Sounding Board: The strange history of risk assessment in child welfare

Part 1: The strange history of risk assessment

The history of risk assessment models, tools and practices; competition between available options, with a hype cycle turning into a clearer understanding of when a risk assessment was most useful, and then the gradual abandonment of risk assessment in favor of safety tools and needs/services assessment; the bureaucratization of tools along with their failure to account for caseworker workloads and cognitive processes turning them into checkmarks after the fact instead of guides at the time of decision making. Originally published December 2022

Part 2: The Use of Predictive Analytics in Allegheny County’s Child Protection System

A discussion of algorithmic prediction tools being introduced in child welfare, focused on the Allegheny Family Screening Tool; praise and criticism it received, assessment of the tool itself and assessment of the methods of assessment for screening accuracy. Originally published January 2023

Part 3: Using an Algorithm to Predict Hospital Encounters, Any-Cause Injury & Abusive Injuries at CPS Intake

Some corrections to how the Allegheny Family Screening Tool is used, commentary on the differences between theoretical and practical usage, and the value of AFST and some other models as predictors of child injury or of family need for services – and an argument that child emergency room visits and hospitalizations should be a higher-valued metric in child welfare. Originally published in February 2023

See all Dee Wilson commentaries at https://www.deewilsoncon.com/writings