Risk assessment tools that shape life after a third strike

In November 2012, California voters approved Proposition 36, narrowing the reach of the state's Three Strikes law so that a life sentence applies only when the third strike is serious or violent. For thousands of people still serving 25-to-life terms for nonviolent offences, the law opened a narrow door back to court. Walking through that door now depends heavily on a single document: a post-conviction risk assessment, prepared for the judge who decides whether release poses an unreasonable threat to public safety.

The assessment is not a formality. It weighs criminal history, institutional behaviour, programming participation, and a range of static and dynamic factors, then produces a recommendation that prosecutors and defence attorneys argue over. For someone whose file has sat untouched in a central file room since the early 2000s, the process can feel like a fresh trial where the only witness is a statistical model. Researchers in Melbourne, Sydney, and Perth have studied similar instruments in Australian parole boards, and their findings about the limits of actuarial tools travel well to Sacramento and back again.

How the post-conviction review actually works

A petition for resentencing under Proposition 36 typically begins with the inmate or their counsel filing in the superior court of the county where the original judgment was entered. The district attorney then has a statutory window to respond, often opposing the petition or asking the court to consider aggravating factors that go beyond the bare minimum required by the statute. Once the briefing is complete, the court orders an evaluation, usually from the California Department of Corrections and Rehabilitation's Office of Correctional Research, and schedules a hearing. A grant is not automatic even when eligibility is uncontested.

The risk assessment sits at the centre of that hearing. It is conducted by a psychologist or licensed evaluator who reviews records, conducts a clinical interview, and scores the petitioner against validated instruments such as the Static-99R or the Level of Service Inventory-Revised. The evaluator's report travels to the judge along with character letters, programming certificates, and any victim impact statements the prosecution chooses to file. The political backdrop matters too; an examination of the political economy shows that prosecutors sometimes weigh electoral considerations alongside statutory factors, which can colour how aggressively they oppose a petition.

Australian readers will recognise parallels with state parole boards in New South Wales and Victoria, where similar actuarial instruments inform release decisions for long-serving prisoners. The structural difference is that in California the question is framed as resentencing rather than parole, which shifts the burden of proof and the standard of review.

What the tool actually measures

Risk assessments used in these hearings generally fall into two broad families: those that look only at historical factors that cannot change, and those that add dynamic factors that can. Both are designed to estimate the probability of reoffending within a defined window, usually three to five years, but they treat that probability very differently in practice.

A typical post-conviction report will cover several domains, and the strength of the eventual recommendation often depends on how the petitioner presents within them. The list below is drawn from publicly available California evaluation samples and from interviews with court-appointed psychologists.

• Prior convictions, with weight given to the nature and recency of each, including juvenile adjudications that many courts now treat as part of the adult record.
• Institutional adjustment, including serious rule violations, work assignments, and whether the person has held positions of responsibility inside the institution.
• Program participation, especially completion of cognitive-behavioural therapy, substance-use treatment, vocational training, and victim-awareness coursework.
• Release planning, covering proposed housing, family support, employment prospects, and access to community-based treatment.
• Clinical findings from the interview, including substance use history, mental health diagnoses, and the evaluator's own observations of remorse and insight.
• Static and dynamic risk scores produced by validated instruments, which the evaluator interprets alongside the qualitative material.

The interaction between these domains, rather than any single factor, is what shapes the final recommendation. A petitioner with a high static number can still receive a favourable report if the dynamic factors point clearly toward desistance, and the reverse is also true.

Why tools sometimes get misread in court

Evaluators are trained to communicate uncertainty, but the courtroom tends to flatten nuance into a binary. A report that says the petitioner poses a "low to moderate" risk of reoffending is sometimes argued by prosecutors as a moderate risk, and sometimes by defence counsel as a low risk. The same language can carry opposite weight depending on which side reads it. Courts have struggled with this for years, and reform advocates argue that clearer statutory guidance would help.

There is also a well-documented tendency for risk tools to overpredict risk for older petitioners, who statistically reoffend at far lower rates than the instruments were calibrated for. Many third-strike petitioners are now in their fifties and sixties, having served two decades or more, and the actuarial estimates generated by their age cohort rarely reflect the actual trajectory of long-term ageing. Criminologists in Adelaide and at the Australian Institute of Criminology have published on the same age curve phenomenon in their own parole populations, and the lessons translate.

Finally, the assessment only captures what the records contain. If early disciplinary write-ups were never formally expunged, if programming records are missing, or if institutional misconduct was driven by untreated mental illness, the report will inherit those gaps. Petitioners who understand this can sometimes improve their position by gathering supplementary documentation well before the evaluation date.

What the numbers say about released third strikers

The strongest evidence base supporting Proposition 36 came from studies of the population already released under the earlier Romero discretion, which allowed judges to strike prior convictions in the interest of justice. That data showed recidivism well below baseline projections, and it helped persuade voters that the reform would not flood the streets with dangerous offenders.

Subsequent research has only strengthened the picture. A recent review of release data confirms that people released under both the Romero pathway and the post-Prop 36 resentencing process reoffend at rates comparable to, and often lower than, the general parole population. New violent offences are particularly rare, and they cluster heavily in the first six months after release, which has clear implications for supervision planning.

The Australian parallel here is strong: studies by the New South Wales Bureau of Crime Statistics and Research have repeatedly shown that older long-term prisoners released after decades inside do not behave like the actuarial models predicted. Both jurisdictions are learning the same lesson, that actuarial tools built on younger cohorts systematically overshoot when applied to older, long-serving populations. That shared finding is one of the more useful pieces of cross-jurisdictional evidence for reformers in California and, increasingly, for those arguing for similar reforms in states that inherited the original Three Strikes model.

Building a strong case before the assessment

Preparing for a post-conviction risk assessment is less about persuasion and more about documentation. The evaluator will work from a record, and the record is only as good as what the petitioner and their counsel can put into it. Waiting until the petition is granted to start gathering materials usually means racing against statutory deadlines.

The following list reflects what experienced Proposition 36 counsel typically assemble in the six to twelve months before a hearing date, drawing on guidance from the Administrative Office of the Courts and from practitioners who have handled dozens of these matters.

• A certified disciplinary history with explanations for any serious rule violations, ideally supported by treatment notes if misconduct followed untreated mental health episodes.
• Documentation of completed programming, including certificates, progress letters from facilitators, and any post-completion assessments.
• A written release plan with a specific address, a real employment lead or vocational referral, and contact information for an identified family or community support person.
• Letters of support that speak to specific conduct over specific periods, not generic statements of belief in the petitioner's good character.
• A current psychological or psychiatric evaluation if mental health is in issue, particularly if it can establish that symptoms are stabilised rather than dormant.
• A clean record of victim restitution or attempts to make restitution where the original offence included a financial component.

The most common mistake counsel describe is the petitioner who believes that time served speaks for itself. In the post-conviction context, time served only counts when the institution can point to what changed during it.

Petitioners who have an upcoming hearing date should request a copy of their central file from the California Department of Corrections and Rehabilitation before the evaluator is appointed, then compare it against the materials above and fix any gaps with institutional records requests well in advance.