Algorithms, Humans, and Racial Disparities in Child Protection Systems: Evidence From the Allegheny Family Screening Tool

Abstract

Equity concerns are a primary constraint in the adoption of algorithms across various settings. We ask how providing decision-makers with a predictive risk model affects racial disparities in the context of the child protection system. We exploit the implementation of the Allegheny Family Screening Tool (AFST), a tool that aims to help child protection workers decide which allegations of maltreatment to screen in for investigation. We find that the AFST reduced disparities in screening decisions and home removal rates for referrals involving Black versus White children. An analysis of false positive and false negative rates suggests that this reduction in disparities was achieved while improving welfare among both Black and White children.