Findings for large sample studies on intelligence and executive functioning do not change our conclusions about their consistent association with violence. Barker et al. (2007) (n = 698) found violent behavior to be significantly associated with other cognitive tests in multivariate analysis, controlling for theft. Nonviolent offending was not associated with cognitive tests in models controlling for violence. They also report no relationship between other tests of EF and violence (or theft). In the Cambridge Study in Delinquent Development, low verbal IQ was associated with some, but not all, measures of violence from childhood into adulthood. H0gh and Wolf (1983) report significantly higher prevalence of “low IQ” among violent offenders in the Danish study, Project Metropolitan. Low IQ and other intellectual measures have been associated with violence in other large samples as well (e.g., Holland et al., 1981) [n = 390]; Loeber et al., 2002 [the Pittsburgh Youth Study]; Nagin & Tremblay, 2001 [n = 1035]; NICHD, 2004; Seguin et al., 1995 [n = 1,037]). In his analysis of data for 256 male delinquents, Walsh (1987) reported that IQ scores were significantly, negatively associated with violent crime, but positively associated with property crime and he also reports a significant association between the discrepancy between performance IQ and verbal IQ (PIQ minus VIQ) and violent behavior (Walsh et al., 1987).
Not all of the larger sample studies are wholly supportive of our hypothesis. The association between IQ and violent crime was attenuated in a later multivariate test by Walsh and Petee (1987). Walsh et al. (2004) used data from a sample of 672 male inmates and report that the number of violent charges was significantly, negatively associated with IQ, but those categorized as violent did not have a significantly lower score than nonviolent offenders (Walsh, Swogger, & Kosson, 2004).
For verbal ability, the findings mainly hold for large samples as well (Dionne et al., 2003 [n = 562]; Beaver, Vaughn, DeLisi, & Higgins, 2010 [Add Health]). Farrington (1989) reported a pattern of findings where low verbal IQ was associated with indicators of violence, though many of the comparisons are not statistically significant (n = 411). In a study of 1,037 French-speaking boys living in low SES areas in the Montreal area, Seguin et al. (1995) found that those with lower scores on a verbal test were more likely to be on a high physical aggression trajectory, but the association was not statistically significant.
Some other studies using large samples and multivariate analysis have not found the expected association between verbal ability and violence. Bellair and McNulty (2005) also used Add Health data and their model specification included a control for previous wave violence, which we believe to be very conservative for our purposes. They did not find a negative association between verbal ability and violence in their multivariate models, but they do report significant associations between other cognitive tests and violence (Bellair & McNulty, 2010). In another large multivariate study, the authors report a null relationship between verbal ability and physical aggression in another sample (Campbell, Spieker, Vandergrift, Belsky, & Buchinal, 2010).