Janczewski, C. E. & Mersky, J. P. (2016). What's so different about differential response? A multilevel and longitudinal analysis of child neglect investigations. Children and Youth Services Review, 67, 123-132.
This study used data from the American National Child Abuse and Neglect Data System (NCANDS) to examine the effect of Differential Response (DR) on the investigation and substantiation decisions of neglect cases and possible predictors of maltreatment and Child Protection Services (CPS) involvement.
The article defines DR as “not a specific practice model or intervention, but rather a ‘policy orientation’ that alters how cases progress through a CPS system” (p. 123). Cases that are judged to be at low or moderate risk follow a separate track which does not include an investigation, maltreatment disposition, or identification of a perpetrator. Instead, these cases would be referred to services such as counselling, childcare, and transportation with the “goal to reduce maltreatment risk and prevent further involvement in the child welfare system” (p. 123).
Multilevel, multivariate models were used to conduct cross-sectional analyses of neglect investigation records from 284 counties in 39 states in 2010 (n = 997,512). The purpose of these analyses was to assess whether an investigated neglect report was more likely to be substantiated in a county that used a DR approach compared to a county that did not use DR. The authors also looked at whether child level factors including age, sex, race, ethnicity, and prior maltreatment and county level factors including poverty rates, proportion of white children, and population density varied between DR and non-DR counties. Findings indicated that, among neglect cases that were referred to investigation, substantiation rate was 2.4 times higher in DR counties than in non-DR counties. Children with a previous substantiated report were also more likely to have a current report substantiated and this difference was significantly greater in DR counties than in non-DR counties. Child race and ethnicity did not predict substantiation decisions.
Longitudinal analysis was also conducted with data from 2001 to 2010 in 269 counties in 39 states which included 61 counties who used DR for at least six months. A mixed-effect time series approach was used to examine case flow and how this changed with introduction of DR. The results showed that the rate of investigations fell sharply within three years of DR implementation. However, substantiation rates did not change as a result of DR implementation. Instead, analyses indicated differences between DR and non-DR counties emerged before the launch of DR. This result points to the importance of considering pre-existing differences between DR and non-DR counties when assessing the overall impact of DR implementation.
The study used data from NCANDS, which brings together data collected using different methods and reporting standards across states, on a voluntary basis from around the United States. Counties and states with data quality concerns and/or small populations were excluded from the study. Cases that were not related to neglect or had a different disposition than ‘substantiated’ or ‘unsubstantiated’ were also excluded from the analyses. The authors also noted a potential selection bias in that there is an indication of pre-existing differences between DR and non-DR counties and that early DR adopters may not be comparable to other counties.