Findings: Academic Attainment
Finally, we look at the findings on academic attainment. This construct is qualitatively different than others that we have discussed because of the situational element. The “meaning” probably varies across subjects. For example, in young adults, academic attainment may really be a sign of current enrollment in school which has additional implications for criminal opportunity (routine activities) and income. In some cases, the measures signify dropping out of school early, in which case academic attainment may be more of a proxy for academic skill or other problems that induce adolescents to quit school. Academic attainment may also serve as a proxy for learning disability.
Academic attainment is operationalized in a number of ways, including dummy codes for high school drop-out, years of school completed, whether or not the subject is enrolled in school, “some college,” and “college degree.”
Of 13 studies looking at violence as an outcome, 11 report a PoC in the predicted direction, and 6 PoC are also statistically significant, leading us to conclude that the association between academic attainment and violence is highly consistent. There were 5 studies of nonviolent-only offending, and the findings are more divided, which may be taken as burgeoning evidence that academic attainment is a differential predictor of violence. In a published example, Jang and Franzen (2013) used Add Health data and report a large majority of analyses where academic attainment was significantly, negatively associated with violence, but in all their models, it was positively associated with nonviolent-only offending. However, Doherty and Ensminger (2013), who also separately reported models with violent and nonviolent offending, found that both nonviolent offending were significantly, negatively associated with academic attainment (as was violent offending). Associations reported by Cusick et al. (2012) were in the opposite direction to that predicted for both crime types.
Authors of 10 studies compared academic attainment between violent to nonviolent offenders and the findings are mixed, though they do lean in the predicted direction (6 out of 10). If we parse the studies by quality, we point out that these are all small sample studies with the potential for serious sampling influence in comparing violent to nonviolent offenders. Thus, although there are 10 studies in this category, we do not feel that they strongly support one conclusion over another.
Using data from the National Youth Survey, we can directly look at the comparison we want to examine. While the average highest grade completed is slightly lower among those who report any nonviolent offending compared to non-offenders (12.3 vs. 12.8) in wave 7 and this small difference is statistically significant in this large data set (r = - .064*), the difference is considerably larger when comparing any violent offenders to other participants (11.8 vs. 12.8) and that correlation is larger in magnitude (r = - .202**). It is also true that those who report violent behavior in wave 7 have lower academic attainment than those who report only nonviolent offending. (The average academic attainment among those who self-report violence in the last year is 11.8; it is 12.7 among those who report theft only.) If we run a logistic regression, we find that violent offenders still have significantly lower academic attainment compared to nonviolent-only offenders, controlling for age, race, gender, marital status, community disorder, and whether or not the participant was receiving welfare or public assistance. In the NYS data set, it is clear that the violent participants have lower academic attainment than the nonviolent offenders, but the magnitude of that difference is not large. Because delinquency also predicts lower academic attainment (e.g., Tanner, Davies, & O’Grady, 1999), it is important to understand that earlier delinquency may be responsible for leaving school early. In the absence of that information, a firm stand cannot be taken about these findings.