The Big Picture
What We Did
Researchers conducted a study with 730 probationers in Houston, TX, using mobile neurocognitive assessment software that included tests for impulsivity, empathy, aggression, and other traits linked to reoffending. Data was collected between 2017 and 2019. Machine learning models were applied to the results to validate the tool’s predictive accuracy, focusing on its ability to predict recidivism using receiver-operator characteristic curves.
What We Found
1
Predictive Validity Comparable to Traditional Assessments:
2
Reduction of Bias and Subjectivity:
3
Efficiency and Accessibility:
Why This Matters
The study underscores the potential of neurocognitive assessments in reforming risk prediction within the criminal justice system. By focusing on cognitive traits and removing bias-laden variables, this tool provides a more equitable method for predicting reoffense, which could significantly impact sentencing, probation decisions, and the allocation of rehabilitative resources.