Measuring Crime and Criminality by John MacDonald

Author:John MacDonald [MacDonald, John]
Language: eng
Format: epub
Tags: Social Science, Criminology
ISBN: 9781351506403
Google: Pq00DwAAQBAJ
Barnesnoble:
Goodreads: 14316860
Publisher: Routledge
Published: 2011-07-15T00:00:00+00:00

Conclusion

This chapter has sought to demonstrate the usefulness of a research approach that differs from the conventional practice of the social sciences. It is conventional to make assumptions strong enough to yield point identification and then argue the merits of those assumptions relative to alternatives. We seek instead to bound disagreements among researchers by examining what can be learned about treatment effects under alternative models of treatment selection and by layering progressively stronger assumptions about which there is likely to be less consensus.

In this spirit, what conclusions can be drawn from our empirical findings? The answer plainly depends on the assumptions one is willing to make. If no assumptions are made about the treatment selection process, the results mostly provide guidance in the negative. The bounds on the treatment effects do not identify their signs and place relatively weak restrictions on their magnitudes. However, for those who have a favored model of treatment selection, the results are far more informative. If one’s favored model of judicial sentencing behavior is outcome optimization, the results suggest that residential treatment exacerbates recidivism. If one favors the skimming model, the results suggest that residential treatment has an ameliorative effect on recidivism. Imposition of the district exclusion restriction strengthens each of these opposing conclusions. Thus our empirical research implies fairly strong conclusions provided that there is a consensus favoring either the outcome optimization model or the skimming model.

That said, we do not see such a consensus among criminologists. In the absence of consensus, our empirical analysis remains instructive. The opposing conclusions arising from the two models of treatment selection demonstrate the value for informed scientific discourse of identifying zones of agreement and disagreement among researchers, as well as the reasons for such agreements and disagreements. In this application, unless researchers can agree on the treatment selection process, analysis of observed sentences and recidivism can do little to settle the question of the sign of treatment effects.

Our work clearly points to the need for further research analyzing how judges actually make sentencing decisions. Although the outcome optimization model and the skimming model differ from one another, they share critical simplifying assumptions—namely, that judges are concerned only with recidivism and that judges have rational expectations about recidivism. It may be that these shared assumptions are unrealistic. To make further progress in the analysis of recidivism using observational data, we need to better understand how sentencing decisions actually are made.

The results also illustrate the value of layering assumptions based on consensus about their validity. Suppose, for example, that one believes the skimming model to be a realistic depiction of the actual treatment selection process. This model alone does not provide a firm basis for signing treatment effects. Only with the additional imposition of the district exclusion restriction were we able to sign treatment effects with full confidence. It should, moreover, be kept in mind that all of the conclusions drawn in this chapter rest on the maintained assumptions that treatment is individualistic and that outcomes and treatments are correctly measured.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.