John's old man suffered from mental illness and his family was involved in a gang. After his parents broke up he lived with his mother, who was a heavy drinker.

Violence was a constant presence in his house. "One time she whacked my brother in the face with a heavy stick."

While mum was at the pub "playing pokies and whatnot", John had to look after himself, and his brothers and sisters. When she had parties, they'd watch the fights through the windows. Often they were without food. John left school having barely started high school, and he learned to steal; mostly he broke into houses. "When you're in a house, never look at the photos. You get that guilty conscience. I never laid an eye on one photo."

He shrugged his shoulders when I suggested his early life had been tough. He hadn't really thought about it. "Crime was normal." But what makes John interesting is that he no longer steals. He is among a small minority of young men who go to prison but who then stay out of trouble. I wanted to know what made him get off our infamous prison conveyer belt.

Advertisement

Among young offenders, the rate of reconviction within five years after release is 91 per cent, much of which happens within the first 12 to 24 months. That means that somehow only 9 per cent of young people who go to prison learn their lesson and stay out of trouble. More court, more prison, more cost, more victims. This story of failure is well traversed but what can we learn from success?

What makes John, whose name I have changed, and the others in that 9 per cent different? And can we replicate what they do to shrink the 91 per cent?

The Department of Corrections engaged me to find the answers. I had to track down 50 people who were sent to prison before 20 but who had successfully stayed out of trouble since their release.

It wasn't easy. Many were simply unreachable, having long since departed old addresses and abandoned old phone numbers. I drove around for days and days up and down the country knocking on doors. Twice, as I sat in my car trying to avoid dogs, I was accused of being an undercover cop.

An old Samoan woman was weeding her driveway. I asked her if she knew where her grandson was. In Australia, she said. We sat in the sun on the edge of her driveway talking. Taking time with people is important, to be polite. And to probe. I wished her well as I left. I'm pretty sure she thought I was a cop too.

When I found someone I was looking for, I was treated generously. Only two people declined to participate. I was welcomed into homes and lives. I teased from people things they would rather forget. It is important, I told them, and I believed it. Each received $50 for talking with me, but for one person that token was extended. Offering me a cup of tea I used the last of his milk, which came from an otherwise completely empty fridge. As he and his kid wished me farewell I gave him everything I had, perhaps another $40, and I wished it was more.

My colleague Ben Elley and I pored over the data. What we found was telling.

All participants took responsibility for their crimes and most were deterred from crime by prison and by upsetting their families - most often their mums. These factors were crucial in making a decision to change but enabling the change came from having professional social support on release - which was absolutely vital - as was breaking away from antisocial peer groups.

These finding are valuable on their own but they also provide important building blocks to further inquiry. Why did prison (and shaming mums) deter these people when it clearly isn't effective for the vast majority? Did the fact these people accepted responsibility for their crimes mean they felt capable of changing their behaviour? What was it about their post-release support options that made a difference? For answers I need a sample of people who reoffend with which to compare results. I need to understand failure. I need to go to prison.

The place of failure is the place where we will find even clearer answers. While it's common (and quite right) that we focus on the victims of crime, sometimes a focus on criminals is just as important.

Often, like John, they are not only criminals but victims as well.

Jarrod Gilbert is a sociologist at the University of Canterbury and the lead researcher at Independent Research Solutions. He is an award-winning writer. A full copy of the report is on his blog at www.jarrodgilbert.com