Multivariate: Control for Parenting
In our introductory remarks, we cited numerous authors who make the case that the effects of poverty on offending might be mediated by parenting. There are two ways this can be addressed in statistical analyses. Authors might hypothesize that the association between poverty and offending is spurious: fully mediated by parenting factors. In this case, they might include control factors measuring parenting in multivariate models. In the second case, they might hypothesize that poverty influences parenting which in turn influences offending. In this case, the association between poverty and offending is not spurious but, instead, indirect. If these dynamics vary by crime type (violent vs. nonviolent), we might see this in a subset of the studies we review here.
Unfortunately, there are few studies in our review that report individual level analyses, where we are most likely to find controls for parenting or parents, and fewer still that do apply such controls. Jarjoura et al. (2002) found that duration of poverty was significantly, positively associated with assault and property offending in models which controlled for cognitive stimulation in the home, “ever lived away from mother" and poor supervision by mother. Maschi et al. (2008), in a model designed to test the impact of trauma, did not find the predicted associations between poverty and property offending or violent offending. This model included parent education and a variety of other factors. We cannot say, at this time, whether the association between poverty and offending is robust to controls for parenting, or operates indirectly by that route.