Abuse and Trauma: Findings and Research Needs
There is consistent evidence that all three forms of abuse studied (physical abuse, neglect and sexual abuse) are correlated with violent behavior. Our summary provides weak support that trauma is differentially associated with violence. Abuse was also correlated with nonviolent offending, so a “dual-dependent” variable comparison does not provide support for the differential etiology hypothesis. We also found, however, that violent offenders report having been abused and traumatized more than nonviolent offenders. This was also the case specifically for physical and sexual abuse. Sexual abuse has consistently been associated with both violent and nonviolent offending, and this conclusion holds when we examine the many studies of females.
A nagging question remains. Early findings reported by Widom confirmed that neglect children were more likely to commit violent crimes later in life. While we conclude that the body of literature on neglect indicates that it is not a differential predictor of violence, our own analysis, published in a separate paper (Savage & Murray, 2015), shows that those reporting neglect victimization were more likely to commit violent offenses in the transition to adulthood. This finding was robust in conservative models, including a control for nonviolent offending. Thus, we would like to see some more careful tests of associations between neglect and violent behavior.
There are many other remaining questions about abuse and its effects on violent behavior in children. First, in spite of so many studies, we cannot yet say what amount of physical discipline is harmful. Because many people spank their children, measuring the degree of physical punishment has important applied significance. It is also unclear whether all age groups are similarly harmed, or whether the effects are stronger at some ages than others.
Relatedly, we recommend pursuing the best measures of abuse through further refinement. Studies of abuse usually suffer from recall biases or oversimplified dummy variables that mix some abused and unabused individuals in what we call “muddy” groupings. This practice disguises what we expect might be dramatic differences if we really compared “abused” to “non-abused” children. Greater use of indicators of severity and chronicity of abuse that account for multiple forms of abuse is needed. We encourage the ongoing separation of different types of abuse to strengthen our understanding of these very different, albeit correlated, phenomena.
There is still no clear consensus about the factors that mediate the association between abuse and violence. One important line of research would further explore Widom’s hypothesis that the effects of abuse may be mediated by alcohol or drug abuse. It would be useful to estimate the proportion of abused individuals (carefully defined) who develop observable substance abuse problems and whether those substance problems lead to violence using a national probability sample. If we find that this is true, then interventions including strong substance abuse prevention components would be warranted. To this end, we need greater exploitation of longitudinal data, drawing the temporal links between abuse experience, substance abuse, and violent behavior.
There are some other mediating factors that have been understudied, such as academic achievement and school bonding, intellectual impairment and brain injury, emotional negativity, other psychological sequelae such as post-traumatic stress disorder, callous-unemotional traits, dissociation and psychopathy. Studies of very serious offenders imply that trauma may be an important intermediary in the association between abuse and violence among abuse victims, but operationalization of trauma in this dynamic must be focused on symptomatology, rather than events, to be helpful.
Trauma per se and its relation to violent behavior has been understudied. Indicators of trauma vary a great deal from to study. We need research that draws the links between specific types of events, the psychological experience of those events and outcomes, and behavioral outcomes. In addition, little is known about the conditions under which violent behavior becomes a consequence of exposure to trauma. There are many experiences that can be traumatic (parent suicide, abandonment, sexual victimization), and we need to estimate the likelihood of violent responses for each of them to make them more useful for targeting and tailoring interventions.
Although we tentatively concluded in Chapter 8 that the effects of abuse on later violence are largely robust to child effects, we believe that important insight could be gained through genetically-informed analyses with violent behavioral outcomes (rather than general measures of externalizing). Such studies would need to distinguish physical abuse, neglect, and sexual abuse to provide more information about the spectrum of effect sizes in this area.