Behavioral Science Critique of HOPE

• Published date: 01 November 2016
• DOI: http://doi.org/10.1111/1745-9133.12256
• Date: 01 November 2016

COMMENTARY

HOPE COLLECTION

Behavioral Science Critique of HOPE

Philip J. Cook

Duke University

The evidence presented in this symposium from five randomized controlled

trials (RCTs) forces consideration of the possibility that Hawaii’s Opportunity

Probation and Enforcement (HOPE)–style probation regime does not perform

better on average than does the traditional regime. I say that with regret because I like

many others were sold both by the RCT evidence from HOPE (Hawken and Kleiman,

2009) and more generally by the idea that deterrence theory could be usefully harnessed

to the project of reducing recidivism. In this policy essay, I consider whether the negative

evidence should in fact be considered a challenge to the theory.

The Evidence

The results of the RCTin Delaware known as “Decide Your Time” (DYT;O’Connell, Brent,

and Visher, 2016) and of the four RCTs funded by the U.S. Bureau of Justice Assistance

known as the HOPE Demonstration FieldExperiment (Lattimore, MacKenzie, Dawes, and

Tueller, 2016) are well documented in this issue. For DYT (an experiment performed with

400 high-risk probationers), there is little difference between the treatment and the control

group with respect to a variety of outcome measures, including the likelihood of at least

one arrest for a new crime over the course of 6 months, 12 months, or 18 months, as well

as the likelihood of incarceration (O’Connell et al., 2016). For the HOPE Demonstration

Field Experiment, the results are similarly discouraging: When the samples from the four

sites are merged for a total sample of more than 1,500, there are no significant differences

between treatment and control on a variety of outcomes (recidivism arrests,new convictions,

revocations, or arrests) after an average of 650 days. A survival analysis found no significant

difference between treatment and control in any of the four sites for the outcome measure

“revocationor arrest,” whereas for revocation alone, there are two sites in which the treatment

accelerated revocation and one in which it delayed revocation (Lattimore et al., 2016).

Direct correspondence to Philip J. Cook, Duke University, P.O. Box 90245, Durham, NC 27708 (e-mail:

pcook@duke.edu).

DOI:10.1111/1745-9133.12256 C2016 American Society of Criminology 1155

Criminology & Public Policy rVolume 15 rIssue 4