An Exploration of the Application of Item Response Theory to the Measurement of Child Abuse in Canada

Abstract

PURPOSE There is a longstanding need for a consistent approach to estimate the prevalence of child abuse to inform programs, policies, and research. A key step in achieving a consistent approach is to examine current differences in the application of cut-offs in the public health reporting of child abuse. METHODS To address this issue, this exploratory study applied item response theory (IRT) to determine standard cut-points for two types of child abuse, namely physical and sexual abuse, using nationally representative data from the 2018 Survey of Safety in Public and Private Spaces. Two different cut-offs from the literature were considered in the IRT models. The first cut-off (i.e., cut-off 1) estimated the proportion of individuals who experienced abuse in childhood (before age 15 years) using a frequency of one or more times. In contrast, a second cut-off (i.e., cut-off 2) was more conservative, requiring experiencing a higher frequency of abuse across all items. RESULTS Model fit statistics and option characteristic curves suggested a good fit for physical abuse using cut-off 2. In contrast, a good fit was obtained for sexual abuse using cut-off 1. CONCLUSION This study provides support towards using cut-off 2 for physical abuse and cut-off 1 for sexual abuse. Consistent approaches to measuring child abuse is important to improve our understanding of the epidemiology of child abuse, as well as approaches to its assessment, treatment, and prevention. Future research should explore the feasibility of using IRT methods to establish cut-offs for other types of maltreatment.