From an article at Science about their work:
Caspi's first collaboration with Moffitt involved a question that had long intrigued him: Do dramatic life events change people, or simply cement who they already are? The team focused on female puberty. Just over half of the Dunedin study subjects were girls, who by age 15 had been given physical and psychological exams since childhood. Previous studies showed that reaching puberty early is especially stressful, and Caspi and Moffitt found that the girls who had the most trouble adjusting to early adolescence were those who had shown behavioral problems in early childhood. That contributed to Caspi's "accentuation" hypothesis—that stressful transitions tend to accentuate who we basically already are.That makes it sound like criminality is genetic, but then there's this:
Since then, their work has often explored the darker side of human nature. Contrary to what the idyllic landscapes in the Lord of the Rings films might suggest, New Zealand is not heaven on Earth. Economic inequality is comparable to that of the United States, and addiction, suicide, and assault rates are similar. By promising strict confidentiality, the research team gets shockingly frank confessions from their subjects, including about brutality and criminal behavior. (A startling example: Women admitted to physically abusing their spouses as often as men, although they generally inflicted less physical damage.) "We have a policy of never interfering and never ratting them out," Moffitt says. "And the New Zealand police, who understand the value of our research, have never asked."
This long, intimate surveillance enabled Moffitt to track a troubled subset of the Dunedin cohort. Delinquent behavior is known to peak between the late teens and mid-20s, mostly in men. By 15, about a third of the study's boys took part in some degree of delinquency, Moffitt found, while a subset offended more frequently. When she looked at data from early childhood, she found that the habitual offenders had been making trouble from age 3, and had arrest records starting before their teen years.
Over the years, Moffitt reported in a series of papers that these boys did poorly in neuropsychological tests (such as verbal skills and verbal memory), measured high for impulsivity, and were likely to engage in substance abuse as they grew older. In 2002, she reported that at age 26 this same group was committing most of the crimes in the community—a pattern that persisted well into their 30s. In short, whereas many boys exhibited "adolescent-limited" criminal behavior, about 5% were "life-course-persistent" offenders. The work had important implications for social work and law, and won Moffitt the 2007 Stockholm Prize in Criminology.
It's commonly known that children who are abused often become violent adults. In the Dunedin group, for example, about half the boys with abusive childhoods grew into men prone to committing crimes.Fascinating.