Data working group proposes 3‑year, ADOC‑custody recidivism definition for Reentry 2030

Becky Cohen, deputy program director in the research division at the Council of State Governments, presented a proposed standard recidivism definition Tuesday: individuals released from ADOC custody who return to ADOC jurisdiction within three years for commission of a new crime. The proposal sets the release event as release from ADOC custody, the follow‑up window as three years and the qualifying return event as a return for a new criminal offense.

The definition was presented as a compromise between two common approaches. Cohen said the working group deliberately widened the release population from ADOC in‑house counts to include those released from ADOC custody (including contracted facilities) while narrowing what counts as a return event compared with ADOC’s most recent public reporting. She noted the group’s proposal differs from the ADOC fiscal year 2021 annual report, which used release from ADOC jurisdiction and counted any return to jurisdiction.

Members immediately raised technical and policy questions. A participant asked how to treat split sentences and when the clock should start for people who serve some time in custody then complete the remainder on probation; Cohen responded that the start point depends on what the measure intends to capture but said Reentry 2030’s aim is to count the population living in the community whether on supervision or not. "We want to look at all individuals who are out in the community as our reentering population," she said.

The group extensively debated whether to include misdemeanors. Several attendees pointed out that ADOC data systems routinely capture felony convictions but often do not ingest misdemeanor sentencing records served in county jails, so many misdemeanor events would not be visible in ADOC data. One participant described how a person sentenced to a short county jail term would not appear in ADOC's systems, even though the event may be a new criminal conviction in the county; another warned that domestic‑violence cases and municipal docketing inconsistencies create similar visibility gaps.

Panelists also discussed technical violations and short returns to custody — sometimes called "dips" or "dunks." Cohen said the working group’s proposed definition would not count short dips back into jail as recidivating events under the qualifying‑event standard the group recommended. Several members urged that the metric remain defensible and transparent so it can be compared across states: "The biggest thing in my opinion is trend lines. Don't look at the static numbers. Look at the trend lines," one participant said.

Speakers described operational limitations in the data. The group noted that courts may docket probation revocation hearings with different point‑designator codes (for example, .50 for a technical hearing or .70 for a regular probation hearing) and that courts are not always required to set these codes; that inconsistency can prevent a reliable algorithmic distinction between technical violations and new convictions. Cohen and others acknowledged the measure will be imperfect given current data linkages but said documenting caveats and improving systems over time are essential steps.

There was consensus on process if not on every definitional detail. Members favored keeping a three‑year follow‑up window — described in the meeting as the most common national practice — and agreed the proposed outward‑facing Reentry 2030 metric could be narrower for comparability while agencies keep disaggregated internal measures (for example, new crime vs. technical violation, misdemeanor vs. felony) to inform policy and practice.

The task force took no formal vote. Cohen said the working group will take the feedback back to the data working group and refine the qualifying return‑event bucket, broaden stakeholder review and return a revised proposal for formal consideration. "We still have to calculate something," she said, urging that the group document caveats and iterate as data systems improve.

What happens next: the data working group will revise the qualifying return event and solicit input from additional stakeholders before the task force considers formal adoption.