Findings on physical abuse are especially important here because of strong theory that links physical abuse victimization with later violent behavior. The evidence is very consistent across studies; physically abused subjects are more likely to commit both violent and nonviolent offenses.
When violent offenders are compared to nonviolent offenders, we find fairly consistent evidence that violent offending has a special relationship with physical abuse. Five out of eleven studies reported a PoC where measures of physical abuse among violent offenders were significantly greater than those seen among nonviolent offenders (Famularo et al., 1990; Loeber et al., 2005; Pollock et al., 2006; Spaccarelli et al., 1995; Tupin et al., 1973) (see Table 8.1). In addition, 4 more studies out of that 11 report a PoC in the predicted direction (e.g., Haapasalo & Moilanen, 2004), and some of these could be taken as supportive of our prediction. In their small sample of Finnish recidivists, Hamalainen and Haapasalo (1996) found that the violent offenders were more likely to report physical abuse overall, but the difference was not statistically significant. The level of certain forms of physical abuse was significantly higher among violent offenders-spanking, being slapped or hit with a first, or being pushed over. Hill-Smith et al. (2002) found that the murderers in their very small sample of convicted adolescents in the United Kingdom were more likely to have harsh parents than burglars matched on age and race, but the differences were not statistically significant in this low power analysis (n = 42). Hughes et al. provided two comparisons (violent delinquents to nonviolent delinquents and homicidal delinquents to nonviolent delinquents). In both cases, the violent offenders had a greater incidence of physical abuse (child injury, burning, poisoning), but the association was significant for only one, so this did not meet our criterion of “a preponderance” (Hughes, Zagar, Busch, Grove, & Arbit, 2009). It should be noted that all the subjects in the sample had been abused, so when they compare those who had been injured to those who had not, they are making a comparison between physical abuse and those who had been physically abused without injury, or neglected, or sexually abused. Ramoutar and Farrington
(2006) report significant differences among females but not males. (The sample consisted of offenders in Trinidad.) Thus the only fully contradictory findings were reported by Ramoutar and Farrington (2006) and McClintic (2003) who reported no significant differences for any comparisons.
These studies as a whole provide resounding support that physical abuse has a special relationship with violent offending. This is not to say that physical abuse is not associated with nonviolent offending. In 11 out of 12 studies, the PoC associating physical abuse and nonviolent offending was statistically significant (e.g., Benda, 2005; Cusick et al., 2012; Lansford et al., 2007; McClintic, 2003; Ramoutar & Farrington, 2006; Rebellon, 2002; Rebellon & van Gundy, 2005; Savage et al., 2014; Sousa et al., 2011; Zingraff et al., 1993).
It should be pointed out that Savage et al. (2014), in an analysis of Add Health data designed to test the present research question, controlled for nonviolent offending in their model and did not find that violence was still associated with physical abuse victimization. The measure of physical abuse was fairly simple, and we used a nonclinical sample which is unlikely to include many serious violent offenders. Tests with other data sets would be needed to draw any firm conclusions.
From this pattern of findings we conclude that, while significant associations between physical abuse and violence are not more consistently reported than significant associations between physical abuse and nonviolent offending, evidence from offenders suggests that physical abuse is more prevalent among them, which provides support for the differential etiology of violence hypothesis.