Methods: The study includes 650 female participants who were arrested for criminal domestic violence perpetration and court ordered to treatment through a local BIP provider located in the Southeastern United States. The participant population mirrored the general population of the state, e.g. 59.1% Caucasian; 99.7% speak English; 68.4% HS graduation; 58% part and full-time employed, 53.2% married or living together; and 73.4% have children. Unlike the general population, 38.4% of the clients have a criminal history and 51.8% reported adverse childhood experiences. T tests, exploratory and confirmatory analyses were conducted to determine if the female responses mirrored the same structural factors as Dutton (1995) found in male responses.
Results: The mean score on the PAS was 67.82 (SE = 21.39), which is significantly higher than Dutton's (1995) reported score for non-assaulting controls (t(681) 6.1406, p < .001; r = .23) and even higher than those who assaulted their spouses (t(711) 2.004, p < .05; r = .08). An exploratory factor analysis was conducted, and we found that an eight factor model was significantly more robust (X2(489) 1857.56, p < .001; CFI = .91; TLI = .87; RMSEA = .066) (Hu & Bentler, 1995) in the female sample than a 3 factor model (X2(627) 4050.94; p < .001; CFI = .78; TLI = .74; RMSEA = .092) suggested by Dutton's (1995) male sample. Confirmatory factor analyzes (CFA) were conducted testing an 8, 7, 6, 4, and 3 factor models. None of the models were robust, e.g. 3 factor model statistics (X2(699) 5926.42; p < .001; CFI = .63; TLI = .61; RMSEA = .11).
Implications: These findings raise significant concerns about assessment tools used with females receiving services through batterer intervention programs. Assessment tools originally developed for men may not be applicable to female populations that are increasing being arrested and mandated to batter intervention treatment.