This is the final post in a series on the effectiveness of faith-based prison programs, based on my recent Alabama Law Review article, Do Faith-Based Prisons Work? (Short answer: no.) Monday’s post introduced the issue, Tuesday’s post surveyed some of the least valid studies, Wednesday’s post critiqued the studies that used propensity score matching and discussed other possible empirical strategies, and Thursday’s post talked about the most valid studies–those that used rejected volunteers as a control group. Throughout, I’ve been putting the faith-based prison research side-by-side with the private schools research, because evaluations of each raise similar methodological problems. The fact that both are voluntary means that they can attract fundamentally different sorts of people, so their good results might be attributable to the higher-quality participants they attract. Today’s post ties the ends together and asks whether there’s any way forward for faith-based prisons.

*     *     *

So, after discarding the faith-based prison studies tainted by self-selection bias, we’re left with two studies that find no effect of faith-based programs, one study that’s too small to be meaningful, and three studies that find some effect, even if the effect that a few of these find is quite weak. And of those three, two aren’t about prisoners at all, but about after-care of released prisoners, and the remaining one shows no significant effect once the prisoners have been released. So we have no study that actually finds a significant effect of an in-prison faith-based program on recidivism.

The picture looks fairly bleak for faith-based prisons. (Of course, this analysis has no bearing on arguments in favor of religious prisons that don’t hinge on their value in reducing recidivism.)

Several literature reviews agree. One says the faith-based prison research

has shown only moderate effects of faith-based programs on outcome measures such as institutional adjustment, incidence of prison infractions, and likelihood of postrelease arrest. This literature also suffers, in some cases, from important limitations such as: small and nonrepresentative samples, a reliance on anecdotal evidence, an absence of theoretical context, limited statistical analysis, and self-evaluation by prison ministry providers.

Another review says that “research on intentional religion is remarkably underdeveloped” and that “[c]ase studies and descriptive studies,” dominated by “subjective interpretations,” are “clearly over-represented,” and calls for the use of “more rigorous methodologies.”

One major contributor to the faith-based prison literature is even more pessimistic about the “more in-depth faith-based programs,” calling “the current trend . . . to put large sums of money and a great deal of public, political, and criminological emphasis” on such programs “mistaken” and arguing that there is no evidence that such programs are better than other correctional treatment programs.

I agree, though I emphasize the self-selection issue more than the other problems. (Indeed, I haven’t even included any anecdotal evidence here because I find it to be of extremely limited [i.e., no] value.)

Is there any hope for the future? I think there may be. Not all faith-based prison programs are the same. Tom O’Connor and Jeff Duncan write that various programs that showed no effect “probably did a good job of faith development but failed to reduce recidivism because they did not follow what are known as the principles of effective correctional treatment, such as criminogenic risk/need, responsivity, family/community context, program integrity, and program delivery type.” On the other hand, they write, the COSA and TOP programs worked because they did follow many of these principles. Similarly, Daniel Mears and his coauthors write that many faith-based programs have failed because they haven’t “articulate[d] a clear statement of program goals and how exactly specific activities will contribute to these goals”; activities have been inconsistently implemented; the different organizations and agencies involved have been uncoordinated; and funding has been insufficient or inconsistent.

If this is the case, the failures of certain faith-based programs may not indicate that the faith-based agenda itself is flawed. Later programs may do better; experimentation may result in the discovery of a more effective model; the programs that have worked well may be replicated in more places. The weak evidence supporting faith-based prisons so far may mean that more experimentation is in order, provided such experimentation can be done consistent with constitutional constraints. “The process of accumulating empirical evidence is rarely sexy in the unfolding, but accumulation is the necessary road along which results become more general.”

Moreover, I’m hopeful that—now that studies are available in each of these categories, and now that a number of critical review articles (including this one) have appeared—faith-based prison researchers will get the hint and pursue valid empirical techniques. They’ll be following in a venerable tradition. According to economists Joshua Angrist and Jörn-Steffen Pischke, empirical economics is going through a “credibility revolution.” “[T]he primary engine driving improvement,” Angrist and Pischke write, “has been a focus on the quality of empirical research designs.” Randomized assignment has been part of the story—we’ve now seen a few faith-based prison studies that used this very approach in the form of using rejected volunteers as a control group. Another part of the story is “natural experiments” or “quasi-experimental” designs where we observe people’s responses to random institutional flukes; this approach hasn’t been used so far for faith-based prisons, but we’ve seen it for education in the “exogenous policy shocks” literature. Even the instrumental-variables literature is better than it was: researchers proposing an instrument now typically take seriously the need to justify why a particular instrument is valid.

*  *  *

Can such experimentation be done consistent with constitutional constraints? As I’ve mentioned above, one faith-based prison program, Prison Fellowship Ministries’ InnerChange Freedom Initiative, was struck down on Establishment Clause grounds in 2006, so at least some faith-based programs are vulnerable.

As I explain elsewhere, a constitutional faith-based prison program will have to comply with the following requirements:

  • Its religious content must be significantly watered down, so that one can’t find “religious indoctrination.”
  • It must be chosen by a process that is neutral as between religious and non-religious programs. Thus, the process that chose it must have been capable of selecting a secular program.
  • There must be at least one, and possibly several, comparable secular programs.
  • The program must not only be formally voluntary but also not offer significantly greater benefits—for instance, a greater possibility of parole or a safer environment—than secular alternatives.
  • Program officials must, at a minimum, not play any role in maintaining order or meting out discipline. But even divesting oneself of these governmental roles may not be good enough to avoid unconstitutionality.

The more ecumenical programs—like the Federal Bureau of Prisons’ Life Connections program, which looks much more like an outgrowth of traditional chaplaincy programs together with visits by “spiritual guides of different faiths”—may yet turn out to be constitutional, but it’s easy to imagine that some of the more “hard-core” advocates of faith-based prisons will be disappointed at this prospect. To many, salvation (and, in this world, rehabilitation) comes through Christ (for instance) alone, so that a rehabilitative program that isn’t allowed to use specifically Christian material may not even be worthwhile. Moreover, program administrators may want to hold on to the ability to expel inmates who aren’t engaging constructively with the material, and the ability to tell an inmate where he can live, or to withdraw some privileges from an inmate, may be “governmental” enough to be unconstitutional when put into the hands of a religious organization, unless the religious organization was chosen by a strictly neutral process that could also choose secular organizations.

In other work, I explain how a system of prison vouchers would solve these constitutional problems. Instead of being assigned to prisons, convicted criminal defendants would receive a voucher, good for one imprisonment, that they would be required to redeem at a participating prison. Any organization, religious or secular, would be able to run a prison, provided it complied with certain minimal security-based requirements (related to, for instance, the number of guards, the medical facilities, the food, or physical security). This system, combining inmate choice with strict government neutrality as between secular and religious providers, is consistent with the Establishment Clause for the same reasons that school choice plans are constitutional, as explained in Zelman v. Simmons-Harris (2002).

Of course, a full policy analysis of prison voucher programs is complicated, and I don’t attempt it here. The full cost-benefit analysis goes far beyond religious matters and the question of faith-based prisons. On the plus side, prison vouchers would finally give prisons an incentive to compete on dimensions that are valuable to prisoners and prison reformers alike, such as decent medical care or low rates of assault or rape. On the negative side, it would also give prisons an incentive to compete on socially negative dimensions, like looking the other way as prisoners smuggled in contraband, or, more generally, to the extent that prisons become more pleasant places from inmates’ points of view, the deterrent value of prison would be lessened.

The positives and negatives are both substantial, and I remain agnostic about whether prison vouchers are a sensible reform proposal. I only mention it here, by way of conclusion, as a possible way to allow experimentation with faith-based prisons to proceed without the specter of unconstitutionality.

*  *  *

There remains another problem. Generally, all we know about the comparison groups in the studies described here is that they’re composed of people who didn’t participate in the program. We don’t know what treatment, if any, they received. Suppose, for instance, they received no treatment at all. The study may thus be measuring the effect of treatment resources, not the effect of the program’s religious content. It may be that a faith-based program is better than nothing—this is important because “the lack of reentry programming constitutes a common criticism of reentry practices to date.” But, at the same time, the program may be no better than a comparably funded secular program—which is problematic for policymakers deciding which of several programs to fund. To answer this question, we would need a comparison group of volunteers who were rejected from the religious program and instead assigned to a comparable secular program. To my knowledge, such a study hasn’t been done.

The reality, of course, is that most prisoners don’t get comparable programs. Nor do they all get nothing. Many prisoners “in fact participate in one or more community-based services, even if the intensity of these services may be nominal.” So perhaps the best comparison is to this “business as usual” approach. But that, too, is difficult since the “business as usual” baseline differs from place to place and over time.

This problem, while real, is less serious than the self-selection problem. The trouble with self-selection is that a program that seems to work may in fact be worthless. In fact, a program may appear to work even when it’s positively harmful, as long as the self-selection effect in the other direction is strong enough.

On the other hand, the resources problem tells us that a faith-based program may be no better, and possibly worse, than some secular program. But where will this program come from? The reality for many prison administrators is that the alternative to a religious program sometimes is nothing at all. Religious providers may just be more available than secular providers, and may also cost less to the prison system to the extent that they’re more likely to be subsidized by donations from the outside.

Moreover, if one did compare a religious program against a specific secular program, it would thus answer a specific question of comparative effectiveness, and would be useful to people considering the hypothetical question of where social resources could, in principle, be best spent. But this result would be hard to generalize to comparisons with other secular programs, and moreover, it would be false to the actual choices faced by prison administrators today.

*  *  *

Let’s take the broad view and come back to the education studies that I’ve been using as a point of comparison throughout this Article.

Finally, after decades of research, we have some credible studies estimating the effect of private schools. The best evidence, taken from studies comparing accepted and rejected applicants, indicates that private schools do have a positive effect on the students who attend them, at least for black students and at least for math scores.

On the one hand, one can observe that, next to these results (modest as they are), it’s all the more disappointing that faith-based prisons haven’t shown much in the way of significant positive effects.

But on the other hand, it took decades of research and debate by different groups, each using a slightly different empirical approach—and many finding little to no effect—before we got even the mild results we have on private education. This suggests that we should encourage more research on the matter, in different contexts, using a variety of different empirical techniques.

The result is that, if there’s no strong reason to believe that faith-based prisons work at all, and even less reason to believe that they work better than comparably funded secular programs, there’s also little reason to believe that they don’t work, and in many cases they may be the only available alternative. It’s probably sensible to allow such programs to operate and to allow the process of experimentation to work its course, provided that all this can be done constitutionally.