Random Assignment to Abstinence and Controlled Drinking

Journal of Consulting and Clinical Psychology1984, Vol. 52, No. 3, 390-403

Copyright 1984 by theAmerican Psychological Association. Inc.

Random Assignment to Abstinence and Controlled Drinking:Evaluation of a Cognitive-Behavioral Program

for Problem Drinkers

M. Sanchez-Craig, H. M. Annis, A. R. Bornet, and K. R. MacDonaldAddiction Research Foundation, Toronto, Ontario, Canada

A cognitive-behavioral program with a goal of either abstinence or controlled drink-ing was assessed. The 70 early-stage problem drinkers were randomly assigned toone of the two goal conditions, and within each condition to one of two therapists.On average they received six individual weekly sessions, each lasting approximately90 min. Both groups were taught to indentify risk situations and existing com-petencies, to develop cognitive and behavioral coping, and to assess their progressobjectively. The controlled-drinking group was also taught procedures for moderatedrinking. Over the 2-year follow-up period, no significant differences were foundbetween the groups in reported alcohol consumption. Six months after treatmentdrinking had been reduced from an average of about 51 drinks per week to 13,and this reduction was maintained throughout the second year. Reports of drinkingwere corroborated by independent measures. Although the outcomes of the groupswere similar, controlled drinking was considered to be a more suitable goal; it wasmore acceptable to the majority of the clients, and most of those assigned toabstinence developed moderate drinking on their own.

There have been many research reports overthe past 20 years of alcoholics who developedmoderate-drinking patterns after being treatedin traditional abstinence-oriented programs(e.g., Davis, 1962; Emerick, 1974; Gerard &Saenger, 1966; Polich, Armor, &Braiker, 1981;Sanchez-Craig & Walker, 1982). Alcoholicswho have successfully resumed moderatedrinking on their own tend to be younger, moresocially stable, and less chronic (e.g., Heather& Robertson, 1981; Miller & Caddy, 1977;Polich et al., 1981; Popham & Schmidt, 1976).The overall rates of moderation reported forunselected alcoholics treated with an absti-nence orientation have been modest. In themost extensive study of traditional alcoholism

The authors wish to thank the following colleagues fortheir contributions to the project: C. Milton-Feasby andV. Ittig-Deland for conducting intake and follow-up as-sessments, B. Sisson for medical screenings, Y. Israel forrecommending the biochemical corroborators of alcoholconsumption, and B. Kapur for conducting the laboratoryanalyses. Special appreciation is extended to H. Lei forhis advice on statistical analyses and to D. A. Wilkinson,K. D. Walker, H. A. Skinner, and C. Poulos for their valu-able feedback and support.

Requests for reprints should be sent to M. Sanchez-Craig, Addiction Research Foundation, 33 Russell Street,Toronto, Ontario, Canada M5S 2S1.

programs conducted to date (Polich et al.,1981), 18% of 548 patients were found to bedrinking moderately without apparent prob-lems at 4 years following treatment, comparedwith 28% who had been abstinent at least theprevious 6 months. Much higher success rates,ranging between 60% and 70%, have been re-ported when selected socially stable problemdrinkers have been trained in controlled-drinking techniques (Heather & Robertson,1981; Miller & Hester, 1980).

At present little is known about the com-parative efficacy of abstinence versus con-trolled drinking. Only three studies have beenreported in which the two goals have beenexplicitly included in the treatment programs.Two of the studies involved "gamma" alco-holics (Foy, Rychtarik, & Nunn, 1982; Sobell& Sobell, 1973, 1976) and the third involvedsocially stable "middle-income problemdrinkers" (Pomerleau, Pertschuck, Adkins, &Brady, 1978). The objective of the three studieswas to test the hypothesis that a behavioralprogram with a controlled-drinking goal wouldprove superior to a traditional abstinence-ori-ented program. In all of these studies the goalsof abstinence and controlled drinking wereembedded in very different treatment inter-ventions; thus, the question of whether the

390

ABSTINENCE VS. CONTROLLED DRINKING 391

outcome of the studies was due to drinkinggoal per se or to the nature of the interventionscannot be answered.

To date, no adequately controlled study hasbeen reported on the outcome of clients as-signed to comparable programs differing onlyin treatment goal. The principal objective ofthis study was to attempt such comparisonwith a population of early-stage problemdrinkers. Specifically, the objectives of thestudy were (a) to assess the effect of a cognitive-behavioral program with a goal of either ab-stinence or controlled drinking, using the 6-month outcome for principal comparativeanalyses; (b) to determine the stability of thedrinking measures and measures of social ad-justment over a 2-year period; and (c) to ex-amine the relationship between independentcorroborators of alcohol consumption and self-reported drinking. The treatment approachused is heavily influenced by the cognitivemodel of emotions and psychological stressdescribed by Arnold (1960, 1969) and by La-zarus (1966, 1968). Based on findings of pre-vious controlled-drinking studies, it was hy-pothesized that clients assigned to a goal ofcontrolled drinking would have a better out-come than clients assigned to a goal of absti-nence.

Early findings of this study have alreadyindicated that controlled drinking was a moreacceptable goal than abstinence and thatclients assigned to controlled drinking con-sumed significantly less alcohol during the ini-tial stages of treatment (Sanchez-Craig, 1980).Levels of reported drinking at this time werecorroborated by analyses of repeated psycho-logical testing (Wilkinson & Sanchez-Craig,1981). Clients reporting abstinence or con-sumption of less than one drink per dayshowed improvement in cognitive functioning,whereas clients reporting an average con-sumption of four drinks per day failed to im-prove.

Method

Selection Criteria and Assignmentto Treatment

The 70 clients were selected to participate in a treatmentprogram described as being suitable for heavy drinkerswhose consumption of alcohol had recently begun to in-terfere with work, studies, or other important life areas.In order to ensure that those admitted to the program

had an alcohol problem at an early stage, a set of screeningcriteria were provided to the Assessment Unit of the Ad-diction Research Foundation (ARF). The criteria servedto identify socially stable persons with relatively short his-tories of problem drinking, normal cognitive functioning,and no evidence of medical problems for which moderateuse of alcohol would be contraindicated. Potential can-didates had to agree to attend an average of eight outpatienttherapy sessions and to provide the name of two collateralswho would be willing to discuss the client's progress. (Fora detailed description of these criteria see Sanchez-Craig,1980.)

Clients were recruited from the population presentingfor treatment to ARF, and through advertisements of theprogram in local newspapers in which no mention wasmade of treatment goals. Referrals were also accepted fromthe Donwood Institute in Toronto. Upon a comprehensivemedical and psychosocial assessment, persons meeting theselection criteria were invited to participate in the study;all those approached agreed to participate. The assessmentworker outlined the objectives of the program (except fordrinking goals), and a statement of consent was signed byeach client.

Clients were assigned on a random basis to a goal ofabstinence or controlled drinking. Within each goal con-dition, clients were assigned at random to one of two ther-apists, both of whom had formal training in psychotherapy.Clients were unaware of the alternative treatment conditionand were not informed of their drinking goal until thefirst counseling session. Clients assigned to a goal of ab-stinence were not allowed during treatment to change theirgoal. However, due to ethical considerations clients werepermitted to shift from a goal of controlled drinking toa goal of abstinence. Initial assessment and assignment ofclients to conditions were conducted independently of thetherapists.

Initial Assessment and ClientCharacteristics

The information required to screen clients was obtainedby means of various standard procedures. A physicianassigned to the project ensured that there were no medicalcontraindications for a client to participate in the study.Liver function was assessed by routine laboratory testsincluding determination of serum gamma-glutamyl trans-peptidase (GGT) and serum glutamic-oxalacetic trans-aminase (SGOT). Information was recorded on demo-graphic characteristics, history of previous alcoholismtreatment, consumption of other drugs, and social andvocational status. The Lifetime Drinking History (LDH;Skinner & Sheu, 1982) questionnaire was used to obtaininformation on drinking patterns in terms of typical quan-tities, frequencies, and beverages consumed. A quantitativeindex of consequences related to alcohol abuse was ob-tained by the Michigan Alcoholism Screening Test (MAST,Selzer, 1971; Skinner, 1979), and an index of severity ofalcohol dependence was provided by the Alcohol Depen-dence Scale (ADS; Skinner & Allen, 1982). Estimates ofcognitive functioning were obtained by applying the Ra-ven's Progressive Matrices, a test of visuospatial problemsolving (Raven, 1960); the Benton Visual Retention Test,a measure of short-term memory (Benton, 1963); the DigitSymbol Substitution Test, a subscale of the Wechsler Adult

392 SANCHEZ-CRAIG, ANNIS, HORNET, AND MAcDONALD

Intelligence Scale (WAIS; Wechsler, 1958); and the ClarkeWAIS Vocabulary Test, a measure of verbal ability (Paitich,1970). Functions assessed by the first three tests of cognitivefunctioning tend to be impaired in chronic alcoholics. Incontrast, performance on tests of vocabulary tends to benormal in chronic alcoholics (Parsons & Farr, 1982; Wil-kinson & Carlen, 1980).

The characteristics of clients assigned to the abstinence(AB) and the controlled drinking (CD) condition are pre-sented in Table 1. As a check on the randomization pro-cedures, / tests were performed on the continuous variablesand chi-square tests on the categorical variables. The groupswere found not to differ significantly in social and de-mographic characteristics, level of cognitive functioning,

Table 1Characteristics of Clients in Both Treatment Groups

Variable

Social and

Age (years)MSD

SexMalesFemales

Marital statusSingleDivorced/separatedMarried/common-law

Present accommodationRents house/apartmentOwns house

Years of educationMSD

Weeks worked full time(past 6 months)

MSD

Present income(thousands/year)

MSD

C o E n i ti vc

Abstinence(N = 35)

demographic

33.49.4

71.4%28.6%

31.4%11.4%57.2%

65.7%34.3%

14.53.0

23.26.8

18.415.2

fiinftmnina

Controlleddrinking(N = 35)

36.211.5

77.1%22.9%

40.0%22.9%37.1%

60.0%40.0%

13.33.2

24.14.9

18.511.7

Raven's Progressive MatricesMSD

Clarke WAIS/VocabularyTest

MSD

Digit Symbol TestMSD

Benton Visual RetentionTest

MSD

85.716.4

12.41.9

11.92.3

7.91.3

84.820.6

12.31.9

11.62.0

8.11.3

VariableAbstinence(N = 35)


Alcohol use

Years of problem drinkingMSD

Quantity (past 3 months)MSD

Frequency (past 3 months)MSD

Weekly quantity(past 3 months)

MSD

Level and severity ofdrinking

AbstinentModerate/with problemsModerate/without

problemsHeavy/without problemsHeavy/with problems

Modal beverageBeer drinkersWine drinkersLiquor drinkersDrinks more than one

above

Consequences

MAST scoreMSD

ADS scoreMSD

5.22.6

9.55.4

5.52.0

51.529.3

0.0%0.0%

11.4%0.0%

88.6%

37.1%2.9%

45.7%

14.3%

of alcohol use

19.06.3

14.36.8

4.83.8

9.64.3

5.42.0

50.427.5

0.0%0.0%

11.4%0.0%

88.6%

45.7%2.9%

37.1%

14.3%

18.36.7

13.57.5

Use of other drugs

Cannabis (past year)Barbiturates (past year)Amphetamines (past year)Tranquilizers

(presently using)

31.4%2.9%0.0%

20.0%

40.0%5.7%2.9%

8.6%

Note. WAIS = Wechsler Adult Intelligence Scale; MAST = Michigan Alcoholism Screening Test; ADS = AlcoholDependence Scale.


measures of drinking behavior, alcohol-related conse-quences, and consumption of other drugs (p < .10 in allcases).

The variables related to alcohol use, that is, quantity(mean drinks per drinking day), frequency (mean daysdrinking per week), and weekly quantity (a combined indexof quantity-frequency), were estimated for the 3 monthsprior to treatment. The investigators considered that thisperiod was long enough to detect low-frequency inter-mittent drinking and short enough to avoid problems ofrecall. For the subjects, one drink was denned as 1.5 ozof liquor 40%, 5 oz of wine 12%, 3 oz of fortified wine20%, and 12 oz of beer 5%. Each one of these units containsapproximately 0.60 oz (13.6 g) of ethanol. The variablelevel and severity of drinking was constructed using a meanof 20 drinks per week to separate "moderate" from "heavy"drinkers; the modifier with problems indicated a positiveanswer to one or more items of the MAST or the ADSconcerned with having experienced blackouts or shakes,having missed meals or work because of drinking, or havingbeen frequently intoxicated.

In view of the controversy surrounding controlleddrinking, it is important to note that clients scored sig-nificantly lower on both the MAST and ADS than an ARFoutpatient sample receiving individual counseling and aninpatient sample receiving conventional alcoholism treat-ment (Skinner & Allen 1982; Skinner & Horn, 1984).Compared to the ARF population and to populationstreated in traditional alcoholism clinics in the United States,the sample of this study is less deteriorated. Nonetheless,the mean score of the clients is substantially above thecutoff for "alcoholism" proposed by Selzer (1971). Con-sumption of drugs other than alcohol during the year pre-ceding treatment was rare; clients using tranquilizers (ex-cept for one CD client) reported taking them as prescribed.

Treatment Procedures

The treatments applied in the AB and CD conditionswere identical, except for the training in controlled drinkingthat was introduced in the fourth treatment session toclients in the CD condition. The treatments were appliedindividually in weekly sessions of approximately 90 min.Their basic components were as follows:

Introduction. The first session began with an outlineof the objectives of the program. To summarize, clientswere told that the aim of the program was to teach methodsfor identifying situations associated with excessive drinkingand for developing more appropriate coping responses.According to the assignment, clients were notified thatabstinence or moderation would be the objective of treat-ment and were asked to rate the acceptability of the assignedgoal. Five of the 35 CD clients rejected the goal of mod-eration, and they were told that the procedures in theprogram were also applicable to a goal of abstinence andwere allowed to pursue this goal. Of the 35 AB clients, 23either rejected the goal or expressed reservations about it.Characteristic expressions were "I cannot accept that 1have to give up alcohol altogether" or "I can accept ab-stinence temporarily, but not for the rest of my life." Theseclients were told that the therapist could only teach themstrategies to achieve abstinence. Clients were advised thatthey would be required to complete homework assignmentsand to attend therapy sessions regularly and that failure

to comply with these requirements could result in dischargefrom the program.

Request for an initial period of abstinence (CD conditiononly). Clients assigned to controlled drinking were re-quested to abstain for the first 3 weeks of treatment. Thefollowing rationales were given: Regular consumption ofhigh levels of alcohol tends to dull cognitive abilities thattypically recover again after 3 weeks of abstinence. Becausethe program emphasizes the teaching of principles of self-control, optimum cognitive functioning is desirable. Inaddition, this brief period of abstinence should give anopportunity to identify strategies spontaneously used tocope with urges to drink and social pressures. Recognitionof such strategies would probably save therapy time andeffort because one could capitalize upon existing com-petencies. The AB clients were given the same rationaleswithout a time limit.

Self-monitoring of drinking behavior. During the firstsession all clients were introduced to a drinking diary inwhich they were instructed to record their abstinent anddrinking days. For drinking days space was provided tospecify the number of drinks consumed, types of beverages,drinking companions and environments, and times ofdrinking. The completeness of the records was checkedby the therapist each week, and the information was usedduring treatment to identify risk situations and to developcoping strategies.

Identification of risk situations and existing competen-cies. During the first session all clients were also instructedin the use of two questionnaires: The Drinking EpisodesQuestionnaire, designed to specify incidents of problemdrinking, and the Urges to Drink Questionnaire, designedto identify successful coping with urges to drink. Incidentsof problem drinking were specified in terms of their an-tecedents (physical and social context, client's feeling andthoughts, and functions attributed to alcohol) and con-sequences (those experienced after consuming two-threedrinks and after the drinking episode). As part of home-work, clients were instructed to describe as many incidentsfrom the past as they could remember well. Such descrip-tions were used during treatment to give clients practicein problem solving (i.e., for each incident clients wereasked to generate cognitive and behavioral responses thatcould have prevented excessive drinking). Once a numberof descriptions had accumulated, the therapist used themto identify areas of risk and to ensure that responses foreffective coping were developed. Existing competencies wereidentified by asking clients to record incidents where theyhad been able to cope with strong desires to drink. Thisrequest was made during the first 3 weeks of treatmentwhen all clients were expected to abstain. Such incidentswere specified in terms of their antecedents and the cog-nitive and behavioral responses employed to forestalldrinking. During treatment clients were asked to commenton the appropriateness and effectiveness of their responses.Where responses produced undesirable consequences (e.g.,alienation of friends), alternative responses were considered.

Development of coping strategies. Because most of theincidents of problem drinking were associated with negativeemotions and interpersonal problems, two strategies wereused extensively to deal with these kinds of problems.Both strategies encompassed the same series of steps butdiffered in the types of coping responses that they weremainly designed to develop. The strategy for interpersonal


problems emphasized the development of behavioral copingdesigned to produce direct changes in the social environ-ment. The development of new behaviors was facilitatedby the reappraisal of the aversive social stimuli and theclient's responses, including the use of alcohol to cope.The strategy for negative emotions emphasized the de-velopment of cognitive coping (i.e., self-statements) de-signed to serve the following functions: (a) reappraise neg-ative emotional states and the role attributed to alcohol,(b) counteract rationalizations for drinking, (c) considerthe consequences of drinking, and (d) initiate actions otherthan drinking. This strategy was mainly applied to situ-ations where clients could not pin-point the source of thenegative emotions. Clients were also taught to use thisstrategy to deal with unexpected urges to break the drinkingrules and with the negative feelings and thoughts typicallyassociated with relapses (Marlatt, 1978). Other problemsidentified in treatment (e.g., use of alcohol as a medicationor as a means of recreation, financial difficulties) werehandled by a general problem-solving strategy similar tothe problem-solving procedures described by D'Zurillaand Goldfried (1971). This more general strategy attemptedto develop in clients the attribute of approaching problemsof daily living in an objective manner, of entertaining avariety of possible solutions, and of systematically assessingthe effectiveness of selected solutions.

Controlled-drinking training (CD condition only). Partof the fourth session was spent in formulating the patternof moderate drinking and in identifying aids to facilitateadherence to the selected pattern. Clients were advisedthat moderate drinking typically involves abstinent days,consumption of no more than four drinks on drinkingdays, and consumption that does not exceed 20 drinksper week. Clients were then helped to specify the frequencywith which they would drink, the maximum number ofdrinks they would consume on drinking days, and thetypes of beverages they would drink. Guidelines that clientswere asked to adopt were as follows: Avoid drinking inthe morning, avoid drinking in situations where drinkinghad been a problem, and avoid using alcohol as a copingdevice. Aids to facilitate moderate drinking included thefollowing: pacing of drinking (e.g., by measuring, sipping,and spacing drinks), preparing in advance to avoid heavydrinking (e.g., by deciding on the number of drinks thatwould be consumed on a given situation, by devising strat-egies for coping with potential difficulties, by solicitinghelp from others), and developing enjoyable activities tofill the times previously spent in heavy drinking. The ther-apist ensured that drinking patterns were congruent withthe client's life-style. As treatment progressed drinkingpatterns were modified, when necessary, in order to makethem more suitable.

Termination- Treatment was terminated when clientsachieved proficiency in problem solving and self-moni-toring procedures. Proficiency in problem solving was as-sessed through hypothetical examples. At the end of treat-ment, clients were informed of the follow-up procedures.They were advised of the importance of maintainingdrinking records and were instructed to bring these recordswith them to each follow-up appointment. Clients werealso told that the therapist would be available, if theyrequired further consultation.

Follow-Up Assessments

During the first 6 months after discharge from treatment,attempts were made to interview clients monthly and,thereafter, at 12, 18, and 24 months. All interviews wereconducted face-to-face, and information was obtained forthe interval since discharge, or since the last follow-upperiod. Areas assessed included employment, marital sta-tus, type of accommodation, treatment received from thetherapist or other professionals, consumption of otherdrugs, drinking contexts, consequences of drinking, self-ratings of improvement as a result of treatment (drinkingproblem was improved, did not change, got worse) andself-ratings of satisfaction about the job, relationship tosignificant others, leisure activities, physical health, andthe helpfulness of treatment (a 5-point scale was used,with 5 indicating extreme satisfaction). Consequences ofdrinking were assessed on items drawn from Armor, Polich,and Stambul (1976). Information on drinking was recordedon an abbreviated form of the Lifetime Drinking Historyquestionnaire. The Digit Symbol Substitution Test wasreadministered 3 months after discharge from treatmentto corroborate self-reported drinking. The levels of GGT,SGOT, and HDLc (high-density lipoproteins cholesterol)were also determined at various follow-up points to beused as potential corroborators of self-reported drinking.These tests of liver function have been shown to be sensitiveto levels of reported alcohol consumption (Castelli et al.,1977; Pomerleau et al., 1978; Reyes & Miller, 1980).

At the end of 24-month follow-up period, clients weredebriefed about the objectives of the study. They were firstasked to report if they knew of the alternative treatmentgoal. All clients were then informed of the alternative optionand were told that training in such an alternative wasavailable if they wished to consider it.

Results

Treatment-Related Measures and Self-Reported Drinking

As part of the outcome of the study a num-ber of treatment variables were examined todetermine whether differences existed betweenthe groups in amount of counseling received,time spent in treatment, rate of program com-pletion, after-care by therapist and otherprofessionals, and self-ratings of improvementand the helpfulness of treatment. Changes inquantity and frequency of drinking from in-take to the end of treatment were also assessed.The values on the variables examined for bothtreatment groups can be found in Table 2.

Treatment was open ended in the sense thattreatment was not terminated until the clientachieved proficiency in problem solving andself-monitoring procedures. In the AB groupthe number of treatment sessions ranged from2-12, in the CD group, from 3-11. Two AB


subjects and 1 CD subject did not completethe training. Before discharge from treatment,no significant differences were found betweenthe groups in number of counseling sessions,number of weeks in treatment, and rate ofattrition. During the first 6 months after dis-charge from treatment, however, significantlymore AB clients received additional counseling

Table 2Performance of Both Treatment Groups onTreatment-Related Variables

Variable

Within

Counseling sessionsMSD

Weeks in treatmentMSD

Completed treatmentQuantity

MSD

FrequencyMSD

Weekly quantity*MSD

Abstinence

treatment

5.72.0

7.43.4

94.0%

3.83.0

2.42.2

13.617.3

Controlleddrinking

5.61.7

7.32.8

97.0%

3.42.3

1.61.6

6.68.4

Posttreatment discharge

Aftercare by therapists**0 sessions1-2 sessions3-5 sessions

Aftercare by otherprofessionals

0 contacts1-2 contacts3 or more contacts

Self-rating ofimprovement

ImprovedNo changeGot worse

Self-ratings of helpfulnessof treatment

MSD

46.6%20.0%33.3%

86.6%6.7%6.7%

65.5%31.0%

3.4%

4.11.0

65.5%31.0%3.5%

89.7%3.4%6.9%

80.8%15.4%3.8%

4.10.9

Note. Sample sizes for the abstinence and controlled-drinking groups within treatment were 35 and 35, re-spectively; at posttreatment discharge, 30 and 29, respec-tively.* p < .05. ** p < .02.

from their therapist, *2(2, N = 59) = 8.72,p < .02; although one third of AB clients re-quested counseling from three to five times,only 1 CD client requested this amount ofcounseling. On drinking measures (obtaineddirectly from the drinking logs) no significantdifferences were found between the groups forquantity or frequency. However, a significantdifference was obtained on weekly quantity,with the controlled-drinking group reportingfewer drinks during treatment, ?(68) = 2.12,p < .05. On self-ratings of improvement andthe helpfulness of treatment obtained 6months after discharge from treatment, nosignificant differences were found between thegroups.

Pre-post changes in quantity, frequency, andweekly quantity were assessed using separatemixed-model analyses of variance (ANOVAS),with one group factor (abstinence vs. con-trolled drinking) and one repeated measure(intake vs. termination of treatment). For eachof these variables the analyses showed no sig-nificant group effect and Group X Time in-teraction, but a significant time effect forquantity, F(\, 68) = 109.2, p < .001; for fre-quency, F(\, 68) = 138.9, p < .001; and forweekly quantity, F = 175.5, p < .001. Fromintake to the termination of treatment themean number of drinks consumed per drink-ing day was reduced in both groups by ap-proximately six drinks; the frequency ofdrinking per week was reduced by 3 days inthe AB group and by almost 4 days in the CDgroup. On weekly quantity this represents amean reduction of 38 drinks for the AB groupand 44 drinks for the CD group.

Comparison Between Intake and6-Month Outcome

For the following reasons the 6-month out-come was selected for principal comparativeanalyses: First, during this period drinking datawere collected more frequently than in sub-sequent follow-up periods during which clientswere interviewed only once. The mean numberof interviews achieved for CD clients over the6 months was 4.0 (SD = 1.9) for AB clientsand 3.9 (SD = 1.6) for CD clients. Second,70% of the clients had maintained drinkinglogs. During subsequent follow-up periods

396 SANCHEZ-CRAIG, ANNIS, BORNET, AND MACDONALD

drinking logs had been dropped by most clientsand consumption data had to be obtained bythe recall method. Evidence from the presentstudy has shown that data obtained fromdrinking logs are significantly more valid thandata obtained by recall (Sanchez-Craig & An-nis, 1982b). Third, 6-month data could beconsidered to be less affected by events of dailyliving (e.g., changes in employment or maritalstatus, health problems), which may have sig-nificant influence on drinking behavior. Fi-nally, a higher rate of client follow-up period(84%) was achieved at 6 months than at anysubsequent period.

Included in these analyses were 59 clients.From the initial sample of 70, 5 AB clients (4males, 1 female) and 6 CD clients (5 males,1 female) were not interviewed. Four of theseclients could not be located, 4 refused to col-laborate, and 3 kept postponing their appoint-ments.

Self-reported drinking. To test for differ-ences in drinking behavior between intake andthe 6-month follow-up period, the followingvariables were examined: quantity, frequency,weekly quantity, and level and severity ofdrinking. The values on these variables forboth groups at 6 months after discharge fromtreatment can be found in Table 3. The firstthree variables were analyzed using mixed-model ANOVAS. For each variable the analysesshowed no significant group effect andGroup X Time interaction, but did show ahighly significant time effect for quantity, F( 1,57) = 153.3, p < .001; for frequency, 7^(1,57) = 71.2, p < .001; and for weekly quantity,F = 13 3.6, p < .001. From intake to 6 monthsthe mean number of drinks consumed perdrinking day was reduced from almost 10drinks to 4 drinks; the frequency of drinkingper week was reduced in each group fromabout 5.5 days to 3 days. On weekly quantitythis represents a mean reduction of 39 drinksfor the AB group and 36 drinks for the CDgroup.

On level and severity of drinking, chi-squaretests showed no significant differences betweenthe groups at the 6-month follow-up period.At 6 months (like at intake), a mean of 20drinks per week was used as a cutoff to separate"moderate" from "heavy" drinkers. To be cat-egorized as a moderate or a heavy drinker"without problems," the client was allowed to

Table 3Drinking Outcome and Social Adjustment atDischarge 6 Months After Treatment

VariableAbstinence(N = 30)


Alcohol useQuantity

MSD

FrequencyMSD

Weekly quantityMSD

Level and severity ofdrinking

AbstinentModerate/problem

freeModerate/with

problemsHeavy/problem freeHeavy/with problems

Modal beverageBeer drinkersWine drinkersLiquor drinkersDrinks more than one

above

Social

Weeks worked full timeMSD

Marital statusSingleDivorced/separatedMarried/common-law

Present accommodationRents house/

apartmentOwns house

Self-ratings ofsatisfaction

JobMSD

LeisureMSD

Relationships tospouse/parent

MSD

Physical healthMSD

3.72.4

2.81.8

12.413.5

6.7%

43.3%

23.3%0.0%

26.7%

0.6%25.0%14.3%

7.1%

adjustment

22.08.5

36.7%13.3%50.0%

46.7%53.3%

3.41.1

2.91.1

3.71.1

4.10.8

4.22.2

3.12.2

14.512.7

3.4%

37.9%

31.0%17.2%10.3%

57.1%14.3%25.0%

4.6%

22.98.1

27.6%24.1%48.3%

37.9%62.1%

3.51.4

3.21.2

3.81.0

3.70.9


report up to six episodes of intoxication duringthe 6-month period, but none of the followingconsequences of heavy drinking: blackouts,shakes, morning drinking, or missed meals orwork because of drinking. Pre-post changeswere assessed using Wilcoxon matched-pairssigned-rank tests. For both groups the testsshowed that from intake to the 6-month follow-up time there was a significant shift from the"heavy/with problems" category into the twomoderate categories (for the AB group, z =4.02, p < .001; for the CD group, z = 4.37,p < .001).

The approximate quantities and frequenciesof drinking reported by clients in the variouscategories were as follows: moderate/withoutproblems, 2.5 drinks, 2 days per week; mod-erate/with problems, 4.5 drinks, 3 days perweek; heavy/without problems, 5 drinks, 6days per week; and heavy/with problems, 7drinks, 5 days per week.

Measures of social adjustment. The vari-ables selected to represent social adjustmentcan be found in Table 3. Differences in weeksworked full time were tested using a mixed-model ANOVA. This analysis showed no sig-nificant effect for group, time, and for theGroup X Time interaction. Similarly, no sig-nificant changes were observed in either groupfrom intake to the 6-month follow-up time intype of accommodation (as indicated byMcNemar tests) or in marital status (as in-dicated by Wilcoxon tests). To test for changesin marital status the proportion of clients whowere married or living in common-law werecompared to the proportion of clients whowere single, divorced, or separated. In self-ratings of satisfaction about the job, leisureactivities, relationship to spouse/parent, andphysical health, t tests showed no significantdifferences between the groups.

Comparison Between 6-Month Outcome andOutcomes at 12, 18, and 24 Months

Follow-up rates. Follow-up rates droppedfrom 84% at 6 months to 74% at 12 months(25 AB and 27 CD clients), 74% at 18 months(26 AB and 26 CD clients), and 73% at 24months (26 AB and 25 CD clients). No sig-nificant differences were observed betweenconditions in rates of follow-up for any of theperiods studied. (2 CD clients died before the

18-month follow-up time of causes unrelatedto posttreatment drinking.)

In order to test for the association betweenclient characteristics and compliance with fol-low-up procedures, three groups of clients werestudied: those attending all four follow-up ses-sions (n = 43), those attending some (n =18), and those attending none (n = 9). One-way ANOVAS conducted on the intake andtreatment variables included in Tables 1 and2 showed significant differences only for theMAST score, F(2, 67) = 4.21, p < .02, andnumber of counseling sessions, F(2,67) = 3.52,p < .001. Clients who complied with all fourfollow-up interviews had significantly lowerscores on the MAST and attended significantlymore counseling sessions than did clients whohad no follow-up interview.

Self-reported drinking. In order to answerthe question of whether the drinking levelsobserved at 6 months were also observed onsubsequent follow-up periods, a multivariateapproach for repeated measures was used toexamine the drinking variables involved inthe 6-month outcome. Included in these anal-yses were the 43 clients who complied withthe four follow-up interviews (21 AB and 22CD clients). Among the 16 clients excludedfrom this analysis, 9 were interviewed threetimes, 5 were interviewed twice, and 2 wereinterviewed only once. In terms of intake andtreatment characteristics, the excluded groupdiffered only in having a higher proportion ofsingle/divorced/separated clients (x2 = 3.88,N = 70, p < .05); there were no differencesin alcohol-related measures.

Changes in quantity, frequency, and weeklyquantity were assessed using multivariateanalyses of variance (MANOVAS) for repeatedmeasurements. For each variable the measuresof the four follow-up periods were transformedinto the following variables: (a) the overallmean, (b) the difference between 12 and 18months, (c) the difference between 24 monthsand the average of 12 and 18 months, and (d)the difference between 6 months and the av-erage of 12, 18, and 24 months. The last threetransformations constitute a variation of theHelmert contrasts (Bock, 1975).

For quantity, the analysis showed that thedifferences among the four follow-up periodswere not statistically significant for both treat-ment groups. Moreover, the groups did not


differ significantly in mean quantity over the2 years. In other words, in both treatment con-ditions the mean number of drinks consumedper drinking day remained around four drinksfrom 6 months to the end of the second year.For frequency the analysis showed that thedifference among the four follow-up periodswas statistically significant. The significant dif-ference was found between the frequency ofdrinking reported at 6 months and the meanfrequency obtained for the last three follow-up periods, F(l, 41) = 6.43, p < .02. In bothgroups, there was an increment of 0.45 (SE =. 18) for the last three follow-up periods, whichrepresents less than half a day per week. How-ever, the mean frequency over the 2 years didnot differ between the groups. No significantdifferences for group and time were observedfor weekly quantity.

To test for changes in level and severity ofdrinking, Friedman two-way ANOVAS wereconducted for each treatment group. Theseanalyses indicated that from 6 months to theend of the second year no significant changesoccurred in either group in proportion ofclients categorized as abstinent, moderate/withproblems or without problems, and heavy/withor without problems.

Measures of social adjustment. A MANOVAfor repeated measurements was used to analyzeweeks worked full time. This analysis paral-leled the procedures used in analyses of theconsumption variables previously reported.The analysis showed that the difference amongthe four follow-up periods was statistically sig-nificant. The significant difference was foundbetween weeks worked full time at 6 monthsand the mean of the last three follow-up pe-riods, F(l, 41) = 5.20, p < .05. In both con-ditions after 6 months the mean number ofweeks worked full time decreased by 1.53(SE = .67). However, the mean weeks workedfull time over the 2 years did not differ betweenthe groups. Cochran Q tests conducted for eachtreatment condition indicated no significantchanges from 6 months to the end of the secondyear in marital status or in type of accom-modation.

Corroborators of Self-Reported DrinkingCollateral reports. Clients were asked to

provide the names of two collaterals who could

be contacted about their drinking over the fol-low-up period. Where two collateral reportswere available, the report of the collateral whohad more frequent contact with the client wasselected for the analysis. If the frequency ofcontact of both collaterals was the same, thecollateral who provided the most complete setof data was selected. At the 6-month follow-up period, collateral reports were available for38 clients; 68% of these collaterals had dailycontact with the client. At the 12-month fol-low-up period, reports were available for only23 clients, with 64% of these collaterals havingdaily contact with the client. Comparativeanalyses of the reports by clients and collateralsat 6 and 12 months were conducted on thefollowing variables: quantity/day, frequency ofdrinking/month, number of times intoxicated/past 6 months, drinking style (abstinent, socialdrinking, problem drinking on sprees, andsteady problem drinking). The first three vari-ables were examined using mixed-model AN-OVAS, with one group factor (daily contact pairvs. nondaily contact pair) and one repeatedmeasure (client self-reports and collateral re-ports). This model was considered most ap-propriate because it allows comparisons of dif-ferences for individual pairs. Analyses at 6 and12 months indicated no significant differencein the reports of clients and their collateralson any of the three variables, for both dailyand nondaily contact groups. In addition, theaverage of the paired reports did not differbetween the two groups. On drinking style,Cohen's Kappa showed a moderate degree ofagreement between client and collateral reportsat 6 months (K = .48); there was 100% agree-ment between collateral and self-reports forthe four abstinent clients, 67% agreement forclients reporting social drinking and problemdrinking on sprees, and no agreement withthe report of the 1 client reporting steadyproblem drinking. At 12 months, Cohen'sKappa indicated no agreement between clientand collateral reports of drinking style.

Tests of cognitive functioning. The DigitSymbol Substitution Test was administered atadmission and at the 3-month follow-up pe-riod. All scores were age scaled. Based on thenumber of drinks reported for the 3 weeksprior to retesting, 39 males and 13 femaleswere categorized as abstinent, moderate (M =


29.6, range 1-63 drinks) or heavy drinkers(M= 89.7, range 66-136 drinks). Scores atthe 3-month follow-up period were analyzedusing an analysis of covariance (ANCOVA), withthe intake Digit Symbol scores as covariate.The analysis showed a significant differenceamong the drinking categories, F(2,49) = 4.52,p < .02. On average, the abstinent and mod-erate groups improved by 1.9 and 1.8, whereasthe heavy-drinking group only improved by0.40.

Biochemical markers. As further corrob-oration of self-reported drinking, GOT, SGOT,and HDLc were determined at various follow-up points, whenever clients agreed to give ablood sample. The advantage of a compositeindex of GGT and HDLc in discriminatingabstinent/light, moderate, and heavy drinkershas already been published (Sanchez-Craig &Annis, 1981). These three groups were formedaccording to the number of drinks reportedby 40 males at 6-12 months for the 3 weeksprior to the blood test. Although GGT wasfound to be more sensitive in detecting theabstinent/light and the heavy-drinking group,HDLc was more sensitive in detecting themoderate group. The overall rate of correctclassifications was 62.5%. The advantage ofcombining GGT and HDLc was indicatedagain with data obtained from 31 males at the18-24-months follow-up periods. The numberof drinks reported over the 3 weeks prior tothe blood test were found to be significantlycorrelated with levels of GGT, r(29) = .44,p < .01, but not with levels of HDLc, whichtended to be elevated in moderate drinkers.The SGOT was insensitive to drinking re-ported by males at any point. Also, no sig-nificant relationships were found between lev-els of the three biochemical indexes anddrinking reported by females (N = 15 at 6-12 months; N = 12 at 18-24 months).

DiscussionThe results of this study did not support

the hypothesis that assignment to a goal ofcontrolled drinking would produce a betteroutcome than assignment to a goal of absti-nence. The AB and CD clients were equallysuccessful in reducing their drinking to mod-erate levels during the 2-year follow-up period.This finding was validated using three sets of

independent corroborators of alcohol con-sumption. However, it can be argued that con-trolled drinking was a more suitable goal be-cause (a) it was more acceptable for the ma-jority of the clients, (b) CD clients drank lessduring treatment, (c) AB clients developedmoderate drinking on their own, and (d) moreAB clients requested and received additionalcounseling from their therapist after dischargefrom the program.

The absence of an experimental effect maybe attributed to the manner in which the treat-ment for abstinence was presented, to the na-ture of some of the tasks involved in the train-ing, or to the clients' determination to drinkin moderation independently of assignment togoal. It may be argued that the abstinencetreatment was not presented in the strongestpossible manner. Although AB clients werealways told that abstinence was the goal of theprogram, they were not told that their problemwas a disease or that lifelong abstinence wasessential. If such an approach had been fol-lowed, an experimental effect might have oc-curred because of an increased dropout ratein the AB group. Pomerleau et al. (1978) havedocumented such an effect. It is also possiblethat some of the tasks permitted AB clientsto infer that moderation was acceptable. Forexample, the specification of incidents ofproblem drinking involved identifying con-sequences experienced after consumption oftwo-three drinks (usually positive conse-quences) and after the drinking episode (usu-ally negative consequences); based on self-monitoring records, therapists tended to applythe problem-solving strategies to those days inwhich drinking had been heaviest. These pro-cedures may have encouraged clients to assumethat the therapist accepted drinking that oc-curred in moderation. An admission criterionwas that the applicant should believe thatdrinking in moderation was an achievable goal.Most of the clients assigned to abstinence re-jected this goal from the outset. Hence, it ispossible that AB clients took from the programonly those notions that would help themachieve a goal that they always had in mind.An important question to raise is why the spe-cific training in controlled drinking did notconfer an advantage to the CD clients afterthe completion of treatment.


The high rate of program completion inboth groups and the reduction in alcohol con-sumption achieved by many AB and CDclients suggests that the treatments offered wereacceptable. At the end of treatment alcoholconsumption had been reduced by 74% in theAB group and by 87% in the CD group. Thegreater reduction in the CD group can be ex-plained by the fact that during the first 3 weeksof treatment more of the CD clients (40% vs.24%) complied with the request of abstinence,and fewer of them reported heavy drinking(Sanchez-Craig, 1980).

Overall, the percentage of successful clients(abstinent plus moderate drinkers) in the AB(73%) and the CD (72%) group is comparableto percentages reported in studies in whichselected socially stable problem drinkers weretrained in controlled-drinking methods (Heath-er & Robertson, 1981; Miller & Hester, 1980).These percentages would still be comparableif a reduction of about 10% is made to adjustthe success rate of this study, so as to reflectclients missing at follow-up interviews as"failures." It should be noted, however, thatthe average intake level of consumption re-ported for clients in the present study (51drinks per week or 693.6 g of ethanol) is higherthan the average levels reported for clients inother controlled-drinking studies (Alden, 1978;Lovibond, 1975; Miller, 1978; Miller, Grib-skov, & Mortell, 1981; Miller, Pechacek, &Hamburg, 1981; Miller & Taylor, 1980; Miller,Taylor, & West, 1980; Pomerleau et al., 1978;Vogler, Weissbach, & Compton, 1977). Whenadjustments are made for differences betweencountries in the amount of ethanol allowed ina standard drink (0.50 oz [ 11.7 g] in the UnitedStates and 0.60 oz [13.5 g] in Canada), theaverage intake level for clients in the presentstudy is from 10% to 40% higher. It wouldappear that controlled-drinking strategies maybe equally effective with early-stage problemdrinkers who have fairly high intake con-sumption levels.

For most outcome measures there was auniform stability for both groups over the 2-year follow-up period. Measures of social ad-justment were stable, except for a decrementin the number of weeks worked full time ob-served after the 6-month follow-up period.

This decrement, however, was small (1.5weeks) and may be accounted for by the wor-sening economic conditions within the countryduring the period of 1980-1982, when mostof the follow-up sessions at 12, 18, and 24months were conducted. On the consumptionmeasures, although quantity remained stableat about four drinks per drinking day, fre-quency showed a significant increment, raisingfrom an average of 2.8 days per week at 6months to 3.3 days for all successive follow-up periods combined. This increment, how-ever, did not significantly affect the meanweekly quantity, which remained stable atabout 13 or 14 drinks per week on average.This finding suggests the importance of ex-amining quantity and frequency separately, aswell as combined, in treatment outcome re-search. In the present study it is difficult tosay what the clinical significance might be ofthe small increment in mean frequency ofdrinking per week (0.45 days) observed for thelast three follow-up periods combined. It maysimply reflect the consolidation of new drink-ing practices following a phase of self-exper-imentation after treatment. Another possibleexplanation for the disparity may be that self-estimates of how often one consumes alcohol(frequency) and how much is actually con-sumed on those occasions (quantity) tend tobe orthogonal dimensions, with frequency in-creasing and quantity decreasing with age(Hartford & Mills, 1978; Vogel-Sprott, 1974).

The independent corroborators of alcoholconsumption employed in the study providedconsiderable support to the validity of clients'reported drinking. Valuable informationgained from the use of the corroborators wasthe finding that consumption of an average offour drinks per day precluded significant im-provement on tests of cognitive functioning,both during treatment (Wilkinson & Sanchez-Craig, 1981) and at the 3-month follow-upperiod. Similarly, this level of consumptionwas associated with significant but not ab-normal elevations of GGT. This suggests thatin controlled drinking conservative goalsshould be recommended to clients. Both theDigit Symbol Test and the biochemical mark-ers used in the study appear to have sufficientsensitivity to validate group data on recent


consumption, but not to corroborate drinkingfor individuals. Because most clients were in-terested in the results of the tests of cognitiveand liver function, the potential effect that ob-jective feedback of this sort may have in mo-tivating some problem drinkers to maintainabstinence or low levels of alcohol consump-tion is worth investigating.

The rates observed in this study of partic-ipation in follow-up interviews (between 73%and 84%) were lower than initially anticipated.Compared to more chronic problem drinkerswith whom the authors had previous experi-ence (Sanchez-Craig & Walker, 1982) the pres-ent sample tended to be less compliant withfollow-up expectations and to be more likelyto protest intrusions into their times and pri-vacy. The rate of participation in follow-upinterviews could have been higher if some in-terviews had been conducted over the tele-phone. However, a decision was made at theinception of treatment not to conduct tele-phone interviews because they would precludeobtaining independent corroborators of con-sumption. In retrospect, perhaps this decisionwas too restrictive. At least for clients who didnot directly refuse to cooperate with follow-up requests, information could have been ob-tained over the telephone.

In summary, the results of this study in-dicate that most CD clients achieved mod-eration of alcohol use and that most AB clientsfailed to abstain but nonetheless moderatedtheir drinking. Although no significant differ-ences were found between the AB and the CDgroup in amount of reported alcohol con-sumption over the 2-year follow-up period, aquestion that remains unanswered is whetherthe drinking outcome in either group is mainlyattributable to client background character-istics or to program factors (e.g., assignmentto goal, drinking during program). In an initialevaluation of this study (Sanchez-Craig & An-nis, 1982a), a significant positive relationshipwas observed between drinking (i.e., meanweekly quantity) reported at intake and at the6-month follow-up period in the AB group.In the CD group, however, the two sets of mea-sures were not significantly correlated. Thisfinding suggests that different factors may becontributing to the similar outcomes of the

groups. In subsequent analyses attempts willbe made to identify predictors of outcomedrinking.

Epidemiological studies indicate that thenumber of persons drinking at the level of theclients in this study constitute the larger pro-portion of the problem-drinking population(Cahalan, Cisin, & Crossley, 1969; Cahalan &Room, 1974). However, treatment proceduresaimed at this group are very scarce. The ma-jority of the treatment services are explicitlydirected to alcoholics and tend to place de-mands on their clients, such as acknowledge-ment that one is an alcoholic, commitment toa goal of lifelong abstinence, and absence fromwork or family responsibilities to participatein treatment. These demands tend to be un-attractive or unacceptable to the type of personrecruited into the present program. Becausethe program was very successful in retainingclients in treatment and produced a satisfac-tory outcome for the large majority of the par-ticipants, similar programs are indicated as ameasure of secondary prevention of alcoholproblems. Further research is clearly neededto assess the minimum intervention that wouldbe effective in treating this population.

References

Alden, L. (1978). Evaluation of preventive self-managementprogramme for problem drinkers. Canadian Journal ofBehavioural Science, 10, 258-263.

Armor, D. J., Polich, M., & Stambul, H. B. (1976). Al-coholism and treatment. Santa Monica, CA: Rand Cor-poration.

Arnold, M. B. (1960). Emotions and personality (Vols. 1and 2). New York: Columbia University Press.

Arnold, M. B. (1969). Human emotion and action. In J.Mischel (Ed.), Human action (pp. 167-197). New York:Academic Press.

Benton, A. (1963). The revised visual and retention test:Clinical and experimental application. Ames: StateUniversity of Iowa.

Bock, R. D. (1975). Multivariate statistical methods inbehavioral research. New York: McGraw-Hill.

Cahalan, D., Cisin, I. H., & Crossley, H. M. (1969). Amer-ican drinking practices. New Brunswick, NJ: RutgersCenter of Alcohol Studies.

Cahalan, D., & Room, R. (1974). Problem drinking amongAmerican men. New Brunswick, NJ: Rutgers Center ofAlcohol Studies.

Castelli, W. P., Gordon, X, Hjortland, M. C, Kagan, A.,Doyle, J. T, Hames, C. G., Hulley, S. B., & Zukel,


W. J. (1977). Alcohol and blood lipids: The cooperativelipoprotein phenotyping study. Lancet, 2, 153-155.

Davis, D. L. (1962). Normal drinking in recovered alcoholaddicts. Quarterly Journal of Studies on Alcohol, 23,94-104.

D'Zurilla, T. J., & Goldfried, M. R. (i 971). Problem solvingand behavior modification. Journal of Abnormal Psy-chology, 78, 107-126.

Emerick, C. D. (1974). Areview of psychologically orientedtreatment of alcoholism: I. The use and interrelation-ships of outcome criteria and drinking behavior followingtreatment. Quarterly Journal of Studies on Alcohol, 35,523-549.

Foy, D. W., Rychtarik, R. G., & Nunn, B. L. (1982).Controlled drinking training effects for chronic veteranalcoholics: A randomized trial. Paper presented at theAmerican Psychiatric Association Convention, Toronto.

Gerard, D. L., & Saenger, G. (1966). Outpatient treatmentof alcoholism: A study of outcome and its determinants.Toronto, Canada: University of Toronto Press.

Hartford, I. C., & Mills, G. S. (1978). Age-related trendsin alcohol consumption. Journal of Studies on Alcohol,39, 207-210.

Heather, N., & Robertson, 1. (1981). Controlled drinking.London: Methuen.

Lazarus, R. S. (1966). Psychological stress and the copingprocess. New York: McGraw-Hill.

Lazarus, R. S. (1968). Emotions and adaptation: Con-ceptual and empirical relations. In W. Arnold (Ed.),Nebraska Symposium on Motivation (Vol. 16, pp. 175-270). Lincoln: University of Nebraska Press.

Lovibond, S. H. (1975). Use of behaviour modification inthe reduction of alcohol-related road accidents. In T.Thompson & W. S. Dockers (Eds.), Application of be-haviour modification (pp. 399-406). New York: Aca-demic Press.

Marlatt, G. A. (1978). Craving for alcohol, loss of control,and relapse: A cognitive-behavioral analysis. In P. E.Nathan, G. A. Marlatt, & T. Loberg (Eds.), Alcoholism:New directions in behavioral research and treatment(pp. 271-314). New York: Plenum Press.

Miller, W. R. (1978). Behavioral treatment of problemdrinkers: A comparative outcome study of three con-trolled drinking therapies. Journal of Consulting andClinical Psychology, 46, 74-86.

Miller, W. R., & Caddy, G. R. (1977). Abstinence andcontrolled drinking in the treatment of problem drinkers.The Journal of Studies on Alcohol, 38, 986-1003.

Miller, W. R., Gribskov, C. J., & Mortell, R. L. (1981).Effectiveness of a self-control manual for problemdrinkers with and without therapist contact. The In-ternational Journal of Addictions, 16, 1247-1254.

Miller, W. R., & Hester, R. (1980). Treating the problemdrinker: Modern approaches. In W. R. Miller (Ed.), Theaddictive behaviors: Treatment of alcoholism, drug abuse,smoking and obesity (pp. 11-142). Oxford: Pergamon.

Miller, W. R., Pechacek, T. E, & Hamburg, S. (1981).Group behavior therapy for problem drinkers. The In-ternational Journal of Addictions, 16, 829-839.

Miller, W. R., & Taylor, C. (1980). Relative effectivenessof bibliotherapy, individual and group self-control. Ad-dictive Behaviors, 5, 13-24.

Miller, W. R., Taylor, C. A., & West, J. (1980). Focusedversus broad-spectrum behavior therapy for problemdrinkers. Journal of Consulting and Clinical Psychology,48, 590-601.

Paitich, D. (1970). The Clarke Automated PsychologicalExamination (CAPER). The Ontario Psychologist, 2, 304-314.

Parsons, Q A., & Farr, S. P. (1982). The neuropsychologyof alcohol and drug use. In S. Filskov & T. Boll (Eds.),Handbook of clinical neuropsychology (pp. 320-365).New York: Wiley.

Polich, J. M., Armor, D. J., & Braiker, H. B. (1981). Thecourse of alcoholism: Four years after treatment. NewYork: Wiley.

Pomerleau, O. F, Pertschuck, M., Adkins, D., & Brady,J. P. (1978). A comparison of behavioral and traditionaltreatment methods for middle-income problem drinkers.Journal of Behavioral Medicine, 1, 187-200.

Popham, R. E., & Schmidt, W. (1976). Some factors af-fecting the likelihood of moderate drinking by treatedalcoholics. Journal of Studies on Alcohol, 37, 868-882.

Raven, J. C. (1960). Guide to the standard progressivematrices. London: H. K. Lewis.

Reyes, E., & Miller, W. R. (1980). Serum gamma-glutamyltranspeptidase as a diagnostic aid in problem drinkers.Addictive Behaviors, 5, 59-65.

Sanchez-Craig, M. (1980). Random assignment to absti-nence or controlled drinking in a cognitive-behavioralprogram: Short-term effects on drinking behavior. Ad-dictive Behaviors, 5, 35-39.

Sanchez-Craig, M., & Annis, H. M. (1981). Gamma-glu-tamyl transpeptidase and high-density lipoproteins cho-lesterol in problem drinkers: Advantages of a compositeindex for predicting alcohol consumption. Alcoholism:Clinical and Experimental Research, S, 540-544.

Sanchez-Craig, M., & Annis, H. M. (1982a). Initial eval-uation of a program for early-stage problem drinkers:Randomization to abstinence and controlled drinking.Paper presented at the American Psychological Asso-ciation meeting, Washington, DC. (Available from Ad-diction Research Foundation Library. Substudy No.1251)

Sanchez-Craig, M., & Annis, H. M. (1982b). Self-moni-toring and recall measures of alcohol consumption:Convergent validity with biochemical indices of liverfunction. British Journal of Alcohol and Alcoholism, 17,117-121.

Sanchez-Craig, M., & Walker, K. (1982). Teaching copingskills to chronic alcoholics in a coeducational halfwayhouse: I. Assessment of programme effects. BritishJournal of Addiction, 77, 35-50.

Selzer, M. L. (1971). The Michigan Alcoholism ScreeningTest: The quest for a new diagnostic instrument. Amer-ican Journal of Psychiatry, 127, 1653-1658.

Skinner, H. A. (1979). Multivariate evaluation of theMAST. Journal of Studies on Alcohol, 40, 821-844.

Skinner, H. A., & Allen, B. A. (1982). Alcohol dependencesyndrome: Measurement and validation. Journal of Ab-normal Psychology, 91, 199-209.

Skinner, H. A., & Horn, J. L. (1984). Guidelines for usingthe Alcohol Dependence Scale. Toronto, Canada: Ad-diction Research Foundation.


Skinner, H. A., & Sheu, W. (1982). Reliability of alcoholuse indices: The Lifetime Drinking History and theMAST. Journal of Studies on Alcohol, 43, 1157-1170.

Sobell, M. B., & Sobell, L. C. (1973). Individualized be-havior therapy for alcoholics. Behavior Therapy, 4, 49-72.

Sobell, M. B., & Sobell, L. C. (1976). Second-year treatmentoutcome of alcoholics treated by individualized behaviortherapy: Results. Behavior Research and Therapy, 14,195-215.

Vogel-Sprott, M. (1974). Denning light and heavy socialdrinking: Research implications and hypotheses. Quar-terly Journal of Studies on Alcohol, 35, 1388-1392.

Vogler, R. C., Weissbach, T. A., & Compton, J. V. (1977).Learning techniques for alcohol abuse. Behavior Re-search & Therapy, 15, 31-38.

Wechsler, D. (1958). The measurement and appraisal ofadult intelligence. Baltimore, MD: Williams & Wilkins.

Wilkinson, D. A., & Carlen, P. L. (1980). Relation ofneuropsychological test performance in alcoholics tobrain morphology measured by computed tomography.In H. Begleiter (Ed.), Advances in experimental medicineand biology: Biological effects of alcohol (Vol. 126, pp,683-700). New York: Plenum Press.

Wilkinson, D. A., & Sanchez-Craig, M. (1981). Relevanceof brain dysfunction to treatment objectives: Shouldalcohol-related deficits influence the way we think abouttreatment? Addictive Behaviors, 6, 253-260.

Received August 1, 1983Revision received October 31, 1983

Correction to Knight

In the article "On Interpreting the Several Standard Errors of the WAIS-R:Some Further Tables" by Robert G. Knight (Journal of Consulting and ClinicalPsychology, 1983, Vol. 51, pp. 671-673), the final column of Table 2 is entirelyincorrect. The figures in the fourth column from the right-hand side were erroneouslyrepeated in the final column. A corrected version of Table 2 is presented below.The assistance of Fred M. Grossman in drawing attention to this error is appreciated.

Table 2Abnormality of Verbal-Performance IQ Discrepancies on the WAIS-R

Age range (in years)% population obtaining

a given or a greaterdiscrepancy

50252010521.1

16-17

7.512.914.418.422.026.229.037.1

18-19

7.813.214.818.922.626.829.838.1

20-24'

7.613.014.518.722.326.529.437.5

24-34, 45-54,65-69

6.911.813.216.920.223.826.734.2

35-44

6.310.812.115.518.422.024.431.2

55-64

6.611.412.616.219.323.125.532.7

70-74

7.913.414.919.222.827.129.938.4

Note. WAIS-R = Wechsler Adult Intelligence Scale-Revised.

Documents

Random Assignment to Abstinence and Controlled Drinking