Child welfare predictive risk models and legal decision making

Area of Science:

• Legal decision-making
• Child welfare policy
• Algorithmic risk assessment

Background:

• Child welfare agencies globally utilize algorithmic predictive modeling for decision support.
• Current models are primarily used internally by agency staff for risk assessment.
• Predictive models have not been integrated into legal proceedings or judicial decision-making.

Purpose of the Study:

• To evaluate the impact of predictive risk models on legal decision-making in child welfare.
• To assess how these models influence attorneys and judges in removal and placement cases.

Main Methods:

• A randomized vignette survey was administered to legal professionals (lawyers, judges, law students).
• Participants reviewed complex foster child removal and placement scenarios with varying predictive risk model scores.
• Decisions were analyzed to determine the influence of the risk scores.

Main Results:

• High predictive risk scores consistently altered legal decisions regarding child removal and placement.
• Medium and low risk scores also demonstrated a significant, though less consistent, influence on legal decisions.
• Structural equation modeling confirmed the impact of risk scores on judicial outcomes.

Conclusions:

• Predictive risk models can significantly affect legal decision-making in child welfare.
• Findings have implications for the development, implementation, and legal education surrounding these models.
• The study highlights the need for careful consideration of algorithmic tools in judicial processes.