Reliability and validity of the Japanese version of the child abuse potential inventory abuse scale

Introduction: The present study examined the reliability, validity, and cutoff scores of the Japanese version of the Child Abuse Potential Inventory (CAPI) abuse scale, which screens for parents at risk of child abuse. Methods: Samples consisted of 1,809 parents, 109 students, and 33 child abusers in Japan. The CAPI was administered to all participants, and twice to the student sample at a 2‐week interval. Internal consistency was evaluated by Cronbach's α, and construct validity by principal factor analysis with the parent sample. Test‐retest reliability was assessed with Pearson's r with the student sample. With regard to predictive validity, the cutoff scores and classification rates were calculated through discriminant analysis between abusers and matched non‐abusers. Results: Internal consistency (Cronbach's α=0.88), test‐retest reliability (Pearson's r=0.93; two‐tailed P<0.001), and predictive validity (overall diagnostic power=90%) were all highly satisfactory. Regarding construct validity, the six‐factor structure of the original version was not replicated; only three factors were obtained. The discriminant analysis showed the basic cutoff score as 159. The conservative cutoff score for the upper 5% of the parent sample was 218. Discussion: The Japanese version of the CAPI abuse scale showed highly satisfactory internal consistency, test‐retest reliability, and predictive validity. The construct correspondence with the original version was more compromised. This version of the CAPI identified parents with scores of 159 or greater as being at risk of abusing a child.