A key goal of community corrections (probation, parole or supervised release, juvenile supervision) is to encourage desistance from crime. It tries to do so using a mix of supervision and services. But success rates are mediocre, especially for those leaving prison, who are more likely than not to return within three years as a result of either violations of release conditions or new crimes.
HOPE-style probation - close monitoring with swift, certain, and minimal sanctions - demonstrably outperforms routine supervision-and-services. The question is why. And that leads me to step away from my usual policy-analyst role and consider doing actual research. A sketch of the idea is at the jump; comments and references would be most welcome.
HOPE can be thought of in terms of more effective deterrence: substituting swiftness and certainty for severity. But it may also influence client behavior through a different mechanism: by improving their capacity to shape their own behavior in light of their long-term goals. Many studies demonstrate that measures of self-command strongly predict social, educational, health, and criminal-justice outcomes, even controlling for SES and measured IQ. Four-year-olds who succeed in waiting fifteen minutes before eating a marshmallow when given the inducement that those who wait get a second marshmallow as well, are more likely to finish high school and less likely to go to prison than otherwise similar four-year-olds who can’t wait that long.
A variety of concepts from behavioral economics, personality psychology, social psychology, and criminology relate to self-command: locus of control, self-efficacy, impulsivity, hyperbolic discounting, delay of gratification, and Conscientiousness (from the Big Five) and some of its subscales. Some appear to be relatively stable personal traits; others seem to be more state-like and domain-specific. There’s evidence that some programs (such as the Nurse-Family Partnership) can influence self-command when applied to young children; much less is known about the malleability of that characteristic in later childhood, adolescence, or adulthood.
A striking finding from the Hawaii probationer interviews was the clients’ high level of satisfaction with the program – surprising in what was, after all, basically a tightening of control – and their use of language suggesting that they were taking responsibility for their own outcomes. It’s possible, but by no means demonstrated, that HOPE outperforms routine probation partly because it creates an island of predictability in what are otherwise highly chaotic personal environments. The random severity of punishments for probation violations reproduces many of the characteristics of bad parenting; we don’t know whether HOPE, which replaces randomness with predictability and thus moves the actual locus of control inward, might affect the psychological locus of control.
That suggests a fairly straightforward study:
1. Find a jurisdiction with an established HOPE or Sobriety 24/7 program running successfully and ready to expand.
2. Select eligibles based on official data (without talking to the subjects). Look for substance abuse problems and serious criminal history; perhaps also non-compliance on probation.
3. Ask for consent to psych battery plus two or three one-hour interviews over two years plus maybe hair at the end.
4. Non-consenters still get randomized into treatment (HOPE or 24/7) or control (probation as usual).
5. Consenters get as much assessment as we can get them to submit to: LSIR or whatever the agency uses for risk/needs assessment; some measure of substance abuse; baseline measures of personal and social functioning (physical/mental health, residence, employment, family status, individual social capital if we can measure it); locus-of-control measure, impulsivity measure; measure of ability to defer gratification (marshmallow-test equivalent for adults); perhaps measure of “criminal thinking” from cognitive-behavioral therapy.
6. Consenters then get stratified and randomized.
7. Outcome measures for everyone:
- Arrests
- Time-behind-bars
- Six-month and one-year compliance snapshots (on the street, not absconded, no positive tests in previous two months).
8. Outcome measures for consenters:
- Repeat personal and social inventory.
- Repeat risk and needs assessment
- Repeat substance abuse measure
- Repeat locus-of-control
- Repeat impulsivity
- Repeat marshmallow test
- Repeat “criminal thinking”
- Accuracy of perception of conditions of probation
- Perceived fairness and predictability of probation
Research questions:
1. What are risk and protective factors for success on HOPE?
2. Is there an identifiable group that would have done better with a quick treatment mandate?
3. Are there systematic changes in self-command measures in the experimental group compared to the control group? If so, do they exceed those associated with desistance from drug abuse? Do they correlate with perceptions of the probation process?
4. Do those changes correlate with other outcomes?
Double the number of possible participants;t explain to them that there’s a 50-50 chance that they get into the program (a la Empey & Erickson), but that they will be first in line if the program works.
I wonder if some of the 4-year-olds don’t believe that there will ever be a second marshmallow. The question of trust would seem to underlie the marshmallow test, in addition to self-control. And it might not be different for adults either.
I also wonder if somehow the HOPE program is taken as an indication of someone giving a dang about people who aren’t used to the experience? Which makes me worry about how they will do once the supervision ends, because out in the world, very few people do give a darn, and I’m sure it’s even less if you’ve got a record.
Survey design was never my forte, but might there be a problem with 3 and 4? Would there be a difference of some importance between people who opt in v. those who don’t? I’m just asking. I have no idea! But I wonder if trust would be an issue there too.
And to piggyback on what MikeM said, put a chance for a decent job at the end of this and you’d get buried with applicants. ‘Course, no one has any decent jobs these days…
Also, I was going to say that in a future test, you might even want to try the use of actual positive reinforcement of some sort too. But then you will get in trouble with conservatives since, as we all know, trying to help other people is now a sin.
As one toiling in the amazingly unproductive trenches of substance abuse treatment….This is exactly what I recently realized as potentially an amazingly effective treatment rubric: the HOPE program (with reported 80% durable effectiveness) paired with motivational incentives (including - if possible, but not limited to - job offers upon successful program completion). Motivational incentives are reported to have good results (never had the resources to use ’em effectively); with the logical consequences of HOPE to extinguish negative behavior, and tangible incentives to reinforce healthy, sustainable behavior, we have what I think would be the best possible approach for real, comprehensive, lasting, and possibly remarkable change from drug abuse - which is, after all, in essence heavily reinforced behavior. Actually, that’s what it is - heavily reinforced behavior, and until this country pulls its moralizing coconut out of its pimply keister and realizes this we will never - never - get a handle on the huge, exploding problem of substance misuse.
Point 4 is how it ought to happen, but getting jurists to agree to randomization is difficult. For some reason, judges are almost as likely to suffer from delusions of omniscience as physicians.
It is also possible that your IRB will take issue with point 4. It’s stupid, and they shouldn’t protest — the non-volunteers are going to go on some monitoring system, after all. It doesn’t matter from any perspective that I can understand that it matters how they are placed in the system they are placed in.
It would matter if there were reason to believe that one system would be substantially superior for them. In that case, you are obligated to put them in the superior system. And I suppose that is arguably the case: HOPE outperforms conventional systems. But it does so on a population basis, and we don’t know what works best for individuals. Improving the efficiency of the system requires useful classification models. Those models require research.
OK, an ignorant question: in point 3, what’s with the hair?