Add Health Data
Because few of the studies compared violent to nonviolent offenders, and model specifications in the multivariate tests of the association between attachment and violence were not exactly as we would design them, we ran some simple analyses to test the differences directly looking to a widely-used data set for insight. In the National Longitudinal Survey of Adolescent Health (Add Health), maternal attachment can be computed on the basis of two items rated by the adolescent subjects: “How close do you feel to your mother?” and “How much do you think she cares about you?” The average maternal attachment among those who reported any violence in wave 1 was 9.34 compared to 9.21 for those reporting nonviolent offenses only. Correlations between maternal attachment and frequency of violent offending were negative, but smaller in magnitude (r = -.022) than they were for frequency of nonviolent offending (r = -.147). In a multiple regression analysis of the standardized outcomes, to enhance the comparability of the coefficients, maternal attachment is significantly, negatively associated with frequency of violent offending (B = -3.035**), but the coefficient is also very large for nonviolent offending (B = -8.061**). In a logistic regression, testing whether maternal attachment was different for violent offenders compared to nonviolent-only offenders, we find no significant difference.
In summary, the Add Health data set reveal no evidence to suggest that violent offenders have lower levels of attachment than nonviolent ones, or that attachment is more strongly associated with frequency of violence than it is with nonviolent offending.