This article has some good insights into why that is and why previous studies have been biased.

https://www.theatlantic.com/magazine/archive/2015/04/the-irr...

tldr: Most people who respond to surveys asking about the efficacy of the program are sober and skew the results, the anonymity of people in programs makes it difficult to identify who are potential candidates for study/surveys for external researchers, and no central authority monitoring programs so little standardisation between them meaning some groups may be more effective than others (this is somewhat reflected in a comment above about trying other groups which may mean that the real effectiveness comes from the community it brings, not the program itself, but that's pure speculation on my part)

Here's another

https://www.salon.com/2014/03/23/the_pseudo_science_of_alcoh...

Don't they give out tokens for days sober? That volume could be informative.

Well, yeah, but you don't have to collect them. I haven't gotten a token since I passed the one year mark, and it's been six.