Evaluating disability bias in the Allegheny Family Screening Tool: implications for child welfare decisions

Child welfare systems are increasingly adopting algorithmic decision-making (ADM) tools, but there has been limited work assessing the impact of such tools on people with disabilities. We address this gap by examining the Allegheny Family Screening Tool (AFST) for disability-related bias. We obtained variables and coefficients used in the AFST from the Allegheny County Department of Human Services. We used data from the American Community Survey, Medical Expenditure Panel Survey, and National Survey on Drug Use and Health to assess whether disabled parents and children would receive worse AFST scores compared to their non-disabled counterparts. Parents and children with disabilities were significantly more likely to be penalized by all elements of the AFST scoring we examined. These include socioeconomic indicators, healthcare utilization patterns, involvement with public benefits programs, and contact with the criminal legal system. Despite containing no explicit disability measures, the AFST effectively encodes disability status through proxy variables, potentially subjecting disabled parents and their children to disproportionate scrutiny and intervention. The AFST may reinforce rather than remedy historical biases against parents with disabilities, raising important questions about the role of algorithmic tools in public sector decision-making.