Poverty: Findings and Research Needs
Although poverty was consistently associated with nonviolent offending in the studies reviewed here, it was more consistently associated with violent offending. This was true when aggregate indicators of concentrated disadvantage or indexes of resource deprivation were used rather than of standard measures of poverty, which has implications for future research. Too few studies using individual-level data were available to allow us to draw any conclusions at that level of analysis, though our own analysis suggests that violent individuals have significantly lower income than nonviolent-only ones, even controlling for frequency of offending (see Chapter 9). Given the consistency across analyses, in particular in a set of direct tests and multivariate analyses, we conclude that poverty is likely to be a useful differential predictor of violence and strongly recommend further research on this issue. The said research must take into account the fact that poverty has also been consistently associated with nonviolent offending and must employ statistical methods to distill the effect on violence.
Unlike some of our previous chapters, we found no comparisons of income (or poverty) using violent and nonviolent offender samples. This would be a useful addition to the literature.
Some findings have suggested that violence may be more associated with extreme poverty than nonviolent offending, which indicates that more studies using disaggregated measures of poverty (as used by Krivo and Peterson, 1996) or allowing for curvilinear effects could be used to determine the levels at which poverty matters most in the etiology of violence. Blunt measures, such as the percentage of the population under the poverty level, are likely to be “muddy.” Some individuals categorized in the poverty group will have enough income to be comfortable, such as those who are supported by others, people who supplement their official, reported earnings with under-the-table (or illicit) earnings, those receiving welfare benefits, or those living in areas where the cost of living is low. Other individuals may not fall under the official poverty threshold but may live in dire circumstances.
A major empirical question not addressed in our review is the matter of whether poverty that persists for a significant period of time will have a greater effect through its impact on child development (e.g., Jarjoura et al., 2002). This idea has not been discussed more fully here because it has not been developed extensively in the criminological literature, but it has been discussed in the literature on developmental psychology. In fact, findings by Pagani, Boulerice, Vitaro, and Tremblay (1999) suggest that a “change and process” model of poverty in childhood and adolescence may be the most appropriate way to estimate the full impact of poverty on children’s behavior problems. A few studies indicate that the differential influence of poverty on violence may be limited to metropolitan samples and/or US samples, so more international studies and more nonmetropolitan samples could be used to elucidate this question. Finally, it is likely that poverty exerts its effect under certain circumstances and not others so the testing of interactions is important. For example, biological effects may be enhanced under conditions of poverty, the effect of poverty may be attenuated when parenting is warm and sensitive, and those who have low family income but live in communities where overall income is not low may be less likely to commit violent offenses. These questions have not been adequately answered.
Important control variables for individual-level studies of the association between poverty and violence probably include parent education, parental bonds, and biological factors. Important control variables at aggregate levels probably include community factors. It is also important to test indirect effects because it could be the case that poverty affects parenting which in turn affects the child’s outcomes; simply controlling for the parenting factor in the model might mask an important impact of poverty. Indirect effects of poverty on violence are likely to be many, and include those through parenting; school factors such as achievement, attachment, and attainment; biological factors including pre- and perinatal factors; and exposure to violent peers.