Selection effect

Please tell me that Denver's corrections officials aren't using these sorts of stats as evidence of the effectiveness of the Making Choices program in reducing the recidivism rate:

Over the past 10 years, the program has proven highly effective for the 450 women who have taken the training.

Prison officials say the recidivism rate is 54 percent for Colorado's prison population in general, but for women who graduated from Making Choices, it is just 12 percent. It drops to 8 percent for those who take a follow-up booster program, which 122 women have done.

"When I heard that, it blew me away," said warden Travis Trani, who arrived here months ago from a job as warden of Limon Correctional Facility.

One important, but unstated, piece of data would be the recidivism rate among female inmates as a whole.  (This study suggests that recidivism rates among female inmates are much lower than those of male inmates.)  But beyond that, how are inmates selected into the Making Choices program?  It may be that those who choose to enroll are already much more motivated than the average inmate to improve their behavior.  Unless they're being selected at random and forced to attend, we really have no idea whether it's an effective program from the evidence presented in this article.