papers.ssrn.com/sol3/papers.cfm?abstract_id=2981749
Three questions arise concerning deterrent effects related to severity. The first is whether increased severity yields measurable reductions in crime. The second is whether a credible case can be made for deterrent effects of any specific level of severity. The third is whether punishments, especially prison sentences, have specific deterrent effects for the people who receive them. The answers are “possibly, at best, a little,” “no,” and “no.”
***
Weisburd, Einat, and Kowalski (2008) conducted a randomized field experiment on strategies for inducing payment of court-ordered fines. The key finding was that a believable threat of immediate incarceration produced higher rates of payment, even when the incarceration was for a short period. Hawken and Kleiman (2009) conducted a randomized experiment on a Hawaiian program in which probationers who failed drug tests or committed other violations of conditions received certain but short periods in jail as a consequence. The probationers in the experiment had fewer positive drug tests, missed appointments, arrests, and imprisonments than did those in a control group. There are reasons to be skeptical about the findings, but for purposes of analysis I accept them here*.
* The National Institute of Justice funded a series of replications that were evaluated using randomized assignments of eligible offenders to treatment and control groups. The new evaluations did not confirm the conclusions of the original project and concluded that the programs were ineffective (Lattimore et al. 2016; O’Connell, Brent, and Visher 2016).