Do the alternatives to jail actually work? Sometimes — and only if you count honestly. Here is what the research shows, read the way it should be: across everyone who enters a program, not just the ones who finish.
“Close the jail and use alternatives” and “alternatives are soft and don’t work” are both too simple. Diversion, drug courts, supportive housing, and crisis response are neither a cure nor a failure in the research — their effects turn on program design, who enters, and one thing almost every headline gets wrong: how many people actually finish.
A lot of research reports outcomes only for people who graduate a program, quietly dropping everyone who was terminated or dropped out. That inflates the result, because completers start out different — more stability, more support, less prior justice contact. The honest measure counts everyone who entered (an “intention-to-treat” baseline), and reports the completion rate alongside any success figure.
So the useful question isn’t “did the program work?” It’s “for whom, at what completion rate, and compared to what?” Every number below is read against that standard. Full sources on our References & evidence page.
Police divert someone suspected of a low-level, drug- or poverty-driven offense into case management instead of booking. In Seattle’s original program, participants had 58–60% lower odds of arrest than the arrest-as-usual group4, and ~1.4 fewer jail bookings and ~41 fewer jail days per year, with lower legal costs5. Crucially, these used a comparison group — not graduates only.
Court-supervised treatment in lieu of a sentence. Meta-analyses find adult drug courts reduce recidivism (roughly 50% → 38%)7 and cut the chance of incarceration on the instant offense (50% → 42%) — but do not reduce total time behind bars, because people who fail out get the long sentence8.
The prosecutor withholds charges in exchange for completing conditions. A four-jurisdiction study (5,040 people) found large drops in conviction and jail sentences and a trend toward less re-arrest at two years10; a broader review of police/prosecutor diversion finds reduced recidivism and cost, though little effect on behavioral health itself11.
Housing chronically homeless people reliably cuts jail time, bookings, and low-level “livability” arrests — but the average effect on crime overall is smaller and depends who you house. We cover this in depth on the References page and in the ecosystem thesis.
The results depend heavily on which model. Mobile crisis and civilian-led response show the clearest gains — a mobile-crisis model significantly cut later arrests12, and Denver’s civilian community-response pilot reduced targeted low-level crime by 34%13. Police co-response reduces immediate arrests and involuntary detentions but shows weak long-term effects — a randomized trial found no significant difference on key outcomes1415 — and a meta-analysis found Crisis Intervention Teams have no measurable effect on arrests16. Building a crisis center is not the same as choosing the model the evidence supports.
The consistent finding: most of the savings come from reduced criminal-justice costs, not healthcare21 — and they only materialize when the treatment on the other side of diversion actually exists.
These are different things. Legal pressure (treatment as an alternative to jail) keeps people in treatment longer — a mandate raised completion odds by 71% in one study22, and alternative-to-imprisonment entrants were retained longer than voluntary ones23 — and since retention drives outcomes, that matters.
But involuntary confinement for addiction is a different bet, and the evidence is weak. A systematic review of 354,000+ people found little high-quality evidence that involuntary treatment beats voluntary care24; a meta-analysis found mandated treatment largely ineffective in custodial settings while voluntary treatment worked across settings25; and a large Vancouver study found no significant improvement for coerced participants over voluntary or no treatment26. (In Washington, involuntary commitment generally requires danger to self or others or grave disability — it is legally a last resort, not a first tool.) “Just force them into rehab” is neither legally simple nor clearly effective.
Diversion, probation, and monitoring often carry fees and legal financial obligations that fall on mostly poor people. Those debts are themselves tied to housing instability and repeat court contact27, and low-level enforcement of poverty offenses churns people through the system without resolving anything28. An alternative that replaces a cell with a bill can recreate the cycle it was meant to break.
Well-designed diversion — voluntary or lightly pressured, with real treatment access, low fees, and honest accounting of who finishes — can cut recidivism and cost less than jail. Poorly designed — coerced into confinement, underfunded, or fee-driven — can match or worsen the revolving door. The useful question is never “jail or alternatives.” It is which design, for whom, and does the treatment on the other side actually exist.
Two local implications follow directly. First, the county’s crisis center is only as good as the model it runs — the evidence favors mobile/civilian crisis response over police co-response or CIT alone, so how it’s staffed will decide whether it diverts people or just relocates them.
Second, booking restrictions are not diversion. Releasing someone with nothing attached — no case management, no treatment, no housing — is the one thing the evidence consistently says fails. Diversion works when something replaces the jail bed; it fails when nothing does.