To go back to the intermediate facility idea, for instance, the correctional services have many people at the moment who go from RTCs back to the general population. Take a group of 300 people who are going to that population, divert 150 of them randomly to the intermediate facility, and find out which one works better for those two groups. That would be a research-type way to do it.
On December 10th, 2009. See this statement in context.