The Effectiveness of Alcoholics Anonymous

by Agent Green.

(Mail for Agent Green published here.)
(Orange's response, and Green's rebuttal, here.)


Everybody is entitled to their own opinions, but not their own facts.
- Senator Patrick Moynahan, quoted at

This page gives referenced information (below) on Alcoholics Anonymous and also provides counter-arguments to a viciously biased anti-AA site known as the Orange Papers. References are at the bottom of the page. Enjoy!


Alcoholism is a contentious topic. Some people swear that it's a disease and that one drink could eventually kill an alcoholic. Others vow that all problem drinkers need is a little determination to control their drinking. The fact is, both these views are true depending on severity.

Some problem drinkers just need their mothers or a couple of hangovers to persuade them to cut back. Others can use willpower and support from friends to kick the habit, or at least limit their intake. Some ruin their families, their jobs, their finances, and their lives, and end up drinking themselves to death. Why won't they stop? Is it pure stubbornness and moral weakness, or are they in the grip of some strange disease or disorder?

Research has shown that there is no sharp distinction between alcoholics and non-alcoholics. Instead, alcoholism can be rated from zero to extremely severe according to the number of symptoms. People are spread out all along the scale, just like with blood pressure or IQ. We're in the 21st century, but the sad fact is, there is no medical cure for severe alcoholism. All medicine can do is provide detox services and crisis management and help people to heal themselves.

Discussing Alcoholics Anonymous

Research on AA is always a challenge. Does AA have a high success rate, or do people just show up when they are ready to stop drinking? Does AA have a low success rate, or is that only because the more severe cases are attracted to AA? It is very hard to compare AA to non-AA people because AA is so accessible that most problem drinkers are likely to have attended at least one AA meeting. For these reasons, most responsible researchers will clearly state that research on AA is based on indirect evidence. So when Agent Orange quotes someone that "...there is a paucity of scientific studies supporting the superior effectiveness of AA," what he forgets to mention is that there is also not much research showing AA does not work. AA is simply a tough topic to research.

AA is a loose organization of many small, independent groups. The AA main office provides a suggested program of recovery, but there is no central authority. AA does not keep membership lists or collect dues beyond "passing the basket" at meetings (and no one is turned away for not contributing). AA is based on alcoholics helping each other: according to the program, in order to stay sober, experienced members need to help other alcoholics.

Me and Agent Orange

AA often provokes extreme reactions, perhaps because by the time people get to AA they are in desperate need of help. If AA helps them, they swear by AA for the rest of their lives. If AA doesn't help them, they sometimes become angry critics. It's hard to find objective opinions, even in the research world. And this brings us to Agent Orange's criticisms. I will focus on Orange's page entitled "The Effectiveness of the Twelve-Step Treatment" ( (I will not, however, attempt to deal with all 100+ pages of it.)

What I respect about Orange is:

What I do not respect about Orange is:

Based on research that I will discuss below, my beliefs are:


For the record, here is my own "Q&A on AA":

Q: Are all AA members abusive cult extremists, like Agent Orange says?
A: No.

Q: Are some AA members abusive cult extremists?
A: Yes.

Q: Are there emotionally mature AA members who can offer guidance and share their experience, strength, and hope?
A: Yes.

Q: So what's an alcoholic to do?
A: If you are having trouble stopping drinking, and you are considering AA, maybe try going to, say, six AA meetings, talk about your experience, and listen to others. If you think that AA may be able to help you, keep going back and get a sponsor you can trust. If not, there are other options out there.

My personal self-statement:

AA - Only a 5% success rate??

You will notice that Orange makes a big deal of AA's 5% success rate, which he claims is universally acknowledged. I would like to show that this figure is actually only in Mr. Orange's head.

Research by Vaillant

I took the trouble to scan some of the pages from Vaillant's book so that you can follow the discussion yourself. PDF is here, 0.7 MB. I also added some non-biased notes as a guide to the dense technical discussion.

Orange writes a lot about psychiatrist George Vaillant, author of The Natural History of Alcoholism Revisited
, which he wrote at Harvard. One of Vaillant's studies was of 100 severe alcoholics who sought help at a clinic. After discharge from hospital detox, these alcoholics were followed for 8 years, during which time they had unlimited access to a local network of halfway houses, drop-in centers, detox units, and integrated mental health facilities. All patients were encouraged to attend twice-weekly outpatient meetings, which in turn encouraged AA attendance. The final outcomes (p 191) after 8 years or death were:

Fully 95% of the subjects had one or more relapses during the 8-year study, and twenty-nine had died - roughly three times the death rate expected for non-alcoholics of the same age. Seeing these results, it is hard to deny that severe alcoholism is difficult to cure. But contrary to Orange's summary, AA actually helped reduce the death rate. Here is where I see Orange falsifying the results:

  1. Where the results showed that access to the health facilities didn't work, Orange claims that AA didn't work.
  2. Orange claims that every relapse counts as a failure for AA, even with patients who recovered at the end, and even though not all patients attended AA.
  3. Orange ignores results that showed positive effects of AA.

This is plain dishonesty. Orange, you have a banner at the top of your page that says "Everybody is entitled to their own opinions, but not their own facts." So what's your excuse? Let's deal with the above points one by one.

(1) Was the study really about AA?

This was a study of patients who came to a clinic in distress and then had a health network made available to them. AA attendance was encouraged but was optional and was followed to varying degrees by the study subjects. At the end of 8 years, only 32 patients had attended AA meetings 100 or more times, for a mean of 600 visits. (Note that over 8 years, 100 meetings averages out to once a month, and 600 meetings, to every 5 days.) Those who did attend AA had higher recovery rates.

(2) Did AA really fail 95% of the time?

95% of patients had relapsed at some time during the study, even though many of these eventually attained sobriety. Orange counted this as a 95% failure rate for AA. His argument was that since AA recommends complete abstinence, even one relapse must be considered a total failure. Sorry, Orange, I don't buy that argument, and I hope others don't either. It's well known that most severe alcoholics only get sober after many relapses, to the extent that relapses can be considered part of the recovery process. So don't pretend that's a failure. When an alcoholic gets sober, personally I see that as a success, no matter how long and difficult the road. The goal is "Progress not perfection." And anyway, this was a study of a health network, not AA.

(3) Were the results negative for AA?

Results that Orange must have forgotten to mention (Vaillant 1995, p 187 - 197) were that increased AA attendance was associated with a higher rate of sobriety, and AA helped many alcoholics who would otherwise been predicted not to stop drinking. For example, half of the stable remissions, but only two of the chronic alcoholics, had made 300 or more visits to AA (Vaillant 1995, p 196). So contrary to Orange's statements, AA attendance was associated with positive results.

(4) Summary of Vaillant, and Orange's conclusions

  1. 100 severe alcoholics were studied for 8 years after an initial detox. They had free access to halfway houses, drop-in centers, detox units, and integrated mental health facilities.
  2. AA attendance was encouraged, though not all patients went to AA.
  3. At the end of 8 years, Vaillant wrote that recovery rates were dismal.
    - 95% of patients relapsed at some time during the eight years,
    - only 29% had achieved stable remission by the end.
  4. Conclusion: there is still no effective medical cure for severe alcoholism. However, AA attendance was associated with a higher success rate (p 196 & 358).
  5. Based on the number of relapses in point 3 above, Agent Orange declares a 95% failure rate for AA.
  6. In his discussion of AA's effectiveness, Agent Orange 'forgets' to mention the pages where Vaillant discusses AA's effectiveness (point 4 above).
  7. Vaillant's results show that AA helps. Orange somehow claims that AA does not help.


Does AA's retention rate indicate 95% failure?

Orange quotes an internal AA survey as evidence that 95% of people who begin going to AA meetings will have left after a year. In reality, the graph shows that 74% will leave within their first year not 95% - Orange either doesn't know how to read a frequency distribution graph, or is willfully presenting it dishonestly.

The graph (page 12) is very simple. The researchers went into different AA meetings and asked the people there how long they had been attending, they plotted the results for those within their first year on this graph by monthy averages. So it shows, 19% of people were in their first month, 13% their second, 10% in their third and so on up to 5% in their twelfth month. Orange claims the 5% of people in their twelfth month indicates that 95% had left after a year (oblivious to the fact that the other 95% in the survey was comprised of those sitting in the same room and with less than 11 months time attending meetings). Agent Orange is in need of a math lesson.

This survey is the other source for Orange's purported 5% success rate. The graph actually shows that 26% of people who try an AA meeting for the first time are still attending AA after the first year, the attrition is from 19% (those in their first month) to 5% (those in their twelfth), and therefore around 74%.

Orange also claims that this 74% attrition is an AA failure rate. I will now use Orange's spectacularly warped logic to prove that exercise is unhealthy. Watch carefully!

  • After one year, 74% of people who began work-out routines at a gym are no longer using the gym.
  • Therefore gyms have a 74% failure rate.
  • Therefore exercise is unhealthy!

It is indeed true that only 26% of visitors to AA stay more than a year, and AA has shown some concern about this statistic. But:

Orange's 'failure' statistics turn out to be at best ignorance and at worst flat out lies.

Antiquated and Irrelevant References

Orange emphasizes studies by Ditsman from 1967 and Brandsma from 1979. Why does Orange rely on decades-old results? There is plenty of newer research. The answer is that Orange selects the studies that he likes, even though they are practically museum pieces.

Another problem: these studies were not on regular, voluntary AA members, so it's no wonder AA showed limited success with them. Orange quotes as follows:

In other words, the people in these studies were forced to attend AA. They don't seem like the most unbiased samples, do they?

So what's the word on the street?

Let's sum up what some experts say for a balanced view, both pros and cons, of AA.


"Certainly, AA is not a magic bullet for every alcoholic. In my own follow-up studies, there were a few men who attended AA for scores of meetings without improvement." (2005, p 434).

"The implication from [my studies] is that a great many severely alcohol-dependent Americans, regardless of their social or psychological makeup, find help for their alcoholism through Alcoholics Anonymous." (1995, p 388)

"Alcoholics Anonymous appears equal to or superior to conventional treatments for alcoholism, and the skepticism of some professionals regarding AA as a first rank treatment for alcoholism would appear to be unwarranted." (2005, p 431).


[Why are randomized studies of AA difficult? Use Quakerism as an example.] "Randomizing people to attendance or non-attendance at a Quaker meeting would not be good science, and would be unlikely to tell us much about the impact of Quakerism on the lives of those people who are non-randomly drawn towards it... Up to the present, science has not found a research approach to overcome the Quaker-meeting conundrum. Without losing critical dispassion, we can however probably do a bit better than a "don't know at all" kind of answer. With the lack of a full and final proof admitted, but on the basis of the available research, it is not unreasonable to conjecture that AA probably works, in some way or other, for not less than 50 per cent of the troubled drinkers who make contact with it. While we wait for the exact science, it seems evident that the reason why people sit around in church halls, hospitals, prisons, and all manner of other settings to attend AA and talk its talk is because this fellowship meets their needs." (2002, p 116-7.)


"AA, like all other therapies for alcoholism, is limited. ... [Criticisms include] AA's being rigid, superficial, regressive, inspirational, fanatical, stigmatizing, and focusing only on alcohol. The rigidity is more likely to lie in individual members than the AA program itself." (p 597)

"Many efforts have been made to assess the effectiveness of AA attendance... In studies of AA from the 1940s to the early 1970s, sampling difficulties and other methodological problems were immense. Nevertheless, the findings indicated that thousands of AA members had achieved sobriety through AA." (p 597)

"AA involvement correlates favorably with a variety of outcome measures. Those patients who attend AA before, during, or after a treatment experience have a more favorable outcome in regard to drinking. In the few studies available that assess the outcome on other variables, AA involvement is associated with a more stable social adjustment, more active religious life, internal locus of control, and better employment adjustment. Increased ethical concern for others, an increased sense of well-being, and increasing dependence on a Higher Power with less dependence on others also have been described. Finally, there is a positive relationship between outcome and extent of AA participation. Outcome is more favorable for those who attend more than one meeting per week and for those who have a sponsor, sponsor others, lead meetings, and work Steps Six through Twelve after completing a treatment program." (p 592)

American Psychiatric Association

"The effectiveness of AA, per se, has not been evaluated in randomized studies. However, other sources of information provide growing support for the utility of AA and 12-step-oriented treatments as well as the efficacy of professional therapies such as TSF [12-step facilitation] that are aimed at motivating patients to participate in AA. In addition, a large number of studies have documented that greater AA participation is associated with greater rates of abstinence from alcohol as well as with better drinking outcomes. Thus, most patients should be encouraged to attend at least several AA meetings to ascertain the appropriateness and utility of AA in helping them remain alcohol free."

- Practice Guideline for the Treatment of Patients With Substance Use Disorders, p 98-99. (PDF, 2.2 MB)


I'm glad I did my own reading on AA. If I'd trusted Agent Orange, I might have believed that AA doesn't work!


Overviews of AA Effectiveness

Orange's references discussed above

Further Online Reading

URL This page was last updated December 5th, 2013. Feedback welcome and appreciated: