Disentangling Substantiation: The Intersection of Race, Poverty, and Risk On the Substantiation Decision in Child Welfare

Background: Although research has documented a significant disparity between African American and White children in substantiations of maltreatment, it has been less successful in revealing the source of this disparity. At issue is whether race alone accounts for this disparity, or whether other factors play a stronger explanatory role. Several studies have examined additional factors that may explain this disparity and findings have been mixed regarding the role of race. Some have found no effect for race when controlling for other factors, while others have. Perhaps most important among potential explanatory factors is the role of family income given the relationship between poverty and maltreatment. Yet, few studies have included income when examining maltreatment. Furthermore, no previous studies examining this disparity have included measures of risk, although studies have documented that caseworkers' assessment of risk is a significant predictor of substantiation. Thus, any effort to understand the effect of race on substantiation must control for the effects of poverty and risk. This study explores the substantiation decision while controlling for these and other relevant factors to better understand the effect of race on this decision point.

Methods: The sample used in these analyses included substantiated reports of maltreatment from the Texas Department of Family and Protective Services from 2003 through 2005 (n=186,182). Primary variables used to predict substantiation included race/ethnicity, annual household income, and caseworkers' risk assessment scores. Additional variables included factors that had been found in previous research to predict substantiation. Two multivariate logistic regression analyses were conducted to examine how race, income, risk, and other potential explanatory variables predict the substantiation decision. The initial model included income and other relevant factors. The second model added risk assessment scores to see if assessed risk contributed to the substantiation decision over and above other included variables.

Results: In the first model, no effect for race was observed. African Americans showed no statistically significant differences from Whites in likelihood of substantiation. However, income was a significant predictor of substantiation. Specifically, the lowest income group (less than $10,150) was 95.6% more likely than the highest income group (greater than $40,550) to have a report substantiated (OR=1.956, p<0.001). However, when risk was added to the model, findings changed considerably. In this model, race contributed significantly to the substantiation decision with African Americans 14.8% more likely (OR=1.148, p<0.001) to have reports substantiated when contrasted with Whites. Further, the effect of income was reversed, with the lower income groups having decreased likelihood of substantiation. Specifically, the lowest income group was 6.4% less likely than the highest income group to have a report substantiated (OR= 0.936, p=0.042).

Implications: These findings suggest a complex interplay between race, assessed risk, and income in the decision to substantiate maltreatment allegations. Without controlling for risk, being lower-income was highly predictive of a substantiated report and race was not predictive. But when risk was included, lower-income groups showed decreased odds of substantiation, and a small effect for race emerged. Implications of these findings and areas for future research will be discussed.