Opinion: Determining likelihood of re-offending shouldn’t be a judicial guess

This is also one of the decisions that our system gets wrong most often.

The lack of rigour associated with predictions of future offending underlines the empirical wasteland that is our criminal justice system.

The approach that is normally taken to community protection is unduly simplistic. When community protection is held to be relevant, it strongly inclines in favour of a prison term being imposed.

However, there are several ­unstated premises that underpin the pursuit of community protection. First, there is only a need to imprison offenders who would have offended during the period of incarceration.

Second, any protective benefit from prison is only transient given that 99 per cent of offenders will be released from prison and when they do get out, 46 per cent are ­returned to prison within two years.

Part of the reason for this is that emerging evidence shows that sending offenders to prison makes them more likely to re-offend.

Given the importance of the prediction relating to whether an offender will reoffend, you would assume that this decision would be informed by empirical data. ­Regrettably, this is typically not the case.

The process by which courts normally determine an offender’s recidivism risk is unstructured ­judicial assessments. This is a ­polite terminology for judicial ­intuition. There is no methodological process that judges use to evaluate an offender’s risk of dangerousness.

Typically, a disproportionate emphasis is accorded to an offender’s prior criminal history. It is not surprising that these decisions are barely more accurate than a toss of a coin.

Research, new learning and evidence-based practices drive progress and evolution in all other disciplines and industries. It is an abuse of the rules of science and a violation of the otherwise uncontradicted law of human progress for decisions of such enormous public importance to be based on the sentiments of judges.

The community would be ­outraged if doctors were permitted to make clinical decisions on the basis of their gut instincts, or engineers built bridges without ­recourse to validated design ­formulas.

It is incongruous that the law should stand apart from the need for important decisions and practices to be made on the basis of the best available evidence.

There are clearly sounder options for evaluating the need for community protection in relation to offenders than relying on the predictions of judges.

And on this front, the world’s largest incarcerator is leading the way. Stung by 40 years of exponential growth in prison numbers, which resulted in many American states spending more on prisons than higher education, the US is ­finally retreating from the practice of mass incarceration.

Paradoxically, it is the Republicans led by Donald Trump who are now intelligently and systematically reducing prison numbers through a process of evidence-based deliberative reforms.

These changes include bifurcating offenders into sexual and violent offenders who receive no penalty discounts, and other ­offenders whose sentences are now being reduced. The changes are so pronounced that they involve letting thousands of offenders out of federal prison early if they are unlikely to reoffend.

This decision is to be informed by a carefully developed algorithm that uses actuarial modelling to determine an offender’s risk profile. The scheme is to be rolled out pursuant to the FIRST STEP Act (an acronym for the Formerly Incarcerated Reenter Society Transformed Safely Transitioning Every Person Act), which was passed in December 2018.

It has been described as the most significant piece of sentencing legislation in the US for more than two decades.

The risk and needs algorithm that determines an offender’s risk of reoffending will be similar to other tools that are already ­increasingly used in the US. These tools incorporate not only static factors (such as prior criminal history), but also dynamic variables, including employment status and history, education and family ­relationships.

These tools have been shown to be far more accurate than ­judicial judgments and predict ­recidivism with an accuracy level of over 70 per cent.

It is a crime that sentencing courts in Australia are not required to undertake a risk and needs assessment appraisal of all serious ­violent and sexual offenders.

False positive assessments of likely reoffending cost the community more than $100,000 a year in prison costs. False negative assessments are worse.

Professor Mirko Bagaric is Director of the Evidence-Based Sentencing and Criminal Justice Project at Swinburne University of Technology. This opinion piece was originally published in The Australian. 

Media enquiries