CONTROLS FOR PARENT FACTORS AND PARENTING
The practice of controlling for parenting factors in studies of intelligence and crime is uncommon. Perhaps this is due to the assumption that parenting practices do not affect native intelligence, which is only partially true, and the lack of attention to potential child effects on parenting. So far, we do not see major changes in coefficients in models controlling for parenting factors, and so far, associations between intelligence and violence have largely withstood controls which are mainly limited to parent education (e.g., Nagin & Tremblay, 2001; NICHD, 2004; Piquero, 2000). Miura (2009) controlled for “changes in the person who brought up the child” and child abuse victimization, and reports mixed findings which did not change from the simple correlation estimates.
Our findings suggest very consistent inverse associations between measures of intelligence and violence overall. They also provide fairly clear evidence that violent offenders have lower intelligence levels than nonviolent offenders. This was true for full scale IQ, verbal ability, and, surprisingly, performance intelligence as well. Although much has been made in reviews about deviation scores, we found only a small number of studies reporting associations between deviation scores and violence-too few to draw firm conclusions. It remains to be seen if the association is robust with respect to academic achievement and other school factors.
Taken as a whole, we can also say that the pattern we observed suggests that poor EFs are also associated with violence, though it appears that this is true for some EFs and not others. The same is true when we examined offender studies- there is a definitive pattern of lower scores on EFs among violent offenders. We draw no firm conclusions on any particular EFs because even when there are a reasonable number of studies, they often used very small samples. The pattern of findings suggests that measures of planning, problem solving, attention, cognitive control, and combined measures of EF are the most promising as differential predictors of violence. The evidence on memory is not as clear. One remaining threat to these conclusions is that studies comparing violent to nonviolent offenders may have several biases related to sampling and selection, leaving plenty of room for future research.
Some mysteries remain. First, the associations between indicators of intellectual function and violence among females are much less consistent. We conclude from these studies that we do not really know whether it holds for females or not. There are surprisingly few studies of nonviolent offending, and we discovered that no generalization can really be made at this time about whether nonviolent offending is actually associated with intelligence or EF. Studies by Barker et al. (2007, 2011) suggest that measures of EF were positively associated with theft trajectories. The contrast in findings was stark in that report and the authors concluded: “It is obvious that the aggregation of these 2 types of behavior in a total antisocial score ... may seriously impede understanding the etiology of the disorders” (Barker et al., 2007, p. 597).
In Chapters 12 and 13, we will discuss our recommendations for future research on this and all other topics, and the policy implications.