"Education, Faculty of"@en . "Educational and Counselling Psychology, and Special Education (ECPS), Department of"@en . "DSpace"@en . "UBCV"@en . "Olson, Gregory Scott"@en . "2009-02-21T00:00:00"@en . "1993"@en . "Master of Arts - MA"@en . "University of British Columbia"@en . "This study was part of a larger one called The Alcohol Recovery Project. The\r\noverarching aim of the present investigation was to broaden understanding\r\nregarding the role of the therapeutic alliance in the treatment of alcoholics within\r\na systemic perspective. Using Pinsof and Catherall\u00E2\u0080\u0099s (1986) conceptualization of\r\nthe affiance, a process-outcome study was conducted within both individual and\r\nconjoint formats. A sample of 63 alcoholic men and 21 non-alcoholic female\r\npartners was divided into three different treatment groups, two for individuals\r\nand one for couples. Several aspects of the affiance were investigated including:\r\n(a) the strength of the alliance in different treatments; (b) the impact of drinking\r\nbehavior on the alliance; (c) the differential power of the affiance in predicting\r\noutcome in the individual therapeutic format compared to the conjoint one; (d)\r\nthe impact of gender on alliance strength; and (e) the relationship between split\r\namd intact affiances and therapeutic improvement. Results of the study\r\nindicated: (a) that the strength of the affiance was nearly identical in two very\r\ndifferent therapeutic approaches; (b) that binge and chronic alcohol-use patterns\r\nserved as potent predictors of the correlation between alliance and dyadic\r\nadjustment gain; (c) that the strength of the relationship between the affiance and\r\ntherapeutic gain was not significantly different in individual as opposed to\r\nconjoint treatment (recognizing limitations in using the scales to compare\r\nmodalities); (d) that women formed significantly stronger affiances than men;\r\nand (e) that the theoretical constructs of split and intact alliances could be\r\nempirically identified. The complexities inherent in interpreting results\r\nsurrounding a process variable like the affiance became very evident as did the\r\nneed for more study of the construct. Implications of the results were discussed,\r\nand recommendations made for future directions of research."@en . "https://circle.library.ubc.ca/rest/handle/2429/4905?expand=metadata"@en . "3439771 bytes"@en . "application/pdf"@en . "THE ROLE OF THE THERAPEUTIC ALLIANCEIN THE TREATMENT OF ALCOHOLIC FAMILIEStyGREGORY SCOTT OLSONB.A., Montana State University, 1983A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF ARTSinTHE FACULTY OF GRADUATE STUDIES(Department of Counselling Psychology)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIANovember 1993\u00C2\u00A9 Gregory Scott Olson, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of_______________________The University of British ColumbiaVancouver, Canada/ /Date ,iDE-6 (2/88)AbstractThis study was part of a larger one called The Alcohol Recovery Project. Theoverarching aim of the present investigation was to broaden understandingregarding the role of the therapeutic alliance in the treatment of alcoholics withina systemic perspective. Using Pinsof and Catherall\u00E2\u0080\u0099s (1986) conceptualization ofthe affiance, a process-outcome study was conducted within both individual andconjoint formats. A sample of 63 alcoholic men and 21 non-alcoholic femalepartners was divided into three different treatment groups, two for individualsand one for couples. Several aspects of the affiance were investigated including:(a) the strength of the alliance in different treatments; (b) the impact of drinkingbehavior on the alliance; (c) the differential power of the affiance in predictingoutcome in the individual therapeutic format compared to the conjoint one; (d)the impact of gender on alliance strength; and (e) the relationship between splitamd intact affiances and therapeutic improvement. Results of the studyindicated: (a) that the strength of the affiance was nearly identical in two verydifferent therapeutic approaches; (b) that binge and chronic alcohol-use patternsserved as potent predictors of the correlation between alliance and dyadicadjustment gain; (c) that the strength of the relationship between the affiance andtherapeutic gain was not significantly different in individual as opposed toconjoint treatment (recognizing limitations in using the scales to comparemodalities); (d) that women formed significantly stronger affiances than men;and (e) that the theoretical constructs of split and intact alliances could beempirically identified. The complexities inherent in interpreting resultssurrounding a process variable like the affiance became very evident as did theneed for more study of the construct. Implications of the results were discussed,and recommendations made for future directions of research.iiiTable of ContentsAbstract iiList of Tables viAcknowledgement viiiChapter 1: Introduction and framework 1Background of process and outcome research ITherapeutic affiance 2Process and. outcome 4A new process-outcome paradigm 4Present considerations 5Alcohol factors 5Definition of terms 6Research problem and questions 9Chapter 2: Review of the literature 10Background and definition of the affiance 10Definition in individual psychotherapy 10Summary 16Definition in family therapy 17Research on the therapeutic alliance 23Overview 23Alcohol affiance studies 25Relationship of affiance and outcome 25Individual research 25Family research 26Timing of measurement 28Pretreatment client variables 28Marital distress 29Pretreatment symptomatology 30Pretherapy social adjustment 31Alcoholism 32Treatment factors 37ivGender and the alliance.40Intact and split alliances 4Client, therapist, or observer report 43Summary and extensions of the literature 46Research hypotheses 47Chapter 3: Methodology 52Design of the study 52Treatments 52Experiential Systemic Therapy 53Supported Feedback Therapy 53Sample 54Therapists 55Division of sample into drinking types 56Data collection procedures 58Instruments: Conceptual and psychometric data 59Dyadic Adjustment Scale 59Therapeutic Affiance Scales 62Psychometric characteristics of the ITAS and CTAS 66Reliabffity 66Validity 68Distribution 68Summary 69Psychometric characteristics of the TAS in the present study 70Revised scales 73Data analysis techniques 77Chapter 4: Results 81Statistical procedures 81Results of analyses 82Hypothesis one 82Hypothesis two 83Hypothesis three 86Hypothesis four 90Hypothesis five 93Adjunct Analyses 95VChapter 5: Discussion and limitations.98Summary of results 98Discussion and implications of results 100The alliance and treatment for men 101The alliance for men in individual and conjointmodalities of ExST 103Gender and the affiance in conjoint therapy 105Intact and split alliances 106Drinking patterns and the affiance 108Pretreatment distress and the alliance 112Limitations 112Methodological Limitations 113One-time measurement 113Source of report 113Early termination 114Statistical 1mM 114Conceptual Issues 114Therapist variables 114Conjoint and individual treatment comparisons 115Conjoint and unilateral couples therapy 115Implications for future research 116Clinical areas for research 117Individual versus family therapy 117Therapist variables 117Gender of clients 117Intact and split affiances 118Alcohol and the alliance 118Cultural implications 119Areas to promote further research 119Continued use of Therapeutic Affiance Scales 119Process studies 120References 121Appendixes 130viList of TablesTable 1: Demographics of sample.55Table 2: Demographics of therapists 56Table 3: Means, Standard Deviations and i-test values:Variables for Binge and Chronic Drinking Types 57Table 4: DAS subscale descriptions and Cronbach reiabifities 60Table 5: Cronbach\u00E2\u0080\u0099s alpha for pre and posttreatment DAStotal and subscales 61Table 6: Number of items on affiance subscales 64Table 7: ITAS/CTAS: Subdimensions and number of items 66Table 8: ITAS/CTAS: Subscale intercorrelations and internalconsistencies (Cronbach\u00E2\u0080\u0099s Alpha shown as diagonalelements) 71Table 9: ITAS/CTAS: Subdimension intercorrelations andinternal consistencies 72Table 10: Item selection for revised TAS using transformed -scores 75Table 11: Revised ITAS/CTAS: Subscale intercorrelations andinternal consistencies (Cronbach\u00E2\u0080\u0099s Alpha shown asdiagonal elements) 76Table 12: Cronbach\u00E2\u0080\u0099s estimate of reliabffity for DAS-gain scoresand Pearson correlation coefficients: DAS-pre with DAS-post.... 77Table 13: Pearson Correlation Coefficients for men in individualtreatments: Alliance with DAS-pre and DAS-post 82Table 14: ITAS means, standard deviations, and t-test values:For drinldng types in individual therapy 83Table 15: Means, standard deviations, and i-test values:ITAS and DAS for drinldng types in SFT and ExST-Ind 85Table 16: Pearson Correlation Coefficients for drinking types:Alliance with DAS-gain in SFT and ExST-Ind 86Table 17: Means and standard deviations for ExST men:ITAS-R, CTAS-R and DAS Scores 87Table 18: Repeated Measures ANOVA for ExST men: Analysisof DAS by Treatment, Affiance Level, and Occasion 88viiTable 19: DAS-gain: Cell means in dichotomized alliance levelsfor ExST men 89Table 20: Pearson Correlation Coefficients for ExST men:Alliance with DAS-gain subscales 89Table 21: Means, standard deviations, and .-test values:DAS scores for men and women in ExST-Cpl 91Table 22: Pearson Correlation Coefficients for couples:CTAS with DAS scales 92Table 23: Pearson Correlation Coefficients for couples:Men\u00E2\u0080\u0099s DAS scores with partners\u00E2\u0080\u0099 92Table 24: Mean differences and standard deviations for splitand intact alliances: DAS-pre and DAS-post 94Table 25: Pearson Correlation Coefficients for intact and splitaffiances: Men\u00E2\u0080\u0099s DAS scores with women\u00E2\u0080\u0099s DAS scores 95Table 26: Frequency contingency table: Partners\u00E2\u0080\u0099 affiance levelsand drinking type 96Table 27: Frequency contingency table: Partners\u00E2\u0080\u0099 affiance levelswith partner level of discrepancy 97viiJAcknowledgementsI would like to express my gratitude to a number of individuals withoutwhose emotional. and/or technical support this thesis would never have crossedthe finish line. To my friends Andrew and Alan, thank you for the many timesyou offered your support, encouragement, and distraction. To my fellowresearchers on the TARP project, I am grateful for our meetings and that sense ofshared desperation that helped to \u00E2\u0080\u009Cnormalize\u00E2\u0080\u009D the process. To those who assistedme with the statistics\u00E2\u0080\u0094Rachel, Warren, and Cheryl--thank you. And last, mythanks go to committee members, Doctors Friesen, Coriry, and Havercamp, whocreated space in the midst of very busy schedules to answer questions and who attimes had to believe for me that this project was not impossible.1Chapter 1INTRODUCTION AND FRAMEWORKThe concept of the therapeutic alliance has been examined from the earliestdays of individual psychoanalysis (Freud, 1958a; Rogers, 1957; Sterba, 1934; Zetzel,1956). Interest in the affiance in individual therapy has continued up to thepresent time, therapists and researchers alike investigating the significant, evenpivotal role it plays in the therapeutic endeavor (Barrett-Lemiard, 1962; Bordin,1979; Greenson, 1967; Horvath & Symonds, 1991; Luborsky, 1976; Marziali, 1984).The focus of the present study is to examine in a systematic fashion the role thetherapeutic affiance plays in a new intervention program applied to bothindividual and marital treatment formats with an alcohol-dependent population.To lay a foundation for the study, a brief review of the background ofpsychotherapeutic research that involves process variables will be undertaken,followed by definitions and research issues.Background of Psychotherapeutic Process andOutcome ResearchSeveral reviews and studies indicate that all approaches to individual andfamily therapy are more or less equally successful (Bennun, 1989; Gurman &Kruiskern, 1981; Smith, Glass, & Miller, 1981). Horvath and Greenberg (1986) notethat, \u00E2\u0080\u009CAttempts to demonstrate differential treatment effectiveness have not beenoverly productive (Smith & Glass, 1977) and this has led to a renewed interest ininvestigating the effects of common factors in several forms of psychotherapy\u00E2\u0080\u009D (p.530).Traditionally, the mechanisms of therapeutic change have been assumed to2be located in the implementation of the techniques of particular theoreticalapproaches. Technical skills have been emphasized over relationship ones.Bordin (1979) was one of the earliest researchers to predict that \u00E2\u0080\u009Cthe strength ofcollaboration between patient and therapist may have more to do with theeffectiveness of the therapy than the particular methods chosen\u00E2\u0080\u009D (p. 255). Thoughtheoretical interventions can suggest methods to promote change, they do notinstruct in the \u00E2\u0080\u009Chow-to\u00E2\u0080\u0099s\u00E2\u0080\u009D of creating and sustaining therapeutic relationships.Bennun (1989) writes, \u00E2\u0080\u009CManagement of symptoms should be considered asimportant as the management of the therapy relationship\u00E2\u0080\u009D (p. 252). Simplyanalyzing treatments, comparing approaches, and examining outcomes neglectsprocess aspects that are inherent in every therapeutic relationship.Rice and Greenberg (1984) recommend the adoption of a new researchparadigm, one that studies the structure of therapeutic interactions andemphasizes the processes surrounding therapeutic change. Researchmethodologies assisting in the understanding of how and why change takes placeare urgently needed. With such methodologies, \u00E2\u0080\u009Cone could speculate with agreater degree of certainty that particular factors account for some of thedifferences in outcome\u00E2\u0080\u009D (p. 249). Researchers of individual therapy have begunexploring process aspects in treatment and have found them to account for someof the differences in therapy (Greenberg & Pinsof, 1986). It is becomingincreasingly clear that relationship variables may account for more of thevariance in change than specific treatment factors (Beutler, Crago, & Arizmendi,1986; Pinsof, 1989).The Therapeutic AffianceThe affiance is one process variable that is generating a great deal ofexcitement within the psychotherapy research community (Bordin, 1979; Hartley3& Strupp, 1983; Horvath & Greenberg, 1986; Luborsky, 1976; Marziali, 1984). Theconstruct is seen as having the potential to be both a potent predictor of outcomeas well as an organizing and focal construct for subsequent process research.While individual psychotherapy research can point to over a half century ofwork on the affiance, it is only in the last decade that attention has begun focusingon the function of the affiance in the area of marriage and family counseffing(Davatz, 1981; Pinsof & Catherall, 1986; Rutan & Smith, 1985). Very little processresearch in these settings has been carried out. Gurman, Kniskern, and Pinsof(1986) could find only one published major review of family therapy processresearch, that of Pinsof\u00E2\u0080\u0099s (1981). Even this was mainly a \u00E2\u0080\u009Cprimer on processresearch\u00E2\u0080\u009D intended to improve the methodological and conceptual quality of suchresearch (p. 597). To date, only one instrument has been designed to measure theconstruct of the alliance within a systemic framework (Pinsof & Catherall, 1986).That so few process studies have been undertaken in conjoint or familycontexts makes good sense in light of the complexities inherent inconceptualizing the interaction processes that involve more than one client.Gurman et al. (1986) write, \u00E2\u0080\u009CFamily therapy not only represents an additivecomplication, but actually creates a new interpersonal gestalt that is notadequately addressed by the methods, procedures, and designs of individualtherapy process research\u00E2\u0080\u009D (p. 597). Fisher, Kokes, Ransom, Phillips, and Rudd(1985) explain the need for much more discussion and thinking surroundingmeasurement theory and technique \u00E2\u0080\u009Cbecause family research is based on themeasurement of an elusive, multi-individual, transactive unit\u00E2\u0080\u009D (p. 222).Understandably, process research involving the affiance has mainly taken placeonly within individual therapy contexts. Marziali (1984) urges that the workcontinue. Only as the affiance factor is more precisely delineated can other4characteristics of the patient-therapist interaction be brought into sharper focus.Process and OutcomeThe daunting challenge to family process researchers is discovering how thetherapist system facifitates change (Heatherington & Friedlander, 1990a; Pinsof,1989). The primary research task is to identify significant relationships betweenprocess and outcome variables. Most past research has failed to adequately linkthese two elements. As a result, the specific components of treatments that areconsistently related to outcome have not been identified. Bennun (1989) notes,\u00E2\u0080\u009CThe field has yet to find the salient variables, powerful instruments to measurethe salient variables, and research strategies that will uncover the links theybelieve exist in practice\u00E2\u0080\u009D (p. 599).In the past, a strong dichotomy has existed between process and outcomeresearch. Process was considered to include the range of events occurring withinthe actual marriage and family session, and outcome was left for what occurredoutside the session after therapy had ceased. Process goals were those a couplecould achieve inside the therapy room, whereas outcome goals were those to beachieved without (Dryden & Hunt, 1985b). With such a strong division, the twoareas of research had serious limitations. Gurman (1982) observes that outcomestudies rarely defined explicitly the actual processes of the different family therapymodels. In the absence of clearly defined treatments, the study of the changeprocess becomes extremely difficult.A new process-outcome paradigm. Sounding very similar to Rice andGreenberg (1984), Gurman et al. (1986) stress the need for new definitions and anew understanding of process research within the marriage and family field.They emphasize that it is primarily conceptual, not methodological,understanding that is needed. New ways of thinking about psychotherapy and5psychotherapeutic change are required for a process or change perspective.Individual therapy process research is cited as a good example of this newemphasis occurring.Pinsofs (1989) idea of the new way of thinking includes the blurring ofprocess-outcome distinctions. He argues that focusing on posttreatment outcome,the \u00E2\u0080\u009CBig 0,\u00E2\u0080\u009D has blinded researchers to the change moments throughout thecourse of therapy. With the quest for the \u00E2\u0080\u009CBig 0\u00E2\u0080\u009D modified, he suggests that allresearch be viewed as process research, looking for the many \u00E2\u0080\u009Clittle o\u00E2\u0080\u0099s\u00E2\u0080\u009D along theway (p. 56). The therapeutic alliance is one of those very important \u00E2\u0080\u009Clittle o\u00E2\u0080\u0099s\u00E2\u0080\u009Drequiring much more study in the conjoint and family contexts.Present ConsiderationsGreenberg and Pinsof (1986) believe the emergence of affiance theory infamily therapy represents a move towards developing \u00E2\u0080\u009Cgeneric\u00E2\u0080\u009D variables thatwill integrate specific (technical) factors and general (relationship) ones fromacross the spectrum of research and therapy contexts. The trend in family therapyresearch seems to be less on competition between models and more on attemptsat finding empirical research tools that can apply across theories and researchmethodologies (Gruman, 1986). The hope is that variables such as the therapeuticaffiance will become common threads making the entire field of psychotherapyresearch more coherent. The focus is on discovery rather than verification, theaim being to find those factors leading to greater efficacy in therapy.Alcohol FactorsAlthough this study does not intend to provide an extensive examination ofalcohol-related issues relative to the therapeutic alliance, certain points need to beaddressed. Gurman et al. (1986) note that despite the widespread problems6associated with alcohol abuse, as well as family therapists\u00E2\u0080\u0099 clinical involvementin treating the problems, very few empirical studies of the efficacy of any ofseveral common family-involved treatments (e.g., marital, family, or multifamilygroup therapy) have taken place (p. 578). Almost no empirical data examining thealliance with an alcoholic population exists. No empirical research has directlyconsidered the effect that alcohol might have upon the formation of the affiance.In fact, most alliance studies have screened out alcohol-dependent subjects fromtheir samples altogether. The few that have allowed alcoholics to participate havenot in their discussions focused upon the issue of alcohol-dependency. Ofconcern in the present research is how alcohol dependency impacts the formationand functioning of the therapeutic alliance for alcoholics and their partners.Definition of TermsKey concepts used in the study are defined below.Family therapyGurman et al. (1986) offer this definition: \u00E2\u0080\u009Cany psychotherapeutic endeavorthat explicitly focuses on altering the interactions between or among familymembers and seeks to improve the functioning of the family as a unit, or itssubsystems, and/or the functioning of the individual members\u00E2\u0080\u009D (p. 565). Couplestherapy is a subclass of family therapy. The definition is not as concerned with thedirect participants of therapy as it is with the conceptual framework and goals oftherapy (Pinsof, 1989).SystemsKey systemic principles outlined by Gurman et al. (1986) include (a) anemphasis on patterns, relationships, and information--change in one subsystem7effects change in other systems; (b) circular causality; (c) the assumption that thewhole does not equal the sum of the parts; and (d) the belief that there is noindependent, objective, knowable reality--instead, multiple realities areconstructed by participants and observers (p. 568).Unilateral couples therapyCouples counselling carried out unilaterally shares with traditional maritaland family therapy the objective of altering the system within which theattending partner functions (Bennun, 1985). Unlike individual psychoanalysis,greater emphasis is placed on marital issues, and whenever personal issues areviewed as contributing to dyadic distress, the matter is placed as much as possiblein the context of the relationship. Interventions are designed with the entiresystem in mind, the assumption being that if change in the sole partner\u00E2\u0080\u0099s systemcan be achieved, that change will affect the couple system as well. Though theentire family is typically addressed in systems theory, in this study, the focus wason the couple dyad only.Process researchThis study uses Greenberg and Pinsof\u00E2\u0080\u0099s (1986) description, \u00E2\u0080\u009CThe study of theinteraction between the patient and therapist systems\u00E2\u0080\u009D (p. 18), bringing togetherconcepts emerging from individual therapy process research and selected ideasfrom family therapy. The goal is discovering how change is facffitated by theinteraction between the two systems. \u00E2\u0080\u009CProcess research covers all of the behaviorsand experiences of these systems, within and outside the treatment sessions,which pertain to the process of change\u00E2\u0080\u009D(p. 18).Family therapy process researchThe definition of \u00E2\u0080\u009Cprocess research\u00E2\u0080\u009D can be applied to individual or familytherapy research. To narrow the focus specifically to \u00E2\u0080\u009Cfamily therapy process8research,\u00E2\u0080\u009D Pinsof (1989) integrates Gurman et al.\u00E2\u0080\u0099s (1986) and Greenberg andPinsof\u00E2\u0080\u0099s (1986) definitions:Family therapy process research studies the interaction between therapist andfamily systems. Its goal is to identify change processes in the interactionbetween these systems. Its data include all of the behaviors and experiencesof these systems and their subsystems, within and outside of the treatmentsessions, that pertain to changes in the interaction between family membersand in their individual and collective levels of functioning. (p. 54)Key components of the definition that Pinsof highlights include a focus onsystems and subsystems, an emphasis on interaction, and a concern with changeprocesses.Alcoholic familyA family in which alcohol is the central organizing principle around whichthe family is structured (Steinglass, Bennet, Wolin, & Reiss, 1987). \u00E2\u0080\u009CAlcoholicfamilies must contend with a condition that (1) is chronic; (2) entails the use of apsychobiologically active drug; (3) is cyclical in nature; (4) produces predictablebehavioral responses; and (5) has a definable course of development\u00E2\u0080\u009D (p. 10).Therapeutic allianceThis construct, central to the present study, wifi be defined in the LiteratureReview.9Research Problem and QuestionsWithin the conjoint context, few empirical findings exist regarding (a)variables that impact therapeutic alliance formation, (b) the impact the affiancehas upon treatment outcome, (c) the function of the alliance in conjoint therapyas opposed to individual counselling, and (d) split and intact affiances within thecouples context. Additionally, the present project examines two new treatments--Experiential Systemic Therapy (ExST) and Supported Feedback Therapy (SFT)-both designed as systemic interventions for alcohol-dependent individuals andcouples. Testing the role of the therapeutic alliance within the new approacheswill be one step toward their validation.The purpose of the present study is to address the following general researchquestions:1. Do type of treatment and client gender result in significant differences in thestrength of the therapeutic affiance?2. Do drinking patterns of alcoholics result in significant differences in thestrength of the therapeutic alliance?3. To what extent does the strength of the therapeutic affiance facifitate orinteract with dyadic adjustment outcome measures?4. Is the strength of the alliance-outcome relationship significantly different inthe individual and marital formats within the same treatment approach?5. To what extent does agreement between spouses on the strength of thealliance interact with outcome?10Chapter IIREVIEW OF THE LITERATUREIn this chapter, the background of the therapeutic affiance will be traced inboth individual and marriage and family contexts. The conceptual definitionused for this study will be developed in the process. Next, research on the alliancewithin individual and family formats will be explored. Particularly important forthe present project will be research findings regarding (a) the alliance and itsrelationship to outcome, (b) the impact pretreatment variables have on allianceformation, (c) treatment factors affecting the alliance, (d) gender effects upon theaffiance, and (e) who the most appropriate sources are for alliance ratings. Finally,the research hypotheses for the study wifi be introduced.Background and Definition of the Therapeutic AllianceThe Definition in Individual PsychotherapyThe notion of the therapeutic affiance has permeated individualpsychotherapy theory from its inception. Defining the construct has been andremains an enormous challenge. As early as 1913, Freud (1958a) described twokinds of attachments patients could have with their analysts--neurotic ones thatinterfered with cooperation, and healthy ones that induced cooperation. Freudbelieved that the positive, reality-based aspects of a relationship provided afoundation for a therapeutic partnership against a patient\u00E2\u0080\u0099s neurosis.Though he did not attempt to define the affiance explicitly, Freud (1958b) atleast initiated the beginning stages of its conceptualization. He theorized that aneffective therapist makes a collaborator out of a patient in two ways: first, bymeans of rapport--e.g., using client transference based on early affectionate and11helping relationships; and second, through technique--e.g., removing resistances,or avoiding moralizing or premature interpretations (Frieswyk et al., 1986). Theimportant point in definitional terms is that Freud viewed the affiance as a dyadicinteraction involving both therapist technical skill as well as patient variables. Agood alliance would also result in each participant having warm feelings for theother (Tichenor & Hifi, 1989).Later conceptualizations of the affiance were developed from object-relationstheorists. Sterba (1934) conceived of the affiance as the way for clients to becomeso identified with an analyst\u00E2\u0080\u0099s ego that they could develop an observing ego forthemselves. Using their \u00E2\u0080\u009Cborrowed\u00E2\u0080\u009D ego, they could reflect intellectually on theirproblems and develop a \u00E2\u0080\u009Cself-object differentiation in order to use an object ratherthan to fuse with the object\u00E2\u0080\u009D (Rutan & Smith, 1985, p. 195).Zetzel (1956) postulated that the alliance was a necessary component for aclient to go through the analysis of transference issues safely. The first to coin theterm therapeutic alliance, she claimed that the client\u00E2\u0080\u0099s identification with thetherapist provided a rational, mature base from which irrational, transferentialissues could be examined (Tichenor and Hifi, 1989). Zetzel\u00E2\u0080\u0099s work with borderlinepatients influenced her to alter Freud\u00E2\u0080\u0099s conception of the affiance as beingcomposed of patient factors and therapist technique. Observing the difficulty herpatients had in establishing rapport because of early developmental experienceswith affectionate others, she highlighted the therapist\u00E2\u0080\u0099s responsibffity to create astrong base for the client (Frieswyk et al., 1986). Empathy and support becamemore important than the traditional therapist skills of making interpretationsdesigned to resolve patient defenses. By deemphasizing patient responsibifities,Zetzel blurred the construct for many and attracted criticism from within her ownpsychoanalytic camp.12Rogers (1957) and his associates greatly influenced thinking on the construct.Like Zetzel, their focus lay on therapist responsibifities. For client change tooccur, they hypothesized that conditions of empathy, unconditional positiveregard, and congruence were essential and sufficient. Initial studies based onthese therapist-offered conditions seemed to suggest that they were indeedeffective across a range of therapeutic contexts. Later investigations, however, didnot find the predicted relationship between the conditions and outcome (Horvath& Greenberg, 1986). Further research examined the benefits to the client whenthis kind of therapeutic approach was used (Barrett-Lennard, 1978, 1985).Recent reviews suggest that the complex counsellor-client relationshipinvolves more than therapist variables alone (e.g., Gelso & Carter, 1985; Mitchell,Bozart, & Krauft, 1977). Responding to Rogers\u00E2\u0080\u0099 emphasis, Horvath and Greenberg(1986) argue:A model that posits the psychotherapist as the major variable influencingboth the process and the outcome.. . is inherently incomplete. It is bothlogically and clinically apparent that a conceptualization of the therapyprocess that includes the contributions of both the client and the therapist,potentially, could account for more of the process and outcome of therapy.(p. 531)Increasingly, the premise is held, similar to Freud\u00E2\u0080\u0099s view, that both therapist andclient make important contributions to the formation of any effective therapeuticendeavor (Horvath & Symonds, 1991; Marziali, et al., 1981).Greenson (1967) introduced yet another consideration to the emergingalliance definition. He depicted three nonmutually exclusive components of therelationship: (a) the working alliance, (b) the transference relationship, and (c) the\u00E2\u0080\u009Creal\u00E2\u0080\u009D relationship (Freud had spoken of a similar \u00E2\u0080\u009Ccordial relationship\u00E2\u0080\u009D). Theworking alliance--discussed in greater detail later\u00E2\u0080\u0094focuses on the shared task of13treatment unique to the counselling setting. The real relationship, that whichdevelops and grows beween a counsellor and client during therapy, includes atherapist\u00E2\u0080\u0099s decency, openness, and friendly approach to patients. It constitutes the\u00E2\u0080\u009Creal\u00E2\u0080\u009D aspects of therapy, as opposed to the \u00E2\u0080\u009Cunreal\u00E2\u0080\u009D transference elements. Gelsoand Carter (1985) maintain that the real relationship is nourished by thetherapeutic bond, but eventually supercedes that bond. Researchers generallyagree that the real relationship facifitates clients\u00E2\u0080\u0099 developing attachments to theirtherapists. Whether it should be considered a part of the technical scrutiny of thetherapeutic alliance, however, is very controversial (Frieswyk et al., 1986).According to Greenson (1967), the central feature of the alliance is thepatient\u00E2\u0080\u0099s collaboration with the therapist. The term working alliance, first used byGreenson to depict the collaborative element between counsellor and client,became a focus around which researchers began attempting to arrive at a moreempirical understanding of the construct. Greenson theorizes that clientcontributions to the affiance include the motivation to deal with a problem, thewillingness to cooperate and follow suggestions, and a sense of helplessness.Therapists participate by being empathic, straightforward, nonjudgmental, andcompassionate, as well as having understanding and insight. Importanttreatment conditions are establishing the notion that therapy is a joint venture,and creating a sense of regularity and consistent goals (Tichenor & Hifi, 1989).Greenson provided a more complete way to conceptualize the affiancedefinition. In the past two decades, a number of theorists have added their ownperspectives regarding a definition. Horwitz (1974) evaluated clinical results withborderline patients of the Menninger Foundation Psychotherapy Research Projectand concluded that the \u00E2\u0080\u009Cinternalization\u00E2\u0080\u009D of the affiance, as evidenced by \u00E2\u0080\u009Casustained collaboration\u00E2\u0080\u009D with the therapist, was the key to change (p. 37). Theinternal image of the therapist aids clients in fighting panic, poor self-esteem, or14catastrophic reactions to separation. Interpersonally, the internalization ismanifested by a client\u00E2\u0080\u0099s improved capacity to develop and sustain healthyrelationships.The Penn Psychotherapy Project (Luborsky, Crits-Cristoph, Alexander,Margolis, & Cohen, 1983) further refined the conceptualization of the construct bydistinguishing between Type I and Type 2 alliances. In Type I affiances, patientsexperience their therapists as warm, helpful, and supportive. They either feelchanged by therapy or experience such rapport with the therapist that they areconfident they can overcome their problems. The authors acknowledge that thisdefinition is directly traceable to Freud\u00E2\u0080\u0099s views of the transference. Type 2alliances are marked by the patient and therapist working together in a jointstruggle against the problem. Key aspects of Type 2 alliances are first, a sharedresponsibifity for working out the treatment goals and second, the patient\u00E2\u0080\u0099s abifityto do what the therapist does. When a patient demonstrates abffities similar tothe therapist\u00E2\u0080\u0099s\u00E2\u0080\u0094able to do for self what was done together--a Type 2 alliance hasformed. Luborsky (1976) observed that patients who showed the mostimprovement developed Type 2 relationships (Marziali et al., 1981).Frieswyk et al. (1986) made use of Type 1 and Type 2 distinctions to establish astrong case for the elimination of certain aspects of the traditional affiancedefinition rooted in Freud\u00E2\u0080\u0099s conceptualization. A broad psychoanalyticdefinition, they argue, leaves no way to distinguish the alliance from transferenceand technique. The definition, therefore, should be limited to \u00E2\u0080\u009Ca patient variablespecifically defined as the patient\u00E2\u0080\u0099s collaboration in the tasks of psychotherapyseparate from the patient\u00E2\u0080\u0099s experience of being helped and distinct from therapistcontributions to the affiance [italics added]\u00E2\u0080\u009D (p. 35). The patient\u00E2\u0080\u0099s experience ofthe relationship--Type 1 affiance (e.g., transference issues of trust, sense ofacceptance, optimism)--as well as technique or treatment variables (e.g., empathy15or skill) are thus removed. These variables are hypothesized to be importantcontributors to the formation and maintenance of the affiance, but are not a partof the alliance itself. The Type 2 alliance, according to Frieswyk et al. (1986), is thebest definition: \u00E2\u0080\u009Cthe extent to which the patient makes active use of the treatmentas a resource for constructive change\u00E2\u0080\u009D (p. 36). The authors acknowledge thatnearly all investigators include therapist contributions as well as patientexperiences when referring to the affiance, but they believe this confuses theconstruct.Process researchers would disagree strongly with Frieswyk et al. (1986),balking at the idea of eliminating therapist variables from the affiance equation(just as they would have rejected Rogers\u00E2\u0080\u0099 deemphasis on patient variables).Particularly within systems thinking, the suggestion to remove the contributionsof one person in a dyadic interaction is absurd, given notions of patterns andrelationships and circular causality.Bordin\u00E2\u0080\u0099s (1979) landmark operationalizing of the construct has been widelyreceived and has probably done more to further research on the alliance than anyother work. His ideas serve as a foundation for this study. He defines the alliancein terms of the agreement that exists between client and counsellor on the goalsand tasks of therapy. Additionally, he specifies a bonds component representingthe affective, experiential elements in the counsellor-counsellee relationship.Like Greenson (1967), Bordin continued the trend begun in the analytic literatureby blending both patient and therapist factors in his definition. In addition, hebroadened (or blurred, depending upon the author) the meaning of patientcollaboration to include underlying attitudes and experiences (e.g., trust andattachment).Unlike humanistic theorists, Bordin does not require therapists to fit intothree necessary and sufficient conditions to create a strong affiance. His16conceptualization does not dichotomize technical and relationship factors and isconsequently more readily applicable to the broad field of treatments. Therelationship is viewed as a general factor, whereas technique is approached as aspecific factor that varies from model to model. This view can accomodatebehaviorists\u00E2\u0080\u0094who may emphasize tasks and goats more than bonds\u00E2\u0080\u0094as well asperson-centered clinicians--who may work for a strong bond and put less energyinto defining explicitly the tasks and goals of therapy. In either case, specifictechnical factors must be combined with a relational bond to be effective.Summary. Definitional ambiguities remain while work on the affiancecontinues. Rutan and Smith (1985) use the description \u00E2\u0080\u009Cgrowthful contract withthe therapist\u00E2\u0080\u009D to summarize Freud\u00E2\u0080\u0099s positive transference, Zetzel\u00E2\u0080\u0099s therapeuticaffiance, and Greenson\u00E2\u0080\u0099s working alliance (p. 195). Rogers (1957) emphasizestherapist variables, Luborsky et al. (1983) focus on the client\u00E2\u0080\u0099s experience, andFrieswyk et al. (1986) highlight the client\u00E2\u0080\u0099s collaboration in the tasks ofcounselling. More recently, Kokotovic and Tracey (1990) have described thealliance as \u00E2\u0080\u009Cthe feeling that both participants care for each other\u00E2\u0080\u009D along with thewillingness and abifity to work productively toward a shared goal (p. 16). The lackof consensus on the construct\u00E2\u0080\u0099s definition represents one of the principalmethodological limitations of the present research. Bordin\u00E2\u0080\u0099s (1979)operationalization has been the most helpful (Horvath, 1981; Horvath &Greenberg, 1986; Pinsof & Catherall, 1986).In spite of the differences, wide agreement exists in a number of areas. Basicto the alliance definition for most researchers are the common ingredients ofmutuality, engagement, and collaboration, whereby both therapist and client havethe capacity to participate in the therapeutic contract (Frieswyk et al., 1986;Horvath & Symonds, 1991; Luborsky, 1976; Marziali, 1984). Consistent is the17assumption that \u00E2\u0080\u009Cthe work of therapy cannot proceed in the absence of aconstructive therapeutic alliance\u00E2\u0080\u009D (Marziali et al., 1981, p. 361). Even behavioristswho once eschewed relationship variables, claiming that technique was theexplanation for change, are now acknowledging the place of the relationship as anecessary base for the effective employment of behavioral procedures (GeLso &Carter, 1985). Jacobsen (1981) writes, \u00E2\u0080\u009CThe most frequent impediment tosuccessful BMT (Behavioral Marital Therapy) is not the unskillful application ofbehavioral technology, but rather relationship skill deficits on the part oftherapists attempting to implement the technology\u00E2\u0080\u009D (p. 584).Clinicians and researchers alike share the belief that if the alliance is notmechanism for change, it is at least a necessary pre-condition for change(Tichenor & Hill, 1989). Others go further, claiming that the alliance may be themain vehicle of change, the primary mediating variable or vehicle of successdetermining the outcome of therapy and, consequently, the primary task oftherapy (Bordin, 1979; Horwitz, 1974; Pinsof & Catherall, 1986). Bordin (1980)emphasizes that the alliance is not therapeutic in and of itself; rather,collaboration provides clients with the opportunity to benefit from a therapist\u00E2\u0080\u0099sinterventions (Horvath & Marx, 1990). A pantheoretic construct, the therapeuticalliance contains concepts that are strong common elements in nearly alltherapeutic approaches, for every therapist attempts to build and maintain somesort of relationship with clients (Horvath & Symonds, 1991).The Definition in Family TherapyDoes the nature of the therapeutic affiance differ, theoretically, in couplesand family versus individual therapy where the concept originated? What does itmean for a therapist to be aiigned with a couple or a group? These are theconceptual questions confronting process and outcome theorists in the conjoint18and family fields.Though the marriage and family field is much younger than individualpsychotherapy, many family theorists and practitioners have written about thealliance. Minuchin (1974) refers to it as \u00E2\u0080\u009Cjoining\u00E2\u0080\u009D the family; Davatz (1981) speaksof \u00E2\u0080\u009Cconnecting\u00E2\u0080\u009D with the family; Ackerman (1966) mentions \u00E2\u0080\u009Cestablishing auseful rapport;\u00E2\u0080\u009D and Sluzki (1975) discusses the managing of the \u00E2\u0080\u009Ccoalitionaryprocess\u00E2\u0080\u009D (cited in Pinsof & Catherall, 1986). In Gurman and Kniskern\u00E2\u0080\u0099s (1981) IhHandbook of Family Therapy, an even greater variety of descriptions areprovided, the alliance taking a prominent position in nearly every model\u00E2\u0080\u0099sdescription of effective therapy: Duhi and Duhl (Integrative Family Therapy)describe it as a \u00E2\u0080\u009Cmatch\u00E2\u0080\u009D or \u00E2\u0080\u009Cfit\u00E2\u0080\u009D (1981, p. 489), Framo (intergenerational) as thedegree of \u00E2\u0080\u009Cconnectedness\u00E2\u0080\u009D (1981, p. 142), Jacobsen (behavioral) as a \u00E2\u0080\u009Ccollaborativeset\u00E2\u0080\u009D (1981, p. 567), and Whitaker and Keith (intergenerational) as \u00E2\u0080\u009Cbasic empathy\u00E2\u0080\u009D(1981, 210).Efforts to arrive at an acceptable systemic definition for the affiance haveoften been short-lived. Gurman et a!. (1986) describe the typical process asconsisting of family researchers developing an instrument, using it one or twotimes, and then abandoning process research altogether because of thecomplexities and the early stage of the field\u00E2\u0080\u0099s development. Few are building onothers\u00E2\u0080\u0099 work, and methodological quality remains poor. Gilbert, Christensen, andMargolin (1984) point out another difficulty. Measures of family alliances haveattempted to operationalize the construct in a variety of ways, most often usingsimilarities in verbal reports of family members, similar attitudes, or similardecision-making responses. The authors question whether suchoperationalizations of interpersonal alliances are adequate. Besides yieldinghighly inconsistent results, the use of family interactions as a means of inferringfamily affiances (content-free measures of family process) presents significant19interpretive problems.The complexities related to the therapeutic alliance in conjoint and familycontexts are many. Marital therapists are presented with the challenge ofdeveloping and maintaining an affiance with an individual husband and anindividual wife as well as with the couple system as a whole (Bourgeois,Sabourin, & Wright, 1990). Just as each individual client has the potential toestablish a relationship and emotional bond with a therapist, so does each familymember who is seen in couple or family treatment. Heatherington andFriedlander (1990b) observe that \u00E2\u0080\u009Cthe multi-person context significantlycomplicates its conceptualization and its study. Each individual\u00E2\u0080\u0099s alliance withthe therapist has the potential to affect and be affected by every other individual\u00E2\u0080\u0099sor subsystem\u00E2\u0080\u0099s relationship\u00E2\u0080\u009D (p. 299). Thus, rather than simply having anindividual alliance with one person, family therapists must also be alert tosubsystem alliances within the family, such as with parents or children, as well as awhole system alliance between the therapist and the family as an entity in itself(Heatherington & Friedlander, 1990b; Pinsof, in press).Heatherington and Friedlander (1990a) note two major differences in thealliance construct in marital and family versus individual therapy. First, becausemembers in the former treatment share the therapist, the affiance may not be asintensely personal. Second, the affiance a therapist has with one or two familymembers impacts the affiance with other family members in a circular, reciprocalfashion. Interventions with one member can have important indirectimplications for members not addressed who assess their feelings about therapyand treatment in light of their observations of the therapist\u00E2\u0080\u0099s interactions withothers. Indeed, \u00E2\u0080\u009Cno single dyadic or triadic (patient-therapist) alliance can beconsidered in isolation\u00E2\u0080\u009D (Pinsof & Catherall, 1986, pp. 138-139).Rutan and Smith (1985) introduce two additional potential alliance20differences in couples counselling. First of all, in individual therapy, clients areremoved from their interpersonal systems, thereby making it easier for them torisk exposure and achieve new perspectives. With couples, any vulnerabilitydiscussed is omnipresent 24 hours a day, possibly making the sharing ofvulnerabilities more difficult. Ironically, partners in couples therapy are asked tobe the most open with one another during their times of crisis--when they aremost likely to feel cautious about disclosing aspects of themselves.A second consideration unique to the marital therapist is the pseudo-alliance(Rutan & Smith, 1985). A couple may be highly cooperative because the partnersare so terrified about what is happening in their relationship. Such an \u00E2\u0080\u009Callianceof desperation\u00E2\u0080\u0099 ultimately interferes with the therapeutic relationship becausemembers use cooperation to defend themselves from fully examining theirrelationship, a prospect they fear they may not survive. Attempts by the therapistto create curiosity and interest in exploration result in compliance rather thandiscovery. A true therapeutic relationship does not form, although theappearance of collaboration might suggest otherwise.Jacobson (1981) raises the issue of triangulation as an importantconsideration when speaking of alliances in marital therapy. Often a therapistmust make a deliberate effort to create an alliance that will counteract a naturalalliance with one spouse in order to recruit the other. The more enthusiasticspouse may not get the same kind of attention that the more skeptical partnerdoes. Rosenbaum and O\u00E2\u0080\u0099Leary (1986) go further, warning that therapists must beon guard, not only to avoid an affiance with one partner, but also to avoid anyappearance of such an alliance, which could alienate one of the spouses andthreaten the therapeutic relationship with the couple system as a whole.Dryden and Hunt (1985a) focus on therapist responsibifities with regard toforming marital alliances. They cite Barker (1984) who argues that the marital21therapist must be aware of at least seven different systems: (1) the marital dyad;(2) the wife as an individual; (3) the husband as an individual; (4) thosedependent on the couple; (5) society; (6) the authority that sanctions the work ofthe marital therapist; and (7) the therapist him or herself. Dryden and Hunt(1985b) add therapist\u00E2\u0080\u0099s duties of achieving empathic symmetry (making eachpartner feel equally accepted, supported, understood), spatial symmetry(positioning oneself equally close to each partner), temporal symmetry (givingeach an equal opportunity to speak), and moral symmetry (not allowing blame tobe lodged on the shoulders of one (pp. 148-149).Addressing specific dimensions of the affiance, Dryden and Hunt (1985b)suggest that it is best to speak of a \u00E2\u0080\u009Cmatrix of goals\u00E2\u0080\u009D in conjoint counseffing. Theyexplain:[Spouses have] their own goals for themselves, their partner and therelationship. Furthermore, each partner has perceptions of what the other\u00E2\u0080\u0099sgoals are in these three areas and what these goals should be. Then there aregoals that represent a consensus between the partners. Finally the therapist\u00E2\u0080\u0099sgoals for each individual and the couple need to be considered and added tothe matrix. (p. 150)A final conceptual issue is how to handle the controversial issue oftransference in couples therapy (Gurman, 1978; Pinsof, in press). One partner maybe coming fi treatment (resistance), whereas the other is coming to form arelationship in order to be j treatment. Gender issues become signficant, thedifferent sexes of the partners interacting with the sex of the therapist inpotentially different ways.Given the conceptual difficulties, it is not surprising that research on themarital affiance has moved slowly. Only until recently have attempts been madeto begin developing an explicit clinical theory concerning the role of the alliance22in marital or family therapy (Dryden & Hunt, 1985a; Gurman, 1981; Rutan &Smith, 1985). Pinsof and Catherall (1986) write that \u00E2\u0080\u009Cadding the alliance conceptto the theoretical base of family therapy ifiuminates and brings into focus a criticalaspect of therapy that has existed in a theoretical twilight\u00E2\u0080\u009D (p. 138).Pinsof and Catherall\u00E2\u0080\u0099s (1986) three Therapeutic Alliance Scales--theIndividual Therapeutic Alliance Scale (ITAS), the Couples Therapeutic AllianceScale (CTAS), and the Family Therapeutic Alliance Scale (FTAS)--are the onlyexisting alliance instruments that recognize the unique complexities of coupleand family contexts as well as the systemic realities inherent even within\u00E2\u0080\u009Cindividual\u00E2\u0080\u009D counselling. They offer significant new contributions to the allianceconceptualization while building upon Bordin\u00E2\u0080\u0099s (1979) work from individualpsychotherapy research. In this way, the scales are integrative, providing a bridgebetween theory in individual and marriage and family fields. Individual andfamily counselling are situated within the same systemic framework, the theoryrecognizing the many systems at work in alliance formation regardless of whethera therapist is working with one or several clients.The way the scales are able to provide this integration is found in the theoryon which they are founded. Unlike instruments developed for individualpsychotherapy research, the ITAS, CTAS, and FTAS conceive of a collaborationnot just between two people, but between two systems: (a) the client system,which includes \u00E2\u0080\u009Call the human systems. . . that are or may be involved in themaintenance or resolution of the presenting problem;\u00E2\u0080\u009D and (b) the therapistsystem, which includes all the people who are involved in treating the patientsystem (Pinsof, 1986, p. 139). These definitions are meant to be broad in order toextend beyond those directly involved in therapy to anyone who influences thechange process or who may be impacted by the treatment (Pinsof & Catherall,231986). With the help of these instruments, the empirical work that is so neededwithin the field of marriage and family can proceed. Pinof and Catherall\u00E2\u0080\u0099s scalesmay provide the field with the same kind of impetus for research that Bordin\u00E2\u0080\u0099s(1979) groundbreaking work did two decades ago in individual psychotherapyresearch.Research on the Therapeutic AllianceOverview: Individual and Family Alliance ResearchArriving at a working definition is an essential first step for empiricalresearch to begin. Theoretical developments clarifying distinctions between theaffiance and other relationship variables such as transference or the realrelationship have provided further impetus for research (Gelso & Carter, 1985;Horvath & Symonds, 1991; Luborsky, 1976).In the last 15 years, a number of instruments have been developed tomeasure the alliance within the individual therapeutic context (Marziali et al.,1981; Tichenor & Hill, 1989). Various centers of research have sprung up: theUniversity of Pennsylvania (Luborsky, 1976); Vanderbilt University (GomesSwartz, 1978; Hartley & Strupp, 1983); the Langley Porter Institute (Marmar,Horowitz, Weiss, & Marziali, 1986; Marziali et al., 1981); and the University ofBritish Columbia (Horvath, 1981). Each has a distinctive approach based onsomewhat different assumptions of what the alliance is, what role it plays intherapy, and who the most appropriate sources are for its assessment (i.e., client,therapist, or impartial observer). Though Horvath and Symonds (1991) foundresults suggesting that unique aspects of the alliance are being tapped by thevarious instruments, they are tentative in their conclusions because, in spite ofthe differences, there is also evidence demonstrating that the intercorrelation24among several measures is substantial including strong intercorrelations atsubscale levels (Safran & Wailner, 1991; Tichenor & Hifi, 1989).Affiance research within individual treatment modes has covered a fairlybroad range of therapeutic contexts (Horvath & Symonds, 1991). Investigationshave drawn from psychodynamic, cognitive, and experiential therapy (e.g.,Greenberg & Webster, 1982; Luborsky, 1976; Safran & Wailner, 1991). Clientpresenting issues have included schizophrenia and depression among others (e.g.,Frank & Gunderson, 1990; Rounsavifie et al., 1987; Tichenor & Hifi, 1989). Bothshort and long-term interventions have been tested (Frank & Gunderson, 1990;Kokotovic & Tracey, 1990), and several different perspectives on themeasurement of the affiance have been used--those of clients, therapists, andindependent observers (Marziali, 1984; Tichenor & Hill, 1989).In spite of the heightened attention to the importance of the therapeuticalliance within individual counseffing, systematic research in the context ofcouple and family therapy has been seriously lacking. Until 1986, research wasalmost exclusively rooted in and focused upon individual psychotherapy theory(Pinsof & Catherall, 1986). Though nearly all marriage and family therapistswould consider the therapeutic affiance to be an essential part of successfultherapy, few empirical examinations have taken place.Until Pinsof and Catherall\u00E2\u0080\u0099s work (1986), clinical theory regarding the role ofthe affiance in marital therapy was lacking. Because of the conceptual difficultiesin the conjoint and family contexts, affiance instruments developed andemployed for use within individual psychotherapy research are not appropriatefor family research. Conceiving of the affiance as applying solely to the therapistand individual client, they do not account for the systemic implications thatworking with more than one client at a time creates (Pinsof and Catherall, 1986).To date, Pinsof and Catherall\u00E2\u0080\u0099s scales constitute the only existing measures25designed to empirically assess the affiance in individual, couple, and familytherapy from a systemic perspective (Bourgeois et at, 1990). Several studies havemade use of these scales (Bourgeois et al., 1990; Heatherington & Friedlander,1990a, 1990b; Gruman, 1986; Johnson & Greenberg, 1985), and others havedesigned their own measurement methods (Bennun, 1989; Gilbert et al., 1984).Alcohol alliance studies. Striking in both individual and family research isthe scarcity of empirical alliance research with alcoholics. In nearly all theindividual studies reviewed, potential subjects with alcohol problems wereexcluded (Eaton, Abeles, Gutfreund, 1988; Horvath & Marx, 1990; Horvath, Marx,& Kamman, 1990; Marmar, Weiss, & Gaston, 1989; Marziali, 1984; Marziali et al.,1981; Tichenor & HiIl,1989). One study reported nothing with regard to whetheralcohol was a factor in subject selection (Kokotovic & Tracey, 1990). In the familyliterature, again several projects excluded drug and alcohol (Bourgeois et al., 1990;Gruman, 1986; Johnson & Greenberg, 1985). One report (Gilbert et al., 1984) didnot mention alcohol. Only Bennun\u00E2\u0080\u0099s (1989) study included alcohol-dependentparticipants--18 couples\u00E2\u0080\u0094and Heatherington and Friedlander (1990a, 1990b) didnot exclude famifies with substance intoxication. Thus, only 2 of the 14 affiancestudies reviewed did not screen out alcoholics. Clearly, empirical studies takingalcohol into consideration as a potential factor in the functioning of thetherapeutic affiance are needed.Relationship of Therapeutic Alliance and OutcomeIndividual research. Given the theoretical importance of the alliance to thetherapeutic process, the major focus of empirical work, has been first, to study itspower as a predictor of outcome and second, the variables important to itsformation. Horwitz (1974) reviewed process, outcome, and follow-up results inthe long-term Menninger Clinic study of 42 patients undergoing two different26treatments. The major contribution of the study was \u00E2\u0080\u009Cthe indication that thetherapeutic alliance is not only a prerequisite for therapeutic work, but often maybe the main vehicle of change\u00E2\u0080\u009D (cited in Bordin, 1979, p. 255). In the PennPsychotherapy Project (Luborsky et al., 1980), the goal was to identify thesignificant predictors of outcome. The only potent predictor of treatmentoutcome was the therapeutic affiance (Luborsky et al., 1983; Morgan, Luborsky,Crits-Christoph, Curtis, & Solomon, 1982). Marmar et al. (1986) tested fivedifferent scales of the working alliance and found that two of them related tooutcome. Marziali et al. (1981), used several self-report and rater measures anddemonstrated that positive affiances resulted in good treatment outcomes,whereas negative ones led to poor outcomes.Bourgeois et al. (1990) note that the individual research literature reportsamounts of variance outcome accounted for by the therapeutic alliance (asassessed by clients, therapists, and judges) as between 9% and 46%. Some studies,however, did not control for initial levels of symptomatology (Horvath &Greenberg, 1989; Tichenor & Hifi, 1989) or compared most and least improvedscores (Morgan et al, 1982). Studies that controlled for initial symptomatology byusing partial correlations with hierarchical multiple regression resulted in 9% to19% of explained variance (Marmar et al, 1989; Marziali, 1984).In short, the empirical individual psychotherapy research literature on thealliance reveals two areas having broad consensus: (1) the affiance is an essentialingredient in the therapeutic process, and (2) there is a positive associationbetween the alliance and outcome. Orlinsky and Howard (1986) in a review of 35years of process-outcome studies involving some 1100 findings agree that theoverall quality of the affiance is consistently associated with good outcome acrossall process perspectives.Family research. In 1985, Hooper reported that there was no clear review of27the various factors affecting outcome in marital therapy other than Gurman andKniskern\u00E2\u0080\u0099s (1981). Since then, there have been a handful of alliance-outcomestudies in family research. Bennun\u00E2\u0080\u0099s (1989) results suggest that good and poortreatment outcome can be statistically differentiated in terms of the perceptionsthat some members of the family hold about their therapist. Using a scaleevaluating therapists on positive regard, competency, and activity, men\u00E2\u0080\u0099s ratingscorrelated with outcome (from .38 to .71, < .01 for n = 35), whereas women\u00E2\u0080\u0099sratings were found to be less correlated with outcome or not at all.Bourgeois et al. (1990), using Pinsof and Catherall\u00E2\u0080\u0099s (1986) CouplesTherapeutic Alliance Scale (CTAS), found the therapeutic alliance to be aprecurser of treatment outcome, particularly among males. Although treatmentefficacy was not affected by the gender of the subjects, the proportion of varianceexplained by the alliance was higher for men. The alliance, as viewed by females,accounted for 5% of the variance in residualized post Dyadic Adjustment Scale(DAS) scores at < .05. For men, the quality of the affiance accounted for 7% ofDAS residualized scores ( < .01), and 5% and 8% ( < .05 and .01 respectively) ontwo other outcome measures. Overall, the proportion of the variance of outcomeexplained by the quality of the alliance was superior to 8%. The authors observethat their findings are consistent with the growing evidence in individualpsychotherapy research that the early development and maintenance of thetherapeutic affiance is predictive of positive outcome.Two unpublished studies also used Pinsof and Catherall\u00E2\u0080\u0099s scales. Gruman(1986) found no significant correlations between DAS gain scores and the totalCTAS in either of two treatment groups, but did discover that the \u00E2\u0080\u009COther\u00E2\u0080\u009Dsubscale (described later) correlated significantly with all three outcome measures.Significant correlations were found between the CTAS and two other outcomemeasures on all subscale and total scores. Catherail (1984) used all three of Pinsof28and Catherall\u00E2\u0080\u0099s scales and found positive ( < .05) correlations between each of theoverall alliance scale scores and patient progress.Timing of measurement. The literature is clear that the most importantphase of therapy for developing a therapeutic alliance and for predicting outcomeis the intial one. Morgan et al. (1982), using the Penn Helping Affiance RatingMethod, found that by the third session, the affiance instrument had significantpredictive power, accounting for 25% of the variance of outcome measures. Adler(1989) measured the working affiance after each of the first five therapy sessionsand found it to be significantly related to outcome by the third to fifth session onfour outcome measures. Hartley and Strupp (1983) found the alliance topositively predict outcome when it was measured in sessions three to six.Marziali (1984) reported that the association between affiance and outcome cansometimes be determined as early as the first session. Several other studies alsoreport a positive alliance-outcome relationship when early session measurementsare used (Bourgeois et al., 1990; Catherall, 1984; Eaton et al., 1988; Gomes-Schwartz,1978; Horvath, 1981; Horvath & Greenberg, 1986, 1989: Horvath & Symonds, 1991;Kokotovic & Tracey, 1990; Luborsky et al., 1983; Marmar et al., 1989; Marziali et al.,1981; Tichenor & Hill, 1989). Therapists and clients use the first few sessions tofind compatible ways of relating to each other. If this early interaction results inpositive attitudes and a satisfactory bond, clients will likely experience therapy asbeneficial.Pretreatment Client VariablesHorvath and Symonds (1991) observe that though collaboration, mutuality,and engagement are universal concepts among alliance researchers, \u00E2\u0080\u009Ca significantlack of unanimity [exists] with regard to. . . what contributions each participantmust provide for [its] development\u00E2\u0080\u009D (p. 147). Bordin (1979) theorizes that clients\u00E2\u0080\u009929readiness to collaborate with their therapists may have an important influence onthe formation of the affiance. Such readiness could be constituted by a number offactors including not only the degree of distress and dysfunction, but also anxietylevels about dependency, preferred working styles, vividness of memory,tolerance for ambiguity and uncertainty, and the abffity to see the therapist as agood object. Marmar, Weiss, and Gaston (1989) note several pretreatment factorsthat positively contributed to the working affiance for 52 subjects undergoing briefdynamic psychotherapy for pathological grief. These included educationalattainment, motivation for psychotherapy, and interpersonal functioning.In this section, research will be discussed that examines the followingpretreatment variables as they impact aiiiance formation: (a) marital distress; (b)pretreatment symptomatology; (c) pretherapy social adjustment; and (d)alcoholism.Marital distress. Clinical lore would suggest that pretreatment maritaldistress would impair the formation of the therapeutic affiance. Gilbert et al.(1984) write, \u00E2\u0080\u009CTo the extent that the marital affiance is weak, the family no longerhas its most experienced, knowledgeable, and resourceful subsystem fullycontributing to regulatory matters. . . and adaptive needs\u00E2\u0080\u009D (p. 83). Its abffitiesreduced, the family experiences additional stress, presumably making existingdifficulties that much harder to address.These ideas are not borne out in two studies, the first by Bourgeois et al.(1990) in which hierarchical multiple regression was used to test the effect ofpremarital distress on the affiance. Contrary to prediction, the level of maritaldistress, as measured by the four subscales of the DAS, did not predict the qualityof the therapeutic alliance as measured by couples and by therapists. The authorsconclude that marital distress is \u00E2\u0080\u009Cnot a consistent predictor of the therapeutic30alliance\u00E2\u0080\u009D (p. 610), neither facilitating nor impairing its formation. Similarly, in anunpublished study (Gruman, 1986), no relationship was found between pretreatment marital distress (measured by FACES) and the therapeutic affiance.Distressed couples were not significantly different from non-distressed couples oneither alliance or outcome scores. It must be noted that the samples in both of theabove studies may not represent couples in crisis because of selection criteria thatincluded no drugs or alcohol, no immediate plans for separation or divorce, andno primary sexual dysfunction. Participants were recruited who wanted to workon their communication skills and on resolving confficts.Pretreatment svmptomatologv. Several studies show that pretreatmentsymptomatology has only a weak association with affiance ratings (Marmar et al.,1989; Marziali, 1984; Morgan et al., 1982). Pretreatment symptoms studied includeadjustment disorder, posttraumatic stress disorder, major depressive episodes,panic disorders, and uncomplicated bereavement. Even less severe levels ofpretreatment distress do not seem to significantly affect the ability of clients toenter into therapeutic relationships (Kokotovic & Tracey, 1990; Marziali, 1984).Moras and Strupp (1982) found no relationship between psychological health(based on pretherapy ratings of patients\u00E2\u0080\u0099 severity of problems, intensity ofsubjective distress, and adequacy of functioning in social, work, and academicroles) and patients\u00E2\u0080\u0099 contributions to the therapeutic alliance. It must be noted thatthese studies are quite limited in their measures of pathology, especially whenunselected outpatient samples are used (Eaton et al., 1988).In contrast to other research, Eaton et al. (1988) found that the higher apatient\u00E2\u0080\u0099s symptomatology, the lower the positive alliance. Pearson productmoment correlations were computed between alliance scores and pretreatmentsymptom ratings from the SCL-90, and several significant results were obtained.In a sample of 40 subjects, negative patient contributions to the affiance correlated31with phobic anxiety at .36, with obsessive compulsive disorder at .38, and withdepression at .35 ( < .01). Positive patient contributions were inversely related toparanoia (-.39), psychoticism (-.38), and interpersonal sensitivity (-.40) ( < .01).The researchers explain these unique results by pointing out that prior studiesonly measured general levels of pathology, rather than specific symptomdimensions. Additionally, Moras and Strupp (1982) and Morgan et al. (1982) useclinician rather than patient-rated measures of symptoms. Eaton et al. (1988)theorize that therapists may view clients primarily in terms of their personalitiesrather than complaints. Patients, having no such perspective or organizingschema, simply rate symptoms phenomenologically.Pretherapy social adjustment. There seems to be no question in theliterature that clients\u00E2\u0080\u0099 pretherapy social adjustment impacts affiance formation(Eaton et al., 1988; Gelso & Carter, 1985; Luborsky et al., 1983; Marmar et al., 1989;Marziali, 1984; Moras & Strupp, 1982). Morgan et al. (1982) hypothesized that theformation of the alliance \u00E2\u0080\u009Cmay be a result of a patient trait, such as priorrequirements and expectations in relationships, combined with the degree towhich the therapist fits these requirements and expectations\u00E2\u0080\u009D (p. 400). Studiescited above consistently show that clients who have experienced a history ofpositive interpersonal functioning--as measured by instruments such as thePatterns of Individual Change Scales (cited in Marmar, Weiss, & Gaston, 1989) orthe Wisman Social Adjustment Scale (cited in Marziali, 1984)--prior to therapyjoin more easily with their therapists than those having a history of highlyconflictual relationships. The former appear to have a capacity to engage inadaptive social interactions, enabling them to respond to the therapeuticrelationship. The latter \u00E2\u0080\u009Cadd stress to the affiance and demand from therapists acapacity to absorb and manage responses which are ambivalent, confusing, andobstructionistic (Marziali, 1984, p. 422). One\u00E2\u0080\u0099s capacity to form productive32attachments to others and trust others impacts one\u00E2\u0080\u0099s success in forming atherapeutic relationship.Research related to the impact of social functioning on the affiance isparticularly significant for certain client groups. Borderline patients--well-knownin clinical literature\u00E2\u0080\u0094were the object of a study by Frieswyk et al. (1986). Thesepatients have difficulty experiencing others as benevolent and seem predisposedto distrust anyone seeking to help them. Consequently, it is less likely that theywill enter into a good treatment relationship, even though it is just such adevelopment that is so critical. The affiance becomes a key outcome variablebecause it is there that borderline patients must face their panic, depression, lossof self-esteem, and fears of separation that can arise when they attempt to sustaina stable, coherent sense of self and others.The research is also particularly relevant for the marital therapist. Drydenand Hunt (1985a) point out that if one member of a couple has had a history ofgood interpersonal relationships, but the other has not, a therapist may find iteasier to form an affiance with the former and in the process, alienate the latter.Alcoholism. The alcohol literature reveals that a substantial number ofalcoholics fall into the category of having difficult histories with interpersonalrelationships. Most alcohol studies reveal that there are serious conflicts and roledysfunctions in alcoholic families (Moos & Moos, 1984). There is wide agreementthat though alcoholism is an etiologically and developmentally heterogeneousdisorder with a number of different subtypes, \u00E2\u0080\u009Cantisocial alcoholism is a majorsubtype of the disorder and is probably the one having received the greatestattention in clinical and research literature\u00E2\u0080\u009D (Jacob, 1992, p. 325). Research onchildren growing up in alcoholic families shows that of all the child outcomeslinked to inadequate parenting (e.g., lack of affection, high levels of criticism,inconsistent discipline, general lack of involvement), the development of33aggressive, antisocial behavior has the strongest documentation (Jacob, 1992).Moos and Moos (1984) note that the more families of alcoholics reportalcohol consumption and drinking problems, the more members complain ofhaving family arguments, less cohesion, and less agreement. The authorsobserve that increasingly researchers (e.g., Steinglass et al., 1977) are assuming thatalcoholism itself is not the major problem, and that problem drinking may evenserve an adaptive function for families that simply find it difficult to relate to oneanother.These findings are clearly relevant to the present research when consideringthe impact on the formation of the affiance and on outcome that (a) \u00E2\u0080\u009Clessagreement\u00E2\u0080\u009D may have--considering that agreement on tasks and goals constitutesa major portion of the operationalization of the construct--and that (b) antisocialpatterns of relating may have in the bonds subdimension. In the only publishedalliance study this author found (Bennun, 1989) involving a specified alcohol-dependent population (n = 18 couples with one spouse a problem thinker),significant correlations were found between outcome and client ratings oftherapist positive regard, competence, and direct guidance for both the problemdrinker and the spouse (range .47 to .61, <.01). These preliminary results,though requiring more studies for validation, at least indicate that an alcohol-dependent sample formed therapeutic affiances.In an historical overview of the research, Jacob (1992) concludes that thefamily interaction/alcoholism literature is at a very early developmental stage:Alcoholism has been defined, investigated, and treated as an individualproblem throughout most of the past century. Given this background, it isnot surprising that the earliest research from a family perspective involved afocus on individuals within the alcoholic\u00E2\u0080\u0099s family--most notably the spouseand, secondarily, the children.... Most important, the vast majority of work34in this area has been based on psychodynamic, individually orientedconceptual frameworks, despite the repeatedly implied interest ininterpersonal relationships. (p. 320)The literature shows few efforts describing patterns of interaction unique toalcoholic family interactions. Traditionally, alcohol research has focused onpersonality variables, but these efforts have resulted in the conclusion thatthough various personality types can be identified through cluster analysis, noneare unique to alcoholics (Graham & Strenger, 1988). Several studies demonstratethat alcoholics are not a homogeneous population, but have a great diversity insocial background, in personality types, and in onset, severity, and patterning ofstresses and responses associated with the disorder (Finney & Moos, 1979; Graham& Strenger, 1988; Jacob & Leonard, 1988).Because no one particular personality profile for alcoholics has surfaced,researchers are increasingly focusing on alcohol-use patterns (Morey, Skinner, &Blashfield, 1984). Findings are demonstrating that different patterns of drinkingsuggest different patterns of interpersonal relating. Morey et al. (1984) proposethat the interpersonal domain may be enough in itself to classify alcoholics.Three distinct types of drinkers have been identified in the literature (Jacob &Leonard, 1988; Morey et aL, 1984). Type A are early-stage alcoholics and representa very heterogeneous group without major symptoms of alcohol dependence.Type B alcoholics exhibit a moderate dependence, tending to drink on a dailybasis, and are more socially oriented, even gregarious, in their drinking habits.They tend to turn to alcohol as a way to facifitate affective expression in theirimportant relationships.It is the Type C alcoholic that could be presented with the greatest difficultyin terms of the affiance. Type C persons have a severe alcohol dependencycharacterized by loss of control and binge drinking. They are described as schizoid,35impulsive, and socially isolative, with more interpersonal problems than otheralcoholics, including fighting, arguments about drinking, marital disruptions, andthe highest levels of aggression. They show a trend to score lower on thecohesion subscale of the DAS, and their proffle on the MMPI is characterized by apattern suggestive of hostile and alienated individuals (Jacob & Leonard, 1988).Although Types B and C are both well along the alcohol-dependence continuum,they can be differentiated by a number of factors, especially these interpersonalvariables.Using clustering techniques, Babor et al. (1992) identified two types ofdrinking patterns divided along very similar lines to the three described above.Type A drinkers have a later onset and are characterized by a steadier pattern ofdrinking. Type B alcoholics, like the above Type C, have more family risk factors,earlier onset, and a binge drinking pattern.Lift, et a!. (1992) used Babor et al.\u00E2\u0080\u0099s (1992) dual categorization of drinkingtypes and found a significant main effect for patient type in predicting socialbehavior when pretreatment level was controlled for. They suggest that Type Bdrinking patterns predict sociopathic kinds of behavior: impulsive,untrustworthy, unable to learn from experience, incapable of anticipating thereactions of others, insensitive to society\u00E2\u0080\u0099s expectations. Particularlyunresponsive to social approval or disapproval, an important clinical applicationis that they seem to respond better to concrete factors in their environment. Notsurprisingly, Lift et al. observed that Type B alcoholics have better outcomes withstructured coping skifis treatment and worse outcomes with interactional therapy.Type A alcoholics, on the other hand, do better in interactional treament andmore poorly with coping skifis training, perhaps finding the latter too retrictive.The anti-social nature of the Type B alcoholic has implications for the36formation of the therapeutic affiance. Client disposition, including traits likeopenness, motivation, or hostility, is a pretreatment variable that has been shownto influence affiance formation in individual therapy. Bordin (1979) suggestedthat clients\u00E2\u0080\u0099 openness \u00E2\u0080\u009Cto accept a particular goal of treatment may turn out to beintimately linked to capacities or dispositions, which in turn are related to howeasy it is for [them] to collaborate in the particular mode of treatment directedtoward that goal\u00E2\u0080\u009D (p. 256). In one study (Marziali et aL, 1981), those showing theleast positive change appeared to bring \u00E2\u0080\u009Ca negative disposition to the treatmentsituation that persisted.. . and was relatively intransigent to the therapist\u00E2\u0080\u0099s effortsto shore up the alliance\u00E2\u0080\u009D (p. 363). Kokotovic and Tracey (1990) observed that themore hostility as well as the poorer the current and past relationships of a client,the weaker the alliance that was formed. Finally, Marmar et al. (1989) foundclient motivation for psychotherapy to be positively associated with the affiance.These findings must be interpreted cautiously in light of how socialrelationships and hostffity were measured. Ratings done by counsellors regardingthese variables could have been reactive to and biased by therapist perceptionsregarding the alliance. If the counsellor viewed an alliance as strong, he or shecould easily have judged the client\u00E2\u0080\u0099s outside relationships as being better or levelof hostility as being less, regardless of their actual quality. Nevertheless, otherstudies also show that hostile, negative, resistive clients have greater difficultyforming affiances, resulting in poorer outcomes (Frieswyk et al., 1986; GomesSchwartz, 1978; Marziali, Marmar, & Krupnick, 1981; Strupp, 1980).Two family alcohol studies suggest conjoint treatment is superior toindividual treatment (McCrady et al., 1986; Zweben, Pearlman, & Selina, 1988).McCrady et al. (1986) found that the most rapid, long-lasting results occur whenspouses of alcoholics are involved, especially when treatment focuses not only onthe alcohol problem but also on marital behavior. The authors speculate that37most alcoholic marriages are so stressed that unless nonalcoholic partners areactively involved in treatment, their pessimism, anger and past disappointmentsmay lead them to become discouraged too quickly. Zweben et al. (1988) alsoemphasized the critical role of the spouse in effecting and maintainingtherapeutic change. Their assumption is that promoting the interaction andcollaboration between spouses is a crucial step in the resolution of the alcoholproblem. Spouses in this study had first-hand experience of the treatmentprocess. They participated in the development and implementation of methodsto limit or end drinking, provided input and feedback, and supported thedrinking partner by buttressing motivation and resolve. In both of these studies,making use of the marital alliance promoted outcome.To conclude, while the research demonstrates that certain pretreatmentinterpersonal variables are related to the establishment of the therapeutic affiance(e.g., pretherapy social adjustment), relationships between other pretreatmentvariables (e.g.., marital distress, symptomatology, and alcoholism) and the allianceneed further study. Questions for the present study include how alcoholics withvarying social histories and drinking patterns will form therapeutic affiances intwo different treatment approaches. Also, wifi pretreatment marital distressindicating strained social relationship predict low affiance scores?Treatment FactorsResearchers have attempted to detect first, how treatment type affects therelation between alliance and outcome and second, whether treatment type has asignificant bearing on the formation of the affiance.With regard to the first question, Horvath and Symonds (1991), reviewing 20distinct data sets from studies on the alliance, discovered that the type of therapypracticed did not impact significantly the alliance-outcome relationship.38Treatment approaches included psychodynamic, eclectic, cognitive, and Gestaltinterventions. All reported similarly strong alliance-outcome relationships thatwere statistically significant ( < .05). Contrasts were also evaluated, none ofwhich were significant at < .05.The authors propose that perhaps a construct like the therapeutic alliancethat emphasizes collaboration and mutuality will \u00E2\u0080\u009Ccovary\u00E2\u0080\u009D exactly with aspects ofimprovement or gain reflected in outcome. The forming of a positiverelationship with the counsellor, the following of therapeutic tasks, and thecollaboration on goals provide \u00E2\u0080\u009Cde facto evidence of success in the therapysituation.... Collaboration is both a predictive process variable and in vivoevidence of success\u00E2\u0080\u009D (p. 147). As symptoms improve, so often does the quality ofthe alliance (Marmar et al., 1989).As to the second question, the effect of treatment type on alliance formation,no published empirical data demonstrate that certain kinds of treatments result instronger therapeutic alliances than do others. This makes sense theoretically. If(a) all treatment types have positive outcomes for some clients (Bennun, 1989;Gurman & Kniskern, 1981; Horwitz, 1974; Smith et al., 1981), and if (b) positiveoutcomes are predicted by strong affiances, then (c) all effective treatments wifiinclude mechanisms for building therapeutic affiances. If one treatment approachresulted in strong alliances with all clients, then presumably positive outcomewould result for all clients, and that approach would likely replace all others. Asit is, certain interventions appeal to some clients and turn others away.This is not to say that different types of treatment do not produce differenttypes of alliances. Bordin (1979) hypothesizes that the quality and development ofthe working affiance depends in part upon the counsellor\u00E2\u0080\u0099s theoretical approach.He maintains that all genres of psychotherapy contain embedded alliances, sodifferent treatments--marked by different demands made on client and therapist--39will produce different kinds of alliances. Other researchers agree (Dryden & Hunt,1985a; Safran & Wallner, 1991), noting that the real question is not whether theaffiance is more important in one approach, but rather whether the affiance isbeing measured in a maimer appropriate to that approach. The natuie of thebonds will vary as will agreement on tasks and goals, but ultimately, what iscritical is that agreement and bonds are established.Gelso and Carter (1985) suggest that the quality of the working alliance isrelated to the difficulty of treatment, part of which involves client vulnerability.The strength of the affiance is a function of the closeness of fit between the kind ofworking affiance demanded by the treatment and the personal characteristics ofthe client and therapist. A treatment approach that does not fit, will result in astrained affiance in which a client is not in agreement with the tasks and goals oftherapy, and as a result is not very bonded to the therapist. Without agreement,outcome wifi be predictably lower than for the client who trusts the therapist andenthusiastically embraces the objectives and activities of therapy.In one study (Marmar et al., 1989), judges who rated therapy sessions notedthat the more clients were exposed to experiential opportunities, the more theycontributed positively to the affiance. Improvement in interpersonal functioningand symptom improvement also strengthened the affiance. Therapist challengesaddressing patient resistances, on the other hand, seemed to affect the alliancenegatively. It cannot be assumed, however, that treatments that stronglyemphasize the therapeutic relationship (e.g., promoting empathy, or here-andnow experience), will necessarily result in higher alliance measures. Theseapproaches, in fact, may be less desirable to certain clients who prefer beingdirected and who feel threatened by \u00E2\u0080\u009Crelationship\u00E2\u0080\u009D models (Bordin, 1979).Bourgeois et al. (1990) demonstrated that a highly structured group interventionwith a very low emotional emphasis resulted in strong therapeutic affiance scores40and a strong relation of affiance to outcome. Even though the therapy did notemphasize a therapeutic relationship, bonds were strong, and there was highagreement on tasks and goals.In the only other marital study that considered the alliance and treatment(Gruman, 1986), therapeutic approach did appear to affect the strength of theaffiance. An emotionally-focused treatment resulted in significantly strongeralliances than did an interactional systemic treatment. Nevertheless, the form ofthe affiance was similar in both therapies and stayed similar over time, theintercorrelations between subdimensions not changing from the third session tothe final tenth session in either treatment. The author concludes that once bondsand agreement on goals and tasks are established, they remain relatively strong,regardless of treatment type.As in individual research, no family treatment has produced proof ofuniversal effectiveness (Gurman & Kniskern, 1981; Gurman et al., 1986). Infamily research, there has been very little controlled empirical study of\u00E2\u0080\u009Cnonbehavioral\u00E2\u0080\u009D approaches and their salient treatment components. Processdata have almost never been gathered in those studies on which analyses arebased (Gurman et al., 1986). In the absence of enough empirical research,suggesting the effect of treatment on the alliance-outcome relationship or on theformation of the affiance would be premature.Gender and the AllianceIndividual psychotherapy affiance research has not reported on gendereffects. Several theorists and researchers in the family field have commented onthis important factor. Two empirical studies (cited in the alliance-outcomesection) agree that the strength of the alliance is a more powerful determinant oftherapeutic success among men than among women (Bennun, 1989; Bourgeois et41al., 1990). Bennun found that none of the factors rated by women showedsignificant corrrelations with outcome, whereas men\u00E2\u0080\u0099s ratings did. When men ingeneral (n = 35) perceived the therapist to be competent and directive, outcomewas positively rated (.58* and .71* correlation with outcome respectively). Formen identified as patients (a = 10), ratings of positive regard and competencycorrelated with outcome (.71* and .58*), and direct guidance fell to .38 (* = <.01).The 10 identified patients evidently valued direct guidance less and positiveregard more than men in general did.It is perhaps significant that whereas men\u00E2\u0080\u0099s perceptions--whether they werethe identified patients or not--interacted with outcome, women\u00E2\u0080\u0099s perceptions ofthe therapist only interacted with outcome--and then, minimally--if she was theidentified patient. Bennun (1989) notes that the existing theoretical familyliterature supports this data, stressing the importance of the father\u00E2\u0080\u0099s position andparticipation in marital and family therapy (Lewis & O\u00E2\u0080\u0099Brien, 1987; O\u00E2\u0080\u0099Brien, 1988for reviews) and suggesting that women are less powerful in determining theoutcome of a family therapeutic encounter (Ker and McKee, 1981). Men arehypothesized to be more reluctant to go to counselling than their wives, toremain more detached, to be less likely to involve themselves in the process, andto have higher drop-out rates (Bourgeois et al., 1990; Heubeck, Detmering, &Russell, 1986).In terms of suggested treatment approaches, men are theorized to preferactive structured therapies and are comfortable with advice and goal-setting,whereas women are said to prefer more non-directive, reflective styles (Bourgeoiset al., 1990; Dryden & Hunt, 1985a). Men are less likely to see the relevance of thetasks of self-disclosure and self-exploration to the solutions of their problems thanwomen are (Dryden & Hunt, 1985b). They prefer a positive prognosis and atherapist who obviously likes the family (Heubeck et at, 1986).42Bourgeois et al. (1990) attribute their finding that men had a higher affiance-outcome correlation to the fact that their treatment was highly structured andcognitive and set out the tasks and goals in a clear manner. The authors citeBerger (1973) who observes that men\u00E2\u0080\u0099ss social roles do not permit them to expresstheir anxieties and fears without compromising their personal image.Bennun (1989) concludes that \u00E2\u0080\u009Csystems approaches based on the theory thatfamily members, particularly within subsystems, are equal interacting parts inrecursive complementarities, may only be part of the picture\u00E2\u0080\u009D (p. 251). The twostudies described indicate that the wives\u00E2\u0080\u0099 position is not equal to their husbands\u00E2\u0080\u0099in determining outcome. Much more research into this area is needed.Intact and Split AlliancesIntact and split alliances (Heatherington & Friedlander, 1990b; Pinsof &Catherall, 1986) are an important therapeutic consideration that requires muchstudy. With an intact affiance, all family members feel positively toward thetherapist and therapy whereas with a split alliance, some will feel positively andothers feel negatively or neutral. Whether the therapist must maintain a positiverelationship with the strongest member of the family when there is a splitaffiance in order for therapy to proceed remains to be tested (Pinsof, in press).Pinsof and Catherall\u00E2\u0080\u0099s (1986) theoretical concept of the split alliance has beenidentified empirically and now needs refinement. Heatherington andFriendlander (1990b) used one standard deviation difference in ratings from themean difference score to constitute a split affiance. Out of a total of 12 couples and12 families, 5 couples (43%) and 9 families (75%) fell into this category. For twostandard deviations of difference, 14% of the couples and 42% of the familiesqualified.43Gruman (1986) also tested the concept of the split affiance and found36% or 15 of 42 couple partners to have alliance scores within 10 points of eachother, and 19% or 8 to differ by 20 points or more. The mean difference score was14.57 with a standard deviation of 10.13. Gruman also compared the outcomes ofthe couple partners who both scored high with those who both scored low andfound no significant differences. She failed, however, to control for treatmenteffect when combining couples\u00E2\u0080\u0099 scores from two very different treatments.Fisher et al. (1985) notes that difference scores among family members havebeen used to predict a variety of outcome variables. Greater discrepancies betweenpartners have been associated with less satisfaction, more conflict, and morestress. Bennun (1989) found that the greater the discrepancy between parents, thepoorer their clinical outcome. By comparing families having the 10 bestoutcomes with the 10 having the poorest outcomes, Bennun also discovered thatthose with favorable outcomes differed significantly in their perceptions of thealliance from those with unfavorable outcomes.Client. Therapist. or Observer reportTheorists initially speculated that to obtain a proper assessment oftherapeutic process variables, an objective observer should be used, the patientand therapist being too caught up in the process to give an accurate measurement.It was thought that patients would have an \u00E2\u0080\u009Cinherent inclination towarddistorted perceptions of therapeutic interactions\u00E2\u0080\u009D (Marziali, 1984, p. 417). Similarbiases were assumed to contaminate therapists\u00E2\u0080\u0099 ratings. Empirical studies haveshown otherwise, that patient and therapist ratings of the affiance are powerfulpredictors of therapeutic change. Horvath and Greenberg (1986) cite a number ofempirical studies using participant report of client-centered dimensions thatpredicted outcome better than those using third party evaluations (Gurman, 1977;44Lambert & DeJuliio, 1977; Parloff, Waskow, & Wolfe, 1978). Marziali (1984),using three sources of report, concluded, \u00E2\u0080\u009CIt may be that the therapeuticparticipants provide the more authentic versions of the quality of the treatmentrelationship\u00E2\u0080\u009D (p. 422).Results are mixed regarding the interaction of client and therapistobservations. Some studies show agreement on client and therapist assessmentof the affiance (Kokotovic & Tracey, 1990; Marziali, 1984), whereas the majorityreport significant differences in the two perspectives. Horvath and Marx (1990)found that client and therapist reports had different points of reference for bothsession quality and affiance components, and Tichenor and Hill (1989) show thattherapist and client measures on the Working Alliance Inventory (Horvath &Greenberg, 1986) are not significantly related to each other (nor were they relatedto any of the four observer-rated measures used in the study). Marziali et al.(1981) found high internal consistency for their two scales measuring boththerapist and patient contributions to the alliance, and low interscale correlations,supporting the idea of there being separate dimensions of the affiance. In areview of 23 studies using client and therapist reports, Gurman (1977) alsoindicates that therapists and clients have different perspectives on the affiance(Horvath & Greenberg, 1986).Horvath et al. (1990) explain that clients typically view therapy as a way tosolve painful and distressing problems, whereas therapists can approach it from amore general, theory-driven perspective. Consequently, clients may interpret thealliance from the point of view of their distress and how therapy is easing it, andtherapists may perceive the relationship only in terms of how well interventionsseem to be going.Many argue that client report is superior to either therapist or observerpoints of view. Though possibly \u00E2\u0080\u009Csubjective,\u00E2\u0080\u009D self-reports are not biased by45theoretical preconceptions. In addition, self-report can point out non-observableaffective and cognitive components of a client\u00E2\u0080\u0099s experience of the relationship,and provides a very different perspective about the interactive process than eitherthird party measures or therapist assessments of behavior (Gruman, 1986).Luborsky et al. (1983) describe a client who announced after her first session thatshe could be helped by her therapist, not because she agreed with him on the tasksand goals of therapy, but because he reminded her of the man with whom shewas having an affair. In such cases, client report is really the only way inside theworld of the client. The therapist or observer, having no access to her innerthoughts, may have either misattributed her responses to agreement on tasks andgoals, or not observed anything at all.In terms of relationship to outcome, it appears that patient report has morepredictive validity. Morgan et al. (1982) cite the Barrett-Lennard RelationshipInventory and Orlinsky and Howard\u00E2\u0080\u0099s Therapy Session Questionnaire asevidence of the value self-report can have in predicting outcome. Reviewing 13studies in which both client- and therapist-based alliance/outcome relations werereported, Horvath and Symonds (1991) found that in 9 of the 13 cases, clients\u00E2\u0080\u0099ratings were superior predictors of outcome. Marziali et aL (1981) found only thepatients\u00E2\u0080\u0099 contribution to the therapeutic alliance to be predictive of outcome.Orlinsky and Howard (1986) condensed findings from hundreds of process-outcome studies over a 35-year period and concluded,The separate and joint contributions of therapists and patients to variousdimensions of the therapeutic bond have received intensive study.These dimensions were very consistently related to patient outcome. Thiswas especially true when process measures were based on patients\u00E2\u0080\u0099observations of the therapeutic relationship. (p. 365)46To conclude, it appears that client, therapist, and observer reports domeasure different aspects of the affiance, and that client report has the greatestpredictive validity. Employing two or more different measurements on the samedata base would help to ensure validity. The scales used in the present study areclient-reports. The authors\u00E2\u0080\u0099 rationale for designing client-rated measures is givenin the methodology section.Summary and Extensions of the LiteratureThe literature reveals a growing recognition that continued empirical studyinto the therapeutic affiance is needed. The construct itself remains in adevelopmental phase, there stifi being a significant lack of unanimity as to how itoperates and what factors need to be considered in its measurement. What hasachieved broad consensus in both individual and family research is theassumption that the affiance is an integral aspect of therapy, if not as the changemechanism itself, then certainly as a prerequisite for change. Research within thecontext of individual psychotherapy convincingly demonstrates the relevance ofthe alliance to treatment outcome in a variety of treatment and client situations.The area needing a great deal more attention is within the realm of marriageand family. The only instruments using systemic theory as their basis, the ITAS,CTAS, and FTAS, require further study and refinement, their psychometricproperties still relatively untested. Without data rooted in systems ideas, avacui.m has existed not only within the marriage and family field but also withinindividual psychotherapy, where every individual client, though treated alone, isa part of a number of systems.The research shows that no client variable or therapist action taken inisolation predicts either the strength of the alliance or treatment efficacy. Rather,there appears to be a \u00E2\u0080\u009Csubtle interplay of patient characteristics, therapeutic47strategy, and their interactive impact on affiance formation and its relation tooutcome\u00E2\u0080\u009D (Frieswyk et aL, 1986, p. 35). Particularly in family research, much morework remains to be done concerning the relationship of the affiance to clientsymptomatology, social adjustment, marital distress, and substance abuse. Inaddition, the impact of therapeutic approach and of gender on the alliance areimportant fields of inquiry.Clearly, the marital and family therapy contexts require their owndistinctive, theory-driven research into the alliance. The present researchexamines--within a systemic context--the impact of treatment, gender, andalcohol-dependency upon the therapeutic alliance, as well as the extent to whichthe affiance facifitates or interacts with outcome in both individual and couplesformats. Additionally, intact and split alliances from the couples format areanalyzed.Research HypothesesFollowing are the hypotheses for the current study along with briefexplanations of how each will contribute to the literature.Hypothesis OneThe therapeutic alliance as measured by the Individual Therapeutic Alliance Scalewill not be significantly different in Experiential Systemic Therapy (for individuals)and Supported Feedback Therapy.The individual psychotherapy literature indicates that the type of treatmentapproach does not significantly impact the strength of the affiance. Thishypothesis supports that view and uniquely contributes to the literature by48extending the findings to two very different and novel treatments. Additionally,the sample on which the treatments were used was one no previous allianceresearch has yet tested, that of the alcohol-dependent client.Hypothesis TwoDrinking patterns of alcoholics will result in significant differences in the strength ofthe therapeutic alliance, binge drinkers having lower therapeutic alliance scores thanchronic drinkers.This investigation is the first empirical effort known to this author toexplore the relationship of drinking behavior to the therapeutic affiance. Theliterature suggests that binge drinkers have a more difficult time withinterpersonal skills and typically have histories characterized by antisocialbehavior and hostility. Chronic drinkers, on the other hand, are often quitesocial, even described as gregarious. Consistent with this body of literature, thisstudy predicts that chronic drinkers will form stronger therapeutic alliances thanwill binge drinkers.Hypothesis ThreeThe therapeutic alliance will moderate differential gains (as indicated by pre andposttest DAS scores) produced by individual and couples formats of ExST. It ishypothesized that the alliance will positively explain a greater proportion oftherapeutic gain in the individual format than it will in the couples format.This question essentially seeks to apply the clear association found inindividual psychotherapy literature between affiance and outcome to the systemicframework. The question focuses specifically on ExST, attempting to discern whatimpact, if any, increasing the number of clients from one to two has on the49alliance. At a theoretical level, it has been suggested that the complexitiesinherent in family therapy complicate the conceptualization of the alliance(Pinsof, 1989). This question attempts to provide empirical data to substantiatewhat has up to now made sense only at a theoretical level. How the nature of theaffiance differs in family therapy versus individual therapy and whether thealliance is as potent in family therapy, where alliances among clients are theprimary relationships to be developed in therapy are questions related to thishypothesis.No prior published study discovered by this author has comparedempirically the relationship of affiance to outcome for individuals and couplesusing the same treatment. No study has even gone so far as to compare thestrength of the alliance using the same treatment within both individual andmarital formats, let alone examine the relationship of alliance to outcome in thetwo modalities. Based on the theory that couples counselling results in lesspersonal attention for each client individually, the prediction is that themoderating impact of alliance scores on outcome will be lower in the coupl&sformat than in the individual format.Hypothesis FourWithin the couples format of ExST, women will report a higher level of therapeuticalliance than will men.This hypothesis examines the effect of gender upon the alliance in thecouples format. Only three previous studies have considered gender and thealliance in the family context (Bennun, 1989; Bourgeois et al., 1989; Heatherington& Friedlander, 1990b). Because the literature supports the idea that females tendto favor more experiential treatments whereas males prefer more cognitive,50behavioral approaches and are more likely to resist therapy (Hunt, 1984; Heubecket al., 1986), it is predicted that women will have higher scores than men.Hypothesis FiveWith regard to intact and split alliances between partners in the couples format ofExST:(a) overall agreement between partners on the strength of the alliancewill predict a higher outcome than will less agreement; and(b) in cases with split alliances, outcome will be improved when theman has the higher score compared to when the woman does.These predictions are extensions of issues considered in hypotheses three(relationship of alliance to outcome) and four (male and female scores). Theconstructs of split and intact affiances have been described theoretically (Pinsof &Catherall, 1986), and identified empirically (Gruman, 1986; Heatherington &Friedlander, 1990). No published study this author found has examined therelationship of split and intact alliances to outcome.Bennun (1989) observed that the greater the discrepancy between a couple\u00E2\u0080\u0099sperceptions of the therapist, the poorer the couple\u00E2\u0080\u0099s gains in therapy. In 5 (a) thisstudy predicts that the same trend will be seen, namely that intact couples wifimake greater therapeutic gains that will split couples.Number 5 (b) focuses only on split affiances, testing the hypothesis that themost powerful spouse must have the stronger affiance with the therapist in orderfor therapy to be effective (Pinsof, in press). Several studies have shown affianceratings of women to be less correlated with outcome or not at all, whereas men\u00E2\u0080\u0099sratings did correlate (Bennun, 1985; Bourgeois et al., 1989). Other studies havefound women to be less powerful in determining the outcome of a family51therapeutic encounter (Bennun, 1989; Lewis & O\u00E2\u0080\u0099Brien, 1987 and O\u00E2\u0080\u0099Brien, 1988,for reviews). Based on this literature, it is predicted that split alliances withhigher male scores will result in greater therapeutic gain than those with higherfemale scores.52Chapter IIIMETHODOLOGYThis chapter describes the methodology used in the present study, includingthe design of the study, the demographics of the subjects and therapists, thetreatment models employed, and the division of the sample into binge andchronic drinking types. Data collection procedures are delineated, and conceptualand psychometric information on the instruments is provided. Finally, theresearch questions to be tested are presented.Design of the StudyThe present research is one aspect of a much larger study, The AlcoholRecovery Project (TARP) (see Appendix F). Structured according to a repeatedmeasures experimental group design, TARP screened 150 alcoholic men and theirpartners for participation, according to the guidelines set out below. The presentstudy used a process-outcome approach (Heppner, Kivlighan, & Wampold, 1992;Pinsof, 1989) to investigate (a) factors affecting the strength of the therapeuticalliance, (b) the degree to which the strength of the affiance facffitates or interactswith a marital outcome measure, (c) the degree to which the strength of theaffiance-outcome relationship differs in individual and couple formats, and (d)the extent to which agreement between partners on the strength of the affianceinteracts with outcome.TreatmentsIn the present study, two different treatment approaches, both of themsystemic, were used--Experiential Systemic Therapy (ExST) and Supported53Feedback Therapy (SFT). Two different formats of ExST were studied, one usingindividuals and the other couples. SFT was used with individuals only.Experiential Systemic Therapy (ExST). This is an integrative approachdeveloped specifically for the treatment of alcoholics. It can be employed withinindividual and family formats. Cessation of drinking is the primary goal in bothform.ats, but different levels of the alcoholic family system are targeted toaccomplish this. The focus within both formats is on the systemic relationshipsof the alcoholic. In individual therapy, alcohol use and other intrapersonal issuesare examined and related to interpersonal issues involving marital, family, andcommunity contexts. In marital therapy, the emphasis is on intrapersonal andinterpersonal issues and mobilizing couple& resources to cope with theinterpersonal stress that goes with the transition to sobriety. Changing theinteraction pafterns of the dyad is one part of this.This treatment model is experiential, present tense, egalitarian, and uses afairly high level of intensity. Techniques include the use of metaphor,externalization, rituals, and symbols (Friesen, Grigg, & Newman, 1991).Supported Feedback Therapy (SFT). This is a comparison model developedfor individual clients. Designed to contrast as noticeably as possible with ExST, itis a highly structured, action-oriented approach, focusing on behaviors of thealcoholic during the preceeding week. Minimal therapist involvement isrequired other than relationship-building involving empathy and positive regardand directing the client to fill in charts that detail any variations in behavior, suchas alcohol consumption, ease in achieving abstinence, relationship to self, tomarriage, to family, to friends, and to work (Grigg, Friesen, Weir, & Bate, 1991).54SampleThe present study used the 42 individuals and 21 couples who hadcompleted treatment and all research instruments by the time data analysis began,approximately 18 months after TARP began screening clients. The volunteersample was selected from the Lower Mainland and Central Vancouver Island\u00E2\u0080\u0099salcoholic population who met the following requirements during the screeningprocess:\u00E2\u0080\u00A2 families were composed of (a) an alcohol-dependent father who scoredabove the critical cutoff score of 5 on the Michigan Alcohol Screening Test(MAST) (Seizer, 1971), (b) a non-alcohol-dependent mother who scored lessthan 5 on the MAST, and (c) at least one child over the age of four living athome or in regular contact with the couple;\u00E2\u0080\u00A2 both partners complained of significant dyadic distress, each scoring belowthe critical value of 99 on the Dyadic Adjustment Scale (DAS) (Spanier, 1976);\u00E2\u0080\u00A2 though experiencing relational distress, both partners were living togetherduring treatment, and had lived together either as a married or common-lawcouple for at least a year;\u00E2\u0080\u00A2 neither partner had a severe psychiatric disturbance as indicated by anexceptionally high score on either the depression or psychiatric subscales ofthe Symptom Checklist Revised (SCL-90-R) (Derogatis, 1983).During a screening session, subjects filled in the above instruments andseveral others. Those accepted into the project were randomly assigned to one ofthree treatment groups: (a) SFT, (b) ExST-Individual, and (c) ExST-Couples. Forthe present study, results were derived from 20 men treated by SFT, 22 by ExSTIndividual, and 21 couples by ExST-Couples.I-tests and ANOVAS revealed that there were no significant differencesamong the treatment groups regarding age, socioeconomic status, number ofchildren, and number of years married. Using the Alcohol Dependency Data55Questionnaire (ADDQ) (Raistrick, Dunbar, & Davidson, 1983), the DAS, the SCL90-R, and the Family Satisfaction Inventory (FS) (Olson & Wilson, 1982), aMANOVA tested the characteristics of alcohol dependency, dyadic distress,psychiatric symptomatology, and family satisfaction across the three groups.Again, no significant differences were found.Table 1 contains the demographic data on the sample used for this study.Table 1: Demographics of SampleNumber Marital Mean Yrs Joint Averageof Status Together SalaryC\u00E2\u0080\u0099hi11ren (fl\u00E2\u0080\u0099 (1 flflOc;\u00E2\u0080\u00A213% one men49% two 79% mrried 11.83 15.2% $60 (63)29% three 21% c.-law (7.94) 64.2% = $20-596% four 17.6% $19 women 37.58 47% full3% five (63) (8.87) 29% part21-65 18%unempl.TherapistsTwelve therapists were involved in TARP, all of whom were at a Master\u00E2\u0080\u0099slevel or higher in psychology, social work, or a related field, and all having had atleast three years of experience working with clients exhibiting alcohol and drugproblems. Counsellors underwent systematic training in the model they were touse. Their work was supervised on a weekly basis, and training manuals for eachtreatment were made available to help ensure that they were adhering to themodel they were employing.Five therapists delivered the individual and couple forms of ExST, andseven provided SFT treatment. Demographic data for the therapists is providedMean Age(SJ2)39.69(8.73)26- 70EmploymentStatus67% full6% part24% unempi.56in Table 2. Therapists working in the ExST treatment were somewhat older withmore years of experience overall as well as in the number of years working withalcoholics. Overall, counsellors in both formats had a considerable degree oftherapeutic experience, particularly in the field of alcohol.Table 2: Demographics of TherapistsAverage Ave Years Alcohol TherapyAge Experience Experience Gender(SD) (SD) (SD)ExST therapists 42.40 9.2 6.10 2 Female(n = 5) (7.8) (4.55) (2.46) 3 MaleSFT therapists 36.14 7.40 5.43 2 Female(a = 7) (4.45) (4.43) (3.74) 5 MaleOverall 38.75 8.17 5.70 4 Female(a = 12) (6.58) (4.37) (3.25) 8 MaleDivision of Sample into Drinking TypesAlcohol dependency was assessed along multiple dimensions. Instrumentsincluded the Michigan Alcohol Screening Test (MAST) (Seizer, 1971) and theAlcohol Dependency Data Questionnaire (ADDQ) (Raistrick et al., 1983) todetermine severity of dependency. The sample was classified into two groups,binge and chronic, based upon the Binge Chronic Differentiation Scale (BCDS), ameasure consisting of seven items constructed from the widely-used and highly-respected Marlatt Drinking Proffle. (See Appendix A for the BCDS.) Three ratersexamined items from the BCDS and came to a consensus on which category ofdrinking behavior best described each client.The sample for the present study includes 7 binge and 13 chronic drinkers forSFT, 12 binge and 10 chronic drinkers for ExST-Ind, and 11 binge and 10 chronic57drinkers for ExST-Couples. Table 3 contains the results of t-tests measuring thedifferences between the two groups in this study. (Scores for one binge drinkerwere unavailable.) In each case, chronic drinkers\u00E2\u0080\u0099 scores indicated significantlygreater likelihood of turning to alcohol when presented by a particular variable.Table 3: Means, Standard Deviations and t-test values:Variables for Binge and Chronic Drinking TypesDrinking TypeBinge Chronic(=29) (=33)Variable SD-testPhysical Discomfort 18.97 19.53 34.85 29.13 0.01Urges & Temptations 41.67 24.40 59.25 27.65 0.01Pleasant Emotions 48.28 23.72 66.76 22.28 0.003Unpleasant Emotions 50.29 24.14 62.12 29.10 0.09Conflict with Others 40.71 20.83 51.43 28.90 0.10Social Pressure to Drink 55.47 28.89 67.25 24.59 0.09A chi-square analysis also indicated that factors of drinking pattern and placeof drinking (in home or out-of-home) were significantly related ( = .02). Theobserved number of in-home chronic drinkers (66%) and out-of-home bingedrinkers (67%) was higher than would be expected by chance alone.Variables not significantly different between binge and chronic drinkersincluded Standard Drinking Units, Personal Control, Age of Onset of DrinkingProblem, and Problem Drinking in Extended Family. Additionally, mean scoretotals on the MAST and ADDQ were not significantly different. Three chi-squaretests produced insignificant results when comparing binge and chronicfrequencies with Time of Drinking, Solo Drinking, and Use of Other Drugs.58Finally, no significant differences between binge and chronic differences could bedetected among the three treatment modalities of SFT, ExST-Individual, andExST-Couple. It should be noted that these results are not generalLzable to allalcoholics because the sample is made up only of alcoholics stifi living in intactfamilies.Data Collection ProceduresAfter screening, data was collected at four measurement occasions: (a)pretreatment, (b) midtreatment, following 7 to 10 weeks of therapy, (c)posttreatment, after 8 to 10 more weeks of therapy, and (d) follow-up, 15 weeksafter termination. In addition, continuous measures of drinking-related behaviorwere implemented through Weekly Situation Diaries and Post-session Reviews.Details like support group meeting participation, work attendance, and quantityand frequency of alcohol consumption were included in these instruments.Before fifing in any instruments, clients were assured that all answerswould remain confidential and anonymous. Questionnaires were completed inprivate, clients sealing them in envelopes before returning them to theresearchers. Clients understood that the data would be used for research purposesonly and would not be made available to their therapists.Midtreatment data packets included the Therapeutic Alliance Scales (TAS).The therapists themselves had no connection with the administration orcollection of these questionnaires. Midtreatment--after session five--was chosenfor the affiance instrument because of research showing that when measured inthe three to six session range, the affiance is predictive of outcome (Adler, 1989;Hartley & Strupp, 1983; Heatherington & Friedlander, 1990b; Horvath &Greenberg, 1986). As noted in the review of the literature, by this point in time inshort-term treatment, relational patterns will be stabilizing, an affiance59developed, and treatment goals established.Although only alcoholic men were directly involved in the therapy sessionsin the SFT and ExST-Individual groups, women in all three treatments carefullymonitored themselves, their relationship, and family dynamics on a weekly basis.Information from the women\u00E2\u0080\u0099s weekly records will be used by TARP, but was notincluded in the sessions and was kept confidential. To help compensate for theconsiderable time commitment required to fill in the many instruments and tominimize attrition, subjects received between $150 and $200 over the course of theproject. All therapy sessions were videotaped.Instruments: Conceptual and Psychometric DataThe scales used in this study were the Dyadic Adjustment Scale (DAS)(Spanier, 1976) and two of the three Therapeutic Affiance Scales, the IndividualTherapeutic Alliance Scale (ITAS) and the Couple Therapeutic Affiance Scale(CTAS) (Pinsof & Catherall, 1986). The DAS was administered at screening andposttreatment, and the TAS at midpoint.Dyadic Adjustment ScaleAdjustment is defined as \u00E2\u0080\u009Can ever-changing process [vs. an unchanging state]which can be evaluated at any point in time on a dimension from welladjusted to maladjusted\u00E2\u0080\u009D (Spanier, 1976, p. 17). Defined as a process, the outcomeof adjustment is operationalized by examining four areas of conjoint living:satisfaction, cohesion, consensus, and affectional expression. Researchers usingthe scale may use one subscale alone without losing confidence in the reliabilityor validity of the measure (Spanier, 1976, p.22). The scale has a theoretical rangeof 0 to 151. Scores under 100 are considered to indicate a high level of distress.60(See Appendix B for complete scale.) Table 4 shows DAS subscale information.Table 4: DAS Subscale Descriptions and Cronbach ReliabilitiesSubscale Definition Reliability(No. items)Consensus the degree to which couples agree on matters .90(13) of importance to the relationshipCohesion the degree to which couples engage in .94(10) activities togetherSatisfaction the degree to which couples are satisfied with .86(5) the present state of the relationship and areconmiitted to its continuanceAffectional the degree to which couples are satisfied with the .73Expression expression of affection and sex in the relationship(4)Reliability of the DAS and its subscales was determined using Cronbach\u00E2\u0080\u0099scoefficient alpha (Anastasi, 1988). Results for the subscales appear in Table 4. Thetotal scale reliability is .96. This data indicates that both the overall scale and itsdifferent parts have sufficiently high reliability to justify their use. A separateassessment of the scale\u00E2\u0080\u0099s reliabifity using the Spearman-Brown average inter-itemformula for internal consistency also resulted in .96 (Spariier, 1976).Reliabifities of pre and posttreatment DAS for the present study, computedusing Cronbach\u00E2\u0080\u0099s coefficient alpha, are listed in Table 5.61Table 5: Cronbach\u00E2\u0080\u0099s Alpha for Pre and PosttreatmentDAS Total and SubscalesScale Pre-DAS Post-DASTotal .91 .93Consensus .85 .87Satisfaction .83 .87Affec/Expres .64 .74Cohesion .82 .79Validity of the DAS was examined in several ways by Spanier (1976). Threejudges evaluated potential items for content validity and included only those thatwere considered to be relevant measures of dyadic adjustment as defined by thefour components of satisfaction, cohesion, consensus, and affection. Criterion-related validity was tested by administering the test to 218 married and 94 divorcedpersons. The sample was recruited using nonprobabifity purposive samplingtechniques, the researchers\u00E2\u0080\u0099 intent being to complete a comprehensive itemanalysis and scale assessment. Each of the 32 items correlated significantly withthe external criterion of marital status. The divorced sample differed significantlyfrom the married sample ( < .001) when t-tests for assessing differences betweenthe sample means were carried out. Construct validity was tested to see whetherthe DAS was measuring the same general construct as another widely-acceptedscale, the Locke-Wallace Marital Adjustment Scale (Locke & Wallace, 1959). Thecorrelation between the two was .86 among married respondents and .88 amongdivorced respondents ( < .001). Factor analysis of the final 32 items furtherconfirmed the scale\u00E2\u0080\u0099s construct validity. Spanier (1976) also notes the scale hasbeen shown to have predictive validity as well as concurrent validity, using thesample described above.In general, Gurman et al. (1986) report that the DAS is one of the two most62widely used self-report marital satisfaction measures. Its strong psychometricproperties are well known and highly respected. The DAS was chosen for thisstudy because what it measures relates directly to the research questions at hand.Gurman et al. suggest the use of pre- and post-measures that relate to the clinicalproblem, the intervention process, and the outcome. \u00E2\u0080\u009CInstruments used shouldarticulate the core theoretical and conceptual dimensions of the hypothesizedtherapeutic process\u00E2\u0080\u009D (p. 607). The way Spanier (1976) operationalizes adjustmentfits well with a study on the alliance, both concepts involving process andsystemic elements in their definitions.A limitation of using the DAS as an outcome measure is that the therapeuticmandate in TARP was to deal with the alcohol dependency of the men ratherthan marital adjustment. Efforts were made in therapy to improve therelationship, but the primary focus remained with the resolution of the alcoholproblem.Therapeutic Alliance ScalesBecause the two alliance scales used in this study operationalize the mainconstruct of interest, this section will offer a detailed examination of theinstruments and the theory on which they are based. The two scales are theIndividual Therapeutic Alliance Scale (ITAS) and the Couple TherapeuticAlliance Scale (CTAS) (see Appendixes C and D for complete scales). The thirdscale, the Family TAS, was not needed, as none of the children of the subjectsparticipated directly in therapy. In SFT and ExST-Individual therapy, only onemember of the client system--the alcoholic father--received direct treatment andfilled in the ITAS. In ExST-Couples therapy, treatment involved both father andmother, and each were given the CTAS to complete.Pinsof and Catherall (1986) designed the scales as self-report measures instead63of therapist or observer-report. They assume the alliance to be primarily anexperiential rather than behavioural phenomenon and believe self-report to bemore appropriate for measuring experience, and observational methods formeasuring behaviors (Pinsof, 1989). Furthermore, they claim the client\u00E2\u0080\u0099sexperience of the alliance has far more predictive potency concerning outcomethan the therapist\u00E2\u0080\u0099s.Pinsof and Catherall (1986) define the therapeutic alliance as \u00E2\u0080\u009Cthat aspect ofthe relationship between the therapist system and the patient system that pertainsto their capacity to mutually invest in, and collaborate on, the therapy\u00E2\u0080\u009D (p. 139).The alliance is not to be equated with the relationship between therapist and client,the latter including all feelings, thoughts and responses that patient and therapistsystem members experience toward each other. Feelings arising from differencesor similarities in political or religious or socioeconomic matters, for example, andsexual feelings are not part of the alliance. Though these aspects may affect theaffiance, they are not defined as essential ingredients or impediments. Twodimensions on the TAS are used to operationalize this definition--Content andInterpersonal System. The Content dimension is represented by three subscalesoriginally conceived of by Bordin (1979): bonds, tasks, and goals.Bonds refer to the quality of the therapeutic relationship. Items designed forthis subdimension measure the extent to which clients allow their therapist tobecome a significant object or person in their psychological lives. Included are theclient\u00E2\u0080\u0099s feelings of being cared for and accepted by the counsellor as well as theclient\u00E2\u0080\u0099s feelings of care for the counsellor.The Tasks portion focuses on the methods and techniques of therapy and theextent to which these are connected to a client\u00E2\u0080\u0099s understanding of her problemsand her desire to change. The way therapy is being conducted, the client\u00E2\u0080\u0099sconfidence level in her counsellor\u00E2\u0080\u0099s abilities, and the client\u00E2\u0080\u0099s perception of the64counsellor\u00E2\u0080\u0099s capacity to understand and help her are all parts of thissubdimension. The task subset will be strongest when the client views her ownand her therapist\u00E2\u0080\u0099s tasks as closely tied to furthering movement toward her goals.Her therapist has made vivid the link between assigned tasks and her sense of herdifficulties and hopes for change (Bordin, 1979).The Goal portion measures the extent to which there is agreement on thegoals of therapy. Clients generally see resolution of their presenting problem asthe primary goal of counselling. Their ability or willingness to collaborate withtheir counsellor will depend on their perception of whether the counsellor isworking toward the same goal.it is the Interpersonal system dimension that makes the TAS unique amongother alliance instruments. Multi-systemic phenomenon are reflected in theitems, addressing not only the therapist and patient affiance, but also therelationships between the therapist and other members of the client\u00E2\u0080\u0099sinterpersonal system. Three categories of the interpersonal system dimension aredefined: self-therapist, other-therapist, and group-therapist. Table 6 summarizesthe number of items for each subscale on the two instruments.Table 6: Number of Items on Alliance SubscalesContent InterpersonalScale Bond Task Goal Self Other GroupITAS 8 11 6 11 7 7CTAS 10 13 6 11 11 765Self-therapist items are identical for both the ITAS and CTAS scales, whereasthe subdimensions for other and group-therapist must be operationalizedseparately because different system members are present in individual and coupletherapy. In the self-therapist subdimension, the three Content subsets of bonds,tasks, and goals are addressed as they relate to the client and therapist. Items read,\u00E2\u0080\u009CThe therapist and I...The Other-Therapist subdimension within the couple therapy context refersto the relationship between the counsellor and the other member of the couple.Items read, \u00E2\u0080\u009CThe therapist and my partner agree on the goals for therapy.\u00E2\u0080\u009D For theindividual therapy context, the other-therapist items read, \u00E2\u0080\u009CThe therapist andsome of the people who are important to me... .\u00E2\u0080\u009C This subdimension assesseshow supportive or critical the client believes her sigiiificant others are regardingher counsellor or therapy.In the Group-Therapist subdimension for couples, items measure therelationship of the therapist with the couple as a system in itself. The items onthe CTAS are operationalized as \u00E2\u0080\u009Cmy partner and myself.\u00E2\u0080\u009D (e.g., \u00E2\u0080\u009CThe therapistdoes not understand the relationship between my partner and myself.\u00E2\u0080\u009D) For theclient in individual therapy, items reflect the client\u00E2\u0080\u0099s perception of therelationship between his whole interpersonal system and the counsellor: \u00E2\u0080\u009Cmyimportant relationships and me.\u00E2\u0080\u009DIn general, the CTAS measures the direct subsystem of the couple in therapy,the emphasis being on the therapist and couple systems and the circularinteractions between and among them. With the ITAS, the clients\u00E2\u0080\u0099 perceptions oftheir indirect subsystems are evaluated in addition to the direct subsystem. Pinsof(1989) stresses that true process research is not limited to the direct subsystems, butincludes the indirect ones as well. The ITAS goes beyond the direct patient66system, linking out-of-session with in-session processes.Six additional subscales are created by the overlapping items of the twodimensions. Table 7 lists this matrix and the number of items in eachsubdimension. It can be seen that \u00E2\u0080\u009Cself\u00E2\u0080\u0099 and \u00E2\u0080\u009Cgoals\u00E2\u0080\u009D have identical numbers ofitems whereas the other four subscales differ.Table 7: ITAS/CTAS: Subdimensions and Number of ItemsContent SubscalesInterpersonalSubscales Tasks Bonds GoalsSelf 5/5 4/4 2/2Other 3/5 2/4 2/2Group 3/3 2/2 2/2Psychometric Characteristics of the ITAS and CTASReliability. Pinsof and Catherall (1986) conducted two preliminary repeatedmeasures design studies on small samples to assess an initial rate-rerate reliabilityof the ITAS and CTAS. The rate-rerate Pearson correlation coefficients for thefirst study were: (a) individual: i = .83 (n = 9); and (b) couple: r = .84 (n = 24).The second study results were: (a) individual: r = .72 (n = 18); and (b) couple: i =.79 (n = 17). All correlations were significant at alpha equals .005.Two other studies examined the internal consistency reliability of the couplesTAS using Cronbach\u00E2\u0080\u0099s alpha (Anastasi, 1988). Johnson and Greenberg (1985)found the CTAS to have a global alpha of .96 ( = 28) and .88, .92, and .85 aiphasfor self, other, and group subscales respectively. Similar results were found byBourgeois et al. (1989) whose estimated reliabilities of the CTAS were .95 (N = 126)for the global alpha and six subscale alphas ranging from .77 to .95. The internalconsistency was also demonstrated by looking at the intercorrelations between the67subscales and between the subscales and the total score. Subscale intercorrelationsranged from .63 to .94. The range of subscale to total was from .80 to .96, all scoresbeing superior to .92 with respect to the global score except in the case of the goalsubscale, where = .80 ( < .05 for all analyses).In a second analysis, Bourgeois et al. (1989) determined correlationcoefficients between the subscale subdimensions and the total score (e.g., Self-Taskwith Self-Goal or with Other-Bond, etc.). Coefficients varied from .33 to .76. Withrespect to correlations between subdimensions and the total score, all coefficientswere greater than .62 ( < .05). Bourgeois et al. conclude that the subscales are sostrongly intercorrelated that the CTAS measures a uni-dimensionalphenomenon, not establishing a distinction between the Content andInterpersonal dimensions as claimed. Additionally, an analysis of principalcomponents, using a factor analysis following an orthogonal rotation did notreproduce the proposed theoretical structure. Bourgeois et al. explain their resultsas confirming Guildford\u00E2\u0080\u0099s (1952) and Roger& (1965) positions that using the sameitems for different subscales does not permit coherent clinical or psychologicalinterpretations. They do stress, however, that this does aQt invalidate using theglobal score (p. 71), and they acknowledge that factor analysis may not be anappropriate procedure for subscales that contain a great number of overlappingitems.These analyses need to be replicated, for in the only other in-depth study ofthe psychometric properties of the scales, opposite conclusions are reached(Heatherington & Friedlander, 1990b). In that study, Heatherington andFriedlander compared the CTAS and FTAS. Like Bourgeois et al., their results(, = 32) show good internal consistency for the CTAS, with high subscale alphacoefficients ranging from .70 to .94, and a global alpha reaching .93. Subscale68intercorrelations ranged from .52 to .84, and subscale to total scale correlationsfrom .77 to .96. The authors explain that high intercorrelations are to be expected,particularly for the Interpersonal subscales because the self and other subscales aresubsystems within the group subscale. Their final conclusion is that the\u00E2\u0080\u009Csubstantial proportion of unaccounted for variance in these correlations(especially for couples) supports their utility as separate subscales\u00E2\u0080\u009D (p. 304). Giventhe discrepancy in conclusions regarding the scales, more research is needed.Validity. Pinsof and Catherall\u00E2\u0080\u0099s (1986) two studies did not disconfirm thevalidity of the scales and did, in fact, uphold their theory of the alliance as beingmore like an evolving state than a stable trait, thus indicating some constructvalidity (p. 146). Heatherington and Friedlander (1990b) also examined theconstruct validity of the CTAS, comparing CTAS Content subscales to the SessionEvaluation Questionnaire (SEQ), a well-established measure of therapeuticimpact. Significantly correlated were session depth/value and tasks. In anunpublished study, Catherall (1984) found the overall scale score correlatedpositively and significantly with therapist\u00E2\u0080\u0099s ratings of progress.Preliminary predictive validity studies have shown encouraging results(Catherall, 1984). Positive correlations were found ( < .05) between each of theoverall alliance scale scores and client progress, indicating at least minimalpredictive validity. Gutterman (1984) found a fair amount of stabifity with theCTAS, discovering little variation in clients\u00E2\u0080\u0099 scores over the first eight sessions oftherapy. Bourgeois et a!. (1990) systematically studied the predictive validity ofthe CTAS by expanding the number of subjects (N = 126), using multiplemeasures to assess outcome, and residualizing the outcome scores by thepretherapy scores. Validity results showed clinical relevance when compared tofindings on the predictive validity of relationship variables in other studies.Distribution. A potential problem in measuring the alliance is the common69incidence of a halo effet. There are several possible explanations. Often clientsare reticent to report anything negative about their therapy or counsellor,resulting in a negative skew in their affiance scores. Another possibifity is thatclients having the poorest affiances may drop out of therapy before the alliance ismeasured, thus eliminating the lowest scores. Additionally, often the mostdistressed clients are screened out of research studies. Bourgeois et al. (1990), forexample, excluded alcohol-dependent clients. Interestingly, Heatherington andFriedlander (1990b), not screening for alcohol-dependency, found a gooddistribution of scores in their study. Another significant factor in many researchprojects is that clients are informed of the goals and tasks of treatment and mustagree to them before being accepted for treatment. Again, those who mightotherwise show lower scores are already eliminated. An important proceduralstep to reduce a ceiling effect is to have researchers administer the scales ratherthan the therapists.Summary. The three forms of the TAS are the only instruments to date thatare founded on a sound conceptual base grounded in systemic theory, therebyoffering potentially significant findings to the systemic field of therapy. In spite ofthe fact that more research needs to be done on the scales--the overall factorialstructure and the validity of the six theoretical dimensions have yet to be verifiedempirically\u00E2\u0080\u0094what has been shown to date is encouraging. The psychometricqualities of the instruments appear satisfactory, the literature indicating the scaleshave predictive power and are able to describe levels of collaboration on goals,tasks, and bonds. Coefficient aiphas are sufficiently large, pointing to goodinternal consistency. The answer to the central question of whether the specifiedbehavior truly reflects the construct of interest seems to be yes.70Psychometric Characteristics of the TAS in the Present StudyGiven the controversy regarding whether the scales measure ahomogeneous construct or whether they reflect the six subdimensions theorizedby Pinsof and Catherall (1986), this study included an examination of thecorrelations between the subscales.The range on item responses using a 7-point Likert scale revealed reasonablevariabifity, superior to previous studies. Mean ranges were from 4.91 to 6.11, andfrom 4.88 to 6.29 on the ITAS and CTAS respectively with standard deviationranges of 1.21 to 1.88 and 0.85 to 1.47. Examination of the frequency distributionsand estimates of skewness suggest fairly normal distributions. (See Appendix E.)Table 8 shows the correlation matrices for both scales, including the alphasfor subscales and total in parentheses. This is the first such data on the ITAS, andthe results, which show very high internal consistency, are similar to priorresearch on the CTAS and FTAS. Subscale intercorrelations ranged from .70 to .91and subscale to total scale from .87 to .94. Overall, goals had the lowest correlationlevel, both when compared to the bonds subscale and to the total. Tasks had thehighest subscale intercorrelation (with self) and the highest correlation with thetotal score. Alpha levels were also very adequate, ranging from .83 to .92 with aglobal alpha of .97.The CTAS data, similar to previous studies on the scales, shows highsubscale intercorrelations ranging from .74 to .92, and subscale to total scale from.90 to .94. Alpha levels ranged from .82 to .93, the global alpha reaching .96. Theseresults indicate an even higher degree of internal consistency than in either of theprevious two studies on the CTAS.The CTAS, like the ITAS, showed the goal subscale to have the lowestcorrelation with the total and the tasks to have the highest. Bourgeois et al. (1989)found the exact same pattern. It is also noteworthy that the couples\u00E2\u0080\u0099 weakest71association was between self and other ( = .74), suggesting that respondents\u00E2\u0080\u0099perceptions of their partners, as compared to their own involvement in therapy,was less similar to any of the other subscale relationships. The correlationsshowing the greatest differences between the two scales were group-goal, and task-goal, both having a weaker association (.07) on the ITAS than on the CTAS.Table 8: ITAS/CTAS: Subscale Intercorrelations and Internal Consistencies(Cronbach\u00E2\u0080\u0099s Alpha shown as Diagonal Elements)Bond Task Goal Self Other Group TotalBond (ITAS) (.90) .77 .70 .83 .85 .90 .91(CTAS) (.87) .83 .77 87 .86 .87 .93Task (ITAS) (.89) .77 .91 .88 .87 .94(CTAS) (.93) .84 .90 .89 .92 .97Goal (ITAS) (.83) .83 .84 .79 .87(CTAS) (.87) .83 .86 .86 .91Self (ITAS) (.92) .80 .81 .94(CTAS) (.92) .74 .88 .93Other (ITAS) (.83) .88 .94(CTAS) (.92) .83 .93Group (ITAS) (.83) .94(CTAS) (.85) .95Total (.97)(.96)(all < .001, ITAS, = 56; CTAS,.= 68)A final test measuring the internal consistency of the instruments took thecombined subscales that, according to Pinsof and Catherall (1986), form separatesubdimensions. This correlation matrix is presented in Table 9.The range of subdimension intercorrelations was quite similar on the two72instruments, from .01 to .75 on the ITAS, and from .16 to .75 on the CTAS. It wasalso much more extreme, the correlations much smaller. This can be expectedgiven the reduced number of items in each subdimension. On the subscale tototal scale, the ranges were .24 to .80 (ITAS) and .11 to .76 (CTAS). Differences canbe seen in those subdimensions that did not correlate as highly in one instrumentas in the other. For example, bond-group had the strongest correlation with goal-group on the ITAS and the weakest on the CTAS. Similarly, bond-group to totalhad the highest correlation on the ITAS and the lowest on the CTAS.Table 9: ITAS/CTAS: Subdimension Intercorrelationsand Internal ConsistenciesTskOth.52.54.31.32.47.69\u00E2\u0080\u0094 .50.62Total.61.51.55.56.80.11*.78.76.65.75.30.63.24*.34\u00E2\u0080\u0094 .50 .43\u00E2\u0080\u0094 .23 .61BndSlf BndOth\u00E2\u0080\u0094 .50\u00E2\u0080\u0094 .53BndGrp.51.21\u00E2\u0080\u0094 .73-- .63TskSlf.68.75.66.24-- .62-- .41TskGrp.31.60.25*.39.33.65.30.60.20.53G1S1f.26.55.36.34.30.29.49.36.30.22*BndSlfBndOthBndGrpTskSlfTskOthTskGrpG1S1fGlOthGlGrpGlOth GlGrp.36 .62.47 .18*.50 .36.50 .02*.52 .75.56 .16*.59 .64.27 .48.45 .35.58 .28.36 .27*.53 .27.43 .01*.18*.31-- .32\u00E2\u0080\u0094 .33-- .67.60(*= 42> .05; for all other correlations, < .05)73It is important to note that the subdimensions did not appear to relate toeach other in any kind of predictable way as the authors may have hoped.Patterns of higher correlations within similar Content or Interpersonalsubdimensions were not evident. These results add further weight to Bourgeoiset al.\u00E2\u0080\u0099s (1989) conclusion that the instruments measure a uni-dimensionalphenomenon, not distinguishing between Content and Interpersonaldimensions. The results of the present research will therefore be founded on theglobal scores only.Revised scalesOne of the exploratory interests behind the present study was to examinepotential differences in the interaction of the alliance with outcome in conjointversus individual treatment. No other study has attempted to compare theconstruct across these two modalities. Heatherington and Friedlander (1990b) arethe only researchers to date to compare two of the scales at all (the FTAS andCTAS). They describe the scales as \u00E2\u0080\u009Cparallel self-report measures\u00E2\u0080\u009D (p. 300), notingthey contain identical items with certain necessary changes in wording toaccomodate the differing contexts.In their study, two 2-way ANOVAs found no significant effects for scale orfor gender on the FTAS and CTAS. Though the authors do not conclude that thenature of the alliance in the two modalities is therefore the same, they note thathad differences in scale means been found, a lack of bias in the instrumentsthemselves would have had to be shown. This way, it can at least be assumedthat the scales are not biased. It could perhaps be speculated that the ITAS,containing many similar items and an identical conceptual base, may also beunbiased, though no empirical research exists. to date to substantiate this.74Given that the ITAS has only 25 items and the CTAS 29, revision of thescales was necessary in order to carry out an analysis comparing the individualand couple formats. The following process was used. First, items on each scalewere categorized into their respective subdimensions and compared to see if theyhad a corresponding item on the other scale. Eleven items were found to beidentical on both scales. Twelve more corresponded conceptually with an itemon the other scale, but differed slightly in wording (e.g., ITAS: \u00E2\u0080\u009CThe therapist caresabout my important relationships;\u00E2\u0080\u009D CTAS: \u00E2\u0080\u009CThe therapist cares about therelationship between my partner and myself.\u00E2\u0080\u009D) The remaining three ITAS itemsand six CTAS items, having no corresponding item, were eliminated.Next, an index of comparability was developed using Pearson correlationcoefficients testing the relationship of each of the 12 similar items with the sumscore of the 11 identical items from the same scale. If the correlation between anitem and the 11 items from its own scale was nearly identical to the correlationbetween the item\u00E2\u0080\u0099s pair and the 11 items from the other scale, the 2 correspondingitems could be included in the revised versions.To determine the significance of the two individual correlations, coefficientswere translated to i-scores using Fisher\u00E2\u0080\u0099s z transformation of , and a-tests wererun (Glass & Stanley, p. 311). Table 10 lists each pair of items along with thecorrelation coefficients, the transformed -scores, the -va1ues, and finally theitem rankings\u00E2\u0080\u0094i being the most similar and 12 being the least.Although all 12 i-tests show a high degree of similarity, only the top 9 itemswere chosen, with -values ranging from 0.044 to -1.140. Because the difference inthe correlations of the items are not statistically significant, it can be assumed thatusing them in the two scales will produce comparable measures for researchpurposes. The revised versions are made up of 20 items each, with a possiblerange of 0 to 140. They have the same theoretical subdimensions from Pinsof and75Catherall (1986) original scales. Items from the original scales that make up therevised are (a) ITAS: 1-5, 7-18, 20, 21, and 24; and (b) CTAS: 1-4, 7, 8, 10-14, 16-19,21,24, and 26-28.Table 10: Item Selection for Revised TASUsing Transformed z-scoresITAS CTAS ITAS \u00C2\u00A3 Z r CTAS \u00C2\u00A3 Z. r Z-test values Rankitem item24 19 .612 .712 .605 .701 0.044 18 8 .547 .613 .569 .646 -0.131 213 27 .625 .733 .700 .867 -0.532 316 26 .491 .538 .595 .686 -0.588 49 11 .647 .771 .542 .606 0.655 54 4 .673 .816 .562 .636 0.715 621 10 .673 .780 .757 .986 -0.818 712 28 .404 .429 .614 .716 -1.140 811 13 .547 .613 .727 .922 -1.280 96 6 .495 .543 .739 .948 -1.608 1023 5 .694 .856 .417 .444 1.636 1122 29 .713 .893 .406 .432 1.831 12To provide descriptive data on the revised versions, correlation matrices andaiphas computed for each are listed in Table 11. The data show that the revisedscales uphold the high internal consistency of the original scales. The ITAS-Rsubscale intercorrelations ranged from .66 to .90, and the subscale to total scalefrom .84 to .95. The CTAS-R similarly had high subscale intercorrelations rangingfrom .69 to .92, and the subscale to total scale from .83 to .96. While stifi in theacceptable range, the differences between scales were most extreme in the groupsubscale (e.g., group/goals difference = .15). In both revised versions, self-tasksubscales had the strongest intercorrelation, whereas sell-other had the weakest.Also on both, the \u00E2\u0080\u009Cother\u00E2\u0080\u009D subscale had the weakest correlation with the total.76Table 11: Revised ITAS/CTAS: Subscale Intercorrelations and InternalConsistencies (Cronbach\u00E2\u0080\u0099s Alpha shown as Diagonal Elements)Bond Task Goal Self Other Group TotalBond (.85) .79 .67 .85 .75 .85 .90(.83) .81 .75 .90 .77 .83 .91Task (.88) .73 .91 .80 .85 .95(.91) .79 .92 .82 .91 .95Goal (.77) .80 .76 .74 .85(.83) .85 .78 .89 .91Self (.92) .66 .78 .94(.92) .70 .86 .97Other (.78) .74 .85(.79) .82 .84Group (.74) .91(.83) .95Total (.94)(.95)(all < .001; ITAS, j. = 56; CTAS, . = 68)Cronbach\u00E2\u0080\u0099s aiphas for the global ITAS-R and CTAS-R were .94 and .95respectively. Subscale aiphas ranged from .74 to .92 on the ITAS-R, and from .83to .92 on the CTAS-R. These were slightly smaller than those obtained on theoriginal versions, but stifi indicate very high internal consistency. Pearsoncorrelation coefficients comparing the revised scales with the original versions ineach treatment obtained the following results: (a) SFT: ITAS with ITAS-R, .99; (b)ExST-Ind: ITAS with ITAS-R, 1.00; (c) ExST-Cpl men: CTAS with CTAS-R, .98;and (d) ExST-Cpl women: CTAS with CTAS-R, .99. Given these data, it seems ahigh degree of confidence can be placed in the revised versions.77Data Analysis TechniquesDifference ScoresDifference, or gain-scores, for the DAS-pre and DAS-post scores were used ina number of analyses. Using such scores is a practice often avoided because themeasurement error associated with each testing occasion is effectively doubledwhen the same instrument is used twice in a calculation. Unless the reliabifity ofan instrument is quite high, the doubling of the measurement error wifi decreasethe reliabffity of the findings to the point where very little confidence can beplaced in them. Reliabilities of the DAS-gain scores were computed usingCronbach\u00E2\u0080\u0099s Aiphas from the pre and posttreatment occasions (see Table 5) and thecorrelations between the two (Crocker & Algina, 1986). Results are listed in Table12. The reliability of the total scale is adequate for the purposes of data analysis inthe present study, and the subscale reliabifities can be used with some caution.Table 12: Cronbach\u00E2\u0080\u0099s Estimate of Reliability for DAS-gain Scores andPearson Correlation Coefficients: DAS-pre with DAS-postCorrelation DAS-gainScales DAS-pre /DAS-post reliabilitiesTotal .321* .88Consensus .410* .76Satisfaction .384* 75Affec/Express .252 .58Cohesion .284 .73* <.05,fl=4278Statistical procedures used to address the research hypotheses are describedbelow.Hypothesis OneA t-test was used to assess the difference between the means of scores fromthe two individual treatment groups. Pearson correlation coefficients weredetermined for the two groups, testing the strength and direction of relationshipbetween the alliance and DAS-pre and DAS-post scores.Hypothesis TwoI-tests were used to examine differences between the means of alliancescores from binge and from chronic drinkers in SFF and ExST-Ind. Additional -tests compared DAS-pre and DAS-post scores for each drinking type, and Pearsoncorrelation coefficients measured the relationship between the therapeuticalilance and DAS-gain scores for each.A second series of i-tests studied binge and chronic therapeutic alliances andlevels of dyadic adjustment within SFT and within ExST-Ind to ascertain whethereither treatment was superior for either drinking type. Pearson correlationcoefficients were also used to examine treatment differences for chronic and forbinge drinkers between therapeutic affiance and DAS-gain.Hypothesis ThreeI-tests compared mean DAS-gains and mean alliance levels for theindividual and couple formats for men in ExST. A repeated measures analysis ofvariance was used, between factors being affiance level and treatment, and therepeated measure being pre and post DAS scores. Alliance scores weredichotomized into Low (86 - 114) and High (115- 140). The number of subjects79assigned to each level was 22 and 21 respectively.A final analysis used Pearson correlation coefficients to measure therelationship between the affiance and DAS-gain scores for the conjoint andindividual formats of ExST.Hypothesis FourA dependent group t-test was used to assess the difference between theaffiance means of women and men in the couple format of ExST. Further j-testscompared mean dyadic adjustment levels at pre and posttreatment for the twogenders. Correlations were computed to explore the relationship that men\u00E2\u0080\u0099s andwomen\u00E2\u0080\u0099s respective therapeutic affiances had with DAS subscales.Hypothesis FiveThe mean difference in affiance was determined from the absolute values ofthe difference in men\u00E2\u0080\u0099s and women\u00E2\u0080\u0099s scores. I-tests compared men and womenfrom split and intact affiances on level of affiance and dyadic adjustment.To answer 5 (a), a t-test compared DAS mean gain scores from the 12 intactcouples and the 8 split ones. To find what extent agreement on the affiancecorrelated with agreement on dyadic adjustment, difference scores computedfrom partners\u00E2\u0080\u0099 individual scores on the CTAS were plotted against pre andposttreatment DAS scores. Pearson correlation coefficients were determined tomeasure the relationship of men\u00E2\u0080\u0099ss DAS scores with their partners in both intactand split alliances.For hypothesis 5 (b), two groups were formed from the split couples--thosewith men\u00E2\u0080\u0099s scores highest, and those with women\u00E2\u0080\u0099s scores highest. Only two ofthe eight split affiances showed the man\u00E2\u0080\u0099s alliance score higher than thewoman\u00E2\u0080\u0099s. An independent t-test compared the DAS-gain of the two groups.80Adjunct analyses. Because the nature of the alliance is so complex, andbecause so little family research has studied the construct, two contingency tableswere designed to identify additional alliance patterns (Fisher et al., 1985). In eachtable, couples were divided according to whether both partners scored high on theaffiance, both scored low, or one scored high and the other low. Scores wereconsidered high if they were above the mean for the appropriate gender (men\u00E2\u0080\u0099smean = 163.0; women\u00E2\u0080\u0099s mean = 174.0) and low if they were below these means.In the first table, couples with binge drinkers were compared to those withchronic drinkers. In the second, couples were grouped on the basis of the level oftheir discrepancy on alliance scores. Levels of discrepancy were determined by thefollowing divisions: (a) high discrepancy> 32 points; (b) medium discrepancy> 13and <33; and (c) low discrepancy < 14 points. CM-square analyses were carriedout to determine if significant relationships for the above categories existed.81Chapter IVRESULTSIn this chapter, results from the analyses addressing the research hypothesesare presented. Dyadic Adjustment Scale pre and posttreatment scores are referredto as DAS-pre and DAS-post. Difference scores in dyadic adjustment areabbreviated as DAS-gain. The three treatments--Experiential Systemic Therapyfor Individuals and Couples, and Supported Feedback Therapy--are called ExSTmd, ExST-Cpl, and SFT respectively. The therapeutic alliance scales forindividuals and couples are referred to as ITAS and CTAS; revised versions of thescales are abbreviated ITAS-R and CTAS-R.Following the discussion of the five hypotheses, two adjunct analyseshaving to do with split and intact alliances are presented. The purpose of these isto offer a preliminary examination of constructs that are still very new tomarriage and family research.Statistical procedures. The significance tests that are cited were conductedusing an alpha of .05. However, given the exploratory nature of the study, theabsence of a solid body of research, and the small sample size that increases thepossibifity of making Type II errors, those analyses that result in significancelevels between .05 and .10 will be discussed. The latter findings will be viewedwith some caution. It is hoped that the results of inquiries\u00E2\u0080\u0094whether they are at.05 or .10 levels of significance--will help to identify new questions and promoteareas of investigation for future research.82Results of AnalysesHypothesis OneThe therapeutic alliance as measured by the ITAS will not be significantly differentin Experiential Systemic Therapy (for individuals) and Supported FeedbackTherapy.Alliance strength. The results of the t-test comparing ITAS levels in SFT andExST-Ind showed that, as predicted, there was no significant difference in thestrength of the affiance, (40) = 0.08 (2-tailed). Means were nearly identical at145.91 (j = 15.57) for SFT and 146.37 (J2 = 20.38) for ExST-Ind.Alliance role. The role of the affiance in the two treatments was examinedby the use of Pearson correlation coefficients testing the strength and direction ofrelationship between the affiance and DAS-pre and DAS-post scores. Table 14reports the results. The alliance in both treatments was somewhat inverselyrelated to DAS-pre scores. By posttreatment, a solid correlation approachingsignficance ( = .08) was evident for ExST-Ind compared to a non-significantrelationship for SFT. In sum, the level of posttreatment dyadic adjustment forExST-Ind subjects could be explained in part by the strength of the affiance,whereas there was essentially no relationship for subjects in SFT.Table 13: Pearson Correlation Coefficients for Men in IndividualTreatments: Alliance with DAS-pre and DAS-postTreatment ii DAS-pre DAS-postExST-IndSFT12< .1022 -.139 .315*20 -.032 .01183Hypothesis TwoDrinking patterns of alcoholics will be associated with significant differences in thestrength of the therapeutic alliance, binge drinkers displaying lower therapeuticalliance scores than chronic drinkers.Drinking type and affiance strength. Table 14 contains results from the i\u00E2\u0080\u0094testthat examined ITAS levels for binge and chronic drinkers in SFT and ExST-Ind.Contrary to the research hypothesis, means for the two groups were similar.Table 14: ITAS Means, Standard Deviations, and i-test Values:For Drinking Types in Individual TherapyDrinking Type ii Means SJ2 t-value +Binge 19 148.64 20.89- 0.81Chronic 23 144.09 15.47t j (40) = 1.68 (one-tail)Drinking type and alliance role. Given that affiance means were notsignificantly different, an additional examination focused on the therapeutic rolethe affiance played with respect to the dyadic adjustment of the two drinkingtypes. Several !-tests and Pearson correlation coefficients were used to addressthis inquiry. First, dyadic adjustment levels for binge and chronic werecompared. Results showed that whereas pretreatment dyadic adjustment was notstatistically different between the two drinking types, (40) = -0.88, chronicdrinkers displayed a higher mean level of posttreatment dyadic adjustment thatapproached significance t (40) = -1.39, < .10 (one-tail).84Second, Pearson correlation coefficients measuring the therapeutic affianceand DAS-gain for the two drinking types were significantly different. Bingedrinkers showed a weak inverse relationship (j = -.052) whereas chronic drinkersdisplayed a strong positive correlation of .643 ( <.001). To conclude, for chronicdrinkers, the affiance played a significant role, predicting 41.4% of the variance inDAS-gain, compared to a nonsignificant role for binge drinkers.Drinking type. treatment. and affiance. I-tests and Pearson correlationcoefficients were also used to examine the relationship that SFT and ExST-Indhad with binge and chronic subjects\u00E2\u0080\u0099 ailiance and DAS scores. Results from thet-tests are listed in Table 15. The level of affiance for binge and chronic drinkerswas nearly identical for the two treatments. This was true in spite of the fact thatbinge and chronic drinkers in SFT began treatment with dyadic distress levels thatapproached a significant difference ( = .08), binge drinkers showing moredistress. Just as alliance levels were similar, t-tests for binge and chronic groupson DAS-post and DAS-gain did not show significant treatment differences.85Table 15: Means, Standard Deviations, and t-test Values: ITAS and DASfor Drinking Types in SFT and ExST-IndMeasureTreatment Mean SD t-ValueDrink TypeITASExST-IndBinge 12 148.18 21.79 0.427+Chronic 10 144.20 19.47SFTBinge 7 149.41 20.92 0.707 ifChronic 13 144.01 12.42DAS-preExSTBinge 12 91.08 17.54 -0.068 tChronic 10 91.56 13.09SFTBinge 7 79.81 18.69 1.390 * ftChronic 13 90.52 18.50DAS-postExST-IndBinge 12 102.65 19.04 -0.439 tChronic 10 106.05 16.83SFTBinge 7 94.29 20.42 1.122 ftChronic 13 104.87 19.55DAS-gainExST-IndBinge 12 11.57 21.88 -0.766 +Chronic 10 18.57 20.66SFTBinge 7 14.47 11.81 0.019 +Chronic 13 14.35 16.39*+ k(20).io = 1.325 (1-tail)if (l8) = 1.330 (1-tail), pooled variance86Table 16 includes Pearson correlation coefficients measuring the strengthof the relationship between the alliance and DAS-gain for the two drinking typesin SFT and ExST-Ind. Binge drinkers in both treatments had very small alliance-gain correlations, whereas for chronic drinkers, there were positive correlationsin both formats. Gain was most strongly associated with the strength of theaffiance in ExST-Ind. The difference between alliance-gain correlations for bingeand chronic drinkers was not significant for clients in SFT but it was for ExST-Jnd(< .01).Table 16: Pearson Correlation Coefficients for Drinking Types:Alliance with DAS-gain in SFT and ExST-IndBinge ChronicTreatmentExST-Ind .035 12 .868* 10SFT -.100 7 .328 13* <.001Hypothesis ThreeThe therapeutic alliance will moderate differential gains (as indicated by pre andposttest DAS scores) produced by individual and couple\u00E2\u0080\u0099s formats of ExST: thealliance will explain a greater proportion of therapeutic gain in ExST-Ind than it willin ExST-Cpl.Therapeutic gains, and alliance strength. Two points stand out from the datain Table 17, which reports ExST-Ind and ExST-Cpl mean scores for men. First,counter to the hypothesis, differential gains in the two formats did not occur, theextent of improvement being almost identical. Second, though the mean alliance87level was somewhat higher in ExST-Ind, the difference was not significant,(40) = .837.Table 17: Means and Standard Deviations for ExST Men:ITAS-R, CTAS-R and DAS ScoresTreatment Alliance DAS-pre DAS-post DAS-gainExST-Ind Mean 117.7 91.3 106.0 14.8SD 16.3 15.3 16.8 21.1ExST-Cpl Mean 113.5 82.1 96.9 14.7SD 16.6 16.5 18.3 19.0Alliance-gain relationship for all ExST men. Results from the repeatedmeasures analysis of variance using dichotomized affiance levels are displayed inTable 18 (procedure described in chapter III). The ANOVA revealed aninteraction between the alliance and occasion that approached significance ( =.069). There was no significant three-way interaction of treatment by affiancelevel by occasion, nor was there a significant interaction between occasion andtreatment. The most significant result can be seen in the major change in dyadicadjustment from pre to post occasions ( < .001), pooling treatments and alliancelevels. Most important for the present study is the fact that across the two ExSTtreatments, therapeutic gain was in part a function of the strength of thetherapeutic alliance.88Table 18: Repeated Measures ANOVA for ExST men:Analysis of DAS by Treatment, Affiance Level, and OccasionSource of Variation 1 M..S sig of fBetween subjectsTreatment (Tx) 1 1657.11 1657.11 5.05 .030Alliance Level 1 1004.37 1004.37 3.06 .088Tx by Affiance 1 103.89 103.89 .32 .577Within cells 39 12799.17 328.18Within subjectsOccasion 1 4485.96 4485.96 20.91 .001Tx 1 4.07 4.07 .02 .891Alliance Level 1 747.56 747.56 3.48 .069Tx by Alliance 1 50.46 50.46 .24 .630by OccasionWithin cells 39 8368.11 214.57Even though the interaction of alliance level, treatment, and occasion wasnot significant ( = .630), mean scores relevant to a possible interaction areprovided in Table 19. Cell means of DAS-gains in individual and couplestreatment with dichotomized affiance levels show that stronger therapeuticalliances were associated with higher total mean therapeutic gains (20.51compared to 8.71, = .069). Visual inspection reveals a pattern of change that,although non-sigrificant, was more apparent for ExST-Ind than ExST-Cpl, theformer showing a somewhat larger difference between the mean gains from lowto high affiance levels.89Table 19: DAS-gain: Cell Means in DichotomizedAlliance Levels for ExST MenTreatment_______Individual Couplen) (11)Total 14.75(22)Alliance and gain in couple versus individual formats. Pearson correlationcoefficients were used to compare the affiance-gain relationship across theindividual and couple formats. Table 20 contains the results. Both formatsdisplayed similar significant correlations of the alliance with gains inaffection/expression. For couples there was also a correlation that approachedsignificance between alliance and cohesion gain. Individuals also showedTable 20: Pearson Correlation Coefficients for ExST Men:Alliance with DAS-gain SubscalesTreatment n Consensus Satisfaction Affection Cohesion DASgain gain gain gain gainExST-IndExST-Cpl<.10<.05AllianceLevelLow(86 -114)High(115- 140)6.65(10)21.50(12)Total(Ii)8.71(21)20.51(22)10.58(11)19.32(10)14.74(21)22 .234 .367** .460** .279 357**21 .141 .131 \u00E2\u0080\u00A2457** .299* .23690significant correlations between the alliance and satisfaction gain and withoverall DAS-gain. Not significant were consensus or cohesion gain forindividuals, and consensus, satisfaction, and overall gains for couples.Similar to the pattern indicated in Table 19, Table 20 indicates a somewhatstronger overall alliance-gain relationship in ExST-Ind as compared to ExST-Cpl,although the difference between the two correlations is not statisticallysignificant. For men in the couple\u00E2\u0080\u0099s format, 5.5% of the variance in DAS-gainwas explained by its relationship to the affiance (> .05) compared to 12.7% in theindividual format ( < .05).Hypothesis FourWithin the couples format of ExST, women will report a higher level of therapeuticalliance than will men.Alliance strength. Results of the dependent group i-test showed that, aspredicted, women had a significantly stronger affiance (yj = 174.05, = 19.14)than did men (M = 163.0, = 23.30), (20) = 1.97 (one-tail), < .05.Alliance role. Two sets of subsequent analyses explored gender differences indyadic adjustment relevant to the alliance. In Table 21, findings from the firstanalyses are listed, providing a general overview of men\u00E2\u0080\u0099s and women\u00E2\u0080\u0099s meanDAS scores. Levels of pretreatment dyadic distress were not significantlydifferent, although there was a trend showing women to be slightly moredistressed. DAS-post means for partners were nearly identical, women havingmade significantly greater gains than men.91Table 21: Means, Standard Deviations, and t-test Values:DAS Scores for men and women in ExST-CplMeasure by gender Mean S12 j-valueDAS-premen 82.11 16.52 (19)= 1.30women 75.63 21.29DAS-postmen 96.86 18.33 L (19) = -.027women 97.33 23.59DAS-gainmen 14.74 21.27 (19) = 2.37*women 21.70 25.25*<.05Table 22 contains results from the second set of analyses, one that examinedthe correlation of the affiance with specific subscales and the total DAS scale.Similar to findings from research hypothesis number one (see Table 13), thealliance generally showed a somewhat stronger positive correlation with DASpost scores and DAS-gain than with DAS-pre scores, indicating that pretreatmentconditions were not related to the development of the affiance. women displayednegative, though not significant, relationships with all subscales at pretreatment.men recorded the only significant correlation, that being between pretreatmentdyadic satisfaction and affiance strength. Affection/expression remainednegatively correlated with the affiance throughout treatment for women, but formen this subscale had the strongest positive correlation with the affiance on bothDAS-post and DAS-gain. On DAS-post, men showed significant correlations forfour of the five scales whereas women showed none. In terms of gain, threescales displayed nearly identical correlations for men and women. Onlyconsensus-gain and affection/expression gains demonstrated slight differences.92Table 22: Pearson Correlation Coefficients for Couples:CTAS with DAS ScalesCTASMen Women Men Women Men WomenScales DAS-pre DAS-post DAS-gainConsensus .029 -.160 .118 .139 .076 .294*Satisfaction .366** -.112 .414** .036 .109 .130Aff/Expression -.038 -.262 .530** -.264 .420** -.001Cohesion -.004 -.262 .418** .028 \u00E2\u0080\u00A2335* .229Total Scale .144 -.205 \u00E2\u0080\u00A2359** .049 .198 .219* < .10** < .05Post hoc analysis: Couples\u00E2\u0080\u0099 concordance. An additional analysis examinedthe role of the partners\u00E2\u0080\u0099 affiance in the couples format. Table 23 shows the degreeto which partners agreed with each other on the DAS throughout therapy.Results indicate that the degree to which partners\u00E2\u0080\u0099 scores correlated with eachother increased substantially. Whereas only 9% of the variance in the DAS-prescore from one partner could be explained by the other\u00E2\u0080\u0099s score, DAS-post andDAS-gain scores of one explained 41% and 63% respectively of the variance in theother\u00E2\u0080\u0099s scores. The couples\u00E2\u0080\u0099 concordance\u00E2\u0080\u0094at least when defined in terms ofagreement--grew stronger.Table 23: Pearson Correlation Coefficients for Couples:Men\u00E2\u0080\u0099s DAS Scores with Partners\u00E2\u0080\u0099 1DAS-pre DAS-post DAS-gainL .293 .640* .792*-[fl= 21.0593Hypothesis FiveWith regard to intact and split alliances between partners in ExST-Cpl:(a) overall agreement between partners on the strength of the alliance will predicta higher outcome on the DAS than will less agreement.(b) in cases with split alliances, outcome on the DAS will be improved when thefather has the higher score compared to when the mother does.Considerable differences in measured affiance levels existed between thepartners in the study. Of the 21 couples, 9 represented partners whose alliancescores differed by more than 20 points, and 5 couples scored within 5 points ofeach other. The mean difference in affiance, determined from the absolutevalues of the difference in men\u00E2\u0080\u0099s and women\u00E2\u0080\u0099s scores was 22.67 ( = 15.88).Eight couples displayed absolute differences that exceeded the mean, differencesranging from 29 to 59 points. These eight were considered to have split alliancesin this study. Twelve couples displayed absolute differences ranging from 4 to 16and were considered to have intact affiances.I-tests provided descriptive data for the theoretical constructs of split andintact affiances. Men from split affiances exhibited significantly lower affiancelevels compared to women from split alliances, t (7) = 1.98, < .05 (1-tail,dependent test). DAS-pre and post scores for women and men were notstatistically different in either intact or split groups. A trend could be seen in thatwomen from split alliances displayed mean DAS-post levels (M = 90.03) that werewell below the critical cutoff point of 100, whereas women from intact affianceshad mean scores were slightly above that point (M = 101.78).Intact and split on DAS-gain. Contrary to prediction, the t-test comparingDAS mean gain scores from the 12 intact couples and the 8 split ones resulted in94no statistical difference, t (18) = -0.85. Results nevertheless showed a trend in thepredicted direction, with the mean gain for couples with split alliances (M = 11.38,SJ2 = 15.39) somewhat lower than that for couples with intact alliances GYI = 20.05,= 28.47).DAS and split/intact alliances. To determine if couples with split affiancesalso showed splits on the DAS when compared to intact affiance couples,difference scores between partners on the DAS-pre and DAS-post were examined.Results in Table 24 are based upon absolute values of the differences betweenpartners.Table 24: Mean Differences and Standard Deviations forSplit and Intact Alliances: DAS\u00E2\u0080\u0094pre and DAS-postDAS-pre + DAS-post tSplit Intact Split IntactMeanDifference 21.13 15.78 17.40 11.09SD 17.37 13.96 15.75 8.62+ for split, n = 8; for intact, , = 12Though t-tests showed no statistical differences between the means of splitand intact groups, there appeared to be a trend for partners from split affiancesalso to show slightly greater splits in their perceptions of dyadic adjustment.Pearson correlation coefficients tested this hypothesis by relating men to womenon the DAS-pre, DAS-post, and DAS\u00E2\u0080\u0094gain. Results are listed in Table 25.Correlations for both groups increased in strength and in significance levels.Though they were not significantly different from split affiance correlations,95intact affiance correlations were slightly stronger, suggesting somewhat higherdegrees of agreement.Table 25: Pearson Correlation Coefficients for Intact and Split Affiances:Men\u00E2\u0080\u0099s DAS Scores with women\u00E2\u0080\u0099s DAS ScoresGroup DAS-pre DAS-post DAS-gainIntact 12 .48 \u00E2\u0080\u00A275** .82**Split 8 .44 .49 .66****2<.01Father-high alliances. For hypothesis 5 (b), the eight couples with splitalliances were examined. Comparing the therapeutic gain of two split coupleshaving men\u00E2\u0080\u0099s scores highest with the six having women\u00E2\u0080\u0099s scores highest showedthat, contrary to prediction, there was no significant difference, (9) = -0.28. Giventhe very low numbers to work with, no further analyses regarding this hypothesiswere undertaken.Adjunct AnalysesTwo analyses were carried out to explore distributions for couples comparingaffiance levels with (a) drinking type and (b) alliance discrepancy. In each case, a9-cell contingency table sorted couples into the following categories: (a) \u00E2\u0080\u009Cbothhigh,\u00E2\u0080\u009D indicating both partners scored above the mean affiances of 174 for womenand 163 for men; (b) \u00E2\u0080\u009Cboth low,\u00E2\u0080\u009D indicating both partners scored below theirrespective mean alliances; or (c) \u00E2\u0080\u009Cmix,\u00E2\u0080\u009D indicating one partner scored above themean and the other scored below. Relationships between row and column96variables in these tables were tested for significance using chi-square procedures.It must be emphasized that these cu-square analyses are for introductorypurposes only. The sample is too small to satisy the traditional hypothesis-testingrequirements of the chi-square analysis.Results from the first analysis, in which couples were categorized by alliancelevel and drinking type, appear in Table 26. Whereas no binge couples recordedhigh therapeutic affiances for both partners, five chronic couples did. Sevenbinge couples displayed mixed alliances compared to two chronic. Similarnumbers of binge and chronic couples exhibited \u00E2\u0080\u0098both-low\u00E2\u0080\u009D alliance levels.Table 26: Frequency Contingency Table:Partners\u00E2\u0080\u0099 ADiance Leveis and Drinking TypeDrinking TypeAlliance Level Binge Chronic(11) (10)Both High (5) 0 5Mix (9) 7 2Both Low (7) 4 3The cM-square test indicated that factors of drinking type and alliance levelwere significantly related, 2 (2) = 7.89, < .05. The observed number of chroniccouples in which both partners measured high affiances was higher than wouldbe expected by chance alone. Similarly, fewer chronic and more binge couplesthan would be expected displayed mixed affiances.97Table 27 reports the distribution from the second analysis in which affiancelevel was charted against alliance discrepancy level (described in Chapter III: highdiscrepancy >32 points; medium discrepancy> 13 and <33; and low discrepancy<14 points). Five couples were categorized as both having high alliance scores,seven scored below their respective means, and nine had mixed scores. Sevencouples were located in each of the three levels of discrepancy.Table 27: Frequency Contingency Table: Partners\u00E2\u0080\u0099 Alliance Levelswith Partners\u00E2\u0080\u0099 Level of DiscrepancyLevel of DiscrepancyAlliance Level High Medium Low(7) (7) (7)Both High (5) 1 2 2Mix (9) 5 2 2BothLow (7) 1 3 3All nine possible cells were filled in the sample size of 21 couples. The chisquared analysis showed that there were no more or less observed frequenciesthan what would be expected by chance alone, 2 (4) = 3.54, > .05.98Chapter VDISCUSSION AND LIMITATIONSIn this chapter, a brief summary and discussion of the results is presentedfollowed by clinical and research implications of those results. Limitations of thestudy are described, as are areas for future research.Summary of ResultsHypothesis one, stating that the strength of the affiance would not besignificantly different in the two individual treatments, was confirmed by thestudy. Therapeutic alliance means for individuals in ExST-Ind and SFT werealmost identical. It was also found that the level of posttreatment dyadicadiustment in ExST-Ind subjects could be explained in part by the strength of theaffiance, whereas there was essentially no relationship between these variables forsubjects in SFT.Hypothesis two projected that binge drinkers, would exhibit significantlylower affiance levels than chronic drinkers. Contrary to expectation, nostatistically significant differences were found for drinking patterns on thestrength of the therapeutic affiance. Alcohol-use patterns were neverthelessfound to serve as potent predictors of the correlation between affiance and dyadicadjustment improvement. Chronic drinkers demonstrated a significantlystronger alliance-gain relationship than binge drinkers did.Hypothesis three maintained that individual and couples formats of ExSTwould result in differential gains and that the individual format would display astronger affiance-gain relationship than would the couples format. Neither99prediction was upheld by the study. First, in both conjoint and individualtreatment, overall mean gains were nearly identical. Second, though thetherapeutic alliance displayed an overall relationship with therapeutic gain thatapproached statistical significance in both modalities, the alliance did not explaina significantly greater proportion of therapeutic gain in ExST-Ind compared toExST-Cpl. The pattern of therapeutic improvement and affiance strength wassomewhat more apparent in ExST-Ind compared to ExST-Cpl.Hypothesis four, predicting that women would report significantly strongertherapeutic alliance levels than would men, was borne out by the study. Womenalso made DAS gains that were not only significantly greater than their partners\u00E2\u0080\u0099but were also the most highly correlated with the alliance.Hypothesis five predicted that couples with intact affiances would improvemore on the DAS than those with split affiances. Although no statisticaldifference was found, results showed a trend in the expected direction with intactcouples improving slightly more on dyadic adjustment than split ones.Compared to women from intact affiances, women from split alliances displayedsomewhat higher degrees of posttreatment dyadic distress. It was alsohypothesized that split couples in which men had the stronger therapeuticalliance would show more improvement than split couples in which women hadthe higher alliance score. Contrary to expectation, average DAS gains for coupleswith \u00E2\u0080\u009Cmen-high alliances\u00E2\u0080\u009D were not significantly different from those with\u00E2\u0080\u009Cwomen-high alliances,\u00E2\u0080\u009D though the sample was too small to suggest meaningfulresults.An adjunct analysis revealed that the degree to which partners\u00E2\u0080\u0099 scorescorrelated with each other on the DAS increased substantially over the course oftherapy, indicating a strengthening of couples\u00E2\u0080\u0099 agreement from both intact and100split alliances. Partners from intact affiances displayed significant correlationswith one another on DAS-post scores and DAS-gain. Split couples approached asignificant correlation with each other on DAS-gain.Factors of drinking type and similarity of partners\u00E2\u0080\u0099 alliance level were foundto be significantly related. A large proportion of binge couples had \u00E2\u0080\u009Cmixed\u00E2\u0080\u009Daffiances, and the only \u00E2\u0080\u009Cboth high\u00E2\u0080\u009D alliances were found in chronic couples. Thedegree of alliance discrepancies between partners and the level of affiance did notprove to be significantly related.Discussion and Implications of ResultsThe present discussion of the results is influenced by an intent to explore asmuch as to verify. Consequently, hypotheses may be advanced even though thefindings on which they are based are not at the < .05 level of significance. Inthis regard, Steinglass, Tislenko, and Reiss (1985) make important observations:For me, inferential statistics are valuable in family research because theyhelp ascertain whether suspected patterns of association between variables--variables I am interested in because of their clinical or theoreticalimportance--seem in fact to be present in the families I have tested. In thisregard, it is the strength of the associative pattern that is often more importantto me than the \u00E2\u0080\u009Csignificance level\u00E2\u0080\u009D of the inferential statistic calculated. Thislatter feature of the statistical procedure (reflected in the value) merely tellsme at what confidence level I can conclude that the associative pattern is anonrandom one.... Hence the bottom line for me is that although statisticalsignificance is useful as a guide, it is up to the family clinician/researcher,not the statistician, to decide whether findings from clinical studies are trulysignificant and therefore worth pursuing. (pp. 383-384)101The Alliance and Treatment for MenGreater understanding regarding the interaction of the affiance andtreatment was generated by the study. The two highly contrasting treatments ofSFT and ExST--the latter used in both individual and couples\u00E2\u0080\u0099 formats--displayedno significant differences in the strength of the therapeutic affiance. This findingcoincides with the individual psychotherapy literature that has discovered noevidence to suggest that the quality of the alliance can be predicted by therapeuticapproach. It appears that even when there is a much stronger focus on thetherapist-client relationship, as in ExST, a stronger affiance does not necessarilyresult.One explanation for this may be that the Therapeutic Alliance Scalesmeasure not only the bond between therapist and client, but also their agreementon tasks and goals. In the initial stages of SFT, attention was given to contractingwith clients around the tasks and goals of therapy. Had the affiance subscalesbeen reliable enough to measure differences in bonds, tasks, and goals, researchcould have focused on the three dimensions separately to examine if any werestronger in one treatment. The bond dimension, for example, may have beenmore the strongest affiance component in ExST whereas task and goals could bethe strongest alliance element in SFT,Another possible explanation for the equally strong affiance in the lessrelational SFT is that the \u00E2\u0080\u009Ctake-charge\u00E2\u0080\u009D role of the therapist in this approach mayhave encouraged certain frightened, discouraged clients to enter into acollaboration. ExST, on the other hand, in its emphasis on client self-disclosureand immediacy of affect, may have distanced some participants. Increasing theemotional charge of sessions and allowing conflictual issues to surface from theoutset may have threatened the therapeutic relationship (Bordin, 1979).Although mean affiance levels were nearly identical for the men in ExST102md, ExST-Cpl, and SFT, it was clear that different relationships between theaffiance and the level of dyadic adjustment at outcome existed in the twotreatments. In both the individual and couple formats of ExST, the allianceserved as a potent predictor of outcome, but in SFT it had virtually norelationship with outcome at all. Why equally strong therapeutic relationshipswould be signficantly related to outcome in one approach and not in anotherraises important questions, particularly when pre and posttreatment levels ofdyadic adjustment were nearly identical.Perhaps part of the answer lies in untapped qualitative differences in thealliance. If, as speculated above, the bonds component was stronger in ExST thanin SFT, perhaps it is this aspect of the affiance that is more associated withoutcome on the DAS than is agreement on tasks and goals. ExST\u00E2\u0080\u0099s greaterinterpersonal demands may have created qualitatively different therapeuticaffiances than those that were developed within SFT. Perhaps alliances formedin response to directive therapists focusing on behavior wifi be less potent interms of final levels of dyadic adjustment than those that are formed in thecontext of self-disclosure and present-centered emotional intensity. A questionthat remains to be answered is what factors accounted for the proportion of SFTsubjects\u00E2\u0080\u0099 gains that the affiance explained in ExST.More sophisticated alliance scales will be needed to test these researchquestions. Until then, it can be stated with increasing confidence that all effectivetherapeutic approaches have mechanisms for the development of a therapeuticalliance, but not all result in statistically significant affiance-outcomerelationships. In the new treatment approaches of SFT and ExST, therapeuticalliances of almost identical strength were formed, yet affiance-outcomerelationships were significantly different.103The Affiance for Men in Individual and Conjoint Modalities of ExSTTwo findings stood out with regard to alliances formed by men treated inindividual and couples\u00E2\u0080\u0099 formats of ExST. First, results showed no statisticaldifference in affiance strength, contrary to the literature\u00E2\u0080\u0099s suggestion that thestrength of the affiance would be substantially different in couples as opposed toindividual therapy (Dryden & Hunt, 1985a, 1985b; Heatherington & Friedlander,1990b; Rutan & Smith, 1985). There are a number of possible reasons for thisunexpected result. To begin with, perhaps later measures of the alliance wouldhave captured differences between the two formats, as the alliance stabilized inExST-Cpl--in which more of the focus was on the couple\u00E2\u0080\u0099s relationship--but grewstronger in ExST-Ind--in which the therapist-client relationship was the focus.Second, ExST-Cpl, though technically considered conjoint therapy, neverthelesszeroed in on the alcoholic man as opposed to the couple relationship as a whole.Perhaps the attention and support from both the therapist and partnerstrengthened for the men what in traditional conjoint treatment would havebeen lower therapeutic alliances. Third, it c\u00C3\u00A0uld be that although the measuredstrength of the alliance in the individual and conjoint formats was the same,qualitative differences, again not able to be accessed by the present scales, existed.Finally, it may be that theory regarding the alliance in the two modalities needsrevision. Though it would seem that the affiance should be weaker in couples\u00E2\u0080\u0099therapy, perhaps it is not.The second finding demonstrated that the alliance contributed significantlyto nearly identical total mean therapeutic gains in each modality. A nonsignificant trend somewhat differentiated the two, suggesting a slightly strongerrelationship between the affiance and dyadic adjustment gains in the individualformat compared to the conjoint one. The actual proportion of variance in DASgain scores explained by the alliance was 5.5% in ExST-Cpl and 12.7% in ExST-Ind.104The above findings seem to indicate that the choice between individual orconjoint treatment does not appear to make a significant difference either interms of the quality of the alliance that may develop or in terms of therapeuticgains in dyadic adjustment. Additionally, the significant affiance-gainrelationship in both formats suggests that even clients who abuse alcohol andhave high levels of dyadic distress are able to form and make use of a therapeuticrelationship. The slightly stronger affiance-gain association for the individualformat makes theoretical sense. More research is needed comparing the twomodalities using more traditional individual and conjoint approaches. It may bethat greater differences between the two will be evidenced when more typicalconjoint issues are addressed as opposed to one person being the identifiedpatient. To carry out such research, an alliance scale will be needed that can beused for both formats.An important question is whether the affiance-gain proportions reportedabove have clinical relevance. Bourgeois et aL (1990) offer a convincingargument supporting the clinical relevance of their findings, ones representingeven lower proportions of variance (3% to 10%) of DAS-gain explained by thetherapeutic affiance. First, they note that their figures are comparable to those inother studies of relationship variables (e.g., Binder, Henry, & Strupp, 1987;Orlinsky & Howard, 1986). Second, they point out that the low variabifity inaffiance scores reduces the predictive power of the variable (Pedhazur, 1982).Third, the multidimensionality of the change process is emphasized, highlightinghow unrealistic it would be to expect one factor to account for a large part of theoutcome (Gurman et al., 1986; Orlinsky & Howard, 1986). This reasoning stronglysuggests that the proportion of the variance in outcome explained by the affiancein the present study has clear clinical relevance.An additional point that must be considered, adding even more weight to105the significance of the findings, is that in both modalities, improving the dyadicadjustment of the partners may not have been the therapeutic mandate for allclients. Because the presenting issue was the man\u00E2\u0080\u0099s alcohol-dependency, itcannot be assumed that all clients also wanted to strengthen their relationship.Some, in fact, may have been committed to its dissolution, and therefore a lowerdyadic adjustment at posttreatment would have meant success and not failure tothem. In such cases, clients may have reported strong therapeutic alliances, butlow DAS outcomes and gains. Given this reality, it is all the more significant thatthe overall mean alliance-gain correlations were as high as they were.Gender and the Alliance in Conjoint TherapyThe examination of the impact of gender upon the alliance carries with itimportant implications for marriage and family therapists who relate to bothsexes simultaneously. Because of the absence of a solid body of comparativestudies, interpretations of these results are necessarily speculative.In the present study, women reported significantly stronger therapeuticalliances than men did. Nevertheless, correlations between the affiance andDAS-pre, DAS-post, and DAS-gain levels were not statistically different for menand women. The greatest difference was evident in the correlation betweenaffiance and final dyadic adjustment levels, men displaying a significant affianceoutcome correlation and women not. This is similar to previous studies(Bennun, 1985; Bourgeois et al., 1989), in which only men showed a significantcorrelation between the affiance and outcome. Bourgeois et al. attributed theirresults to the socio-cognitive nature of their treatment. Assuming that mengenerally prefer cognitive, task-oriented forms of therapy, whereas women preferexperiential, they reasoned that the affiance for men was therefore more potent interms of outcome. The present research indicates that this reasoning needs106rethinking because the experiential form of treatment also found men and notwomen to have a statistically significant affiance-outcome correlation. Theseresults raise questions about attributing the strength of the affiance-outcomerelationship to therapeutic approach (something the individual literature hasshown cannot be done).One possible alternate explanation is that the strength of the therapeuticaffiance may be more \u00E2\u0080\u009Cmeaningful\u00E2\u0080\u009D for men than it is for women. Becauseengagement in a therapeutic relationship is typically more difficult for--or moreoften resisted by--men (Bourgeois et al., 1990; Heubeck et al., 1986), it may be moresignificant for them when such a collaboration with a therapist does occur. Theformation of an affiance for men may thus be a more potent signal of actualtherapeutic engagement than it is for women who are more used to relating toothers. Having had less exposure to relationships in general, men mayexperience the therapeutic relationship as more significant and thereforeefficacious in creating change. How to distinguish \u00E2\u0080\u009Cpotency\u00E2\u0080\u009D or \u00E2\u0080\u009Cmeaningful\u00E2\u0080\u009Dlevels for the two genders is a complex task for future study. Larger sample sizesand more traditional conjoint treatment will be needed to make definitiveconclusions regarding gender and the affiance.Intact and Split AlliancesThe fact that 8 of 21 couples had split affiances and 12 had intact onesprovides more empirical support for these theoretical constructs identified byPinsof and Catherall (1986). The trend for couples with intact affiances toimprove somewhat more than those with split affiances is consistent withBennun\u00E2\u0080\u0099s (1989) finding that the greater the discrepancy between partners\u00E2\u0080\u0099perceptions of the therapist, the poorer the clinical outcome. A split affianceindicates that one partner is not in agreement with the tasks and goals of therapy107or does not feel bonded to the therapist while the other partner does agree or doesfeel a bond. It appears that such a discrepancy can interfere somewhat with theprogress of the couple in therapy.There are several explanations as to how this might happen. Heatheringtonand Friedlander (1990a) postulated that the alliance the therapist has with onepartner impacts the affiance with the other partner in a reciprocal manner. Thus,an intervention with one can have indirect results on the other who is observingthe interaction. A split alliance could result in the partner with the weaker-affiance consciously or unconsciously resisting or even sabotaging the work oftherapy. A reluctance by one to participate, for example, could deter the otherfrom being as vulnerable. Also possible with split affiances is the issue oftriangulation (Jacobson, 1981). The person with the weaker alliance may feelthere is a coalition between the partner and therapist. Finally, a split affiance mayindicate that the partners disagree as to what the problem is. For example, thewoman may have considered the man\u00E2\u0080\u0099s alcohol dependency to be the problem,whereas the man may have felt the problem was the woman\u00E2\u0080\u0099s reactivity. Givendifferent problem definitions, discrepancies in therapeutic goals will likely result,affecting one aspect of the alliance. Therapeutic tasks will also be understooddifferently, and ultimately the outcome of therapy may be compromised.A second trend ifiuminating the change process within conjoint therapy wasthat the correlation between partners\u00E2\u0080\u0099 level of agreement about their dyadicadjustment increased for both intact and split couples, suggesting that both weremoving in the direction of an increasingly intact dyadic affiance. Had therapeuticalliance scores been measured at posttreatment, it could have been determinedwhether higher levels of couple agreement regarding dyadic adjustmenttranslated to greater couple agreement on the affiance. According to Pinsof (in108press), it is important to consider the many different kinds of affiances that existin therapy and what relationship they have to one another. Multiple alliancemeasures would be needed to address this issue.An adjunct analysis demonstrated that although alliance level and alliancediscrepancy were not significantly related, the fact that all nine cells of acontingency table (measuring high, medium, and low discrepancies, and both-high, mixed, and both-low alliances) were represented with at least one coupledemonstrates the complexity of the alliance construct. \u00C2\u00BE/here partners scorealong the continuum of low to high, and where one scores in relation to one\u00E2\u0080\u0099spartner are both important considerations for future research. Just because bothpartners have high alliances does not mean they also have high agreementbetween themselves. Five couples represented partners who both scored high onthe alliance, yet only two had high agreement. Similarly, of the seven coupleswho measured \u00E2\u0080\u009Cboth low\u00E2\u0080\u009D alliances, only one had a high discrepancy. It may bethat the level of discrepancy explains more variance in therapeutic gain than theactual level of affiance does.Drinking Patterns and the AllianceBecause the following inquiries are preliminary empirical attempts toexplore the relationship of drinking types and the alliance, interpretations areadvanced tentatively. The findings demonstrate that the relationship is far fromsimple. To begin, two results require some explanation.First, the fact that binge and chronic drinkers displayed no statisticaldifference in the level of therapeutic alliance in both treatments comes as asurprise, given the alcohol literature that indicates that binge drinkers will findinterpersonal relationships more difficult (Babor et al., 1992; Litt et al., 1992; Jacob& Leonard, 1988; Moos & Moos, 1984; Morey et al., 1984) and the affiance literature109showing that pretherapy social distress has a negative impact on affianceformation (Eaton et aL, 1988; Gelso & Carter, 1985; Luborsky et al., 1983; Marmar etal., 1989; Marziali, 1984; Moras & Strupp, 1982). Second, although the affiance wasas strong for binge drinkers as it was for chronic drinkers, its relationship to gainwas significantly different. Both groups made significant therapeutic gains, but itwas the latter that showed a significantly higher positive correlation between theaffiance and improvement. Binge drinkers displayed a negative alliance-gaincorrelation. These findings suggest that the therapeutic role of the affiance fortwo drinking types was different. It is noteworthy that the most potent predictorof the strength of the correlation between alliance and outcome was nottreatment type, but alcohol-use patterns.There are a number of possible explanations for these two findings. First, inexamining the similarity between binge and chronic affiance scores, the systemicnature of the affiance scales must be taken into consideration. Indirectsubsystems of a client are tapped through statements like, \u00E2\u0080\u009CSome of the peoplewho are important to me would distrust the therapist\u00E2\u0080\u009D (Item 16) and \u00E2\u0080\u009CThe peoplewho are important to me would feel accepted by the therapist\u00E2\u0080\u009D (Item 23). Had allitems pertained only to the therapist and client, affiance levels may have beenquite different. The literature (Lift et al., 1992) describes binge drinkers asimpulsive, untrustworthy, and incapable of anticipating the reactions of others-partly because of an insensitivity to others\u00E2\u0080\u0099 expectations. If such descriptions areaccurate, obtaining reliable systemic assessments of the affiance will be moredifficult for the binge drinker. Having sufficient sensitivity to reliably rate patientsystems beyond the direct therapeutic one is challenging at best for many clients.Much more widespread use and refinement of the instruments is needed toaddress this complication.110A second explanation a1ong the same lines of thinking is that, given alimited relationship history, binge drinkers\u00E2\u0080\u0099 estimation of the alliance may havebeen inflated to levels of strength that other, more relationally experiencedpersons would consider weak. Another possibility is that if other relationships ofthe binge drinkers were poor, the therapeutic one may have seemed strong bycomparison. Observer and/or therapist reports of the affiance would be one wayto validate relationship strength. The client\u00E2\u0080\u0099s experience of a strong affiancewould remain valid; however, for purposes of comparison, multiple reportswould improve the reliabffity.A final explanation must deal specifically with the poorer binge affiance-gainrelationship. Perhaps because binge drinkers have presumably had more difficultand unsuccessful personal interactions, they may be less likely to rely uponrelationships in general, having learned to cope in life using other means.Experiencing a strong therapeutic alliance may have been novel, pleasant, orterribly frightening, but knowing how to make use of the relationship and beingable to trust the therapist consistently over time may have been challenges forwhich the binge drinker was not prepared.Speculations aside, it is clear that when working with binge drinkers,clinicians would do well to be aware of potentially weaker affiance-gainrelationships. What variables contribute to therapeutic gain for binge drinkersremains an important area for future research. Perhaps one factor is simplybuilding predictability into the binge drinker\u00E2\u0080\u0099s life by a schedule of counseffingsessions. Providing greater stabffity may create enhanced perceptions of dyadicadjustment.For chronic drinkers, the affiance appears to be a powerful predictor oftherapeutic improvement. Therefore therapists can be more confident inexpecting that strong bonds and agreement on tasks and goals wifi promote111therapeutic change for this drinking type. Though results were not significant,trends in the study did lend support to the idea that different kinds of treatmentapproaches might better facifitate therapeutic growth for the two drinking types(Litt et at, 1992). Chronic drinkers showed slightly greater improvement on theDAS in the ExST counseffing format than binge drinkers, raising the possibifitythat the experiential, interpersonal approach may have been somewhat morecongruent for chronic drinkers. Further research is needed to test this hypothesis.In an adjunct analysis using both the alcoholic men and their femalepartners, drinking type and the similarity of partner& alliance levels were foundto be significantly related. Couples with binge drinkers either both reported lowtherapeutic alliances or had mixed alliances, none showing both partners withhigh affiances. Only with chronic couples did both partners display levels ofaffiance with the therapist that were above the mean. Though the sample sizewas too small to make major conclusions, these findings seem to suggest thatwhen involved conjointly, couples influenced by a binge as opposed to chronicdrinking pattern may have more difficulty forming therapeutic relationships.The women from binge relationships tended to report weaker therapeuticalliances than those from chronic relationships. One possible explanation is thatbinge drinkers may have required more attention from the therapist, leaving lesstime for their partners who consequently would not have experienced as strong abond or even as much agreement on the tasks and goals. Further research using agreater number of couples is needed to verify this hypothesis.112Pretreatment Distress and the AllianceThe lack of relationship between pretreatment distress and the affiance wasremarkably consistent in all the analyses. Because earlier studies screenedalcoholics from their samples, the present research is the first to demonstrate thatpretreatment dyadic distress, due at least in part to alcohol-related issues, did notprevent the formation of strong alliances. This finding extended to bingedrinkers, whose alliance levels were as robust as those of chronic drinker& eventhough pretreatment DAS scores were lower. Similarly, in conjoint therapy,female partners of alcoholics showed somewhat higher levels of pretreatmentdyadic distress, yet developed significantly stronger alliances. The very weakassociation found between pretreatment distress and the affiance is consistentwith the previous work on this subject (Bourgeois et al., 1986; Graman, 1986). Itseems clear that the affiance has far more power as a predictor of treatmentoutcome than it does as a reflector of pretreatment problems, even with highlydistressed populations.LimitationsThis study should be viewed as a process-outcome investigation designed toencourage others to pursue more research in the area. Of interest in this sectionis the generalizabffity of the results. Because therapy occurred in its naturalcontext--the therapists in the study currently employed by the agencies in whichthe counselling took place--generalizabffity is enhanced. Given the sample,results are generalizable to couples consisting of one alcoholic partner withoutpsychiatric problems or other substance abuse disorders and for whom outpatienttreatment is suitable. The methodology of the study has certain shortcomings,and both methodological and conceptual issues require acknowledgement.113Methodological LimitationsOne-time measurement. The therapeutic alliance scale was administered atonly one point in therapy, posing a potential limitation. One-time scores limitconclusions because fluctuations in the alliance over the course of therapy cannotbe detected. Future research is needed to test whether the tearing and repairing ofthe affiance is a critical component of successful therapy (Horvath & Marx, 1990;Pinsof, in press).Theorists and researchers remain divided as to the importance of multipleaffiance measures. One school argues that the alliance varies markedly fromearly to late stages in therapy, even manifesting relatively large between-sessionfluctuations throughout (Horvath, Marx & Kamman, 1990; Horvath & Symonds,1991; Kokotovic & Tracey, 1990; Marziali, 1984). Another school claims that oncethe affiance is established, it remains fairly stable and relatively resistant tochange (Eaton et al., 1988; Luborsky, 1976; Tichenor & Hifi, 1989). Only furtherstudies using multiple measurements can settle the debate.Source of report. This study sought to learn only the patient\u00E2\u0080\u0099s perspective onthe affiance. Future research should solicit multiple perspectives. There areproblems of circularity associated with using the same source for both process andoutcome measures. Clients\u00E2\u0080\u0099 reports of the therapeutic relationship may be morestrongly influenced by how well therapy is corresponding to client expectationsthan by actual therapeutic events (Horvath et aL, 1990; Marmar et al., 1986). It hasbeen observed that as symptoms improve, so often does the quality of the alliance(Marmar et al., 1989). Marmar et al. (1986) recommend the use of both self andobserver reports to overcome what they term the \u00E2\u0080\u009Credundancy of perspective\u00E2\u0080\u009Dthat can occur when clients report on both process and outcome (p. 367). Systemic114principles acknowledge the reality of multiple perspectives, there being no one\u00E2\u0080\u009Ctrue\u00E2\u0080\u009D picture as all persons are limited to only partial perspectives whetherobserving or participating (Pinsof, 1989). Research on the affiance will beenhanced as the unique components of the perceptions of different sources areidentified.Early termination. The present research did not attempt to study subjectswho terminated therapy after having completed the therapeutic affiance scale inthe fifth session. Investigating the strength of early terminators\u00E2\u0080\u0099 affiance scoresrepresents an important area of future research. Of interest is determiningwhether the level of their affiance would have predicted their prematuretermination.Statistical limits. The use of gain scores in this study did not take intoaccount where on the continuum of improvement a subject\u00E2\u0080\u0099s scores fell. Animportant question for future study is whether discrepancies of equal size atdifferent points along the level of alliance strength or dyadic adjustment expressdifferent meanings (Fisher et al., 1985). Additionally, certain analyses, particularlythe latter two involving only couples, are limited in their generalizabffity by therelatively small number of subjects involved.Conceptual IssuesTherapist variables. Therapist variables not accounted for by the studyconstitute potential limitations. The alliance scales are limited because they aredesigned only for patient report (Pinsof, in press). Therapists are a keycomponent of the affiance, each developing the therapeutic relationshipdifferently and each bringing issues into the counselling room that either115facifitate or detract from the forming of a good alliance. Their readiness to carefor or like a client wifi vary. How willing they are to disclose their own feelingsand past experiences may impact the alliance, as may their level of activity orpassivity (Bordin, 1979). Their gender and family life are other potentialmediating factors (Catherail & Pinsof, 1987).Accounting for the amount of variance due to the therapist\u00E2\u0080\u0099s style,personality, gender, and level of experience was not an aim of this study thoughadmittedly these are potentially significant in affecting a client\u00E2\u0080\u0099s perception of atherapist\u00E2\u0080\u0099s abilities.Conjoint and individual treatment comparisons. Using the therapeuticaffiance scales to compare the affiance across conjoint and individual formatsposes certain limitations in the interpretation of the results. W. M. Pinsof(personal communication, May 29, 1993) explains that the three scales are\u00E2\u0080\u009Cconceptually parallel,\u00E2\u0080\u009D designed according to identical theoretical assumptions.Nevertheless, he observes that they are not parallel in the sense of being able to becompared empirically, the most obvious reason being the different number ofitems in each. A more serious issue is whether differences in the wording of theparallel scale items reflect substantive differences in the various treatmentmodalities that may influence the way participants respond (Heatherington &Friedlander, 1990b). More work on the construct is needed to come up with scalesthat are both psychometrically and conceptually equivalent, so that confidentclaims can be generated regarding differences across treatment modalities.Conjoint and unilateral couples therapy. Whether traditional conjoint orunilateral couples therapy actually occurred in TARP presents a potentiallimitation of the study\u00E2\u0080\u0099s generalizabilty. According to the technical definitions offamily and unilateral couples therapy offered in Chapter One, these therapeutic116forms did occur. However, unlike traditional marital therapy where the primaryfocus of treatment is on the interaction of the couple as a unit\u00E2\u0080\u0094the client beingthe relationship--the emphasis in ExST-Cpl was often on alcohol and the alcoholicman. Thus, whereas the usual task of the couples therapist is to connect with thesystem rather than the individual, in ExST-Cpl the alliance between the therapistand the man often took precedence over that between the therapist and couple(Davatz, 1981). As a result, affiance scores from women and men may not betypical of the usual conjoint context.Similarly, traditional unilateral couples therapy, though systemic inorientation, does not normally provide the luxury of having the nontreatedpartner participating through weekly reports and completion of instruments.Because both partners played significant indirect roles in SFT and ExST-Ind, theaffiance measured by the ITAS may be different from what it would be fortraditional unilateral couples therapy. Presumably scores would be higherbecause the father still received all of the therapist\u00E2\u0080\u0099s attention, but he also knewhis partner was very much aware of the tasks and goals of therapy. Overall, thedifferences between the two formats may not be as apparent because of the uniquesituation offered by TARP.Implications for Future ResearchIn general, the findings of this study wifi help to enrich the field of familytherapy by contributing to the small body of literature related to the therapeuticaffiance, especially regarding alcohol. Results from this study suggest directionsfor future research, a few of which have already been highlighted in the previoussection. Following are some of the areas needing further investigation, first thosepertaining more specifically to clinical concerns, and last those areas related more117to research concerns.Clinical Areas for ResearchIndividual versus family therapy. Continued research exploring thedifferences between the individual and family contexts is needed. Discussion at aconceptual level is emerging, and empirical work must follow. How differentaspects of the affiance moderate the process in the two modalities as well as whatkinds of affiances exist across treatments (e.g., Type I or Type II alliances) are twoexisting questions. Already, indications of difference have been spotted,associations between tasks and bonds seeming to be more apparent in familytherapy, whereas tasks and goals appear more related in individual therapy(Heatherington & Friedlander, 1990b). This may suggest a difference in the natureof the goals or in clients\u00E2\u0080\u0099 perceptions of goals in individual versus family therapy.Therapist variables. Pinsof (1989) notes that beyond demographic andpersonality variables, there has been very little attention paid to the therapistsystem and how it pertains to the affiance. Because the individual literatureemphasized therapist characteristics so much through the work of Rogers and hisassociates, family research did not see the need to scrutinize the personalattributes of therapists to the same extent (Hooper, 1985). It seems obvious thattherapist characteristics such as gender, age, and marital status, affect the alliancefor certain clients (Beutler et al., 1986; Dryden & Hunt, 1985a).Gender of clients. The impact of gender on the affiance needs further study.Better subscales would be helpful in determining if, for instance, agreement ontasks between counsellors and females is stronger at the outset of therapy than itis between males and counsellors. Timing is another issue. Perhaps the criticalphase of affiance development differs for men and women (Bourgeois et al., 1990)118Intact and split alliances. The present study makes even more clear howprevalent split and intact alliances are, underscoring the fact that any timecouples\u00E2\u0080\u0099 mean scores are used in family research, discrepancies should be takeninto consideration. A couple with extremely different individual alliance scoresmay have the same mean conjoint score as a couple whose individual scores areboth close to the mean. The differences between the two couples are obscured ifdiscrepancies are not acknowledged.The abifity to identify and creatively utilize the realities of split and intactalliances represents an important area of future research, providing the potentialto influence treatment successes and failures in the conjoint setting. Moresophisticated affiance scales that could identify the particular aspects on whichpartners are split could assist therapists in addressing the disaepancy.Remaining questions include (a) whose report in the conjoint context bestpredicts distal outcomes? (b) whose perspective, the father\u00E2\u0080\u0099s or mother\u00E2\u0080\u0099s, andwhat kind of conjoint score is best for research and for therapy (Pinsof, 1989)? and(c) what is more important for therapeutic gain, a high alliance or a lowdiscrepancy?Alcohol and the alliance. Results from the present study indicate the needfor more research into the relationship of alcohol-use patterns and the affiance.The significant relationship of drinking type and couples\u00E2\u0080\u0099 agreement on thealliance in the present study suggests areas for future study using larger samplesizes. Additionally, slight trends within the individual theapy contexts suggestedpreferred therapeutic approaches for different drinking types. Of the fourindividual therapeutic groups, chronic drinkers in ExST-Ind made slightly highergains than the others, whereas binge drinkers treated by ExST\u00E2\u0080\u0094Ind experienced theleast improvement. Binge drinkers may have found the concrete, structured119approach of SFT combined with the safety of not having to deal as intensivelywith interpersonal issues more appealing. Chronic drinkers, who theoreticallyhave more successful interpersonal histories, may not have experienced the samelevel of threat posed by ExST-Ind and may have felt the structured coping skillstreatment of SFT was too restrictive. Further empirical work is required for theverification of these possibilities.Cultural implications. In the West, it is often believed that strongrelationships will be characterized by agreement, openness, and honest expressionof feelings. A compelling research question is whether therapeutic alliancesrequire different ingredients in other cultures.Areas to Promote Further ResearchContinued use of the Therapeutic Affiance Scales. The use of the ITAS,CTAS, and FTAS should continue, accompanied by ongoing exploration of theirtheoretical underpinnings and the addition of a system of rating scalesincorporating therapist and observer perspectives (Horvath & Symonds, 1991;Pinsof, in press). In particular, the subscale dimensions need refinement. Scaleswith discrete subdimensions would make possible the tracking ofmicrotherapeutic change processes. Empirical verification of theoreticalhypotheses such as \u00E2\u0080\u009CExST-Ind wifi produce stronger bonds than SFT\u00E2\u0080\u009D could thenoccur. Also important would be detecting trends in subscales between intact andsplit alliances--for example, \u00E2\u0080\u009COver what are split couples typically divided--goals,bonds, or tasks?\u00E2\u0080\u009DGetting at qualitative differences in the dimensions, though admittedlydifficult to pinpoint, would be a next step. Pinsof (in press) suggests creating\u00E2\u0080\u009Calliance profiles\u00E2\u0080\u009D that would characterize different therapist-client interactions.120Process studies. Future research on the affiance requires process studies.There needs to be a focus on specific in-session behaviors of therapists and familymembers that are associated with relatively stronger or weaker affiances(Heatherington & Friedlander, 1990a). Frieswyk et al. (1986) suggest an interplaybetween (a) group-comparison research to identify broad patient and therapisttypes of characteristics and their interaction, and (b) discovery-orientedapproaches that focus on intensive single-case studies, systematically exploringinteractions at more molecular levels. Microscopic evaluations of therapeuticchange events could help answer questions like, \u00E2\u0080\u009CWith what type of patient andat what phase of the process do particular classes of interventions facffitate thedevelopment of an affiance and with what short- and long-term impact?\u00E2\u0080\u009D (Rice &Greenberg, 1984, p. 33).The indirect system of the client is yet another, very complex area forresearch. It is already difficult enough doing process research on the directtherapeutic system. Stifi, the systemic perspective demands that the larger systembe addressed (Gurman et al., 1986) The indirect patient system plays a critical partin the process and outcome of all forms of treatment.121ReferencesAdler, J. V. (1989). A study of the working alliance in psychotherapy.Unpublished doctoral dissertaion, University of British Columbia,Vancouver.Anastasi, A. (1988). Psychological testing (6th ed.). NY: Macmillan PublishingCompany.Babor, T. F., Hofmaim, M., DelBoca, F. K., Hesseibrock, V., Meyer, R. E., Dolinsky,Z. S., & Rounsaville, B. (1992). Types of alcoholics, I: Evidence for anempirically derived typology based on indicators of vulnerabifity and severity.Archives of General Psychiatry, 42, 599-608.Barrett-Lennard, G. T. (1962). Dimensions of therapist response as a causal factorin therapeutic change. Psychological Monographs,.Z., 1-36.Barrett-Lennard, G. T. (1978). The Relationship Inventory: Later developmentand adaptations. JSAS Catalog of Selected Documents in Psychology, , 68.(Ms. No. 1732)Barrett-Lennard, G. T. (1985). The helping relationship: Crisis and advance intheory and research. The Counseling Psychologist, ii 278-294.Bennun, I. (1985). Unilateral marital therapy. In W. Dryden (Ed.), Maritaltherapy in Britain. Volume 2: Special areas (pp. 3-23). London: Harper andRow.Bennun, I. (1989). Perceptions of the therapist in family therapy. Journal ofFamily Therapy, 11,243-255.Beutler, L., Crago, M. & Arizmendi, T. (1986). Therapist variables inpsychotherapy process and outcome. In S. L. Garfield and A. E. Bergin (Eds.),Handbook of psychotherapy and behavior change (3rd ed.), (pp. 257-310). NewYork: John Wiley & Sons.Bordin, E. 5. (1979). The generalizabifity of the psychoanalytic concept of theworking alliance. Psychotherapy: Theory, Research and Practice, jj, 252-260.Bourgeois, L., Sabourin, S., & Wright, J. (1989). A study of the interstructure ofthe Couples Therapeutic Alliance Scale. Unpublished doctoral dissertation,University of Montreal, Quebec.122Bourgeois, L., Sabourin, S., & Wright, J. (1990). Predictive validity of therapeuticaffiance in group marital therapy. Journal of Consulting and ClinicalPsychology, 608-613.Catherall, D. R. (1984). The therapeutic alliance in individual, couple and familytherapy. Unpublished doctoral dissertation. Northwestern University,Evanston, IL.Catherall, D. R., & Pinsof, W. M. (1987). The impact of the therapist\u00E2\u0080\u0099s personalfamily life on the ability to establish viable therapeutic alliances in family andmarital therapy. Journal of Psychotherapy and the Family, a, 135-160.Crocker, L., & Algina, J. (1986). Introduction to classical and modern test theory.New York: Holt, Rinehart, and Winston.Davatz, U.. (1981). Establishing a therapeutic affiance in family systems therapy.In A. S. Gurman (Ed.), Ouestions and answers in the practice of familytherapy (pp. 44-49). New York: Brunner/Mazel.Derogatis, L. R. (1983). SCL-90-R: Administrative, scoring and proceduresmanual - II for the revised version and other instruments of psychopathologyrating scale series. Towson, MD: Clinical Psychometric Research.Dryden, W. & Hunt, P. (1985a). Therapeutic alliances in marital therapy: I. Pretherapy influences. In W. Dryden (Ed.), Marital therapy in Britain, Volume 1:Context and therapeutic approaches (pp. 121-143). London: Harper and Row.Dryden, W. & Hunt, P. (1985b). Therapeutic affiances in marital therapy: II.Process issues. In W. Dryden (Ed.), Marital Therapy in Britain. Volume 1:Context and Therapeutic Approaches (pp. 144-168). London: Harper and Row.Eaton, T. T., Abeles, N., & Gutfreund, M. J. (1988). The therapeutic affiance andoutcome: Impact of treatment length and pre-treatment symptomatology.Psychotherapy. 2, 536-542.Fimiey, J. W., & Moos, R. H. (1979). Treatment and outcome for empiricalsubtypes of alcoholic patients. Journal of Consulting and Clinical Psychology,Z, 25-38.Fisher, L., Kokes, R. F., Ransom, D. C.., Phillips, S. L. & Rudd, P. (1985).Alternative strategies for creating \u00E2\u0080\u009Crelational\u00E2\u0080\u009D family data. Family Process, .,213-224.Frank, A. F., & Gunderson, J. G. (1990). The role of the therapeutic alliance in thetreatment of schizophrenia. Archives of General Psychiatry, 4Z, 228-236.123Freud, S. (1958a). On the beginning of treatment: Further recommendations onthe technique of psychoanalysis. London: Hogarth Press. (Original workpublished 1913).Freud, S. (1958b). The dynamics of transference (1912). In J. Strachey (Ed.),Standard Edition, 12, pp. 99-108. London: Hogarth Press and Institute ofPsychoanalysis.Friesen, J. 0., Grigg, 0. N., & Newman, J. A. (1991). Experiential SystemicTherapy: For individuals, couples and families (rev. edi. Unpublishedmanuscript, University of British Columbia, Vancouver.Frieswyk, S. H., Allen, J. G., Colson, 0. B., Coyne, L., Gabbard, G. 0., Horowitz, L.,& Newsom, G. (1986). Therapeutic alliance: Its place as process and outcomevariable in dynamic psychotherapy research. Journal of Consulting andClinical Psychology. 4, 32-39.Gelso, C. J., & Carter, J.. A. (1985). The relationship in counseling andpsychotherapy: Components, consequences, and theoretical antecedents. TheCounseling Psychologist. i, 155-243.Gilbert, R., Christensen, A., & Margolin, G. (1984). Patterns of affiance in non-distressed and multiproblem families. Family Process, 2, 75-87.Glass, G. V., & Stanley, J. C. (1970). Statistical methods in education andpsychology. New Jersey: Prentice-Hall, Inc.Gomes-Schwartz, B. (1978). Effective ingredients in psychotherapy: Prediction ofoutcome from process variables. Journal of Consulting and ClinicalPsychology, 1023-1035.Graham, J. R., & Strenger, V. E. (1988). IVIMPI characteristics of alcoholics: Areview. Journal of Consulting and Clinical Psychology, , 197-205.Greenberg, L. & Pinsof, W. (1986). Process research: Current trends and futureperspectives. In L. Greenberg & W. Pinsof (Eds.), The PsychotherapeuticProcess: A Research Handbook (pp. 3-20). New York: Gui]ford.Greenberg, L. S., & Webster, M. C. (1982). Resolving decisional conflict by Gestalttwo-chair dialogue: Relating process to outcome. Journal of CounselingPsychology. 22,468-477.Greenson, R. R. (1967). The technique and practice of psychoanalysis (Vol 1).New York: International Universities Press.124Grigg, D. N., Friesen, J. D., Web, W., & Bate, C. (1991). Supported FeedbackTherapy Manual. Vancouver. Western Family Learning Institute.Gruman, M. (1986). A study of the therapeutic affiance in couple therapy.Unpublished master\u00E2\u0080\u0099s thesis, Department of Counseffing Psychology,University of British Columbia, Vancouver, B. C., Canada.Gurman, A. S. (Ed.). (1981). Ouestions and answers in the proactice of familytherapy. New York: Brunner/Mazel.Gurman, A. S. & Kniskern, D. P. (1978). Deterioration in marital and familytherapy: Empirical, clinical and conceptual issues. Family Process, 1Z, 3-20.Gurman, A. S. & Kniskern, D. P. (Eds.). (1981). The handbook of family therapy.New York: Brunner/Mazel.Gurman, A. S., Kniskern, D. P., & Pinsof, W. M. (1986). Research on the processand outcome of marital and family therapy. In S. L. Garfield and A. E. Bergin(Eds.), Handbook of psychotherapy and behavior change (3rd ed.), (pp. 565-624). New York: John Wiley & Sons.Gutterman, D. L. (1984). The stability of the therapeutic alliance: A study of thealliance across eight sessions of couples psychotherapy. Unpublishedmanuscript, Departments of Psychology/Psychiatry, Northwestern UniversityMedical School, Chicago, IL.Hartley, D. E., & Strupp, H. (1983). The therapeutic affiance: Its relationship tooutcome in brief psychotherapy. In J. Masling (Ed.), Empirical studies inpsychoanalytic theories (pp. 1-37). Hifisdale, NJ: Eribaum.Heatherington, L., & Friedlander, M. L. (1990a). Complementarity and symmetryin family therapy communication. Journal of Counseling Psychology,.Z1 261-268.Heatherington, L., & Friedlander, M. L. (1990b). Couple and Family TherapyAlliance Scales: Empirical considerations. Journal of Marital and FamilyTherapy, j, 299-306.Heppner, P. P., Kivlighan, D. M., & Wampold, B. E. (1992). Research design incounseling. Pacific Grove, CA: Brooks/Cole Publishing Company.Heubeck, B., Detmering, J., & Russell, G. (1986). Father involvement andresponsibffity in family therapy. In M. Lamb (Ed.), The father\u00E2\u0080\u0099s role: Appliedperspectives (pp. 191-226). New York: Wiley.125Hooper, D. (1985). Marital therapy: An overview of research. In W. Dryden (Ed.),Marital therapy in Britain. Volume 2: Special areas (pp. 254-300). London:Harper and Row.Horvath, A. 0. (1981). An exploratory study of the working alliance: Itsmeasurement and relationship to the therapy outcome. Unpublisheddoctoral dissertation, University of British Columbia, Vancouver.Dissertation Abstracts International, 42., 06A-2503.Horvath, A. 0., & Greenberg, L. S. (1986). The development of the WorkingAlliance Inventory. In L. S. Greenberg and W. M. Pinsof (Eds.), Thepsychotherapeutic process: A research handbook (pp. 529-556). New York:Guilford.Horvath, A. 0., & Marx, R. W. (1990). The development and decay of theworking alliance during time-limited counselling. Canadian Journal ofCounselling, 24, 240-259.Horvath, A.O., Marx, R. W., & Kamann, A. M. (1990). Thinking about thinkingin therapy: An examination of clients\u00E2\u0080\u0099 understanding of their therapists\u00E2\u0080\u0099intentions. Journal of Consulting and Clinical Psychology. , 614-621.Horvath, A. 0., & Symonds, B. D. (1991). Relation between working affiance andoutcome in psychotherapy: A meta-analysis. Journal of CounselingPsychology., 139-149.Horwitz, L. (1974). Clinical prediction in psychotherapy. New York: JasonAronson.Jacob, T. (1992). Family studies of alcoholism. Journal of Family Psychology, ,319-338.Jacob, T., & Leonard, K. E. (1988). Alcoholic-spouse interaction as a function ofalcoholism subtype and alcohol consumption interaction. Journal ofAbnormal Psychology. 2Z, 231-237.Jacobson, N. S. (1981). Behavioral marital therapy. In A. S. Gurman & D. P.Kniskern (Eds.), Handbook of family therapy (pp. 556-591). New York:Brunner/Mazel.Johnson, S. M., & Greenberg, L. S. (1985). Emotionally Focused Couples Therapy:An outcome study. Journal of Marital and Family Therapy. ii, 313-317.Kokotovic, A. M., & Tracey, T. J. (1990). Working affiance in the early phase ofcounseling. Journal of Counseling Psychology,, 16-21.126Lift, M. D., Babor, T. F., DelBoca, F. K., Kadden, R. M., & Cooney, N. L. (1992).Types of alcoholics, II: Application of an empirically derived typology totreatment matching. Archives of General Psychiatry, , 609-614.Locke, H. J., & Wallace, K. M. (1959). Short marital adjustment and predictiontests: Their reliability and validity. Marriage and Family Living, 21, 251-255.Luborsky, L. (1976). Helping alliances in psychotherapy. In J. L. Claghorn (Ed.),Successful psychotherapy (pp. 92-111). New York: l3rumier/Mazel.Luborsky, L., Crits-Cristoph, P., Alexander, L., Margolis, M., & Cohen, M. (1983).Two helping alliance methods for predicting outcome of psychotherapy.journal of Nervous and Mental Disease, LZ1, 480-491.Luborsky, L., Mintz, J., Auerbach, A., Christoph, P., Bachrach, H., Todd, T.,Johnson, M., Cohen, M., O\u00E2\u0080\u0099Brien, C. (1980). Predicting the outcomes ofpsychotherapy: Findings of the Penn Psychotherapy Project. Archives ofGeneral Psychiatry, Z, 471-481.Marmar, C. R., Horowitz, M. J., Weiss, D. S., & Marziali, E. (1986). Thedevelopment of the Therapeutic Alliance Rating System. In L. S. Greenberg& W. M. Pinsof (Eds.), The psychotherapeutic process: A research handbook(pp. 367-390). New York: Guilford.Marmar, C. R., Weiss, D. S., & Gaston, L. (1989). Toward the validation of theCalifornia Therapeutic Alliance Rating System. Psychological Assessment: AJournal of Consulting and Clinical Psychology, j., 46-52.Marziali, E. (1984). Three viewpoints on the therapeutic affiance: Similarities,differences and associations with psychotherapy outcome. Journal ofNervous and Mental Disease, LZZ, 417-423.Marziali, E., Marmar, C., & Krupnick, J. (1981). Therapeutic affiance scales:Development and relationship to therapeutic outcome. American Journal ofPsychiatry, j., 361-364.McCrady, B. S., Noel, N. E., Abrams, D. B., Stout, R. L., Nelson, H. F., & Hay, W.M. (1986). Comparative effectiveness of three types of spouse involvementin outpatient behavioral alcoholism treatment. Journal of Studies onAlcohol, IZ, 459-467.Mitchell, K. M., Bozart, J. D., & Krauft, C. C. (1977). Reappraisal of the therapeuticeffectiveness of accurate empathy, non-possessive warmth, and genuineness.In A. S. Gurman & A. M. Razin (Eds.), Effective psychotherapy (pp. 482-502).New York: Pergamon.127Moos, R. H., & Moos, B. S. (1984). The process of recovery from alcoholism: III.Comparing functioning in families of alcoholics and matched controlfamilies. Journal of Studies on Alcohol, 5, 111-118.Moras, C., & Strupp, H. H. (1982). Pretherapy interpersonal relationship, patients\u00E2\u0080\u0099alliance and outcome in brief therapy. Archives of General Psychiatry, 2,405-409.Morey, L. C., Skinner, H. A., & Blashfield, R. K. (1984). A typology of alcoholabusers: Correlates and implications. JQwnal of Abnormal Psychology. 23.,408-417.Morgan, R., Luborsky, L., Crits-Christoph, P., Curtis, H., & Solomon, J. (1982).Predicting the outcomes of psychotherapy by the Penn helping alliance ratingmethod. Archives of General Psychiatry. 39, 397-402.Olson, D. H., & Wilson, M. (1982). Family Satisfaction. St. Paul, MN: FamilySocial Science, University of Minnesota.Orlinsky, D.E., & Howard, K. I. (1986). Process and outomce in psychotherapy. InS. L. Garfield and A. E. Bergin (Edsi, Handbook of psychotherapy andbehavior change (pp. 311-381). New York: John Wiley & Sons.Pinsof, W. M. (1981). Family therapy process research. In A. S. Gurman & D. P.Kniskern (Eds.), Handbook of family therapy. New York: Brunner/Mazel.Pinsof, W. M. (1989). A conceptual framework and methodological criteria forfamily therapy process research. Journal of Consulting and ClinicalPsychology. Z, 53-59.Pinsof, W. M. (in press). An integrative systems perspective on the therapeuticaffiance: Theoretical, clinical, and research implications. In A. Horvath and L.Greenberg (Eds.), The therapeutic affiance.Pinsof, W. M., & Catherali, D. R. (1986). The integrative psychotherapy affiance:Family, couple, and individual therapy scales. Journal of Marital and FamilyTherapy. 12, 137-151.Raistrick, D., Dunbar, G., & Davidson, R. (1983). Development of a questionnarieto measure alcohol dependence. British Journal of Addiction, Z., 89-95.Rice, D., & Greenberg, L. (1984) Patterns of change: Intensive analysis ofvsvchotherauv nrocess. New York: Guilford.128Rogers, C. R. (1957). The necessary and sufficient conditions of therapeuticpersonality change. Journal of Consulting Psychology. 22, 95-103.Rosenbaum, A., & O\u00E2\u0080\u0099Leary, K. D. (1986). The treatment of marital violence. InN. S. Jacobson and A. S. Gurman (Eds.), Clinical handbook of maritaltherapy (pp. 385-405). New York: Guilford.Rounsavifie, B. J., Chevron, E. S., Prusoff, B. A., Elm, I., Imber, S., Sotsky, S., &Watldns, J. (1987). The relation between specific and general dimensions ofthe psychotherapy process in interpersonal psychotherapy of depression.Journal of Consulting and Clinical Pscholog, , 379-384.Rutan, S. J., & Smith, J. W. (1985). Building therapeutic relationships withcouples. Psychotherapy, 22, 194-200.Safran, J.D., & Waliner, L.K. (1991). The relative predictive validity of twotherapeutic affiance measures in cognitive therapy. PsychologicalAssessment: A Journal of Consulting and Clinical Psychology, , 188-195.Seizer, M. L. (1971). The Michigan Alcoholism Screening Test: The quest for anew diagnostic instrument. American Journal of Psychiatry, 1ZZ, 1653-1658.Smith, M., Glass, G., & Miller, T. (1981) The benefits of psychotherapy.Baltimore: Johns Hopkins University Press.Spanier, G. G. (1976). Measuring dyadic adjustment: New scales for assessing thequality of marriage and other similar dyads. Journal of Marriage and theFamily, , 15-28.Spatz, C., & Johnston, J. 0. (1989). Basic statistics: Tales of distributions (4th ed.).Belmont, CA: Brooks/Cole Publishing Company.Steinglass, P., Bennet, L. A., Wolin, S. J., & Reiss, D. (1987). The alcoholic family.New York: Basic Books.Steinglass, P., Tislenko, L., Reiss, D. (1985). Stabffity/instability in the alcoholicmarriage: The interrelationships between course of alcoholism, familyprocess, and marital outcome. Family Process, 2i 365-376.Sterba, R. (1934). The fate of the ego in analytic therapy. International Journal ofPsycho-Analysis, i, 117-123.Tichenor, V., & Hifi, C. E. (1989). A comparison of six measures of workingalliance. Psychotherapy, 2, 195-199.129Zetzel, E. (1956). Current concepts of transference. International Journal ofPsycho-Analysis. Z, 369-376.Zweben, A., Pearlinan, S., & Selina, L. (1988). A comparison of brief advice andconjoint therapy in the treatment of alcohol abuse: The results of the MaritalSystems study. British Journal of Addiction, , 899-916.Appendix ABinge Chronic Differentiation Scale130THE ALCOHOL RECOVERY PROJECTBINGE/CHRONIC DIFFERENTIATION FORM1. Which pattern best describes your drinking over the last year?[1] Drink heavily (6 or more drinks) every day[2] Drink moderately (3 to 5 drinks) daily[3] Drink a little (less than 3 drinks) every day[4] One day binges[5] Weekend/several day binges[6] Drink heavily (6 or more drinks) a couple of times a week[71 Drink moderately (3 to 5 drinks) a couple of times a week[8] Drink a little (less than 3 drinks) a couple of times aweek[9] Occasional drink but rarely get drunk[10] Never drink2. Which statement best describes your drinking habits over thelast year? (Circle A. B or C):A. A periodic, intermittent drinker (one who drinks heavily on abinge or drinking bout every so often, with periods of little orno drinking between binges) (Complete Section A only)B. A steady, regular drinker (one who continuously drinks more orless the same amount on a day-to-day basis)(Complete Section B only)C. Cannot say (Complete Sections A and B)A. SECTION FOR PERIODIC DRINKERS:1. About how many drinking bouts haveyou had in the last year?______bouts2. About how long does your averagedrinking bout usually last? Hours_____Days3. What is the longest boutyou have ever had?_____Hours Days4. On average, how much time goesby between drinking bouts?_____Days Weeks MonthsB. SECTION FOR STEADY DRINKERS:1. Are there any particular days ofthe week during which you drinkmore than any other days? No Yes> If yes. circle days: H T W Th F Sa SuI- r \u00E2\u0080\u0098Appendix BDyadic Adjustment ScaleSpariier, 1976131QUESTIONNAIRE * 131INSTRUCTIONS:Most persons have disagreements in their relationships.Please indicate below the approximate extent of agreement ordisagreement between you and your partner for each item on thefollowing list over the past four weeks. (Circle the number onthe answer sheet to indicate your response to each item).For the following questions, please use the scale below to rateyour responses:Almost AlmostAlways Always Occasionally Frequently Always AlwaysAgree Agree Disagree Disagree Disagree Disagree1 2 5 6Handling family financesMatters of recreationReligious mattersDemonstrations of affectionFriendsSex relationsConventionality (correct or proper behaviour)Philosophy of lifeWays of dealing with parents or in-lawsAims, goals, and things believed importantAmount of time spent togetherMaking major decisionsHousehold tasksLeisure time interests and activitiesCareer decisions1.2.3.4.5.6.7.8.9.10.11.12.13.14.15.DAS. 1131. bQUESTIONNAIRE *2Use the following scale for the next 7 questions:All the Most of More oftentime the time than not16. How often do you discuss or have you considered divorce,separation. or terminating your relationship?17. How often do you or your mate leave the house after a fight?18. In general. how often do you think that things between youand your partner are going well?19. Do you confide in your mate?20. Do you ever regret that you married (or lived together)?21. How often do you and your partner quarrel?22. How often do you and your mate \u00E2\u0080\u009Cget on each others\u00E2\u0080\u0099 nerves\u00E2\u0080\u009D?23. Do you kiss your mate.a. Every dayb. Almost every dayc. Occasionallyd. Rarelye. Never24. How many outside interests do you and your mate engage intogether?a. Allb. Mostc. Somed. Very fewe. NoneOccasionally Rarely Never1 2 3 4 5 6DAS.2131 cQUESTIONNAIRE * 2.Use the following scale for the next 4 questions:Less than Once or twice Once or twice Once a MoreNever once a month a month a week day often1 2 3 4 5 6How often would you say the following events occur between youand your mate?25. Have a stimulating exchange of ideas26. Laugh together27. Calmly discuss something28. Work together on a projectThese are some things about which couples sometimes agree andsometimes disagree. Indicate if either item below causeddifferences of opinions or were problems in your relationshipduring the past few weeks. (Circle yes or no).29. Being too tired for sex30. Not showing loveFor the next question, please use the following scale:Extremely Fairly A Little Very ExtremelyUnhappy Unhappy Unhappy Happy Happy Happy Perfect1 2 3 4 5 6 7The scale represents different degrees of happiness in yourrelationship. The middle point. \u00E2\u0080\u009Chappy\u00E2\u0080\u009D, represents the degreeof happiness in most relationships.31. Please indicate on the answer sheet which number bestdescribes the degree of happiness, all things considered, ofyour relationship.DAS.3l31QUESTIONNAIRE 41 2.32. Which of the following statements best describes how youfeel about the future of your relationship?a. I want desperately for my relationship to succeed, andwould go to almost any length to see that it does.b. I want very much for my relationship to succeed, andwill do all I can to see that it does.c. I want very much for my relationship to succeed, andwill do my fair share to see that it does.d. It would be nice if my relationship succeeded, but Ican\u00E2\u0080\u0099t do much more than I am doing now to help itsucceed.e. It would be nice if it succeeded, but I refuse to doany more than I am doing now to keep the relationshipgoing.f. My relationship can never succeed, and there is no morethat I can do to keep the relationship going.DAS. 4Appendix CIndividual Therapeutic Affiance ScalePinsof and Catherall, 1986132QUESTIONNAIRE 1\u00E2\u0080\u009932oINSTRUCTIONS: The following statements refer to your feelings andthoughts about your therapist and your therapy right NOW. Eachstatement is followed by a seven point scale. Please rate the extentto which you agree or disagree with each statement AT THIS TIME.If you completely agree with the statement, circle number 1. If youcompletely disagree with the statement, circle number 7. Use thenumbers in-between to describe variations between the extremes.Completely Strongly Agree Neutral Disagree Strongly CompletelyAgree Agree Disagree Disagree1 2 3 4 5 6 7Please work quickly. We are interested in your FIRST impressions.Your ratings are CONFIDENTIAL. They will not be shown to yourtherapist and will only be used for the purposes of the project.Although some of the statements appear to be similar or identical,each statement is unique. PLEASE BE SURE TO RATE EACH STATEMENT.1. The therapist cares about me as a person.2. The therapist and I are not in agreement about the goals forthis therapy.3. I am satisfied with the therapy.4. The therapist lacks the skills and ability to help me with myimportant relationships.5. I trust the therapist.6. The therapist does not understand some of my importantrelationships.7. The therapist understands my goals in therapy.8. Some of the people who are important to me would not agree withthe therapist about the goals of this therapy.9. The people who are important to me would approve of the way mytherapy is being conducted.10. The therapist does not understand me.11. The therapist is helping me with my important relationships.12. The therapist does not understand some of the people who areimportant to me.TAS. I.QUESTIONNAIRE *13. The therapist cares about my important relationships.14. I do not feel accepted by the therapist.15. The therapist and I are in agreement about the way the therapyis being conducted.16. Some of the people who are important to me would distrust thetherapist.17. The therapist has the skills and ability to help me.18. I do not care about the therapist as a person.19. The people who are important to me would think that the therapyis helping me.20. The therapist is not helping me.21. The therapist understands the goals I have for my importantrelationships.22. The therapist does not appreciate how important some of myrelationships are to me.23. The people who are important to me would feel accepted by thetherapist.24. The therapist does not, agree with the goals I have for myimportant relationships.25. The therapist understands what the people who are important tome would want me to achieve in therapy.TAS. I . 2Appendix DCouples Therapeutic Affiance ScalePinsof and Catherall, 1986133QUESTIONNAIRE *INSTRUCTIONS: The following statements refer to your feelings andthoughts about your therapist and your therapy right NOW. Eachstatement is followed by a seven point scale. Please rate the extentto which you agree or disagree with each statement AT THIS TIME.If you completely agree with the statement, circle number 1. If youcompletely disagree with the statement, circle number 7. Use thenumbers in\u00E2\u0080\u0094between to describe variations between the extremes.Completely Strongly Agree Neutral Disagree Strongly CompletelyAgree Agree Disagree Disagree1 2 3 4 5 6 7Please work quickly. We are interested in your FIRST impressions.Your ratings are CONFIDENTIAL. They will not be shown to yourtherapist or partner and will only be used for the purposes of theproject. Although some of the statements appear to be similar oridentical, each statement is unique. PLEASE BE SURE TO RATE EACHSTATEMENT.1. The therapist cares about me as a person.2. The therapist and I are not in agreement about the goals forthis therapy.3. I trust the therapist.4. The therapist lacks the skills and ability to help my partnerand myself with our relationship.5. My partner feels accepted by the therapist.6. The therapist does not understand the relationship between mypartner and myself.7. The therapist understands my goals in therapy.8. The therapist and my partner are not in agreement about thegoals for this therapy.9. My partner -res about the therapist as a person.10. The therapist does not understand the goals that my partner andI have for ourselves as a couple in this therapy.11. My partner and the therapist are in agreement about the way thetherapy is being conducted.12. The therapist does not understand me.TAS. C. 1QUESTIONNAIRE *7- 1331z13. The therapist is helping my partner and me withourrelationship.14. I am not satisfied with the therapy.15. The therapist understands my partner\u00E2\u0080\u0099s goals for this therapy.16. I do not feel accepted by the therapist.17. The therapist and I are in agreement about the way the therapyis being conducted.18. The therapist is not helping me.19. The therapist is in agreement with goals that my partner and Ihave for ourselves as a couple in this therapy.20. The therapist does not care about my partner as a person.21. The therapist has the skills and ability to help me.22. The therapist is not helping my partner.23. My partner is satisfied with the therapy.24. I do not care about the therapist as a person.25. The therapist has the skills and ability to help my partner.26. My partner distrusts the therapist.27. The therapist cares about the relationship between my partnerand myself.28. The therapist does not understand my partner.29. The therapist does not appreciate how important the relationshipbetween my partner and myself is to me.TAS.C.2Appendix EFrequency Distributionfor Revised Therapeutic Alliance Scale134GregChart3dFrequencyDistribution.4772-0AllianceMidpointsdTAS-RandCTAS-R:(mean=117.42,skew=-0.128,kurtosis=-0.996)Page1135Appendix FThis study is one in a series of investigations resulting from a large-scaleresearch project entitled The Alcohol Recovery Project (TARP). Carried out overa period of five years, TARP has received funding from the British ColumbiaAlcohol and Drug Program (now part of the provincial Ministry of Health andformerly in the Ministry of Labor and Consumer Services) and from the BritishColumbia Health Research Foundation (Health Services Research Programme).Other assistance has been extended by the University of British Columbia and theHumanities and Social Sciences Research Services. These funds and other formsof assistance have enabled the completion of this study, as well as others resultingfrom TARP activities. This body of research has been conducted under thegeneral direction of the Principal Investigator, John D. Friesen, Ph.D., and coinvestigator Robert Conry, Ph.D."@en . "Thesis/Dissertation"@en . "1994-05"@en . "10.14288/1.0054049"@en . "eng"@en . "Counselling Psychology"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en . "Graduate"@en . "The role of the therapeutic allance in the treatment of alcoholic famlies"@en . "Text"@en . "http://hdl.handle.net/2429/4905"@en .