Despite the substantial body of literature on racial disproportionality and disparity in child welfare involvement, few studies focus on the cumulative stressors of caregivers of color at risk of child welfare system involvement. Little is known about their ecosystemic context, including child, caregiver, and family characteristics as well as sociodemographic circumstances that predict child protection investigation. This study addresses this knowledge gap by using an ecological model to examine relationships between race, poverty, family characteristics, and relational stressors to determine the independent effect of race on the odds of investigation for allegations of physical abuse, general neglect, and lack of supervision.
408 child-caregiver pairs from the Longitudinal Studies of Child Abuse and Neglect (LONGSCAN) were studied. In order to be included in the subsample, the pairs could not be from a study site that exclusively recruited child welfare system-involved families, had complete data through age 6, and had the same non-foster parent caregiver since study entry at age 4.
The presence of at least one investigation was the dependent variable. Neglect type (general and lack of supervision [LOS]) was derived from CPS records. Caregivers were interviewed about SNAP/TANF usage, depression, alcohol misuse, parenting attitudes, discipline strategies, family and social support, stress, and neighborhood quality. Child behavior was assessed using the Child Behavior Checklist. Chi-square tests were used to examine relationships between race and investigation indicators. For the multivariate analyses, nested logistic regression with variables introduced in blocks in congruence with ecological systems theory was completed using STATA 15SE.
Over half of the subjects were Black, and a majority had income under $15000 a year and/or received food stamps. Based on their proportion in the study sample, Black families were underrepresented among investigations for physical abuse, general neglect, and failure-to-provide, while White families were overrepresented. In the full model for general neglect investigations, significant predictors included poor parenting attitudes (OR: 1.57), child externalizing behavior (OR: 1.57), poor social supports (OR: 1.53), and poor neighborhood quality (OR: 0.66). Investigations for LOS were predicted by parenting attitudes (OR: 1.59) only. The full model for physical abuse was nonsignificant, but in the block containing race and food stamp participation, being Black predicted lower odds of abuse investigation (OR: 0.26) while food stamps predicted higher odds of investigation (OR: 6.63).
Conclusions and Implications:
Contrary to the literature, there were no effects related to race found other than underrepresentation of Black caregivers investigated for physical abuse. This may fit with a risk model of disproportionality: because families of color are exposed to more risks that predict child welfare system involvement, they have more contact. Since every family in LONGSCAN is considered at risk regardless of race, this may explain the contrary findings.
Also interesting is that caregiver factors predicting investigation are relational, and the one child outcome that predicts investigation may result from poor parent-child relationships. Relational challenge and scarcity for caregivers was frequently studied in older literature on neglect; a return to assessment of relational risk may be in order.