Education Factors: Findings and Research Needs
We found that academic achievement has a very consistent, negative association with violence, even in strong multivariate tests. However, there are too few studies to determine with certainty whether specific academic deficits in reading or math are associated with violence (see Table 12.1) or with nonviolent offending. The evidence does so far indicate that violent offenders have academic deficits in all of these compared to nonviolent offenders.
More detail on academic achievement levels that put a child at risk for violence would be useful to identify children for intervention. For example, it may not be the case that those getting Bs are at a greater risk of violence than those getting As, or that those getting Cs are at a greater risk than those getting Bs—it is possible that those at risk for violence are concentrated among those failing school and being held back. We need empirical analyses designed to make this type of distinction.
Genetic researchers might argue that associations between violence and academic achievement may be due to native intelligence. There were too few studies including measures of both constructs in their models to know. A small number of genetically-i nformed studies or studies controlling for IQ could be used to settle this issue. So far, academic achievement remains significantly associated with violence in models controlling for IQ and prior violence, so our concerns on this issue are not strong, but a study or two designed to test this particular question would be useful. Other important control variables in studies of academic achievement and violence include nonviolent offending, family income, and peer influence.
In addition, low school attachment was also unexpectedly, consistently associated with violence across many studies, even in multivariate tests that controlled for economic factors, but it was also consistently associated with nonviolent offending. Very few studies compare violent to nonviolent offenders; those reported so far suggest that violent offenders have lower school attachment than nonviolent offenders. Few studies control for academic achievement or parent education, so the current set of findings are, in our mind, inconclusive. We recommend controlling for grades in estimates of the effects of school attachment.
Parent education is also clearly a risk factor for violence, and there is burgeoning evidence that it is a differential predictor of violence. Findings, so far, indicate that parent education is not consistently associated with nonviolent offending, and the few comparisons of violent to nonviolent offenders also support the differential etiology thesis. Authors should be cognizant that the operationalization of parent education, whether we compare college-educated to those with high school diplomas, or whether we compare those with high-school diplomas to those who dropped out of high school, may influence the results observed, and we recommend that analysts work with those operationalizations to come to some consensus on this point. Further, because parent education cannot influence child behavior directly, a vast and open area of research might ask what educated parents do that suppresses violence in their children. We might speculate, of course, that minimal use of corporal punishment and the conveyance of beliefs and attitudes that favor other approaches to conflict resolution might be at work.
Although too few findings on learning disability (LD) and nonviolent offending have been reported to know if LD is more common in nonviolent offenders compared to non-offenders, in 4 out of 4 studies comparing violent to nonviolent offenders, the violent offenders had greater learning deficits, and in 2 cases the preponderance of findings were statistically significant. This provides weak support that LD, like academic achievement, may be differentially associated with violence. It is completely unclear at this time if any particular learning disabilities are associated with antisocial behavior or violence. The research on LD and violence includes a variety of measures, and there is a need for more specific definitions in order to learn anything of practical value. It is probably not too soon, however, to test whether interventions already in progress to address learning disabilities improve academic performance and result in declines in physical aggression or violence. Determining which age groups are most helped by these interventions would be useful for pinpointing policies designed for this purpose.
In order to understand the potential for using indices of “school problems,” suspensions or expulsions to identify those at high risk for violence, we would need more research. In particular, it would be important to examine data for students who receive suspensions and expulsions for nonviolent acts and follow up over time. Similarly, indices of school problems that do not include violent acts are needed in order to disentangle the dual causal paths. In six out of six studies, school problems were associated with violence, and in comparisons of violent to nonviolent offenders, the violent offenders had greater school problems. So the findings lean in the direction of school problems being a risk factor for violence, but since suspensions and expulsions could be due to violent behavior, more research is needed to address the problem of temporal order. In addition, it may yet be the case that the association between school problem indicators and violence is due to the fact that those who are suspended or truant frequently also have low grades, so more studies where academic achievement is controlled in multivariate models are needed to understand this association.
Low academic attainment is, surprisingly, very consistently associated with violent offending but reports so far suggest that it is not associated with nonviolent offending (see Table 12.1). Academic attainment has been lower among violent than nonviolent offenders in most studies. This combination of evidence suggests that academic attainment is a differential predictor of violence. It is with a little embarrassment that we acknowledge that we cannot say what levels of education would constitute risk or protection because the reports do not provide enough information for us to figure that out. We would anticipate that studies comparing those with college to those without it would have smaller effect sizes than those comparing those who finished high school with those who did not, but this remains an empirical question. In addition, testing interventions such as GED programs to see if they prevent offending in the transition to adulthood, and violent offending in particular, are important—especially interventions for juvenile delinquents.