Measuring Discrimination in Multi-Phase Systems with an Application to Child Protection

Join a thought-provoking conference talk from the Big Data Conference 2024 where Brown University's Peter Hull explores the complex dynamics of racial discrimination in multi-phase decision-making systems. Delve into groundbreaking empirical tools developed to analyze discrimination patterns, specifically focusing on foster care placement decisions within child protective services. Learn how the research leverages quasi-random assignment of hotline call screeners and investigators to examine racial disparities in a dataset of over 200,000 maltreatment allegations. Discover key findings showing Black children face 55% higher likelihood of foster care placement compared to white children with identical potential for future maltreatment. Examine how discrimination compounds across different phases of the system, with call screeners contributing up to 19% of unwarranted disparities and investigators accounting for the remainder. Understand the concerning pattern of disparity concentration in high-risk cases, suggesting potential underplacement of white children in situations with significant maltreatment potential.