OVERVIEW OF THE LITERATURE ON ABUSE AND VIOLENCE
Many studies combine more than one form of abuse (e.g., physical abuse and neglect), and those studies indicate that abuse, in general, is associated with violent behavior in children and adolescents. There is some evidence that abuse may be related to moderate to severe delinquency, but not minor delinquency (e.g., Smith & Thornberry, 1995). Severity of abuse may also distinguish the most violent offenders (e.g., Lewis, Moy, Jackson, Aaronson, Restifo, Serra, et al., 1985).
Physical Abuse, Harsh Parenting, and Antisocial Behavior in Victims
Physical abuse has been associated with externalizing (e.g., Bourassa, 2007), antisocial behavior (e.g., Kim-Cohen et al., 2006), and delinquency in many studies (e.g., Felson & Lane, 2009; Rebellon & van Gundy, 2005; Robertson, Baird- Thomas, & Stein, 2008; Widom & Maxfield, 2001; Williams, van Dorn, Bright, Jonson-Reid, & Nebbitt, 2009). Kent (1976) followed children who had been in foster care in Los Angeles for more than 6 months. He found that abused children were more prone to “excessive disobedience” and aggression toward adults and peers compared to a high-risk nonabused group.
Even mere corporal punishment has been correlated with antisocial behavior and delinquency (e.g., Grogan-Kaylor, 2005; Millar, 2009; Straus, 1991). Gershoff (2002) reports large average effect estimates of the impact of corporal punishment on aggression (d = .36) and delinquency (d = .42), adult aggression (d = .57) and adult criminal and antisocial behavior (d = .42). In a study looking for “transatlantic replicability” of risk factors, Farrington and Loeber (1999) report very strong effects of “harsh discipline” on delinquency in the London sample (odds ratio [OR] = 3.3), but nonsignificant effects of “physical punishment” in the Pittsburgh sample, which suggests that not all levels of spanking are equally adverse. Adding to that point, Hoeve et al. (2007) found that high levels of physical punishment were not associated with moderate or serious delinquency at ages 20-25 in either the Pittsburgh Youth Study data or their Child-Rearing and Family in the Netherlands Study.
A proliferation of studies from many samples shows that physical abuse victimization is significantly associated with later violent behavior and offending (e.g., Barnow et al., 2001; Benda & Corwyn, 2002; Carter, Stacey, & Shupe, 1988; Dutton & Hart, 1992; Fagan, 2005; Fang & Corso, 2007; Felson & Lane, 2009; Haapasalo & Moilanen, 2004; Herrenkohl, Egolf, & Herrenkohl, 1997; Herrenkohl, Mason, Kosterman, Lengua, Hawkins, & Abbott, 2004; Lansford et al., 2007; Maxfield & Widom, 1996; Rebellon & van Gundy, 2005; Spaccarelli, Coatsworth, & Bowden, 1995; Swinford, DeMaris, Cernkovich, & Giordano, 2000; Weeks & Widom,1998; Widom, 1989c). In fact, a recent review concludes that physical abuse is the most consistent abuse-related predictor of later violent behavior (Maas, Herrenkohl, & Scusa, 2008). In Widom’s (1989c) original analysis, those who were physically abused were approximately two times as likely to incur an arrest for any violent offense. Dutton and Hart (1992) report that abused subjects in their sample were three times more likely than the nonabused subjects to commit violent acts as an adult. In another study, the OR for an adult arrest for a crime against a person among individuals who had a physical abuse record was huge: OR = 9.91 (controlling for gender), indicating that those who had come to the attention of authorities as physical abuse victims were nearly ten times as likely to incur such an arrest (Cohen et al., 2002). Felson and Lane (2009) interpret the pattern of offender data they examined as indicative that offenders had modeled their own offending on abusive behaviors to which they had been exposed; in their data set, physically abused subjects were more likely to engage in violent offenses than nonviolent offenses. These large effects are attenuated when controls for potential spurious explanations are applied in some studies, but the effects of abuse are generally robust (e.g., Cohen et al., 2002).