We found very strong indications that academic achievement is a consistent predictor of violence. In addition, other indicators of academic ability, such as reading, and measures of learning and learning disability also lean in this direction, though we have too few studies to say this with confidence about any particular academic skill. Further, measures of school attachment were also very consistently associated with violence, as were indicators of parental education, learning problems, school problems, and academic attainment.
Less certainty was achieved for answering our principal research question: whether these educational indicators predict violence per se, over and above their effect on offending more generally. The strongest support for the differential etiology of violence thesis is found in studies of academic achievement and learning. When we combine the comparisons of violent to nonviolent offenders across indices of grades, reading, math and learning, we see that the vast majority of studies report that violent offenders perform less well on academic measures than nonviolent offenders.
Another good prospect for a risk factor in the differential etiology of violence is broadly defined “school problems" A small number of studies shows that violent offenders have greater problems in school than nonviolent offenders, and our supplemental analyses Add Health data supported this conclusion as well. Important potential confounds have not yet been eliminated as rival hypotheses (GPA, chronic offending).
Parent education is clearly associated with violent behavior in children, and not so far associated with nonviolent offending, so we tentatively categorize is as a differential predictor of violence, but there are too few studies that shed light on the differential effects of parent education on violent compared to nonviolent offending to draw any conclusions from the existing literature. Our own analysis of Add Health data, presented here, however, suggests that parent education is lower among violent than nonviolent-only offenders, so we conclude that this remains a viable differential predictor of violent offending.
The findings about academic attainment are blurry but lean in favor of a differential association with violence. It is fairly clear from the literature that violent offenders have lower academic attainment than nonoffenders but consistent associations between academic attainment and nonviolent offending have not emerged. When compared, violent offenders usually have less schooling than nonviolent-only offenders. Our own analysis of NYS data suggested that the academic attainment was differentially associated with violence, so it remains a “good prospect"
Surprisingly, low school attachment was associated with violent behavior but seemed to be found in both violent and nonviolent-only offenders. There are too few studies to draw a conclusion about the differential etiology thesis, but the dual-dependent variable analysis suggests that it is not supported, so far, in the literature. As we earlier mentioned, we have no reason to expect school attachment to be differentially associated with violence, but the wealth of data indicating a correlation should not be dismissed.
Our findings have raised questions about whether some academic factors, such as reading, parent education and academic attainment, are associated with nonviolent offending at all. We are surprised at the paucity of studies that specifically look at the impact of educational factors on nonviolent-only offending. Our findings also beg the question of whether the association between academic factors and violence is simple and direct, or whether there are important mechanisms or mediating factors. We believe the existing literature already addresses concerns about certain potential confounding factors such as parent education and SES, but, due to the centrality of school in the lives of children, and the opportunities for intervention, detail about how school factors and violence are connected deserves significantly more attention than it has thus far received.