Keeping John Doe on File
The Jane Doe Files, Part 6 of 6
When DNA testing became routine, people hoped for two things. Rapists would be caught and locked away. And maybe, while inside, some of them could be changed. Thirty years of research doesn’t crown a winner. It shows several things that help, each working on a different part of the problem, and one popular policy that doesn’t.
How often do they do it again?
Start with the base rate.
The Bureau of Justice Statistics followed state prisoners released in 2005 for nine years. 7.7% of released sex offenders were arrested for another rape or sexual assault, more than three times the rate of other released prisoners (2.3%) (BJS).
Risk also fades. Among offenders rated high-risk at release, 22% were caught committing a new sexual offense within five years. Among those who had stayed offense-free for ten years, the figure was 4.2% (Hanson, Harris, Helmus & Thornton, 2014).
One essential caveat runs through everything in this post: these are arrests and convictions. In 2023, only 42% of rapes and sexual assaults were reported to police at all (BJS NCVS). Every recidivism number here is a floor, not the true rate.
Does treatment work?
In most studies, yes: treatment cuts reoffending by a quarter to a third. But the effect shrinks in the most rigorous trials, and one flagship program backfired.
| Study | What it covered | Treated reoffended | Untreated reoffended | Change from treatment |
|---|---|---|---|---|
| Holper, Mokros & Habermeyer, 2024 | Meta-analysis, 37 samples, 30,394 offenders | 9.3% | 13.6% | ▼ 32% |
| Gannon et al., 2019 | Meta-analysis, 68 studies | 9.5% | 14.1% | ▼ 33% |
| Schmucker & Lösel, 2015/2017 | 29 well-controlled comparisons | 10.1% | 13.7% | ▼ 26% |
| UK Core SOTP evaluation, 2017 | 2,562 treated vs. 13,219 matched prisoners | 10.0% | 8.0% | 🔺 25% |
Two warning signs sit underneath. The newest review found that randomized trials show smaller effects than weaker study designs, and that studies run by the program’s own developers report bigger effects. A Cochrane review limited to randomized trials found no difference at all (as cited in the UK evaluation above).
Then there’s England. Its flagship prison program, the Core Sex Offender Treatment Programme, ran for decades. When the Ministry of Justice finally evaluated it against a carefully matched comparison group, treated men reoffended more often than untreated ones. The program was scrapped in 2017. Nobody has proven why it backfired.
Treatment isn’t worthless. Programs built on the “risk-need-responsivity” model, which target the highest-risk men and the specific drivers of their offending, perform best in the research (Hanson et al., 2009). But “we’ll fix him inside” is not a promise the evidence lets anyone make.
Being on file with police
Treatment works on the man. A different set of tools works on his odds of getting caught.
One of the strongest findings in criminology is that certainty matters more than severity. As the National Institute of Justice summarizes decades of research: “The certainty of being caught is a vastly more powerful deterrent than the punishment” (NIJ).
For sex offenders specifically, the clearest evidence comes from a study that separated two things usually lumped together as “the registry.” Requiring offenders to register with police was associated with about 13% fewer sex offenses, concentrated among victims who lived near the offender, the people police could most easily connect back to him (Prescott & Rockoff, 2011).
DNA databases test the same idea on a bigger scale. Economist Jennifer Doleac found that US violent offenders whose DNA was put on file were 17% less likely to be convicted of a new crime (Doleac, 2017). When Denmark greatly expanded its database in 2005, offenders who were added cut their reoffending the next year by 42% (Anker, Doleac & Landersø, 2021). They were also more likely to hold jobs and live in stable families.
An honest limit: these studies tracked offenders of every kind, and most of them weren’t sex offenders. Sex crimes made up under 1% of the new offenses in the Danish study, too few for researchers to measure the effect on them separately. Nobody has yet shown directly that DNA registration reduces rape.
Public registries
The public, searchable registry is the policy most voters associate with keeping sex offenders in check. It has the weakest record of all. Studies in New Jersey (Zgoba et al.), South Carolina (Letourneau et al.) and national data (Agan, 2011) found no reduction in reoffending. The same study that credited police registration found public notification may increase reoffending by registrants, possibly by making work and housing impossible.
Telling the neighbors didn’t help. Making sure police knew who he was and where he lived did.
They stack
These aren’t competing options. Therapy tries to change the man. Being on file changes his odds of getting caught. Time, as the numbers above show, wears down his risk on its own. A man can face all of them, and the evidence suggests each does some good.
What stands out is the cost. Testing a rape kit costs about $950, and Cleveland’s analysis found each one prevented about $8,900 in harm (Part 2). But that isn’t money a police department gets back. The bill lands on a police or lab budget, while the savings land on women who are never raped. That mismatch likely helps explain why the step with that payoff, testing the evidence and getting it on file, is one police and prosecutors have repeatedly skipped. Hundreds of thousands of kits sat untested. In Los Angeles, a federal study found the reason wasn’t swamped labs: “investigators and prosecutors had not requested that they be tested” (Peterson et al., NIJ 2012).
Closing the series
DNA was supposed to make rape cases simple. Six posts later, the picture is a chain, and it only works when every link holds.
- A victim has to agree to an exam that can feel like a second assault, knowing it may never decide her own case (Part 1).
- Police have to test the kit, and offenders’ DNA has to be on file (Part 2).
- When the database has no match, someone in his family has to have checked a box (Part 3).
- Even then, DNA can show who was there, not whether she consented, and most rapists are men she knows (Part 4).
- And she has to stay in a process that may ask for her phone, her medical records and months of her life (Part 5).
The links that have broken most visibly are institutional: kits left on shelves, evidence never requested. Those are also the cheapest to fix. The kit a woman agrees to today may not convict the man who raped her. Tested and on file, it may be the reason he doesn’t do it to someone else.
Research and drafting assistance for this piece, including source verification, was provided by Claude (Anthropic). All claims are cited to primary or secondary sources linked inline throughout; the analysis and opinions are my own.