NDARC researcher Professor Donald Weatherburn writes about his latest study on the effect of NSW Cannabis Caution Scheme on the risk of reoffending.

Australia, along with many other countries prohibits the possession or use of many recreational drugs, including cannabis. Whatever its merits as a deterrent to drug use, the prohibition imposes significant costs on those convicted of using or possessing illicit drugs. They include reduced earnings and employment prospects in addition to any punishment they receive. In some countries they also include limited access to housing, travel restrictions and loss of a driving license. 

Australian public attitudes to legalising the use of cannabis are changing rapidly. In 2013 only about a quarter of the Australian population aged 15 and over supported the legalisation of cannabis use. By 2022-2023, the proportion supporting cannabis legalisation had risen to 45%. More than a quarter (28.9%) of those who do not support legalising use of cannabis, either want no action taken against those caught in possession of a small quantity of cannabis for personal use, or no more onerous sanction imposed upon them than a warning or a caution. 

Growing public support for a move away from the prosecution of cannabis users has prompted many governments to introduce alternatives to criminal prosecution. In 2001, for example, the then NSW government introduced a cannabis cautioning scheme. Under this scheme, police have the power to caution rather than charge an offender found in possession of (or using) a small quantity of cannabis, if he or she admits to having committed the offence and consents to being cautioned. Cannabis cautions are currently issued at the rate of around 200 each month. 

The main argument against legalisation of cannabis use (or any other drug) is that the risk of arrest acts as a deterrent to illicit drug use. An obvious question of interest to policy makers and the public, therefore, is whether drug diversion programs increase or reduce the risk of further offending. Very few studies have convincingly answered this question. The problem is that cannabis cautions are not issued at random. Police tend to issue them to offenders who they believe are less likely to reoffend. This makes it difficult to disentangle the effect of the caution from the effect of factors associated with the person receiving it. 

One way around this problem is to look for a situation in which some cannabis users are less likely to receive a cannabis caution than others for reasons unrelated to their risk of reoffending. The rollout of the cannabis cautioning in NSW occurred sooner in some police local area commands (LACs) than others, but we found no evidence that the rollout of cannabis cautioning was related to either the reoffending rate or crime rate in an LAC. This meant that your chance of receiving a cannabis caution depended on which LAC officer caught you using or possessing cannabis but not on whether you had a high or a low risk of reoffending.

To exploit this, my colleague Dr Anais Henneguelle and I created a large dataset of 71,754 cannabis possession/use cases, 53,011 of which resulted in a cannabis caution, and 18,743 of which, though eligible for a cannabis caution, were in fact prosecuted. Using these cases we constructed a statistical model designed to predict the likelihood of receiving a cannabis caution based on the LAC that handled the case and a few other factors of importance (e.g., socioeconomic status, number of previous cannabis cautions, number of previous arrests). We then took the predicted likelihood of getting a caution from this first model and plugged it into a second model, designed to predict the likelihood of reoffending.

Why do this? Recall that, the predicted likelihood of getting a caution is determined (in part) by the LAC that dealt with the case—and the willingness of an LAC to issue a caution is  unrelated to reoffending. This means we now have a measure of the likelihood of getting a cannabis caution that is unrelated to any effect getting a caution might have on the risk of reoffending. If the predicted likelihood of getting a caution is found to be negatively related to the risk of reoffending we can conclude that getting a caution reduces the risk of reoffending. A positive association would tell us that getting a caution increases the risk of reoffending.

So, what did we find? The results revealed that cannabis users given a cannabis caution were, on average, 16% less likely to reoffend (any offence) than those prosecuted, 13% less likely to be rearrested for a drug possession offence and 9% less likely to be rearrested for a cannabis possession offence. Cannabis cautioning is more likely to enhance public safety than to reduce it.

You can watch Professor Weatherburn's presentation on the study findings from the 2025 NDARC Annual Research Symposium below (starting at 2:20).