Los Angeles County and a team of scholars are collaborating on a “Risk Stratification Model” – an example of ‘predictive analytics’, which uses complex algorithms to prioritize risk in child maltreatment reports.
Algorithms are often viewed skeptically by Progressives and leaders of BIPOC communities as potential contributors to “heightened surveillance”, meaning practices that disproportionately screen families of color into child protection.
The LA County project addresses this concern with an oversight group focused on bias, and by piloting the approach in predominantly Black communities. This Sounding Board article describes a related project in Allegheny County, Pennsylvania which positively impacted racial disparities by eliminating bias in 98% of screening decisions.
This tool may also help optimize limited resources, compensate for inexperienced workforces, and prevent families from getting deep into the system.
The challenge is that this is being implemented without additional resources.
For a deeper discussion and analysis, listen to this week’s podcast here or wherever you hear your favorite shows.
Read the transcript of the podcast here.