Predicting re-offending

The Youth on Track model aims to focus its effort on those people most likely to re-offend. A key to the proposal is being able to identify those people as quickly and easily as possible. The NSW Bureau of Crime Statistics and Research (BOCSAR) have developed several models to predict which young people are at highest risk of re-offending. BOCSAR have found that re-offending can be accurately predicted using only a few characteristics of the young person that are easy to find out at the time of the caution:

  • Gender
  • Aboriginal and Torres Strait Islander status
  • The number of prior contacts the person has had with the criminal justice system
  • Whether the caution came from police or from a court.

Risk of re-offending varies considerably with these characteristics. See Figure 2 for predicted re-offending for police cautions and Figure 3 for predicted re-offending for court cautions. A non- Aboriginal and Torres Strait Islander female with no prior contacts with the criminal justice system receiving a police caution has only a 14% estimated likelihood of offending again in the future, whereas it is almost certain that an Aboriginal and Torres Strait Islander male with a long history of offending will re-offend in the future.

  Figure 2 - Predicted re-offending rates for police cautions

 Predicted reoffending rates for police cautions

Figure 3 - Predicted re-offending rates for court cautions

Predicted reoffending rates for court cautions 

Since we know the estimated probability of re-offending for each person, we can show how the offending population is distributed by their re-offending risk. See Figure 4 for a distribution of young people receiving cautions by their predicted risk of re-offending. On average, about 50% of young people receiving cautions will re-offend some time in the following three years. However, we can identify a group for which the re-offending probability is significantly higher.  

 

Figure 4 � Distribution of young people receiving cautions by risk of re-offending

Distribution of young people receiving cautions by risk of reoffending

Cautions & Contacts

 

Cautions & Contacts 

The above table shows the likelihood of re-offending by the number of cautions and then all criminal justice contacts. It shows that a large proporiton of people don't re-offend after a first caution, but following two or three the likelihood increases, but the people who appear at court or go to conference have higher likelihoods of re-offending. 

In order to most effectively divert young people from becoming lifetime clients of the criminal justice system, it makes sense to focus resources away from people at very low risk of re-offending, and towards those that we are quite sure will re-offend. Our proposed model of Youth on Track targets all those offenders with an estimated probability of re-offending of at least 75% (the red bars in Figure 4).