Community Factors in Differential Responses of Child Protective Services
Abstract
Child maltreatment results in over 3 million referrals annually to U. S. child
protective services agencies and an estimated 695,000 children who are determined
to be child maltreatment victims. There are ongoing concerns about the large
volume and complexity of referrals and the appropriateness of an investigative
model that has been criticized as adversarial, intrusive, and inappropriate for some
referrals. In response, a Differential Response Model of child protection has
emerged, with investigative and non-investigative alternative response paths that
better acknowledge the complexities of child maltreatment and child protection. The
purpose of this study was to add to the knowledge base by identifying the
relationships and significance of county-level community variables in the
investigative and non-investigative response paths of the Differential Response
Model.
Secondary data analysis used retrospective child maltreatment data from the
National Child Abuse and Neglect Data System. County-level data on social,
economic, and demographic variables were obtained from the American Community
Survey, an ongoing national survey conducted by the U.S. Census Bureau. The
final dataset included 62,499 cases in 98 counties from Kentucky, Louisiana,
Missouri, North Carolina, and Virginia. Predictor variables included data at child,
county, and state levels. Multilevel modeling procedures were used to build multiple
three-level models to analyze predictors for the binary outcome variable of child
protective services differential response path: alternative response (noninvestigation)
or non-alternative response (investigation).
The final three-level model demonstrated that county-level factors accounted
for 12.30% of the variability in the response path outcome variable. Key results
indicated that the county-level variables of housing vacancy, unemployment, child
poverty, and households with public assistance were significant (p<.05) in predicting
response pathway. Child-level variables (report source, maltreatment type, child
age, race, and number of children in the report), and the state variable of number of
years since implementation of differential response were also significant (p<.05)
predictors in the response path outcome variable.
Results demonstrated that factors from multiple levels and contexts impact
how child protection units in the Differential Response Model respond to
maltreatment referrals. Research using advanced multilevel analytic procedures is
essential for accurate modeling and clarification of variables in nested relationships.
Table of Contents
Introduction -- Review of literature -- Methodology -- Results -- Discussion -- Appendix A. University of Missouri-Kansas City Institutional Review Board Exemption -- Appendix B. Data license: National Child Abuse and Neglect Data System Child File
Degree
Ph.D.