The influence of differential response on decision-making in child protective service agencies

Differential response (DR) profoundly changes the decision pathways of public child welfare systems, yet little is known about how DR shapes the experiences of children whose reports receive an investigation rather than an alternate response. Using data from the National Child Abuse and Neglect Data System (NCANDS), this study examined the relationship between DR implementation and decision outcomes in neglect cases, as measured by investigation, substantiation, and removal rates in 297 U.S. counties. Multivariate regression models included county-level measures of child poverty and proportions of African American children. Path analyses were also conducted to identify mediating effects of prior decision points and moderating effects of DR on poverty and race's influence on decision outcomes. Results indicate that compared to non-DR counties, those implementing DR have significantly lower investigation and substantiation rates within county populations but higher substantiation rates among investigated cases. Regression models showed significant reductions in removal rates associated with DR implementation, but these effects became insignificant in path models that accounted for mediation effects of previous decision points. Findings also suggest that DR implementation may reduce the positive association between child poverty rates and investigation rates, but additional studies with larger samples are needed to confirm this moderation effect. Two methods of calculating decision outcomes, population- and decision-based enumeration, were used, and policy and research implications of each are discussed. This study demonstrates that despite their inherit complexity, large administrative datasets such as NCANDS can be used to assess the impact of wide-scale system change across jurisdictions.