According to this research, childhood issues associated with anxiety, including nervousness and social withdrawal, may prevent a certain subgroup of teens from acting criminally, at least until adulthood. Once adulthood is reached, however, the effect appears to wear off. From the article:
The authors found that being nervous and withdrawn shielded boys against committing criminal acts during adolescence, but, after the age of 21, it no longer held them back. Compared with early onset offenders, late onset criminals were more nervous, had fewer friends from ages 8 to 10, and were less likely to have had sexual intercourse by the age of 18. Compared with nonoffenders, those who turned to crime later in life were more anxious at school from ages 12 to14 and very neurotic by age 16.
I haven't read the actual research, just the article, but this makes some intuitive sense, if we broadly consider two different antisocial categories based on psychopathy (despite psychopathy being a continuous rather than discrete variable). Individuals loading high on psychopathy will generally experience relatively little anxiety, both generally as well as specifically towards potential consequences. This is in comparison to individuals who may have a predisposition towards antisocial behavior, but for reasons other than psychopathy. The anxiety the non-psychopathic adolescents experience appears strong enough to prevent acting out, at least until they reach adulthood.
From an intervention perspective, the article's suggestion that certain efforts may be effective in preventinglate-onset antisocial behavior in this group, if we can address soem of the underlying factors prior to adulthood. I'd certainly give this much more credence with this subgroup than with the psychopathic subgroup, who (at best) it can be argued require a whole different set of interventions that (as of now) haven't been shown to be statistically effective.