Social Psychiatry, Vol. 1, (4), 195-204, 1967.

Alcoholics Anonymous: The Anatomy of a Self Help Group

Griffith Edwards, Celia Henaman, Ann Hawker, and Valerie Williamson Alcohol Impact Project, Institute of Psychiatry, Maudsley Hospital, London, S.E., 5.

Summary. 1. 40 out of 45 A.A.groups in London cooperated by asking all members attending meetings during one week to complete a questionnaire – 306 questionnaires were returned. – 2. Males accounted for 81% of membership, females 19%. Mean age was 45.7 years. 58% were married, 19% single, 20% divorced or separated and 3% widowed. Present Social Class distribution was I (Upper) 9%, I1 – 26%, I11 – 50%, IV – lo%, V – 4%. Mean scores on the short scale of the Maudsley Personality Inventory were Neuroticism, 8.59 (significantly high) and Extraversion, 7.97 (not differing significantly from normals). – 3. Mean age at which drink “became a problem” was 28.5 years for men and 33.9 years for women (significantly older). Only 33% of members had not received hospital treatment for alcoholism. Mean duration of A.A. membership was 28.8 months but 42% had been sober for less than 6 months. 82% of members believed that their G.P.s knew of their alcoholism and 51%, their employers. – 4. The frequency of 23 consequences or accompaniments of excessive drinking is given. – 5. A.A. is to some extent an autonomous treatment facility which produces maintained sobriety, but a greater emphasis is needed on its possibly larger and more unique role as a supportive fellowship for the relapsing alcoholic, and as an organization supplementary rather than alternative to hospital care. – 6. The question of who goes to A.A. can be in part answered by saying that the person who affiliates is able to identify with a clear-cut but partly fantasy-based image, which also serves a vital function in excluding the person who would threaten the group’s homogeneity. Introduction Surprisingly little is known about who goes to Alcoholics Anonymous. In England, the only attempt to collect descriptive data on an A.A. population was that carried out by COOPER and MAULE: they analyzed responses to 76 interviews conducted at 4 open A.A. meetings in London, to which were added replies to 76 postal questionnaires sent in form various parts of the country, but refusal rates were not given. American studies have included one by RITCHIE in which 47 A.A. members were formally and 150 informally interviewed, but in which no actual data were published; a thesis by BOHINCE and ORENSTEEN based on an analysis of replies to 231 questionnaires which were returned out of a total of 582 sent to A.A. members in Minneapolis; and a paper by FLAHERTY et al., describing a study in which 52 of 111 postal questionnaires were returned. JELLINEK’S classical paper on the phases in the drinking history of alcoholics was based on 158 replies – a 10% response rate. A more recent American report written by an A.A. member, Bill C. has given a fuller and probably more accurate account of who

goes to A.A. than previously available, but this study was of alcoholics showing some regularity of attendance at meetings, rather than of the total A.A. population. Confident interpretation of the results of all these previous studies is to some extent hampered, it seems, by the uncertain nature of the sampling. Purpose of the Present Investigation

The purpose of this project was, firstly, to provide a description of the A.A. population in London. Secondly, however, the aim was to go beyond the mere descriptive phase and to use the data collected to gain insight into the dynamic interaction between personality and environment which may determine success in giving up drinking, regularity and length of A.A. attendance, and even the incidence of certain complications of drinking. To this end as much quantifiable information as possible was collected. Some preliminary results appeared in the Lancet, August 13, 1996. Method A) Design of Questionnaire

The final design of the questionnaire was determined after pilot study on 40 ex-hospital patients had shown the most comprehensible wording. The eight page duplicated schedule then contained 15 items on drinking history and A.A. attendance, 22 items on the complications of excessive drinking, and 23 demographic and social items. The 12 questions of the Short Scale of the Maudsley Personality Inventory were also included: this instrument measures the trait of Neuroticism and Extraversion, each on a 12-point scale. Social stability, both at present and at the time of joining A.A., were rated on a 4-point scale derived from STRAUS and BACON. One point was allotted on each of the following criteria: a) Subjects held a steady job for at least the 3 immediately preceding years. b) Residential immobility for at least the 2 preceding years. c) Living in own home or in that of relatives or friends. d) Married and living with spouse. Social class was determined on the basis of occupation using the Registar General’s Classification of Occupations: information was also elicited on income, this being used as a guide to classification (e.g. with housewives where occupation was not otherwise stated). Anyone unemployed or retired was classified by previous occupation. The remaining items in the questionnaire, rather than being set out here, will be indicated in the section reporting results which follows. *Copies of the questionnaire can be obtained by writing to the Alcohol Impact Project .

B) Dirhibutwn and Collection of Questionnaires

There are held in London each week 45 A.A. meetings, the arrangements for each meeting being the responsibility of the local Group Secretary. A letter sent to each Secretary explained the purpose of this study, stressed its confidentiality, and asked how many questionnaires would be required for complete coverage. Batches of questionnaires were sent to each co-operating secretary for use at all meetings held during the week of Monday, 7th to Sunday, 13th December 1964 inclusive. Secretaries were asked to ensure that every member present at any meeting complete a questionnaire, unless one had been filled in at an earlier meeting in the same week. The plan adopted was that, at the close of the meeting, the secretary should collect the completed forms and send them to the research office, rather than permitting members to take the questionnaires home (with the attendant uncertainties of individual co-operation). RESULTS A) Co-operatwn

Co-operation was obtained with 40 (89%) of the 45 groups, which should have held meetings during the week of the study. Failure of 5 groups co-operate was due to difficulty in making contact, and it seems likely that some of these groups had indeed ceased to function. The secretary of one group preferred to leave the return of questionnaires to individual members. As a result of the approach to individual A.A. members coming to each meeting, by the group’s own Secretary, the number of members of any group failing to complete a questionnaire was very small – the precise figure could not, however, be determined. A total of 306 completed questionnaires was returned. B) Data

(The notation will be used throughout to indicate 95% confidence limits on the mean). 1. Sex: Of the 306 members 248 (81%) were men and 58 (19%) were women. 2. Age: The mean age was 45.7 1.7 for men, 45.6 2.1 for women, and 45.7 1.1 for the sample taken as a whole. Four percent were aged 20-29, and 22%, 4 2 % , 24%, 7% and 1% were in successive 10-year age-groups up to 70-79. 3. Marital Status: Details of present marital status are set out in Table 1.

Table 1. Marital Status: Percentage of Subjects in eao3 Category Marital status Men Women Total n-248 n-58 n-306

Single 18 26 19 Married 60 46 58 Separated 12 11 12 Divorced 8 8 8 Widowed 2 9 3

4 . Social Class: Enquiry was made into both “present” and “best” social class; the analysis is given in Table 2. Table. 2. Present Social Class and Best Social C k s : Percentage of Members in each Category I I1 111 IV V Not categorized

Total: resent 9 26 50 10 4 1 fest 12 32 45 9 1 1

5. Social Stability: Replies to the questions which comprised the social stability scale were marred by a high incidence of incomplete or unsatisfactory answers: 45 subjects had to be excluded from the analysis of initial social stability and 17 from the analysis of present social stability. Data on the remaining subjects is given in Tables 3 and 4. Table 3. Social Stability. Percentage of Subjects Scoring 0 4 on Scale Derived from S r r ~ u sand BACON Initial Pmmt 4 point Scale Men Women Men Women Score on n-217 n-44 n-238 n-51

Table 4. Social Stabilit ‘score (Straus and Bacon Scale) of A.A. and Two ~ o s ~ i t a ! ~ a m p l e s . Means and 91% Confidence .Limits on Means Smdr Type of Number Mean SS population score

A.A. – Tim: of Male 217 2.5 & 0.2 joining Female 44 2.3 k0.4 Total 261 – A d . Time. of Male A.A. 238 2.5 f 0.2 2.5 f 0.2 mqulry Female 51 2.2 k0.4 Total 289 2.4 k 0.2 DAVIESet a[. (1956) Hospital 39 1.5 k0.5 EDWARDS (1965) Hospital 40 2.4

6. Personality: Findings on the present group, on Eysenck’s original standardisation sample and on a normal population recently surveyed in London (HARE and SHAW, 1965) are given in Table 5. The A.A. population have a significantly higher mean N score (for neuroticism) than Eysenck’s standardisation group (P<.01) and although information was not given by HARE and SHAW on the standard deviation of the N score of their population, it appears probable that here too the A.A. group has a significantly higher score. The mean E score (for extraversion) of A.A. members resembles that of a normal population. HARE and SHAW also reported the percentage of subjects scoring 10,ll or 12 on the N scale, showing a correlation between this and other evidence of neurotic disturbance. Data on the number of subjects scoring within this range is given in the

last column of Table 5: the difference between A.A. and the normal population is striking. Table 5. Scores of A.A. Members and Nonnal Populations on N (neuroticism) and E (extraversion) Scales of Short Form of Maudsley Personality Inventory. Means and 91% Confidence Limits on Means Population Sample Mean Mean % Scoring Size N E N 10, 11. 12

A.A. male A.A. female A.A. total Eysendr’s normals London: New surburb male female Old surburb male female

7. “Age at which drinking first became a problem”: Subjects were asked at what age “drink first became a problem.” No attempt was made to define the terms of the question more closely – each respondent was left to interpret the onset of the “problem” according to his own criteria. For men, the mean age was given was 28.5 1.2, for women 33.9 2.2 and for the total group 29.5 1.1 years. The difference between the means for mein and women is significant (P.tO.05). Distributions are given in Table 6. Table 6. Age in Years at which Drinking First Became a Problem: Distribution b y 10-year Groupings in Percentages

__ _ Age t r o w Men n-248 Women n-56 . – 10-19 1 16.9 1.8 %

20-29 42.3 25.0 30-39 24.6 53.6 40-49 13.3 14.3 50-59 2.4 3.6

8. Hospital Treatment for alcoholism: Both among the men and among the women 60% had received in-patient treatment for alcoholism at least once. Those who had received out-patient treatment comprised 19, 21 and 20% of the male, female and total group respectively. There was some overlap between those who had received in-patient and out-patient care: 33% of both men and women had received neither. 9. Duration of A.A. membership: Distributions are given in Table 7. The mean duration was 50 5.8 months for men, 38 9.3 months for women and 48 5.1 months for the group. Table 7. Duration o f A.A. Membership in Years; D&tribution , in Percentages Duration Men Women 18-241 11-53

0- 6 months – 22. 16 7-12 months 6 7 – – 2 years 4 13 17 23 24 -6 15 16 – -10 8 11 8 2 9 . 10+ 8 2

10. Regularity and frequency of A.A. attendance: Subjects (N=304) were asked which of the following three statements described their pattern of attendance most closely: “Regularly all the time (at least once/month)” 60% “Regularly except for one or two breaks (for more than 3 months)” 30% “Just off and on” 10% An additional question was then asked on present pattern of attendance, with the following alternatives: “Most nights1′ 20% “At least once in most weeks” 739 “At least once in most months” 3% “Just dropping in occasionally” 4% There was no difference between the men and women. 11. Duration of sobriety: The question asked was “How long is it since you last had a drink?” Distributions are given in Table 8: 42% of members had been sober for only six months or less, but the maximum period was almost 15 years. The mean duration was 30.3 4.1 months for men, 21.9 7.8 months for women and 28.8 3,6 months for the total group. Table 8. Duration of Sobriety in Months: Distribution in Percentages Duration Men Women n-248 n-55 0- 6 months 42 40 7-12 months 10 11 – 2 years 9 20 – – 4 years 6 years 14 9 18

– 8 years 8+ 8 6 -56

12. Number of “slips” since joining A.A. : 43% of 293 subjects said they had not “slipped”, by their own definition since joining A.A. : 29% had slipped once or twice, 10% had slipped 3-4 times and 18% had slipped 5 or more times. Subjects with less than 1 months’ membership were excluded fromthe analysis. There was no difference in distribution between men and women. 13. Complications and accompaniments of excessive drinking: The questionnaire contained 23 items which enquired into the occurrence of a variety of complications and accompaniments of excessive drinking. The term “accompaniments” is used advisedly because it cannot, for instance, be assumed that peptic ulcer was always caused by drinking, and the factors contributing to breakdown of marriage are, for example, often extremely complex. In Table 9 which gives the actual wording of the questions, the percentages of subjects answering “yes” to each of them is set out. These items elicited a high response rate, and the number of subjects for whom answers were analyzed is given in the Table. The right- hand column indicates which phenomena showed a significant between-sex difference of incidence.

Table 9. Incidence of Complications and Accompaniments o f Excessive Drinking, with Signifi- \ cance of Between-sex Differences (Signifiancc of Between-sex Differences: ** P < 0.01, * P < 0.05) Men i Women n ’10 n ‘1’

. . . .. .. .. .. .. .. .. .. .. ..-.. H a d “bla&outsn-patches of completely blank memory through drinking Had severe morning shakes ………. ** Lost a job b e c a w of drinking Had D.T.s – that is, shaky, confused and seeing things * Pawned to get money for drinking ……… * Been arrested for being drunk ** Slept rough because of drinking .. .. .. .. .. .. .. .. .. .. * Stolen to get money for drinking ……… “Heard voices’ because of drinkin . . .. .. .. .. .. .. ** Been in a serious fight because of &inking Broken up a marriage because of drinking …… Begged to get money for drinking . . . . . . . . . ** Drunk cheap wine regularly ……….. . . . . .. .. .. .. .. .. .. ** Attempted suicide because of drinking Drunk rough cider regularly Been married more than once . .. .. .. .. .. .. .. .. .. Bern in prison because of drinking Had a stomach or duodenal ulcer ……… Been convicted of drunken driving ……… Been arrested for being drunk 5 or more times. . . . . Drunk methylated or surgical spirit . . . . .. .. .. .. Had an operation for stomach o r duodenal ulcer Had T. B. ……………..

Table 10. Incidence of Certain Complications o f Excessive Drinking: Comparison of Male A.A. Members and Male Hosa pital Alcoholics. Hospital Data from GLATT[14 . Probability Beyond 0.001 Indicated by + ++ and Beyon 0.05 by Significance + Percentages of diicrencc A.A. Hospital

Amnesias Lost job 95 63 78 52 +++ + Suicidal attempt 26 24 – , Imprisonment 22 20 – Drunk Driving Crude spirits 16 14 23 18 – – 14. General Practioners and Employers: Subjects were asked “Do you think your General Practitioner now knows you have a drinking problem?” 80% of men and 88% of women thought that they did. A similar enquiry was made as to the employers’ awareness. 54% of men and 32% of women answered affirmatively. Discussion A) The Method

Considerable problems must be overcome in studying an organization which, in England at least, keeps scanty records if any, and has a particularly well-maintained tradition of anonymity. The relation between A . A . and doctors in London happens, however, to be particularly good, and that it was made clear at the planning stage that if A . A . ‘ s prime purpose of rendering aid to the

individual alcoholic were in no way compromised, every assistance would be given by them to this research. There are always difficulties in accepting evidence gained from a self-administered questionnaire, but this method of surveying the A.A. population was the only practical one interviewing would have been limited to “open” meetings. Neither the reliability nor the validity of the answers can be established, and although this may be of small importance so far as simple

demographic information is concerned, the truthfulness of subjects’ answers on such matters as arrest, imprisonment, regularity of attendance, etc. obviously requires investigation. It is relevant to note, of course, that the majority of interview and case studies are open to the very same criticisms. The immediate collection of completed questionnaires at the meetings resulted in a high response rate, and probably to a large extent reduced the risk of sample bias inherent in the low response rates obtained in a number of earlier studies. It should be emphasized, however, that a shortcoming of the present investigation is that no absolutely precise count of non-responders was possible. Some of the individual items on the schedule – those for instance which were designed to measure social stability – were unsatisfactory. In wording a self-administered questionnaire, problems occur because of the inevitable conflict between the need for clarity and the often opposing demands of brevity. More experience will be required before a totally satisfactory questionnaire is evolved for this type of survey. B) The Results

1. Demographic factors – sex, age, marital status, social class and social stability. A general picture of Alcoholics Anonymous in London can now be sketched. This is an organization in which male exceeds female membership by 4 to 1. A.A. draws its greatest strength from those who are middle-aged and the Social Class of its members is preponderantly Class I11 or above. Almost 60% of members are married and living with spouse, and Social Stability as measured on a derivation of the Straus-Bacon scale, shows that A.A. is far from being a society of the drifting. The extent to which these demographic data convey a “true” picture of alcoholism may be questioned, as opposed to their giving a picture distorted by selection processes inherent in A.A. affiliation. The “trueftpicture of alcoholism in any society – that picture which is undistorted by the idiosyncratic sampling bias of any particular treatment agency – is almost impossible to obtain. Something may, however, be learnt from comparing and contrasting A.A. and hospital populations: not only will common factors point perhaps towards the fttruetf picture, but dissimilarities should reveal the nature of the selection processes. a) Sex: Thus for instance, sex distribution of A.A. members in

the present study (81% male) is in remarkable agreement with Cooper and Maule’s A.A. survey (82% male), while PARR’S survey of General Practice and KENDELL and STANTONWs report on alcoholics attending a hospital clinic but refusing treatment each gave the proportion of men as 69%. DAVIES, in a treatment series from a London hospital found 78% men, but believed that the hospital’s admission policy was biased toward males. Taken together, these figures suggest that A.A. may have a slightly larger male representation than is found among those English alcoholics known to doctors as either requiring or accepting treatment: the difference is interesting but small. b) Social Class: More striking is the difference between the class structure of A.A. and hospital populations. The figures given in Table 2 show that in the present study A.A. was found to draw its membership predominantly from Social Class 111, with underrepresentation at the two extremes – this structure differs significantly (P.tO.01) from the hospital series’ reported by DAVIES et al., GLATT and KENDALL and STANTON. A.A. shows a smaller proportion of Class 1 than any of these hospital series, but A.A. and hospitals are alike in having a very small representation from Social Classes IV and V – a finding in contrast with PARR’s figures on drunkenness arrests and VALLANCE’S report from a Scottish hospital which did not operate a special alcoholism unit. COOPER and MAULE’s findings on A.A.’s Social Class were at variance with the present findings, a fact which is perhaps explained by their sampling method. Seen against the various reports on social class of alcoholics, how are the present results of A.A. to be interpreted? Again, there is really no evidence as to which source is much more representative of the true class distribution of alcoholism, but what can be claimed with certainty is that the unskilled labourer is as unlikely to be found in A.A. as in the alcoholism unit despite his often being found before the magistrate. The explanation is either the likely one that alcoholism is rare in Social Class V, or that A.A., like the hospitals, operates a powerful selection process. The data set out in Table 2 suggests that a slight drift down the Social Class scale may have taken place during the lives of A.A. members, but this change is not significant. In A.A. there must, of course be members whose drinking has led to a lowering of occupational status, butthis study seems to demonstrate that such a consequence is not as common as might have been expected. c) Age: Almost three-quarters (74%) of A.A. members are over 40 – the concentration in the 40 – 49 age group (42%) is striking. Several hypotheses could be put forward to explain the relative dearth of younger people in A.A.: pathological drinking is perhaps in reality a relatively rare condition under 40, or perhaps alcoholics under 40 are not so ready to seek treatment. Again, however, there is the possibility that the A.A. group exercises a covert selection process and that pathological drinking in a subject’s twenties, which may not by then have gone so far as total “loss of control,” fails to conform to the A.A. stereotype of “alcoholism.” When the mean age (45.7 years) of the A.A. members in this survey is compared with that of British alcoholics in various other

samples, it is apparent that the age at which alcoholics are admitted to hospital is very similar to the A.A. mean. Drunkenness offenders are again the exception – on average the court inebriate is ten years younger than the A.A. member. d) Marital status: Marital status (Table 1) is not a differentiating factor when the results of the present survey are compared with DAVIEStsor KENDALLfs hospital alcoholics but among VALLANCEts patients there were significantly fewer who were married. Information on PARR’s drunkenness offenders is unfortunately on this point lacking. e) Social stability: The concept of the “rootednessttof the individual in society is important, but it is not easy to measure. The social stability scale (STRAUS and BACON, 1951) was scored by adding together points given on four different aspects of the subjects social functioning, without any weighting. This scale has already proved its value in a series reported by EDWARDS, but in the form put forward by the authors it is clear that it was not regarded as more than a rough and ready attempt at quantification: further testing, adaptation and statistical analysis would have to be undertaken before it could be much more than a rule of thumb. Even in its present apparently very simple form, the response rates show the scale is not very suitable for a self-administered questionnaire. With these reservations in mind, the data set out in Table 3 and 4 may be interpreted. The mean S.S. score of A.A. members is significantly better than that of DAVIESfs patients but does not differ significantly fromthat of Edward’s patients. Alcoholics who affiliate to A.A. are, it may probably be concluded, no more rootless (and perhaps indeed rather more socially settled) than the type of alcoholic who is today accepted for hospital treatment. The data shows also that on the average a period of A.A. affiliation sees neither a bettering nor a worsening of S.S., but the failure of the scale to detect any change is perhaps due in part to scoring on job and residential stability having long (three and two years respectively) qualification periods. 2. Personality: The extent to which demographic factors differentiate A.A. from the hospital population of alcoholics has been discussed above, and with the single exception of social class no very positive findings emerges. Are personality factors in this respect more revealing? The short scale of the Maudsley Personality Inventory offered the advantage that recent normative data were available on a London population (HARE and SHAW, 1965). The fact that every one of the 306 members of A.A. to whom the questionnaire was given satisfactorily completed the M.P.I. confirmed that this test is very suitable for self-administration. The finding that A.A. members have on average a higher N score than a normal population but a normal E score, parallels data which were obtained when the full-scale M.P.I. was administered to hospital alcoholics by EDWARDS and the conclusion may probably be drawn that neuroticism and extraversion fail to differentiate A.A. from hospital alcoholics. It would of course be premature to conclude that personality factors are unimportant determinants of A.A. affiliation – the factors measured here may simply not be those

which are relevant. In 1961 SEIDEN found a significant difference between A.A. members and other alcoholics on Bender-gestalt scores. The finding that the mean E score of A.A. members is not significantly greater than that of a normal population does something to contradict the idea that A.A. is particularly attractive to the more roistering and exhibitionistic individual. 3. Age at which drinking first became a problem: The criteria which subjects are applying when they answer this question are by no means certain, but more exact wording would involve detailed enquiry into the first occurrence of such happenings as family quarrels, alienation of friends and work impairment, and it seems unlikely that many subjects would easily or accurately date these. To the patient, the seemingly vague question may be the most meaningful. Whatever the basis on which each subject is making a judgement, the fact emerges that men, see their drinking problem as having arisen at an earlier age on average than do women. Inspection of Table 6 shows that 59.2% of men “had a problem” before the age of 30, while for women the corresponding figure was 26.8%, a result with at least two possible explanations. Men may indeed develop addictive drinking at an earlier age than women, and current social mores which condemn male less strongly than female intemperance may be relevant. Alternatively, it is possible that women appear to have a late onset of this “problem,” abnormal drinking being less immediately disruptive in the life of a housewife than it is to the man earning his living: here the “problem” is seen as created not only by the subject’s drinking but also by the manner in which the environment reacts. For the drinking problem first to have arisen in an A.A. member after the age of 50 is rare – only 2.6% of membership had this late onset. This finding provides an interesting parallel to the classical belief that the first onset of neurotic symptoms is seldom in middle age. 4 . Hospital treatment for alcoholism: The failure of either the majority of demographic or the E and N scores to differentiate A.A. from reported hospital series becomes more explicable: to a large extent the two populations overlap. This overlap is, no doubt, partly a reflection of the good relationship which exists in London between A.A. and doctors – in many hospitals it is now routine practice to arrange for all alcoholics admitted to be introduced to an A.A. group, and conversely A.A. sponsors frequently persuade new members to seek psychiatric help. Increasing availability of free specialized treatment for alcoholism within the National Health Service, with the patient receiving sickness benefit and welfare support for his family, bear on the present scene. Thus, in London at least, it is a considerable mistake to picture A.A. now as an organisation largely helping alcoholics who have not had medical treatment: the inquiry conducted by COOPER and MAULE is in close agreement with these findings. To suggest that A.A. is increasingly serving as an aftercare – and perhaps pre-care – organisation, rather than playing its original and more autonomous role, is in no way to lower its valuation.

More research would of course be necessary to assess the relative contribution of psychiatry and of A.A. in the recovery of those who have received both forms of help. 5. Duration of A.A. membership: The mean duration of A.A. membership was over 4 years, and the picture which comes from Table 7 is of an organisation with a large percentage of faithful hardcore members, and a smaller component of transients, most of whom drop out after about 6 months. The Table shows the rapidity of initial drop-out which is then followed by a much slower attrition. Who are these transient attenders? A guess is that they are not only alcoholics who are lost to A.A. because of immediate relapse or insincerity of intention, but include many who find it impossible to identify with the A.A. image. The present results do seem to underline the fact that a considerable number of alcoholics, after a more or less brief attendance at meetings, find that A.A. is not to their liking and cannot help them. The recent report by EDWARDS on a series of 40 alcoholics who were treated in a London hospital, all of whom were given the strongest encouragement to attend A.A. regularly, showed that one month after discharge 13 (33%) and one year after discharge only 5 (13%) of the 40 patients retained any contact with A.A. The conclusion must again be drawn that A.A. is a selfselecting group which recruits according to very stringent rules. 6. Frequency and regularity of A.A. attendance: The range of possible patterns of attendance is wide and varied, and the technical problem is design a framework of questions which will allow these patterns to be fitted into a few simple categories. The answer under the two headings “regularity” and “frequency” point however in the same direction. A.A. seems to consist largely of members who are energetic in their attendance. Someone who goes to A.A. tends typically to go regularly and often and for a long time. Only 10% describe a regularity of “just off and on” and only 7% go to meetings less frequently than “once most weeks.” A member who is lax in attendance is very probably made to feel that he is deviant. The orthodoxy of A.A. teaching is that a serious concern for sobriety necessitates “sticking close to A.A. If Someone who only turns up once a month is certainly, in statistical terms, deviating from the norm of the group behavior. 7. Duration of sobriety and number of slips: The most important findings as regards duration of sobriety is that 42% of members had been sober for no longer than 6 months. This fact must be set against there being only 21% of members who have been in A.A. for six months or less – the explanation of a considerable proportion having had only a short period of sobriety is not therefore the loading by new members. Further inspection of Table 8 shows that just over half A.A. members had been sober for less than a year, and the average of 28.8 months sobriety for the group as a whole is a reflection of the skewed distribution with its long upper tail. The findings on the number of slips which members have experienced since joining A.A. serves further to bring out the point that A.A. is as much a society of alcoholics who are having difficulty in remaining sober as it is one in which they are staying off drink. This interpretation however must not be pushed

too far, for although it is evident that more than half the members have slipped since joining A.A. and 18% more than 5 times, 43% of members have remained completely sober. The total picture which comes from these results suggests that an accurate assessment of A.AIts achievements requires rather a different emphasis from that usually given. A.A. is of course in part a “treatment” organising achieving “cure” or “arrest,” and it is these therapeutic successes which have given it prestige. This emphasis has inevitably led to the neglect of a less dramatic but perhaps more important and more unique achievement: A.A. has created a supportive organisation which accepts and continues to tolerate the relapsing alcoholic who has little ability to maintain long-term sobriety. Although it is clear that relapse does not necessarily lead to rejection, a question unanswered by the present study is how often getting repeatedly drunk is the cause for a long-term member ultimately leaving A.A., as opposed to assured and contented sobriety as the reason. Another question deserving investigation is the ability of A.A. to modify the severity of the “slip” without preventing it. A.A. may assist a man in cutting short the drinking bout, with less damage done than there would have been if A.A, had not helped. A count of “slips” which ignores their severity cannot give the complete picture. 8. Complications and accompaniments of excessive drinking. a) Types of alcoholism: What percentage of A.A. members should, in terms of JELLINEKts classification 1960, be termed gamma, delta or epsilon alcoholics – the three categories which imply pharmacological dependence on alcohol? That part of the schedule dealing with symptomatology would need to be very expanded if a satisfactorily complete analysis of the drinking patterns af A.A. members were to be made, but if a subject’s having suffered from “severe morning shakes” can be taken as evidence of pharmacological dependence on alcohol then 82% of members were chemically dependent. Clinical experience suggests however that it is not uncommon to find a patient whose total story suggests gamma alcoholism, but who denies ever having felt very shaky in the morning: the figure of 82% is a low estimate. b) The severity of damage: As regards the extent of damage incurred, the data in Table 9 present few surprises: that 30% of members had attempted suicide, 59% had lost a job, 34% had broken up a marriage “because of drinking,” 6% had contracted T.V. and 19% had developed peptic ulceration at some time demonstrates once more the familiar connections between abnormal drinking and varieties of pathology. The finding that at least 20% of members had been in prison because of drinking and 45% arrested on drunkenness charges with 13% having 5 or more arrests, suggests that any attempt to establish the image of “the alcoholictfas someone whose behavior is absolutely distinct from that of “the drunk” must be viewed somewhat critically. Even the most casual examination of the statistics drives home the seriousness of the disorder which is being studied: the total suffering which this group of 306 alcoholics have between them undergone is appalling and this perhaps bears again on a particular

aspect of A.A.’s teaching. It is widely held in A.A. that an alcoholic must “experience rock bottom” before being able to accept help, and the severe damage shown in the present analysis to have been suffered by so many A.A. members is, no doubt, the group experience which forms the basis of the group’s assertion. And again this formulation of the “ideal” may be suspected of influencing the selection process: the problem drinker who cannot himself catalogue a series of disastrous experiences may have difficulty in identifying at all. Rock-bottom is today quite often interpreted as a personal experience of the insufferable, rather than as meaning gross degradation, but the idea is still that the alcoholic seeks treatment only in response to an excess of suffering. The hypothesis must however either be modified to postulate a rock bottom shifting in response to offered insights, so that the concept becomes relatively meaningless, or alternatively the hypothesis stands in its original and nihilistic form. Data are available on the incidence of certain items of damage among alcoholics treated in a London hospital by GLATT. Table 10 sets out the comparisons between hospital patients and A.A. members, but it must be noted that the questions asked of the two groups were not in all instances strictly comparable. GLATT, for instance, asked about “prison (at any time prior to admission)” rather than “been in prison because of drinking.” The Table shows that on four of the items listed, including the important one of suicidal attempt, there was no significant difference between A.A. and GLATT’S patients, but with two other items the difference is significant: the incidence of amnesias in A.A. is highly significantly greater than it is among hospital patients (P.<0.001) and significantly more A.A. members had lost a job through drinking (P.tO.05). If these results are interpreted as showing that A.A. tends to select with in some respects severer drinking problems than the average hospital alcoholic such interpretation must be tentative: the same questionnaire with exactly the same wording should be administered to subjects in either setting. c) Between-sex differences: Most of the between-sex incidence differences shown in Table 9 are those which would have been predicted. Men have been in prison, pawned, stolen, lost jobs, been arrested on drunk charges, slept rough and been involved in fights more often than women – in summary the male alcoholic is more socially disordered and more obviously aggressive than the female. The greater incidence of D.T.’s among men is perhaps explained by the man’s life pattern (and imprisonment) more often leading to sudden alcohol withdrawal. That the incidence of suicidal attempt among females exceeds that among male alcoholics accords with other known facts on attempted suicide (STENGEL and COOK, 1958) while the female predilection for cheap wine, a drink which in England is bought in Off Licences )liquor stores), probably reflects the likelihood of the female alcoholic drinking in her home rather than the Pub. 9. General practitioners: The fact that 82% of A.A. members believed their G.P.’s to be aware of their drinking problem is of some interest – the gap between the 1.1 per 1000 adults estimate of alcoholism prelevance in England and Wales which is based on a G.P.

survey (PARR, 1951) and the 11 per 1000 which derives from the Jellinek formula (W.H.O. 1951) is sometimes used to argue that the practitioner knows of only 1 out of 10 of the alcoholics on his list. The A.A. member may however be specially likely to come to the notice of his G.P. because of his willingness to talk about his alcoholism: and if, as is likely, he has also had hospital treatment, the G.P. will usually have been notified. Furthermore, the type of cases which are counted in the Jellinek estimation probably include a majority more mildly affected than this A.A. population. On balance it may seen unwise to assume that the awareness of the A.A. member’s G.P. reflects at all accurately the usual level of awareness of alcoholism in General Practice. Similarly the findings that 51% of employers were thought to know of the A.A. member’s drinking problem cannot be interpreted hastily. The A.A. member’s relationship with his employer is often a very special case, and in many instances there will have been no drinking since the man came to his present job. This result does however hint again at the fact that a considerable proportion of alcoholics – even severely affected alcoholics – are able to keep their drinking problem hidden at work. Final Conclusion A postulate which has been used repeatedly in interpreting the findings of this study is that A.A. selects- that despite the declared policy of having no bar to membership, there are inevitable covert and dynamic selection processes at work. The task of selection, which is orthodox and “led-group” therapy is an important responsibility born by the therapist, is in A.A. somehow carried out by the group itself. Selection is not effected by showing the nonconforming member the door, but by A.A. establishing norms of what the good A.A. member is to be. If the newcomer deviates too far from these norms, he will find identification very difficult, and is then unlikely to be seen at many meetings. Identification is the very essence of the affiliation process. The role played by the Sponsor may sometimes be important, but can be exaggerated: identification is not with any one established member so much as with fragments of a whole series of life histories which are synthesised into identification with the group ideal. The importance of identification in group dynamics was stressed by FREUD, and identification assumes particular importance in the leaderless group which must have a clear and firmlyestablished picture of the ideal member. Although this picture may partly be based on the statistical norm, it derives also in some measure from the group’s fantasy and wishfulfilment. Thus the statistical reality that the average A.A. member is of middle-age and of middle-class origin, is possessed of some social stability, has suffered a great deal in the course of his drinking, and is someone whose alcohol dependence has for many years precluded social drinking, are all facts which provide the basis for the image of the A.A. member as a respectable citizen for whom, unfortunately, one drink is too many. Fantasy then also contributes: the A.A. member is the respectable citizen who can be

clearly differentiated from the drunk, who owes to A.A. a lasting and contented sobriety, and the fact that at least 20% have been in prison, 45% been arrested for drunk, 42% been sober for 6 or less months, 57% experienced a slip, 67% been treated by hospitals, are all realities which will in the interest of the image, be very much played down. The prodigious growth of A.A. bears witness to the efficacy with which this strongly-cut image subserves its function: it is no criticism of A.A. to say that the image which has proved so valuable an identification object for those who have been able to affiliate, has also had the covert effect of exclusion. A group must exclude to achieve the homogeneity needed for group survival, and A.A. in London has excluded those pathological drinkers whose behavior does not conform to one of the “out of control” syndromes, and has largely excluded the young alcoholic and The ~urking-class addict. A danger arises however if the idealised image of the alcoholic created by one self help organisation to serve its particular identification needs, is mistaken for either the true or the total image of the pathological drinker. For instance, A.A.’s identification of “the alcoholic” with the alcoholic addict may turn attention from the very real needs of the nonaddicted drinker. This image requires evidence for return of the addict to normal drinking (DAVIES,1962; KENDALL,1965) to be suppressed. The present report, as has already been stressed, is a study of Alcoholics Anonymous in one particular city; butthis world-wide organisation has so much to tell research workers that it is bound increasingly to attract investigation. Research will produce new facts and will force fresh interpretations, but if the present study is any guide, growth of understanding will only enhance admiration. The astonishing skill with which A.A. has in the process of its growth sensed out and sensitively exploited the dynamics of the leaderless group, offers lessons of no little importance. Acknowledgements. We would like to thank Dr. D.L. Davies, Dean of the Institute of Psychiatry, for his encouragement. Our thanks are also due to 306 members of Alcoholics Anonymous, particularly to the Group Secretaries, who so kindly gave – —- DpcLAaA us their confidence and cooperat& iil tple i-eseaietr L a u n U C ~ C Lve3 UUL “”—I-” 3——- —-: 7

gratitude for the help he offered us in organisation. Miss BARBARA KINSLEY has advised on statistical methods and International Computers and Tabulators have made available free computer time and the great help of Mr. CHRISTOPHER HAYES and Mr. MICHAEL BROWN. The research was supported by a grant from the Bethlem Maudsley Research Fund. Finally, we would like to thank most warmly Miss K. BLICK, MRS> COOPER and MRS> POLGLAZE for their expert secretarial assistance. References BILL, C., The growth and effectiveness of Alcoholics Anonymous in a Southwestern City, 1945-1962. Quart. J. Stud. Alc. 26,279,(1965). BOHINCE, E.A., and A.C. ORENSTEEN: An evaluation of the services and program of the Minneapolis Chapter of Alcoholics Anonymous. Abstract Archive of the Alcoholic Literature prepared by Quart. J. Stud. Alc. 6119 (1950).

COOPER, J., and H. G. MAULE: Problems of drinking: an enquiry among members of Alcoholics Anonymous. Brit. J. Addiction 58, 45 (1962). DAVIES, D.L.: Normal drinking in recovered alcohol addicts. Quart. J. Stud. Alc. 23, 94 (1962). -, M. SHEPHERD, and E. MYERS: The two-year’s prognosis of fifty alcohol addicts after treatment in hospital. Quart. J. Stud. Alc. 17, 485 (1956). EDWARDS, G.: Controlled trial of hypnosis in the treatment of alcohol addiction with an analysis of factors affecting treatment outcome. Quart. J. Stud. Alc. To be published (1966). -, C.HENSMAN, A. HAWKER, and V. WILLIAMSON: Who Goes to Alcoholics Anonymous? Lancet 11, 382 (1966). EYSENCK, H.J.: Manual of Maudsley Personality Inventory. London: London University Press (1959). FLAHERTY, J.A., and H.T. McGUIRE, and R.L. GATSKI: The psychodynamic of the “Dry drunk.” Amer. J. Psychiat. 112, 460 (1955). . FREUD, S .: Group psychology and the analysis of the ego. English trans1 by JAMES STRACHEY, in: Complete works of Sigmund Freud, Vol. 18. London: Hogarth Press. (1921). General Register Office: Classification of Occupations. London: H.M. Stationery Office 1960. GLATT, M.M.: Drinking habits of English (middle class) alcoholics. Acta Psych. Scand. 37, 88 (1961). HARE, E. H., and G.K. SHAW: Mental Health on a new housing estate. Maudsley Monograph 12. London: Oxford Univ. Press (1965). JELLINEK, E.M.: Phases in the drinking history of alcoholics. Analysis of a survey conducted by the official organ of Alcoholics Anonymous. Quart. J. Stud. Alc. 7, 1 (1946). -, The disease concept of alcoholism. New Haven and Connecticut: Hillhouse Press (1960). KENDELL,R.E.: Normal drinking by former alcohol addicts. Quart. J. Stud. Alc. 26, 247, (1965). -, M.C. STATON: The fate of untreated alcoholics. Quart. J. Stud. Alc. In Press. PARR, D.: Alcoholism in general practice. Brit. J. Addiction 54, 25 (1957). -, Offences of drunkenness in the London area: a pilot study. Brit J. Criminol 3, 272 (1962). RITCHIE, O.W.: A socio-historical survey of Alcoholics Anonymous. Quart. J. Stud. 9, 119 (1948).

SEIDEN, R.H.: The use of Alcoholics Anonymous members in research on alcoholism. Quart. J. Stud. Alc. 21, 506 (1961). STENGEL, E., and N.G. COOK: Attempted suicide its social significance and effects. London: Chapman and’Hall 1958. Straus, R., and S.D. BACON: Alcoholism and social stability. A study of occupational integration in 2,023 male clinic patients. Quart. J. Stud. Alc. 12, 231 (1951). VALLANCE, M.: A two-year follow-up study of patients adniitted to the psychiatric department of a general hospital. Brit. J. Psychiatry. 111, 348 (1965). World Health Organisation: Tech. Rep. Wld. Hlth. Org. 42 (1951).

Griffith Edwards, M.A., DiM., D.P.M., Celia Hensman, M.A. Ann Hawker Valerie Williamson, B.A. ALCOHOL IMPACT PROJECT INSTITUTE OF PSYCHIATRY MAUDSLEY HOSPITAL LONDON, S.E.5, ENGLAND.