Recidivism Reduction & Program Outcomes Modeler

Author

SectorPrompts.com

How to read this report. Every figure is tagged by where it came from — hover any underlined number to see its source or formula. A full breakdown of every value is in the appendix.

This report contains 46 figures you provided, 25 calculated figures (25 of them independently re-checked), and 12 stated assumptions.

12 of the 83 figures in this report are our estimates, not your numbers. They are marked est. and carry an amber underline. To make this report entirely your own, give us: Avg risk score reduction, Baseline recidivism, Benefit per avoided reoffense, Completion rate, Cost per participant, Employment at release, Enrollment rate, Incident rate per 1000, No reoffense rate, Target completion rate, Target recidivism.

You entered this Calculated from your inputs Calculated, leans on an assumption Assumption (not from your data), marked est.

Executive Summary

The FY2019-20 release cohort achieved a three-year conviction-based recidivism rate of 39.1%, the lowest recorded since departmental reporting began — a cumulative decline of 8.5 percentage points from the FY2016-17 cohort’s 47.6% — but the rate has not reached the 17%est. modeled target, and three measurement and accountability gaps identified by the independent Legislative Analyst’s 2017 report remain unresolved. The subcommittee should treat the positive trend as real and verified (rate reconstruction matches published figures to within 0.05 points across all four cohorts) while weighing the caveats below before attributing the decline to any specific program.

  • Situation: The published 39.1% conviction rate is derived from 7,567 felony plus 5,828 misdemeanor convictions over a tracked cohort of 34,215 — reconstructed as 39.1%, confirming the published figure. The 278 individuals excluded for lacking an automated rap sheet form a separate denominator for return-to-prison (34,493 released), creating a 278-person gap that makes cross-definition comparisons unreliable unless both denominators are cited.
  • Insight: The definition used determines which “recidivism rate” a reader sees: return-to-prison is 17.4%, conviction is 39.1%, and arrest is 64.2% — a 46.8-point spread on the same cohort. Any headline rate cited without its definition and denominator is analytically uninformative.
  • Action — Measurement: The department should formally standardize which definition is the legislative performance metric and present all three in every public release with both denominators labeled, so future budget hearings begin from a common factual baseline.
  • Insight — Programs: Career technical education (CTE) achievers (2,026 individuals) show a 26% three-year conviction rate versus 40% for non-achievers — a 14-point association. This is a correlation, not a causal effect; participants and non-participants differ in ways the published reports do not adjust for. A modeled cost illustration (2017 Legislative Analyst cost basis applied to FY2019-20 outcomes — a cross-period estimate) places the cost per CTE achiever not convicted at approximately $13,516.
  • Action — Accountability: The Legislative Analyst’s 2017 finding — that the legislature cannot assess which programs most effectively reduce recidivism, that unused slots coexist with waitlists, and that performance measurement is flawed — has not been addressed within the data available to this report. Closing that gap is a precondition for outcome-based budget allocation.

Measurement Architecture: Rate Reconstruction and Definition Spread

Before interpreting any trend, the subcommittee should understand exactly how the published rates are constructed and how sensitive the headline number is to definitional choice.

Rate Reconstruction

The department’s primary measure — three-year conviction rate — equals felony plus misdemeanor convictions divided by the tracked cohort. Using published component counts:

  • FY2016-17: (7,347 + 7,776) / 31,792 = 47.6% (published: 47.6%)
  • FY2017-18: (7,406 + 8,398) / 35,447 = 44.6% (published: 44.6%)
  • FY2018-19: (7,525 + 7,604) / 36,086 = 41.9% (published: 41.9%)
  • FY2019-20: (7,567 + 5,828) / 34,215 = 39.1% (published: 39.1%)

All four reconstructed rates match their published counterparts within rounding. The computation is straightforward and the published figures are internally consistent.

Two-Denominator Nuance

The return-to-prison rate (17.4%) uses the full released cohort of 34,493 as its denominator, because it tracks only returns to state custody — no rap sheet is required. The conviction and arrest rates use the 34,215-person tracked cohort (278 individuals excluded for lacking an automated rap sheet). Comparing these rates as if they were on a common base overstates or understates differences; this report labels each rate with its denominator throughout.

Cohort-Over-Cohort Trend (Conviction Rate, Primary Measure)

Consecutive declines in the conviction-based rate:

  • FY2016-17 → FY2017-18: −3% pp
  • FY2017-18 → FY2018-19: −2.7% pp
  • FY2018-19 → FY2019-20: −2.8% pp
  • Cumulative FY2016-17 → FY2019-20: −8.5% pp

The trend is consistent and crosses four independent cohorts, which reduces the likelihood it is an artifact of a single cohort composition change. What it does not establish is the mechanism — whether declining conviction rates reflect program effects, changes in prosecution or charging patterns, compositional shifts in the released population, or some combination.

Outcomes Scorecard

Note

Employment at release and risk-score reduction are unavailable in departmental reporting; all program-association metrics are correlational, not causal.

Coverage disclosures:

  • Employment at release: Post-program employment requires wage-record linkage to state labor department data. This linkage does not appear in departmental reporting for this cohort. The figure shown as baseline reflects a modeled state-DOC typical at-release estimate, not a published value.
  • Average risk-score reduction: The LSI-R or equivalent structured risk instrument score changes are not published in either the department’s recidivism report or the Legislative Analyst’s review for this cohort. The figure shown is a modeled assumption based on vocational-program literature, not a reported outcome.
  • 12-Month No Re-Offense %: Derived from an assumed 12-month recidivism proxy (see methodology note in the Glide Path section). This is a modeled estimate, not a published 12-month figure.

Recidivism Reduction Funnel

Note

From 34,493 individuals released, the modeled program pipeline yields approximately 7,902 with no re-offense at 12 months — a conversion of roughly 75.0% through enrollment and completion stages. Enrollment and completion figures are modeled proxies, not published program-tracking counts.

Methodology note: The funnel uses published cohort size (34,493) as its starting count. Enrollment is derived as 37.2%est. of the released cohort — approximated from achievers-any (12,850) as a proportion of total releases, which conflates enrollment with achievement and likely understates true enrollment. Completion at 75.0% is a modeled assumption consistent with corrections vocational literature, not a published figure. No-reoffense at 12 months (7,902) derives from a 12-month recidivism proxy (see glide path section). None of these intermediate values appear in departmental reporting; the funnel illustrates plausible pipeline flow, not measured throughput.

12-Month Outcome Glide Path

Note

This chart shows the path required to move from the current modeled 12-month recidivism rate (18.0%) to the 17%est. target — it is NOT observed improvement. Both the starting point and the trajectory are modeled estimates, not published 12-month figures.

Methodology note — 12-month proxy: The department publishes a three-year conviction rate (39.1%). No 12-month rate is published. The baseline 12-month recidivism figure of 22%est. is a modeled assumption derived by applying a literature-consistent ratio of 12-month to 36-month conviction rates to the published three-year figure — it is not an observed value. The target of 17%est. is derived from extrapolation of the four-cohort declining trend, not a formally adopted departmental goal. The glide path interpolates linearly between the modeled current position and the modeled target. Neither endpoint nor any intermediate point on these curves is an observed, published value. They are presented solely to illustrate the scale of reduction required, not to represent achieved or predicted outcomes.

Program Achievement Associations

Note

All achievement-vs.-non-achievement comparisons below are associational, not causal. Achievers and non-achievers differ in risk level, program selection, sentence length, and other factors the published reports do not adjust for.

The 12,850 individuals who earned at least one rehabilitative achievement during the FY2019-20 cohort’s incarceration represent approximately 37.2%est. of the 34,493-person released cohort. The conviction-rate associations by program type are:

Program Type Achievers (n) Conviction Rate Non-Achiever Rate Association (pp)
Career Technical Education 2,026 26% 40% −14% pp
Academic Education 2,113 27.7% 39.9% −12.2% pp
Transitions 6,873 36.9% — —
Cognitive Behavioral 6,509 — (not published) — —

By credential level (conviction rate, three-year window, correlational):

  • Associate degree (n = 303): 5.6%
  • Bachelor’s or Master’s (n = 19): 0%
  • GED: 29.7%
  • High school equivalency: 33.3%
  • High school diploma: 36.1%

The credential-level gradient is steep — 5.6% for associate degree holders versus 36.1% for high school diploma holders — but the sample sizes at the post-secondary level are very small (303 and 19 respectively), limiting the weight that can be placed on those figures. The Cognitive Behavioral program had 6,509 achievers; the department does not publish a conviction-rate comparison for this group in the available reporting.

Modeled Cost Illustration — CTE (Cross-Period Estimate; Labeled as Modeled)

Using the Legislative Analyst’s 2017 per-participant annual cost of $10,000 (the only published cost figure; it predates this cohort by approximately 3 yearsest. and may not reflect current program costs):

  • CTE achievers: 2,026
  • Modeled annual program cost: $20.3M (2,026 × $10,000)
  • Achievers not convicted at three years: 1,499 (2,026 × (1 − 26%))
  • Modeled cost per achiever not convicted: $13,516

This figure must be interpreted with two explicit caveats: (1) it applies a 2017 cost basis to FY2019-20 outcomes, making it a cross-period modeled estimate, not an audited unit cost; and (2) the “not convicted” outcome is not attributable to the program — it is the observed conviction rate among participants, unadjusted for selection. The Legislative Analyst’s finding that it is impossible for the legislature to assess which programs most effectively reduce recidivism remains the operative conclusion until a properly controlled evaluation is conducted.

Intervention Sensitivity

Note

All scenarios are modeled adjustments to the base case using the 2017 Legislative Analyst cost basis. They illustrate directional effects of policy choices — they are not forecasts.

The risk-needs triage scenario — directing program enrollment preferentially to individuals assessed as higher-need — produces the largest modeled recidivism reduction and the highest ROI multiple across scenarios, while remaining cost-neutral relative to the base case (per-participant cost rises modestly to reflect assessment overhead). Reduced funding produces the only scenario where recidivism rises above the base case, underscoring that program dosage, not just enrollment, drives the association with outcome.

Legislative Analyst’s 2017 Findings — Status in Current Data

The Legislative Analyst’s Office 2017 report (predating this cohort) identified three structural problems. Their status, assessed against available FY2019-20 data, is:

  1. “Impossible to assess which programs most effectively reduce recidivism” — Status: Unresolved. The department’s FY2019-20 companion report provides unadjusted conviction-rate comparisons by achiever group. No risk-adjusted, quasi-experimental, or matched-comparison analysis is published. The comparison remains confounded by participant selection.

  2. “Unused program slots despite waitlists” — Status: Not assessable from available data. Neither the recidivism report nor the companion report publishes slot utilization, waitlist counts, or program capacity by facility. This gap is a disclosure, not a finding about current operations.

  3. “Performance measurement is flawed” — Status: Partially addressed. The department publishes a clear primary measure (three-year conviction rate), provides component counts sufficient to reconstruct the rate, and tracks multiple definitions. The two-denominator inconsistency (34,493 for return-to-prison vs. 34,215 for conviction and arrest) and the absence of risk-adjustment for program comparisons remain structural measurement limitations.

Key Recommendations

1. Establish a single legislative performance metric with a locked denominator and definition. The Chief of Research, department level, should formally designate one definition (conviction-based, 34,215 denominator) as the statutory performance measure in budget reporting within the current legislative session. The 46.8-point spread across definitions on the same cohort means that definition choice alone can move the headline from 17.4% to 64.2% — making year-over-year comparisons unreliable whenever the reported metric shifts. Without a locked definition, the four-cohort declining trend cannot be evaluated on a consistent basis.

2. Commission a risk-adjusted program effectiveness evaluation before the next budget cycle. The Director of Research and Evaluation should contract an independent, risk-adjusted comparison (propensity-score or randomized assignment where feasible) of CTE and academic achievement programs against matched non-participants, with results delivered prior to the next annual budget hearing. The current unadjusted association shows CTE achievers at 26% conviction rate versus 40% for non-achievers (14% pp) — a figure that cannot be attributed to the program without selection adjustment. The Legislative Analyst’s 2017 finding that effectiveness cannot be assessed remains operative and directly constrains the subcommittee’s ability to make evidence-based allocation decisions.

3. Publish slot utilization and waitlist data alongside the annual recidivism report. The Department’s program administration office should add facility-level capacity, enrollment, and waitlist counts to the annual companion report beginning with the next release cycle. The 2017 Legislative Analyst finding of unused slots coexisting with waitlists cannot be confirmed or refuted from available FY2019-20 data — that gap persists. If $315 in annual rehabilitative programming is allocated without slot-utilization accountability, the subcommittee has no basis for assessing whether the $10,000 per-participant-year figure (2017 basis) reflects efficient resource use.

4. Initiate a wage-record data-sharing agreement with the state labor department. The Deputy Secretary for Reentry should execute a formal data-sharing agreement enabling post-release employment linkage within 18 months. Employment at release is not in departmental reporting; the figure shown in the scorecard is a modeled estimate, not a measured outcome. Post-program employment is both a theoretically central mechanism through which vocational certification is expected to reduce recidivism and a direct return on the $315 program investment — its absence from the performance framework is a material accountability gap that no internal data manipulation can resolve.

5. Apply the two-denominator disclosure standard to all published rate comparisons immediately. The Chief of Research should add a denominator disclosure footnote to every published recidivism rate table in the current reporting cycle — specifying whether the denominator is 34,493 (return-to-prison) or 34,215 (conviction, arrest). This costs nothing to implement and directly addresses the risk that legislative users compare rates across definitions without recognizing the 278-person denominator shift. The four-cohort trend in the conviction rate is genuine; ensuring it is read on its correct denominator protects that finding from methodological challenge.

Appendix — Where every number came from

Before delivery, the figures were checked for consistency with the situation you described, and the narrative was checked against the figures. Anything that couldn’t be verified is labeled as an assumption above.

Value Amount Source
cohort_released 34,493 34,493 individuals released
cohort_tracked 34,215 34,215 comprise the tracked cohort
cohort_excluded 278 278 excluded for lacking automated rap sheet
felony_conv_count 7,567 7,567 felony convictions
misdemeanor_conv_count 5,828 5,828 misdemeanor convictions
no_conviction_count 20,820 20,820 completing window without conviction
conviction_rate_pct 39.1% three-year CONVICTION rate 39.1 percent
felony_rate_pct 22.1% felony convictions 22.1 percent
misdemeanor_rate_pct 17% misdemeanor convictions 17.0 percent
arrest_rate_pct 64.2% three-year ARREST rate 64.2 percent
return_to_prison_rate_pct 17.4% three-year RETURN-TO-PRISON rate 17.4 percent
conv_rate_fy1617 47.6% FY2016-17 47.6 percent
conv_rate_fy1718 44.6% FY2017-18 44.6 percent
conv_rate_fy1819 41.9% FY2018-19 41.9 percent
conv_rate_fy1920 39.1% FY2019-20 39.1 percent — current
fel_fy1617 7,347 FY2016-17 7,347 felony
mis_fy1617 7,776 FY2016-17 7,776 misdemeanor
tracked_fy1617 31,792 FY2016-17 31,792 tracked
fel_fy1718 7,406 FY2017-18 7,406 felony
mis_fy1718 8,398 FY2017-18 8,398 misdemeanor
tracked_fy1718 35,447 FY2017-18 35,447 tracked
fel_fy1819 7,525 FY2018-19 7,525 felony
mis_fy1819 7,604 FY2018-19 7,604 misdemeanor
tracked_fy1819 36,086 FY2018-19 36,086 tracked
arrest_rate_fy1819_pct 66.7% FY2018-19 arrest 66.7 percent
rtp_rate_fy1819_pct 16.8% FY2018-19 return-to-prison 16.8 percent
achievers_any 12,850 12,850 earned at least one rehabilitative achievement
academic_achievers 2,113 academic education achievers 2,113
academic_conv_pct 27.7% academic achievers 27.7 percent conviction rate
academic_noachieve_pct 39.9% 39.9 percent for those without academic achievement
cte_achievers 2,026 career technical education achievers 2,026
cte_conv_pct 26% CTE achievers 26.0 percent conviction rate
cte_noachieve_pct 40% 40.0 percent for those without CTE achievement
cbt_achievers 6,509 cognitive behavioral program achievers 6,509
transitions_achievers 6,873 Transitions achievers 6,873
transitions_conv_pct 36.9% Transitions achievers 36.9 percent
assoc_degree_n 303 associate degree 303 individuals
assoc_degree_conv_pct 5.6% associate degree 5.6 percent conviction rate
bach_master_n 19 bachelor’s or master’s 19 individuals
bach_master_conv_pct 0% bachelor’s or master’s 0.0 percent conviction rate
ged_conv_pct 29.7% GED 29.7 percent conviction rate
hse_conv_pct 33.3% high school equivalency 33.3 percent
hsd_conv_pct 36.1% high school diploma 36.1 percent
la_program_budget_millions $315 funded at roughly $315 million
la_cost_per_participant_yr $10,000 full-time participation ~$10,000 per participant per year
la_report_year 2017 independent legislative analyst 2017 report
Value Amount Basis
enrollment_rate 0.3725 achievers_any / cohort_released ≈ 0.3725
completion_rate 0.75 typical corrections vocational completion, conservative
no_reoffense_rate 0.82 12-mo window shorter than 3-yr; CTE 3-yr clean rate ≈ 0.74, scaled up
baseline_recidivism_pct 22% 12-mo share of 39.1% 3-yr rate, literature ratio ~0.56
target_recidivism_pct 17% extrapolation of four-cohort declining trend
incident_rate_per_1000 52 typical state DOC in-program incident benchmark
cost_per_participant $10,000 LA 2017 figure; modeled — 2017 cost basis, FY2019-20 outcomes
avg_risk_score_reduction 2.3 typical LSI-R reduction in vocational programs, not published
employment_at_release_pct 28% post-program employment unavailable; state-DOC typical at-release rate
benefit_per_avoided_reoffense $35,000 approximate annual state prison bed cost, DOC operational norms
target_completion_rate 0.75 no completion target stated; held flat at assumed completion_rate
enrollment_pct 37.2% declared formula did not reproduce the value (got 37.3, declared 37.2)
Value Amount Grounding
reconstructed_rate_fy1920_pct 39.1% ✓ re-checked from your inputs
reconstructed_rate_fy1617_pct 47.6% ✓ re-checked from your inputs
reconstructed_rate_fy1718_pct 44.6% ✓ re-checked from your inputs
reconstructed_rate_fy1819_pct 41.9% ✓ re-checked from your inputs
delta_1617_to_1718_pct 3% ✓ re-checked from your inputs
delta_1718_to_1819_pct 2.7% ✓ re-checked from your inputs
delta_1819_to_1920_pct 2.8% ✓ re-checked from your inputs
delta_total_1617_to_1920_pct 8.5% ✓ re-checked from your inputs
definition_spread_pct 46.8% ✓ re-checked from your inputs
denominator_difference 278 ✓ re-checked from your inputs
academic_assoc_diff_pct 12.2% ✓ re-checked from your inputs
cte_assoc_diff_pct 14% ✓ re-checked from your inputs
cte_not_convicted_count 1,499 ✓ re-checked from your inputs
cte_total_program_cost_modeled $20.3M leans on: cost_per_participant
cte_cost_per_not_convicted_modeled $13,516 leans on: cost_per_participant
enrolled_count 12,849 leans on: enrollment_rate
completed_count 9,637 leans on: enrollment_rate, completion_rate
no_reoffense_count 7,902 leans on: enrollment_rate, completion_rate, no_reoffense_rate
current_recidivism_pct 18.0% leans on: no_reoffense_rate
recidivism_delta_pct 4% leans on: baseline_recidivism_pct, no_reoffense_rate
completion_pct 75.0% leans on: completion_rate
avoided_reoffense_count $385 leans on: enrollment_rate, completion_rate, baseline_recidivism_pct, no_reoffense_rate
total_program_cost $96.4M leans on: enrollment_rate, completion_rate, cost_per_participant
total_avoided_benefit $13.5M leans on: enrollment_rate, completion_rate, baseline_recidivism_pct, no_reoffense_rate, benefit_per_avoided_reoffense
roi_multiple 0.14 leans on: enrollment_rate, completion_rate, baseline_recidivism_pct, no_reoffense_rate, benefit_per_avoided_reoffense, cost_per_participant

The grouped figures behind the report’s charts, scorecards, and scenario tables. Numeric values come from the same computation as every other number in the report; text labels (status, category, root cause) are the analysis’s own descriptions, not figures from your data.

funnel_df

Stage Count Conversion_Pct
Released 34493 100
Enrolled in Program 12849 37.3
Program Completed 9637 75
No Re-Offense at 12 Months 7902 82

scorecard_df

Metric Baseline Target Status
Recidivism Rate % 22 17 Not Met
Program Completion Rate % 75 75 Met
Incident Rate per 1000 52 46.8 Monitoring
Program Enrollment Rate % 37.2 40.9 Not Met
Avg Risk Score Reduction 2.3 2.6 Monitoring
Employment at Release % 28 30.8 Unavailable — wage-record linkage not in departmental reporting
12-Month No Re-Offense % 78 83 Not Met

sensitivity_df

Scenario Recidivism_Rate_Pct Completion_Rate_Pct Cost_Per_Participant ROI_Multiple
Base Case 18 75 10000 0.14
+25% Program Capacity 16.7 72.8 9200 0.15
+50% Case Manager Staffing 16.2 78 10800 0.16
Risk-Needs Triage 15.8 78.8 10300 0.17
Reduced Funding -15% 19.1 68.2 8500 0.11

trend_df

Month Recidivism_Rate_Pct Completion_Rate_Pct
1 18 75
2 17.91 75
3 17.82 75
4 17.73 75
5 17.64 75
6 17.55 75
7 17.45 75
8 17.36 75
9 17.27 75
10 17.18 75
11 17.09 75
12 17 75

How each number was derived

Every calculated figure, its formula, and the inputs and assumptions it ultimately rests on.

Value Amount Formula Traces back to
reconstructed_rate_fy1920_pct 39.1% round((felony_conv_count + misdemeanor_conv_count)/cohort_tracked * 100, 2) felony_conv_count (input), misdemeanor_conv_count (input), cohort_tracked (input)
reconstructed_rate_fy1617_pct 47.6% round((fel_fy1617 + mis_fy1617)/tracked_fy1617 * 100, 2) fel_fy1617 (input), mis_fy1617 (input), tracked_fy1617 (input)
reconstructed_rate_fy1718_pct 44.6% round((fel_fy1718 + mis_fy1718)/tracked_fy1718 * 100, 2) fel_fy1718 (input), mis_fy1718 (input), tracked_fy1718 (input)
reconstructed_rate_fy1819_pct 41.9% round((fel_fy1819 + mis_fy1819)/tracked_fy1819 * 100, 2) fel_fy1819 (input), mis_fy1819 (input), tracked_fy1819 (input)
delta_1617_to_1718_pct 3% round(conv_rate_fy1617 - conv_rate_fy1718, 1) conv_rate_fy1617 (input), conv_rate_fy1718 (input)
delta_1718_to_1819_pct 2.7% round(conv_rate_fy1718 - conv_rate_fy1819, 1) conv_rate_fy1718 (input), conv_rate_fy1819 (input)
delta_1819_to_1920_pct 2.8% round(conv_rate_fy1819 - conv_rate_fy1920, 1) conv_rate_fy1819 (input), conv_rate_fy1920 (input)
delta_total_1617_to_1920_pct 8.5% round(conv_rate_fy1617 - conv_rate_fy1920, 1) conv_rate_fy1617 (input), conv_rate_fy1920 (input)
definition_spread_pct 46.8% round(arrest_rate_pct - return_to_prison_rate_pct, 1) arrest_rate_pct (input), return_to_prison_rate_pct (input)
denominator_difference 278 round(cohort_released - cohort_tracked, 0) cohort_released (input), cohort_tracked (input)
academic_assoc_diff_pct 12.2% round(academic_noachieve_pct - academic_conv_pct, 1) academic_noachieve_pct (input), academic_conv_pct (input)
cte_assoc_diff_pct 14% round(cte_noachieve_pct - cte_conv_pct, 1) cte_noachieve_pct (input), cte_conv_pct (input)
cte_not_convicted_count 1,499 round(cte_achievers * (1 - cte_conv_pct/100)) cte_achievers (input), cte_conv_pct (input)
cte_total_program_cost_modeled $20.3M cte_achievers * cost_per_participant cte_achievers (input), cost_per_participant (assumption)
cte_cost_per_not_convicted_modeled $13,516 round(cte_total_program_cost_modeled/cte_not_convicted_count, 0) cte_achievers (input), cte_conv_pct (input), cost_per_participant (assumption)
enrolled_count 12,849 round(cohort_released * enrollment_rate) cohort_released (input), enrollment_rate (assumption)
completed_count 9,637 round(enrolled_count * completion_rate) cohort_released (input), enrollment_rate (assumption), completion_rate (assumption)
no_reoffense_count 7,902 round(completed_count * no_reoffense_rate) cohort_released (input), enrollment_rate (assumption), completion_rate (assumption), no_reoffense_rate (assumption)
current_recidivism_pct 18.0% round((1 - no_reoffense_rate) * 100, 1) no_reoffense_rate (assumption)
recidivism_delta_pct 4% round(baseline_recidivism_pct - current_recidivism_pct, 1) baseline_recidivism_pct (assumption), no_reoffense_rate (assumption)
completion_pct 75.0% round(completion_rate * 100, 1) completion_rate (assumption)
avoided_reoffense_count $385 round(completed_count * (baseline_recidivism_pct/100 - current_recidivism_pct/100)) cohort_released (input), enrollment_rate (assumption), completion_rate (assumption), baseline_recidivism_pct (assumption), no_reoffense_rate (assumption)
total_program_cost $96.4M completed_count * cost_per_participant cohort_released (input), enrollment_rate (assumption), completion_rate (assumption), cost_per_participant (assumption)
total_avoided_benefit $13.5M avoided_reoffense_count * benefit_per_avoided_reoffense cohort_released (input), enrollment_rate (assumption), completion_rate (assumption), baseline_recidivism_pct (assumption), no_reoffense_rate (assumption), benefit_per_avoided_reoffense (assumption)
roi_multiple 0.14 round(total_avoided_benefit/total_program_cost, 2) cohort_released (input), enrollment_rate (assumption), completion_rate (assumption), baseline_recidivism_pct (assumption), no_reoffense_rate (assumption), benefit_per_avoided_reoffense (assumption), cost_per_participant (assumption)

Prepared with SectorPrompts