The International Journal of the Addictions Vol. 15(1), 839-848, 1980. Factors Related to “Stablew and “Unstablew Affiliation with Alcoholics Anonymous

Joseph Boscarino, PhD. Social Research Division Market Opinion Research Detroit, Michigan 48226

Abstract The study suggests that A.A. affiliates who are among the more seriously impaired when they come to A.A. tend to make less stable members of this organization. Also, affiliates who are younger, male, lower in SES, have more slips, are in A.A. shorter time, and less involved in A.A. activities tend to be less stable members. Implications for treatment are suggested. Previous research indicates that alcoholics who are more seriously impaired at the onset of treatment tend to do worse after treatment than those less seriously impaired (c.f. Costello, 1975a. 1975b: Armor et al., 1976. pp. 87-91: Kish and Hermann, 1971). Specifically, studies show that alcoholics who have higher levels of alcoholic symptomatology, are of lower socioeconomic status, and have lower levels of social stability (as measured by employment, residence, and marital status) tend to have lower remission rates after treatment. These findings are consistent with the more extreme observation that alcoholics living on skid row are usually more difficult to rehabilitate (cf. Pittman and Gordon, 1967: Pittman and Gillespie, 1967; Wiseman, 1970: Zimberg, 1974). HYPOTHESIS Based on these earlier findings, it is anticipated that more geriously impaired alcoholics would do no worse in A.A. than those less seriously impaired. In this respect, one study on A.A. has previously found that A.A. affiliates who have never been hospitalized for alcoholism prior to affiliation tend to have fewer “slips” in A.A. in comparison to those who have been hospitalized (C.,1965). Another study also found that “physical stability” prior to A.A. affiliation is positively related to “full-fledgen membership in this organization (Trice and Roman, 1970). Thusthere is some direct support for theassumption that the degree of impairment prior to A.A. affiliation is related to the degree of improvement in A.A. It is also anticipated that lower socioeconomic status affiliates would also do worse in A.A. There are several reasons for this assumption. One reason is because, as noted above, lower SES alcoholics are known to generally do worse after treatment. This is believed due to the fact that they have higher degrees of alcoholism impairment. For example, the epidemiological studies by Cahalan and his associates have consistently found that lower SES drinkers have more serious drinking problems than higher

SES drinkers (Cahalan et al., 1969; Cahalan, 1970; Cahalan and Room, 1974). Also, A.A. has been known to have a conspicuous absence of lower-class members (see Alcoholics Anonymous, 1972; Leach, 1973). Although Lofland and Lejeune (1960) have found no “overt” discrimination against such persons, it is suspected that a certain “associative bias” exists which tends to attract working and middle-class members, and repel lower-class members from affiliating with this organization. Another factor which is believed to affect A.A. membership is age. The Cahalan studies mentioned above show that younger drinkers actually have more serious drinking problems overall than older drinkers. In addition, because the average A.A. member is often middle-aged (Alcoholics Anonymous, 1972), it is believed that younger A.A. members may often find affiliation more difficult. For these reasons it is suspected that younger affiliates will do worse in A.A. than older affiliates. Another important factor related to A.A. affiliation should be sex. Most research (e-g., the Cahalan studies) indicates that men tend to have significantly more alcohol-related problems that women. Thus it is suspected that they, too, would make less successful A.A. affiliates. Finally, common sense suggests that affiliates who are less involved in A.A. activities, have not accepted A.A.’s basic belief system, have been in A.A. a shorter period of time, and have had more “slips” while going to meetings would also do worse in A.A. Stated in more formal terms, the following hypothesis is now suggested: “A.A. affiliates who are younger, male, lower in SES, have more drinking problems and previous treatment, who participated less in A.A. activities, not accepted basic beliefs, affiliated for a shorter period of time, and had more slips while going to meetings will tend to do worse in A.A. ” It is hypothesized that these nine independent variables can significantly predict the outcome of affiliation with this “treatment” organization. METHOD

Data to test this hypothesis have been drawn from mail-return questionnaires completed by A.A. members. The questionnaires were distributed at 10 different A.A. groups in one state to every A.A. member who agreed to take one. The groups ranged from working-class to upper middle-class and from suburban to urban. The questionnaires were distributed after the meetings by one researcher and the help of several A.A. members. A total of 155 questionnaires was returned which represented 44% of the questionnaires actually passed out. The dependent variable selected to reflect the A.A. affiliate’s overall ”outcomef’with this organization is the number of times the affiliate claimed to have “dropped out” (NUMDROP) of A.A. in the course of his or her affiliation. It is felt that this variable would be closely related to successful or nonsuccessful affiliation with A.A. by providing a measure of the “stability” of that affiliation over time. The main assumption here is that if the affiliate could not maintain a stable membership, he or she could not be successful member of A.A. Participant observations for a year in this organization qualitatively substantiated this

assumption. The nine independent variables used to predict this independent variable are operationalized as follows. AGE is the age when the member first came to A.A. SES is a modified version of Hollingshead’s (1965) two-factor index used as a full interval scale. Directionality is reversed so that a low score signified low status, and a high score, high status. SEX is represented by scoring men as 1 and women as 2. TREATMENT is the total amount of previous treatment for alcoholism. PROBLEMS is a score reflecting the number of alcohol-related problems encountered in key life areas (e.g., job, legal, and health areas). IDEOLOGY is a Likerttype score which reflected basic aspects of A.A. ideology (e.g., abstinence, religiosity, and proselytization). A.A.-TIME is the total amount of time the affiliate actually spent in A.A. in the course of affiliation. PARTICIPATION is a score reflecting the affiliate’s involvement in A.A. activities )e.g., times help office, did twelfth-step work, and set-up or cleaned-up at meetings). SLIPS represent the number of times the affiliate had a drink while affiliated with A.A.(for a more detailed description of these scores the interested reader is directed to Boscarino (1977, pp.245-250) RESULTS

The purpose of this analysis is to predict the number of times the A.A. affiliate will drop out of A.A. (NUMDROP) from the nine predictor variables delineated above. As was noted, the major assumption in this connection is that this is related to the “stability” of an affiliate’s association with A.A. and, furthermore, that this stability is related to “successful” affiliation. These assumptions should be kept in mind throughout the analysis presented below. To understand how these variables are related, a correlation matrix was constructed and is now presented in Table I. Examination of this matrix reveals that NUMDROP is significantly related to age (-.15, p.<.05), TREATMENT (.39, p.<.001), PROBLEMS (-14,p.<.05), and SLIPS ( . 3 6 , P.<.001). It is also related, although not significantly, to sex (-.lo, N>S.), IDEOLOGY (.07, N>S>), and A.A.- TIME (>11, N.S.). For the most part, these relationships support hypothesis. Affiliates who were younger when they first came to A.A., who had more drinking problems, who had more treatment prior to A.A., who had more slips while going to meetings, and who were men tended to drop out of A.A. more often. As can be seen, however, not all of the hypothesized relationships seem to hold up. For example, the correlation coefficient between NUMDROP and SES should have been negative. The same is true of the correlation between NUMDROP and IDEOLOGY. The correlation coefficient of NUMDROP with A.A.-Time and PARTICIPATION is also unexpected. According to the hypothesis, these two variables should have been significantly negatively related to NUMDROP. Part of the reason why some of these independent variables are not related to the dependent variables in the predicted manner could be due to the complex way they are intercorrelated. For example, men have more drinking problems j = – .45, P < ,001 ), but they also have more time in A.A. , r 4 , = – 16, < ,05), Time in A.A. on

the other hand, is positively related to not only the dependent variable ( I , . = 1 I . S ) but A.A. ideology as well = P < 05). which in turn is positively related to the dependent variable ( r , – = .07. N S . ) . ‘ These kinds of intercorrelations are suggestive of complex “suppressor” effects between some of the variables (cf. Cohen and Cohen, 1975,pp.87-91). In a univariate analysis, these will conceal the true relationship between variables. If this is true, then one can expect some of the contradictory relationships found in Table I, and more complex multivariate techniques must be used.

Table 1 Correlation Matrix of NUMDROP with Independent Variables ( N = 155-135)

(1) (2) (3) (4) (5) (6) (7) (8) (9 (10)

(1) NUMDROP – -.IS’ .O 1 -.I0 .39′ .14* .07 .I1 – .04 .36′ (2) AGE – .05 .IS’ .02 -.03 .17′ .05 .08 .08 (3) SES – .04 -.09 – .07 .02 -.01 .08 -.08 (4) SEX – -.I2 -.4SC .I1 -.16′ .08 -.03 (5) TREATMENT – .23b .05 .I0 -.I2 .27′ (6) PROBLEMS – .I0 .19b .08 .13′ (7) IDEOLOGY – .IS’ .35′ .05 (8) AA-TIME – .65′ .03 (9) PARTICIPATION 4 – – .02 (10) SLIPS –

Mean 1.26 38.22 45.48 1.29 4.34 7.24 34.60 4.31 14.29 1.41 SD 3.29 9.74 12.93 .46 4.73 4.77 4.65 5.68 1 1.80 .51 *p < .05. bp < .01. ‘ p < ,001.

Carrying the analysis one step further, a multiple regression was performed using the nine independent variables as predictors of NUMDROP (see Table 2). Variables were entered in the three-step hierarchical manner shown in Table 2. Altogether these nine variables explained 28% of the variance, which is significant As can be seen, only AGE, TREATMENT, and SLIPS emerged as significant (all pc.01 here). As insightful as this regression is, it is suspected that some of the hypothesized relationships are still being weakened by the complex suppressor effects mentioned. To test this possibility, a factor analysis was performed. The result of this analysis is now presented in Table 3. Examination of the factor loadings within these two factors, indeed, lends support to the suppression hypothesis. As can be seen, when all ten variables are reduced to two factors. NUMDROP loads relatively low on the first and high on the second factor. Hence, Factor I is labelled “stable affiliation” and Factor I1 “unstable affiliation.” Once this discrimination is made, closer inspection reveals that

Table 2 Multiple Regression Predicting Number of Times Affiliate Dropped out of A.A. ( N U M D R O P ) (N = 135)” –

r R R’ BETA F

I. AGE SEX SES 11. TREATMENT PROBLEMS 111. SLIPS IDEOLOGY AA-TIME PARTICIPATION – —

=Total df = 91126. Total F = 6.26. p < .05. bp < .01.

Table 3 Factor Analysis Relating NUMDROP with Independent Variables ( N = 135)

Factor loadings’

Variables I % I1

NUMDROP AGE SEX – SES TREATMENT PROBLEMS SLIPS IDEOLOGY AA-TIME PARTICIPATION

‘Principle components with Varimax rotations.

the size or direction of the factor loadings for all nine independent variables now support the original hypothesis. For example, AGE loads positive on Factor I and negative on Factor 11. SEX loads a high negative on Factor I1 and a relatively low negative on Factor I. SES loads a low positive on Factor I and a negative on Factor 11. TREATMENT loads a low positive on Factor I and a high positive on Factor 11. PROBLEMS loads moderately high on Factor I but it loads nearly twice as high on Factor 11. SLIPS loads a much higher positive on Factor I than on Factor I. IDEOLOGY loads a high positive on Factor I and a low negative on 11. Also, A.A.-TIME loads a high positive on Factor I and low on 11. Finally, PARTICIPATION loads a very high positive on Factor I and a high negative on Factor 11. These factor loadings are quite remarkable because all nine independent variables are found to vary with the dependent variable in the direction predicted by the original hypothesis. The sizes of some of the loading are smaller than expected, but nevertheless they all vary in the hypothesized direction. Thus the expectation

of complex suppression effects seems to be substantiated. Factor analysis appears to have helped reduce the complex relationships between variables and therefore control for the expected suppression effects. The result of this is that a more accurate picture of the relationships between these variables has emerged. DISCUSSION It was hypothesized that A.A. affiliates who were younger, male, lower in SES, had more drinking problems and previous treatment, participated less in A.A. activities, accepted fewer basic A.A. beliefs, affiliated a shorter period of time with A.A., and had more slips while going to meetings would tend to do worse in A.A. Initial analysis partially supported the hypothesis. However, it was suggested that the intercorrelations between variables were confounding the relationship between the independent and dependent variables and that multivariate analysis was needed. A factor analysis was performed in an effort to bring out a more accurate relationship between these variables. The result of this analysis indicated that all nine independent variables were in fact related to the dependent variable as originally hypothesized. In conclusion, the study reported here lends continuing support to the finding that those more seriously impaired tend to do worse in treatment. A.A. appears to be no exception to this tendency. It also suggests that the affiliate’s social background, as well as the degree of involvement in this organization, also plays an important role in successful membership. These findings, while limited because of the size and method by which the sample was obtained (e.g., mail-return questionnaires) are nevertheless significant. They help shed light on the types of alcoholics which may be less successful with an A-A, affiliation. These findings should have important relevance for clinicians who refer alcoholic patients to A.A. For example, knowing that an individual has most of the characteristics associated with less stable A.A. membership, clinicians should take special measures if they recommend this patient attend A.A. meetings. It must be strongly emphasized, however, that these findings do not mean A.A. affiliation should not be used for a certain type of patient. They do suggest that if a patient has most of the characteristics associated with unstable membership, that additional efforts should be made to maximize the effectiveness of this organization. In particular, care should be taken in sending the individual to a group close to the individual’s own social background. In addition, special arrangements could be made for A.A. members to escort or meet this individual at an A.A. meeting, or at least pay additional attention to this person when he or she arrives. If none of these is possible, then clinicians should at least spend additional time preparing this individual for an A.A. meeting. Finally, it should be mentioned that a serious problem with this type of research is that it is based on a cross-sectional analysis. This, together with the mail-return questionnaire method used, renders the reported findings here tentative. Longitudinal research which actually follows A.A. members over time is urgently

needed in this area. Of course, this is difficult to accomplish because of A.A.’s resistance to outside researchers. A possible solution to this problem is that researchers study the A.A. affiliation process through an outside referral agency that sends patients to A.A.. Ditman et a1.(1967) have had some success doing this with court-assigned probationers. Clearly more research, and especially the longitudinal kind, is needed before more accurate inferences can be drawn. REFERENCES ALCOHOLICS ANONYMOUS. Profile of an A.A. meeting. A.A. World Services. 1972. ARMOR, D.J., POLICH, J.M., and STAMBUL, H.B. Alcoholism and Treatment. Santa Monica. California: Rand Corp., 1976. BOSCARINO, J. Alcohol Career Patterns in Alcoholics Anonymous: A Systemic Approach to Alcoholic Defined Behavior. Unpublished PhD Dissertation. New York University, 1977. C., B. The growth and effectiveness of Alcoholics Anonymous in a Southwestern city, 1945-1962. Q.J. Stud. Alcohol 26:279-284, 1965. CAHALAN, D. Problem Drinkers. A National Survey. San Francisco: Jossey-Bass, 1970. CAHALAN, D., CISIN, I.H., and CROSSLEY, H.M. American Drinking Practices: A National Study of Drinking Behavior and Attitudes. New Brunswick. New Jersey: Rutgers Center of Alcohol Studies. 1969. CAHALAN, D. and ROOM,R. Problem Drinking Among American Men. New Brunswick, New Jersey: Rutgers Center of Alcohol Studies, 1974. COHEN, J. and COHEN, P. Applied Multiple Regression Correlation Analysis for the Behavioral Sciences, New York: Wiley, 1975. COSTELLO, R.M. Alcoholism treatment and evaluation: In search of methods. Int. J. Addict. 10: 251-275, 1975a. COSTELLO, R.M. Alcoholism treatment and evaluation: In search of methods, 11: Collation of two year follow up studies. Int. J. Addict. 10: 857-867, 1975b. DITMAN,K.S., CRAWFORD, G.G., FORGY, E.W., MOSKOWITZ, H. and MacANDREW,C. A controlled experiment on the use of court probation for drunk arrests. Am. J. Psychiatry 124: 160-163, 1967. HOLLINGSHEAD, A.B. Two-Factor Index of Social Position. New Haven, Connecticut: Privately printed, 1965. KISH, G.B. and HERMANN, H.T. The Fort Meade alcoholism treatment program: A follow-up study. Q.J. Stud. Alcohol 32: 628-635,1971. LEACH, B. Does Alcoholics Anonymous Really Work? In P.G. Bourne and R. Fox

(eds.) Alcoholism Progress in Research and Treatment. New York: Academic, 1973,pp. 245-284. LOFLAND,J. and LEJUNE, R.A. Initial interaction of newcomers in Alcoholics Anonymous: A field experiment in class symbols and socialization, Soc. Probl. 8:102-111, 1960. PITTMAN, D.J. and GELLESPIE, D.G. Social policy as deviancy reinforcement: The case of the public intoxication , offender. In J. Pittman (ed.) Alcoholism, New York: Harper and Row, pp. 106-124, 1967. PITTMAN, D.J. and Gordon, C.W. The chronic drunkenness offender. In D.J. Pittman (ed.) Alcoholism, New York: Harper and Row, pp.99-105, 1967. TRICE, H.M. and ROMAN, P.M. Sociopsychological predictors of affiliation with Alcoholics Anonymous: A longitudinal study of “treatment success.” Soc. Psychiatry 5: 51-59, 1970. WISEMAN,J.P. Stations of the lost: The Treatment of Skid Row Alcoholics. Englewood Cliffs, New Jersey: Prentice-Hall, 1970. ZIMBERG, S. Evaluation of Alcoholism Treatment in Harlem. Q.J. Stud. Alcohol 35:550-557, 1974.