Rates of Child Welfare Interactions: Methodological Challenges and Policy Implications

Recent synthetic cohort analyses using data from the National Child Abuse and Neglect Data System (NCANDS) suggest that the childhood incidence (CI) of actual maltreatment in the US is over 12 percent (Wildeman, et al., 2014). In terms of magnitude, this places the CI of maltreatment several orders above childhood cancer (18 per 100,000) (CDC, 2015) and within the range of childhood obesity (also over 12 percent in the general population) (Cunningham, 2014) - both of which have been referred to as an “epidemic”. More recent analyses suggest that over one-third of children experience an investigation of maltreatment by age 18 (Kim, et al., 2017). Both of these findings challenge conclusions that have been historically drawn from annualized indicators of child welfare activity. For example, the Children's Bureau (CB) 2015 report of child maltreatment suggests that the annual rate of substantiated allegations of maltreatment is approximately 9 per 1,000 children – a 172% discrepancy from the cumulative incidence rate reported by Wildeman, et al. While researchers should expect a cumulative estimate to be higher than an annualized estimate, the magnitude of the difference is striking. Synthetic cohort estimates of maltreatment stratified by race/ethnicity also challenge established understanding of disproportionate representation. Specifically, both Wildeman and Kim find Native American children (typically the most disproportionately represented group in US jurisdictions) to have a risk of child welfare activity similar to Caucasian children. The contrast between these new studies and traditional understandings of child welfare system involvement raise questions about methodological issues inherent in synthetic cohort calculations (e.g. migration) and NCANDS data itself (e.g. the validity of race/ethnicity variables). The answers to these questions have important implications for policymakers as they attempt to understand the incidence of child maltreatment.

This roundtable will bring together researchers who have conducted independent state-based replications of the synthetic cohort analyses cited above using other demographic techniques. Panelists will include, researchers from the Partners for Our Children, the Children's Data Network, and the Alaska Department of Health and Social Services. The panelists will facilitate a discussion, with each other and the audience, sharing the findings of their recent analyses, how these findings compare with NCANDS-based synthetic-cohort strategies, challenges that have been encountered, and lessons learned.

Specific topics will include: 1) methodological approaches to understanding cumulative child maltreatment rates; 2) data concerns when conducting this type of research at the state and national level; 3) techniques to deal with the question of migration; 4) inconsistencies observed by racial groups; 5) qualitative differences in child welfare system activity for young children vs. adolescents; and 5) the policy implications for the results of this research. Attendees will be encouraged to ask questions throughout the roundtable and participate in discussions.