Search

First Step Act: An Early Analysis of Recidivism

About the Analysis

Passed in 2018, the First Step Act (FSA) was designed to reduce reoffending among people leaving federal prisons. This analysis estimates recidivism rates among individuals released from federal prison prior to the FSA who had similar risk profiles and were tracked for similar periods of time as those released under the Act.

Though not an evaluation of the FSA or its implementation, the brief discusses several factors that individually or collectively may account for lower recidivism among people released under the FSA.

Key Findings

  • Recidivism rates were an estimated 37% lower for people released under the FSA.
  • People released under the FSA incurred an estimated 3,125 fewer arrests.
  • People released under the FSA accounted for an estimated 0.02% of national arrests.

About the Author

Dr. Avinash Bhati is the founder and CEO of Maxarth LLC, a data science company providing creative solutions to real-world problems and opportunities. Bhati earned a Ph.D. in econometrics from American University in 2001. He is interested in all aspects of data science, including predictive modeling, quasi-experimental evaluations, micro-simulation models, and synthetic data tools. Bhati has developed and validated numerous pretrial and post-adjudication risk assessment instruments.

Recent Posts

Veterans Justice Commission

Can Suicide be Predicted for Justice-Involved Veterans?

Military veterans have more than twice the risk of suicide death as non-veterans. This brief examines a federal program designed to identify veterans with the highest suicide risk in order to provide enhanced outreach and support. It finds that the prediction model had low accuracy for identifying veterans who died by suicide.

Read More »

SAFETY AND JUSTICE FOR ALL

We have accomplished a lot together in our first five years, but we are just getting started. Will you support the Council as we build bridges across ideological divides and craft consensus for solutions?