You are here: Home / Research / Projects / Evaluability Assessment of Circles of Support and Accountability (COSA)

Evaluability Assessment of Circles of Support and Accountability (COSA)

COSA is a community-based reentry program aimed at sex offenders with a high risk of reconviction. Justice Center researchers conducted evaluability assessments of 5 COSA programs across the U.S. to determine which sites are prepared for more extensive evaluations.

Project Team

  • Principal Investigator: Ian Elliot, Ph.D. (
  • Co-Principal Investigator: Gary Zajac, Ph.D.
  • Research Assistant: Courtney Meyer, M.A.

About the Project

  • NIJ-funded evaluability assessment of COSA prison reentry programs in U.S.
  • COSA is a community-based program aimed predominantly at sex offenders with the highest risk of reconviction.
  • Offenders are provided with 4-5 community volunteers who provide social support while challenging risky behaviors and modeling pro-social behavior, overseen by related professionals.

Research Questions

  • Do U.S. COSA providers implement comparable programs?
  • Can U.S. COSA programs contribute to a rigorous multi-site outcome, cost, and customer satisfaction evaluation?

Project Details

  • Assess implementation at 5 COSA providers across the U.S. for intended application of the COSA model, actual program operations, data management, and challenges to evaluation.


  • Vermont's COSA program demonstrated high program fidelity; Fresno and Lancaster programs showed adequate fidelity.
  • 5 potential obstacles to conduct a successful experimental evaluation of COSA were identified (choice of outcomes; significant differences in program implementation; core member selection issues; sample size, site capacity, and low baselines of recidivism; and ownership of data).
  • 3 recommendations for future evaluative activity include: conduct an experimental evaluation of Vermont's COSA program; conduct an experimental evaluation that combines Vermont and Fresno programs; or allow fledgling sites to develop and conduct a multi-site evaluation of COSA in the future.

Final Report