Predictors of Violent or Serious Delinquency by Age Group: A Comparative Ranking1


Researchers Mark W. Lipsey and James H. Derzon (1998) examined predictors of violent or serious delinquency in adolescence and early adulthood. Applying the procedures used for a meta-analysis, Lipsey and Derzon compiled information from published and unpublished research into a database that indexed the strength of the relationship between the predictor variable and the criterion variable in terms of effect sizes. Through a statistical analysis, the relative strength of different types of predictor variables was measured at different ages, and procedures were used to control for methodological differences between studies. The first goal was to determine which predictors seen at adolescence had the strongest empirical associations with subsequent violence or delinquency. The second goal was to identify which of those associations were of sufficient magnitude to help identify at-risk juveniles to receive intervention.


The table below lists the predictors of violent or serious delinquency at ages 6-11 and ages 12-14 in the order of significance determined by the statistical analysis and in groups based on estimated aggregated effect size.

The most interesting comparisons follow:

  • The best predictors of violent or serious delinquency differ according to age group. A juvenile offense at ages 6–11 is the strongest predictor of subsequent violent or serious delinquency even if the offense did not involve violence. For the 12–14 age group, a juvenile offense is the second most powerful predictor of future violence. Substance abuse is among the best predictors of future violence for children ages 6–11 but one of the poorest predictors for children ages 12–14.

  • The two strongest predictors of subsequent violence for the 12–14 age group—the lack of social ties and involvement with antisocial peers—have to do with interpersonal relations. The same predictors, however, are relatively weak for the 6–11 age group.

  • Relatively fixed personal characteristics are the second- and third-rank predictors of subsequent violence for the 6–11 age group. The ages 12–14 group has a heavier representation of behavioral predictors of subsequent violence (i.e., general offenses, aggression, and school performance).

  • Broken homes and abusive parents are among the poorest predictors of subsequent violence for both age groups.

  • The significance of antisocial peers and substance abuse is reversed in the two age groups. Whereas having antisocial peers is a strong predictor for the age 12–14 group, it is a weak predictor for the age 6–11 group.

Implications for Intervention

For an intervention to be effective, the targeted risk factors must be amenable to change. The strongest predictors of subsequent violence for both age groups are relatively malleable factors. Because they are cumulative, the second rank of variables for the 6–11 age group, the effects of antisocial parents and socioeconomic status, may not be very amenable to change—and gender is not subject to change. The predictors in the first, second, and third rank (except for male gender) for juveniles ages 12–14 are malleable.

Because many of the strongest predictors of subsequent violence can be changed, they offer possible targets for successful intervention. This suggests that disrupting early patterns of antisocial behavior and negative peer support is a promising strategy for the prevention of violence and serious delinquency.

For more information about the meta-analysis discussed here, please see Lipsey and Derzon, 1998.

Table 1

1 This section is based on "Predictors of Violent or Serious Delinquency in Adolescence and Early Adulthood," by M.W. Lipsey and J.H. Derzon, in Serious and Violent Juvenile Offenders: Risk Factors and Successful Interventions, edited by Rolf Loeber and David P. Farrington (Sage Publications, Inc., 1998).

Previous Contents Next

Predictors of Youth Violence Juvenile Justice Bulletin April 2000