Attachment and the Differential Etiology of Violence
If attachment problems are differentially associated with physical aggression and violence, we would have three expectations. First, empirical associations between low attachment and violence should be more consistent in the literature than associations between attachment and nonviolent offending. Second, studies which report findings for both violent and nonviolent offending should report stronger associations for violent offending. Third, violent offenders should report lower attachment to parents than nonviolent offenders.
Attachment and Nonviolent Offending
Most studies report similar findings for violent and nonviolent offending (Alarid et al., 2000; Benda, 2005; Buist et al., 2004; Cheung & Cheung, 2008; Gottfredson et al., 1991; Johnson, 1979; Liska & Reed, 1985; Marcus & Betzer, 1996; Owens- Sabir, 2007; Ozbay & Ozcan, 2008; Rankin & Wells, 1990; Sarracino et al., 2011). A large group report negative coefficients estimating the association between attachment and nonviolent offending as being more common than they are for violent offending, which is the opposite of our prediction (i.e., Chui & Chan, 2011; Fagan et al., 1983; Gardner & Shoemaker, 1989; Jang & Franzen, 2013; Ozbay & Ozcan, 2006; Peacock et al., 2003; Sousa et al., 2011).
Breaking these findings down by attachment measure, there was just one study using an attachment “category” and nonviolent offending; its findings were ambiguous. There was only one study estimating the association between caregiver sensitivity and nonviolent offending, which is unfortunate since this is such an important indicator of attachment. That study reported associations between measures of parental sensitivity and nonviolent offending and the findings were statistically significant. Studies using attachment scales or indicators of parental “bonding” did provide numerous estimates of associations between attachment and nonviolent antisocial behavior, and these show that attachment is inversely associated with nonviolent offending with approximately the same consistency as it is with violent offending. Nineteen of these 32 studies reported a PoC that was both in the right direction and statistically significant, which is inconsistent with the idea that attachment problems may have a special association with violence.
There were only three studies in which the association between parental loss and nonviolent antisocial behavior was reported. In two of the studies, the coefficients were largely in the right direction, but not statistically significant, the PoC of the other study was in the opposite direction to that predicted. This may be taken as suggesting the possibility that the effect on nonviolent offending is less strong than that on violent offending (where 3 out of 8 studies reported a PoC with statistically significant coefficients in the predicted direction).