We re-derived a published figure from the public record.
California's corrections department reports a 39.1 percent three-year recidivism rate for the FY2019-20 release cohort, and the report reconstructs it exactly: (7,567 felony + 5,828 misdemeanor convictions) ÷ 34,215 tracked = 39.15%. The same people over the same three years also measure 64.2% by arrest and 17.4% by return-to-prison: a 47-point spread that comes entirely from the definition, all three published by the same department.
The situation
One cohort, three published rates, and a program table with honest caveats.
A legislative budget subcommittee reviews corrections outcomes: the department publishes cohort recidivism under three definitions, program-completion outcomes, and an independent legislative analyst has questioned whether program effectiveness can be assessed at all.
We entered the published cohort counts and program-achievement figures, and had the report reconstruct the published rates, show the definition sensitivity, and label every completer-versus-non-completer comparison as an association, not a causal program effect.
What we supplied
Only public inputs, each one sourced.
Inputs come from the department's statewide recidivism report and its program-achievements companion (FY2019-20 cohort), plus the legislative analyst's 2017 program review.
| Input | Value | Source |
|---|---|---|
| Released / tracked cohort | 34,493 / 34,215 (278 excluded: no rap sheet) | CDCR statewide report |
| 3-yr conviction rate (primary) | 39.1% (7,567 felony + 5,828 misdemeanor) | CDCR statewide report |
| 3-yr arrest / return-to-prison | 64.2% / 17.4% | CDCR statewide report |
| Prior cohorts | 47.6% → 44.6% → 41.9% → 39.1% | CDCR report series |
| Education achievers' conviction rate | 27.7% vs 39.9% without (−12.2 pts) | CDCR program report |
| Career-technical achievers | 26.0% vs 40.0% without (−14.0 pts) | CDCR program report |
| LAO program review (2017) | ~$315M funding; ~$10K per participant-year; effectiveness "impossible to assess" | LAO report 3720 |
Source: CDCR Statewide Recidivism Report, FY2019-20 cohort · download the source document we used.
The report
The finished, decision-ready output.
This is the actual rendered report, interactive charts and all. Nothing was edited after generation.
Reconcile
Our output vs. the published figures.
| Metric | Published | Ours (computed) | Match | Note |
|---|---|---|---|---|
| Published conviction rate | 39.1% (CDCR) | (7,567 + 5,828) ÷ 34,215 = 39.15% | ✓ exact | All four cohorts reconstruct the same way |
| Definition spread | All three rates (CDCR) | 17.4% → 39.1% → 64.2%: 46.8 points | ✓ exact | Same people, same window; the definition does the work |
| Credential ladder | Per-credential rates (CDCR) | Associate degree 5.6% · bachelor's/master's 0.0% (n=19) | ✓ | Labeled as associations, not causal effects: participants differ from non-participants |
The two-denominator nuance is stated wherever rates are compared: arrest and conviction rates use the 34,215 tracked cohort; the return-to-prison rate uses all 34,493 released. Time-to-case-resolution and post-program employment require court and wage-record data that departmental reporting doesn't publish: disclosed, not estimated.
Build the same report from your own numbers.
You supply the figures you already have, any gap is filled by an estimate labeled as one, and every number comes back labeled by where it came from so you can put your name on it and show your work.
Generate your report, $199