Less imprisonment is no doubt a good thing: more policing is not

If policy makers were to take account of evidence, as Steven N. Durlauf and Daniel S. Nagin (2011, this issue) argue, then American crime-control policies would look very different. Here is how and why. Prison sentences have few general deterrent effects, if any. The effects of imprisonment on individual deterrence are most likely perverse; people sent to prison tend to come out worse and more likely to reoffend than if they had received a lesser punishment. Some things police do, however, can reduce crime rates. American jurisdictions should reduce their use of imprisonment and increase their investment in policing, thereby lowering crime, victimization, and imprisonment rates as well as saving money simultaneously. Everybody—offenders, prospective victims, and taxpayers—wins. Yes, probably, concerning imprisonment and possibly concerning policing, but serious problems of unwanted unintended consequences would need to be overcome. The conclusions and proposals are based on in part on Nagin’s recently completed comprehensive reviews of the literatures on general deterrence (Apel and Nagin, 2011) and on the effects of imprisonment on released inmates’ subsequent offending (Nagin, Cullen, and Jonson, 2009). Durlauf and Nagin (2010) again surveyed the deterrence literature as well as the literatures on the effects of variations in police numbers, per capita spending, and crackdowns on crime. Durlauf and Nagin (2011) pull all that work together and add to it. The deterrence surveys conclude that credible evidence indicates that changes in sentencing laws sometimes produce deterrent effects but that the effects are highly contingent, depending on threat communication and patterns of implementation; are not easily replicable; and provide an insufficient basis on which to build sentencing policies. The prison effects survey concludes that no credible evidence suggests that imprisonment reduces reoffending by released offenders. To the contrary, tentative but not yet conclusive