How a Bush-promoted Christian prison program fakes success by massaging data.
The White House, the Wall Street Journal, and Christian conservatives have been crowing since June over news that President George W. Bush's favorite faith-based initiative is a smashing success.
When he was governor of Texas, Bush invited Charles Colson's Prison Fellowship to start InnerChange Freedom Initiative, a Bible-centered prison-within-a-prison where inmates undergo vigorous evangelizing, prayer sessions, and intensive counseling *. Now comes a study from the University of Pennsylvania's Center for Research on Religion and Urban Civil Society reporting that InnerChange graduates have been rearrested and reimprisoned at dramatically lower rates than a matched control group.
For those who know how hard it is to reduce recidivism, the reported results were impressive. Colson celebrated the report by visiting the White House for a photo op with the president. House Majority Leader Tom DeLay issued a triumphal press release. The Journal smacked critics of faith-based programs for "turning a blind eye to science" by opposing InnerChange. The report heartened officials in the four states that have InnerChange programs and buttressed President Bush's plan to introduce the Christian program in federal prisons.
You don't have to believe in faith-healing to think that an intensive 16-month program, with post-release follow-up, run by deeply caring people might be the occasion for some inmates to turn their lives around. The report seemed to present liberal secularists with an unpleasant choice: Would you rather have people "saved" by Colson, or would you rather have them commit more crimes and go back to prison?
But when you look carefully at the Penn study, it's clear that the program didn't work. The InnerChange participants did somewhat worse than the controls: They were slightly more likely to be rearrested and noticeably more likely (24 percent versus 20 percent) to be reimprisoned. If faith is, as Paul told the Hebrews, the evidence of things not seen, then InnerChange is an opportunity to cultivate faith; we certainly haven't seen any results.
So, how did the Penn study get perverted into evidence that InnerChange worked? Through one of the oldest tricks in the book, one almost guaranteed to make a success of any program: counting the winners and ignoring the losers. The technical term for this in statistics is "selection bias"; program managers know it as "creaming." Harvard public policy professor Anne Piehl, who reviewed the study before it was published, calls this instance of it "cooking the books."
Here's how the study got adulterated.
InnerChange started with 177 volunteer prisoners but only 75 of them "graduated." Graduation involved sticking with the program, not only in prison but after release. No one counted as a graduate, for example, unless he got a job. Naturally, the graduates did better than the control group. Anything that selects out from a group of ex-inmates those who hold jobs is going to look like a miracle cure, because getting a job is among the very best predictors of staying out of trouble. And inmates who stick with a demanding program of self-improvement through 16 months probably have more inner resources, and a stronger determination to turn their lives around, than the average inmate.
The InnerChange cheerleaders simply ignored the other 102 participants who dropped out, were kicked out, or got early parole and didn't finish. Naturally, the non-graduates did worse than the control group. If you select out the winners, you leave mostly losers.
Overall, the 177 entrants did a little bit worse than the controls. That result ought to discourage InnerChange's advocates, but it doesn't because they have just ignored the failure of the failures and focused on the success of the successes.
The Penn study doesn't conceal the actual poor outcome: All the facts reported above come straight from that report. But the study goes out of its way to put a happy face on the sad results, leading with the graduates-only figures before getting to the grim facts. Apparently, the Prison Fellowship press office simply wrote a press release off the spin, and the White House worked off the press release. Probably no one was actually lying; they were just believing, and repeating as fact, what they wanted to believe. It's hard to know for sure what those involved were thinking: Study author Byron Johnson canceled a scheduled interview at the last moment. The White House didn't respond to requests for comment.
InnerChange program manager Jerry Wilger says he doesn't know much about research, but he doesn't think it's fair to count the performance of the people who dropped out of his program against him, a fair-sounding objection that misses the point entirely. If InnerChange's 177 entrants were truly matched to the control group but ended up having more recidivism, then either the apparent success with the graduates was due to "creaming" or the program somehow managed to make its dropouts worse than they were to start with. If the program genuinely helped its graduates and didn't harm its dropouts, and if the whole group of entrants was truly matched to the controls, then the group of 177 should have done better than the controls. And they didn't.
So, the feel-good winners-only analysis simply isn't worth the paper it's printed on. Only the full-group analysis (known technically as "intent-to-treat," a holdover term from its origins in medical research) has any real value. And on that analysis, the program has a net effect of zero or a little worse than zero. That makes it a loser.
John DiIulio, an intellectually serious advocate of faith-based programs who was the first director of the Bush administration's faith-based initiatives and the founder of the Penn research center, acknowledges frankly the results weren't what a supporter of such programs would have hoped for. But he points out that a single study almost never provides a convincing yes or no answer on a program concept. "The orthodox believers point to a single positive result and say it proves faith-based programs always work. The orthodox secularists point to a single negative result and say it proves faith-based programs never work. They're both wrong."
The poor result of InnerChange doesn't mean that no faith-based prison program could work, but it does mean that this one hasn't, at least not yet. It joins a long line of what seemed like good ideas for reducing recidivism that didn't pan out when subjected to a rigorous evaluation. Maybe my own pet, literacy training, wouldn't do any better in a real random-assignment trial. But that's why you do evaluations; they tell you things you didn't want to hear. If you're honest, you listen to them.
And if you're smart, you don't listen to the political advocates of "faith-based" this and that when they say they're only asking us to support programs that have been "proven" to work.
Correction, Aug. 6, 2003: The article originally and incorrectly described InnerChange Freedom Initiative as "a fundamentalist prison-within-a-prison." The Prison Fellowship regards itself as being part of the evangelical tradition rather than the fundamentalist tradition. (Return to the corrected sentence.)
Mark A.R. Kleiman, an expert on crime control, teaches public policy at UCLA. His Weblog is markarkleiman.blogspot.com.
Illustration by Robert Neubecker.