Are County-Level Child Welfare Risk and Protective Factors Associated with Child Welfare Involvement?: A Multi-Level Investigation of Child Welfare Equity

Background. Important questions have been raised about the appropriate type and level of child welfare intervention. Some argue that we have too much paternalistic, punishment-oriented intervention and too little preventive, supportive intervention (Shdaimah, 2009). Concerns are compounded by racial disparities and biases that harm Black families (Roberts, 2022). Concerns are also compounded by widely-varying rates of child welfare involvement in states, counties, and neighborhoods. Widely-varying levels of child welfare involvement should be scrutinized and explained to address inequities and injustice. This study aims to: (1) assess the association between an index of child maltreatment risk and protective factors and child maltreatment report and victimization rates for most U.S. counties, and (2) identify counties in which the level of child welfare involvement is out-of-sync with local child maltreatment risk and protective factors.

Methods. The study involved analysis of county-level data from 2012-2015. Child maltreatment report and victimization data from the National Child Abuse and Neglect Data System were merged with data from the U.S. Census, the State Child Abuse and Neglect (SCAN) Policy Database, and other publicly-available sources for all U.S. states and all counties with at least 1,000 residents (n = 3,005). A county-level risk and protective index was computed involving ten variables previously found to be associated with child welfare involvement, including economic, service-access, social support, and health-status indicators. Bivariate associations were assessed between the child welfare risk/protective index, child maltreatment report and victimization rates, and county demographic characteristics including rurality and racial/ethnic composition. Multilevel regression models assessed the association between the risk/protective index and child welfare involvement while accounting for nested county and state level variables, including state child welfare policies. Quartiles of the level of risk/protective factors and child welfare involvement were computed and compared across counties.

Results. The county-level risk/protective index (coded so higher scores reflected greater risk) was positively associated with both child maltreatment report and victimization rates at a bivariate level and in multilevel regression models controlling for county and state demographic and policy characteristics. Contrasting with the general pattern, some counties in the highest risk/lowest protective factor quartile were in the lowest quartile for child maltreatment reports (n = 126) and victimization (n = 97) (many in North Dakota, South Carolina and Texas). Likewise, some counties in the lowest risk/highest protective factor quartile were in the highest quartile for reports (n = 99) and victimization (n = 97) (many in Oklahoma and Indiana).

Discussion. Child welfare researchers have previously identified areas of the country in which child welfare system involvement is disproportionate to an area’s racial/ethnic composition (Maguire-Jack et al., 2015). This study builds on previous research by identifying areas of the country in which child welfare involvement is disproportionate to child welfare risk and protective factors of the geographic area. Geographic inequities in child welfare involvement are often attributed to local needs and norms. Yet rigorous analysis accounting for local characteristics, needs, strengths, and policies, such as the present study, can challenge longstanding assumptions and raise new questions about child welfare equity.