Marriage Reduce Crime? A Counterfactual

¶ … Marriage Reduce Crime? A counterfactual Approach to Within-Individual Causal Effects" by Sampson, Laub and Wimer (2006), the authors explore the possibility that men who enter marriage are less likely to continue with criminal activities after the fact, or indeed are less likely to commit a crime than unmarried men. To investigate this, 500 high-risk boys were followed from their youth to the age of 32, with 500 additional candidates being followed to age 70. The finding was that marriage resulted in an approximately 35% reduction of crime than the non-married state.

The basis of the paper is the bias that may result from marriage selection. Although outcomes have suggested that marriage has a reducing effect on criminal activity, the interpretation might also be that men who are not involved in crime in the first place are more likely to be chosen as marriage partners than those who are. Hence the results are difficult to pinpoint. The authors were concerned with addressing this potential pitfall.

To do this, 500 young men were investigated as they entered adulthood with a high risk of continued criminal activity, with a follow-up study for these men to the age of 70. By choosing men with a history of criminal activity, the bias related to marriage partner choice is eliminated. This enabled the investigators to determine the variants within individuals before and after marriage, rather than simply comparing married to non-married individuals.

Assuming the positive effect of marriage on the tendency to desist from crime, the authors identify four factors that creates this association for marriage: 1) The social bonds that form as a result of marriage; 2) significant changes in everyday routines and patterns; 3) direct social control by the female spouse, where she insists on the husband spending a certain amount of time within the family, being a good breadwinner, and so on; and 4) the change in sense of self that often comes with marriage, i.e., the young man now feels "grown up" and ready to take the responsibility of marriage and providing for his family. This can only be effectively done by means of a steady and legal employment, while desisting from crime. The authors hold that these are major factors playing a role in married individuals desisting from crime to a greater degree than their non-married counterparts.

Although it is uncertain whether cohabitation is less likely to inhibit criminal activity that marriage itself. Nevertheless, it is found that both cohabitation and marriage tend to reduce the use of marijuana and binge drinking. While the authors then found that marriage did have a positive effect upon crime reduction, the question to be addressed by the rest of the paper was whether this effect was causal.

Before moving to the body of their research, the authors make the point that the focus of this particular research is men and their likelihood to benefit from marriage. The authors follow this by questioning the equality of benefit for both partners. Even while criminally-minded men benefit, their female spouses may suffer ill effects that would not have been the case had they remained unmarried. This is a good basis for future study of another dimension of the effect of marriage on criminal activity.

Do determine the causality of marriage as a deterrent for crime, the authors use counterfactual methods, or causal inference. This means that observational data are used to determine the causal relationship between marriage and crime reduction.

The inverse probability of treatment weighting (IPTW) is a statistical method that helped the authors handle the consideration of bias. The population for study was then chosen accordign to factors such as measured itnelligence, competence, self-control, and temperament as relevant to marriage selection. Other, more traditional factors were also included as potential predictors for marriage, such as employment and educational attainment, although most of the men investigated were high school dropouts. The authors go on to specifically discuss the measures that are relevant to marriage selection, and how individual differences played a role. The outcomes showed that family-background measures were less important during ages 17 to 32, although this factor grew in importance towards later life. Indeed, all 25 possible predictions appeared significant, including age, marriage history, criminal-deviance, etc.

The authors added three additional factors to the IPTW model in order to merge it with a generalized hierarchical model for longitudinal data. These include the conception of crime as a rare event, unexplained variation between individuals in the propensity to offence, and the variation across time and individuals in incarceration.

Another interesting question the authors pose concerns the potential effects of the quality of marriage upon criminal behavior. It was found that deviant spouses had a negative effect upon the behavior of their criminally-minded men. However, even with the proportion of deviant spouses in the sample, the findings indicated a significant reduction in the criminal activity of men who are married by age 25, as opposed to those who are not. Specifically, the number comes to about a 35% reduction.