Findings: Studies Controlling for Child Effects
Among studies accounting for child effects in this review, the least conservative control for particular child characteristics in a multivariate model. Heck and Walsh (2000) controlled for verbal IQ and found significant associations between maltreatment and violent offenses in a probationer sample. (They also found an association with property offenses). Lansford et al. (2007) control for child temperament and health and report mixed findings. Mersky and Reynolds (2007) used data from a longitudinal study in Chicago and report that “any maltreatment” was associated with petitions for violence controlling for low birth weight (the association between any abuse and “non-violent petition only” was not significant in any of these models) (Mersky et al., 2012). This set of studies suggests that the association between abuse and violence is so far robust with respect to “child effects,” but because the controls were narrowly circumscribed, we cannot rely on them to eliminate this rival hypothesis.
In other studies, broader variables might be construed as controls for “child effects.” As discussed above, controlling for a child’s behavior at a previous point in time can be seen as a conservative indicator of a child effect, and the association between abuse and violence has been surprisingly robust to that control. Pears and Capaldi (2001) report that the association between parental experience of physical abuse was significantly associated with physical punishment of their own children, controlling for “early childhood difficulty” of those same children. Using National Youth Survey data, Rebellon (2002) found a significant effect of abuse on violence, controlling for offending in the previous wave which should, effectively, control for a host of predisposing conditions (also see Rebellon & van Gundy, 2005). In a study by Spaccarelli et al. (1995), serious physical abuse was no longer significantly associated with violence in a model controlling for numerous child characteristics: self-control, autonomy, aggressive control coping, and self-image enhancement. We conclude at this time that the association between abuse, especially physical abuse, and violent behavior of victims is largely robust in studies controlling for child effects.