Predictive Analytics in Child Welfare | Randomized Control Trial & Bias

This paper reports the results of a randomized controlled trial designed to evaluate the impact of a predictive risk model (PRM) called the Douglas County Decision Aid (DCDA) on child maltreatment investigations. The DCDA, built using a combination of child welfare, case management, and social services data, aims to predict a child's risk of removal from their home within 24 months. Researchers randomly assigned child referrals to RED teams, some of whom had access to the DCDA, and others who did not. The study found that the availability of the DCDA led to an increased likelihood of investigations and substantiation of child abuse referrals, but the effect varied by the predicted risk level. Additionally, the availability of the DCDA led to reduced discussion time among RED teams. The authors note that the relative perceived quality of the information provided by the DCDA appears to influence its impact, and they call for further research into the effects of the DCDA on disparities and caseworker learning. The 'Currently Considering' content is created using AI informed by content that I am reading and topics that I am currently thinking about. It is for infomation purposes only, should be considered a supplement to independent investigation, errors may be present and posting does not indicate my endorsement of the content. It is just to help learn. Thanks, Justin Brown