Latent class analysis of new self-report measures of physical and sexual abuse
- Author(s): Nooner, Kate Brody
- et al.
The Longitudinal Studies in Child Abuse and Neglect (Longscan) have developed measures to assess how pre- adolescents perceive maltreatment to allow for a broader, more ecologically valid understanding of these experiences. These measures assess physical and sexual abuse in 12-year-old children, with the potential to capture experiences that Child Protective Services (CPS) and caregiver self-report may not identify. To integrate these self-report measures into a broader conceptualization of maltreatment, the following research questions are addressed: a. What are the latent class profiles of the Longscan self-report measures of physical and sexual abuse? Demographic factors (e.g., gender, ethnicity, and study site) will be included in these analyses. b. What is the agreement of youth self-report of physical and sexual abuse with CPS reports of those types of abuse? Data from 845 children participating in Longscan who have completed the 12-year-old interview and have complete CPS record reviews are used to identify distinct classes of children. Latent class analysis, a latent variable framework that groups children into classes based upon their self-report, is employed to explore the class profiles for physical abuse, sexual abuse, and combined physical and sexual abuse. From these analyses, the best fitting models are determined and agreement is then evaluated in relation to CPS reports of physical abuse and sexual abuse. Results from the latent class analyses point to a 2-class solution for physical abuse, a 3-class solution for sexual abuse, and a 4-class solution for combined physical and sexual abuse. Follow-up analyses indicate that CPS reports and youth self-report agree at a rate of: 65% for the physical abuse model, 83% for the sexual abuse model, and 64% for the combined model. In cases where self-report and CPS reports differ, children tend to under-report abuse. Overall, the models reveal that 12-year-old children respond in a nuanced manner when asked to self-report abuse. Therefore, youth self-report of abuse can be meaningfully classified and incorporated to existing ecological-developmental models of child maltreatment