower levels of anxiety, provided additional support for the concurrent validity. The unidimensional ICU12 total score was associated both to delinquency and rule-breaking behavior as observed for the Callousness factor and to lack of anxiety and reduced fear sensitivity as observed for the Uncaring factor. The ICU items aimed at capturing lack of prosocial emotions, seem to need revision. Future research should assess the predictive validity and clinical relevance of unidimensional versus factor models of the ICU.