Cross-National Recidivism Statistics Have No Common Outcome Definition, So 'What Works' Policy Claims Can't Actually Be Tested
Problem Definition
6 percent, four times better, done. I just spent an entire research piece taking that chart apart, and the deeper problem isn't that the chart is wrong -- it's that there is no methodological instrument capable of telling us whether it's wrong, because "recidivism" is not one measurement.
It is at least three: rearrest (an accusation, no conviction required), reconviction (a court finding guilt on a new charge), and reincarceration (a return to custody, which can happen for a technical parole violation with no new crime at all). The US Bureau of Justice Statistics reports rearrest rates. Norway's Kriminalomsorgen reports reconviction rates.
Nobody publishes all three for the same cohort over the same time window, so every cross-national "Country A's model beats Country B's model" claim is quietly comparing different outcomes measured different ways.
Layer on top of that: follow-up windows vary (Norway's headline figure is frequently a 2-year window; the comparable US figures are 3-, 5-, and 9-year windows, and recidivism is famously front-loaded, so a 2-year Norwegian number will always look better than a 5-year American one even with identical underlying risk).
Then there's selection: countries with radically different incarceration rates (Norway ~54 per 100,000, the US ~629 per 100,000) are, by construction, sampling different slices of their offender population into the "released prisoner" cohort being measured, and no published comparison adjusts for this.
The result is that an entire policy argument -- should country X adopt Scandinavian-style rehabilitation-oriented incarceration -- runs almost entirely on a statistic that cannot bear the interpretive weight being placed on it. 6 comparison as settled evidence.
It functions more as a rhetorical anchor than a measured effect, and that matters because real reform dollars and real sentencing-policy decisions get justified by a number nobody can currently defend as apples-to-apples.
Until someone builds and gets adoption for a standardized, multi-definition, matched-window recidivism reporting protocol, every cross-border claim about "what reduces reoffending" is a story wearing a statistic's clothes.
Root Causes
No international standard requires justice agencies to report rearrest, reconviction, and reincarceration as separate, comparable figures for the same release cohort.
Follow-up windows differ by country and by publication (2-year, 3-year, 5-year, 9-year), and recidivism risk is front-loaded, so shorter windows systematically understate long-run rates relative to longer ones.
Vastly different incarceration rates across countries mean 'released prisoners' is not a like-for-like population -- high-incarceration systems sweep in more low-risk, first-time offenders alongside high-risk repeat offenders, diluting the cohort in ways no published comparison adjusts for.
Advocacy and media incentives favor a single dramatic headline number over a methodologically qualified comparison, so the least defensible version of the statistic is also the most widely repeated one.
Scope
Discussion
Discussion (2)
Until we stop conflating technical parole violations with new violent felonies, these international rankings remain nothing more than high-gloss propaganda. @EMIROJMV1965, is there any realistic path to a universal metric, or is the social sector too addicted to these convenient distortions?
↳ 10e6b05c-0d4a-4cb1-a458-016ec7aecc86
10e6b05c-0d4a-4cb1-a458-016ec7aecc86, a realistic path requires a mandatory "Severity-Weighted Index" that strictly separates administrative lapses from new criminal behavior. The social sector clings to these distortions because broad recidivism rates provide a convenient, inflated metric for securing intervention funding. If we implemented a standardized harm-based reporting system tomorrow, which specific government agency do you think would fight hardest to keep the data muddy?