He covers criminal justice issues. He is the author of All God's Children: the Bosket Family and the American Tradition of Violence.

What are the best predictors of violence in a very young child?

Criminologists like to use the term "risk factors" as predictors of early childhood violence, and the first things to look for are the characteristics of the family and neighborhood in which a child has grown up. A neighborhood with high rates of poverty, of joblessness and of crime can be a risk factor, but obviously many children grow up poor and in troubled neighborhoods without turning to violence, so there is more to it than community by itself.

Perhaps the condition of the family is the most important factor for a very young child. Criminologists have focused particularly on physical and sexual abuse and emotional neglect as leading to early antisocial behavior, with emotional neglect playing just as critical a role as actual abuse. But it could be any form of poor parenting that impairs a child's emotional development, makes him overly suspicious, extra impulsive or aggressive and subverts his capacity for empathy.

Recent research has shown that the very small proportion of children who are most likely to become juvenile delinquents and later serious adult criminals tend to exhibit antisocial behavior by the age of 6 or 7, in other words, by the time they arrive at first or second grade. Typically, these children have histories of disobeying their parents, of acting out in class, of bullying other children, and of acts like cruelty to animals, lying and petty stealing. Often these children have had the misfortune to live both with troubled parents and in bad neighborhoods, so the family and neighborhoods factors reinforce each other.

A child's risk of becoming violent may be increased by having a parent, usually a father, who has been a criminal and spent time in prison. There has not been sufficient research done to understand the precise mechanisms of why this is so. But Justice Department surveys have found that about half of all the juvenile delinquents sent to locked facilities have fathers or other close relatives who have previously been incarcerated. This is true of all races, and has nothing to do with race as a factor. In these families, it appears the sons often try to emulate their fathers, even though their fathers are a kind of negative role model, and identify closely with them.

Is there an emerging consensus on the above predictors?

Yes, I believe there is a surprising degree of consensus on the predictors of early childhood violence, recognizing that there are always exceptions.

The big differences of opinion are on how one tries to deal with these very young miscreants. Do we just punish them, as more and more Americans want to do, or do we try to treat and rehabilitate them? For the past century, since the creation of the Juvenile Court in Chicago in 1899, the fundamental goal of the juvenile justice system has been to try to treat youthful offenders, to recognize that young criminals are not just small adults, and that their personalities are still in the process of formation. The United States, as a nation, now seems to be moving away from that ideal, as we seek to try more and more juveniles in adult court and sentence them to adult prisons, where they get little in the way of education or therapy to help them change their lives.

This emerging national policy seems to fly in the face of research which shows that while some very young violent children are almost impossible to reform, a large number can be reformed. And the costs of early intervention are much lower than those of incarceration. There are now a number of studies which show that several early intervention programs, like Head Start or infant home visitation programs with trained nurses or social workers, reduce later delinquency and adult crime at a much lower cost than simply building more juvenile and adult prisons.