The Power of the Therapeutic Relationship

Amer. J. Orthopsychiat. 61(3), July 1991

THE POWER OF THE THERAPEUTIC RELATIONSHIP

Elsa Marziali, Ph.D., and Leslie Alexander, Ph.D.

A review of the psychotherapy research literature of the last decade shows that considerable advances of clinical significance have been made toward defining and measuring components of the treatment relationship. The relevance of the therapeutic alliance for predicting outcome in diverse models of treatment is emphasized, and the implications of the findings for clinical training, practice, and research are discussed.

istorically psychotherapists have rec- tice the alliance evolves through an integra- H ognized that the client-therapist relation of the contextual and interventive tionship provides the context for all treat- aspects of the client-therapist interactions; ment processes. While definitions of the consequently, the alliance is viewed as play- nature and function of the therapeutic rela- ing a central role in determining the out- tionship have evolved and varied over time, come of treatment. the importance of the relationship in the The need for empirical validation of prac- treatment encounter has remained virtually tice has been articulated by both practition- unchallenged. What has varied is the cen- ers and clinical investigators (Ivanoff, trality of specific factors of the relationship Blythe, & Briar, 1987; Luborsky, 1987). A in defining the parameters of therapy. From series of meta-analytic studies have shown a psychodynamic perspective, the thera- that a variety of treatment models are ef- pist's role in developing and managing the fective (Lambert, Shapiro, &Bergin, 1986; treatment relationship is considered to pro- Smith & Glass, 1977; Videka-Sherman, vide both the context and the mutative agent 1988). The studies reviewed employed through which change occurs; that is, the group designs that explored psychody- enhanced understanding of the treatment re- namic, behavioral, cognitive-behavioral, lationship is applied to understanding mal- and problem-centered treatment methods. adaptive aspects of the client's troubled re- The consistent conclusion drawn from the lationships. In contrast, from a behavioral, reviews is that all treatments are effective, problem-solving orientation, a friendly re- regardless of their theoretical orientation or lationship develops in tandem with the strat- technique. One major gap in these studies egies used to achieve the goals of treat- is the paucity of attention paid to process ment. In this case, the therapist-client variables (Videka-Sherman, 1988). That is, relationship is viewed as a necessary con- which factors within a treatment model ex- text for the technical interventions. In prac- plain the obtained outcomes? Which fac-

Submitted to the Journal in October 1990. The authors are at: Faculty of Social Work, University of Toronto, Ont. (Marziali); and Graduate School of Social Work and Social Research, Bryn Mawr College, Bryn Mawr, Pa. (Alexander).

© 1991 American Orthopsychiatnc Association, Inc 383 384 THERAPEUTIC RELATIONSHIP tors are common to several treatment ap- influenced most subsequent attempts to de- proaches? Which factors are unique to any scribe the qualities of the therapeutic rela- given treatment model? tionship. This article examines process variables Bordin's (1979) writings about the treat- that tap important dimensions of the thera- ment relationship provide the clearest de- pist-client relationship. Recent studies in the scriptions of three key dimensions that best research literature on psychotherapy have portray the therapist's and client's respec- found that considerable advances have been tive contributions to the evolution of the made toward defining and measuring com- therapeutic alliance. A productive therapeu- ponents of the treatment relationship. A va- tic relationship includes 1) the client's and riety of measures of the therapeutic rela- therapist's agreement on the goals of ther- tionship have been developed and tested in apy, 2) the client's and therapist's agree- treatment-outcome studies (Barrett-Len- ment on the tasks needed to achieve the nard, 1962; Gomes-Schwartz, 1978; Hart- agreed-on goals, and 3) the development of ley & Strupp, 1983). Reviewed are studies an interpersonal bond. Bordin's formula- that clearly show the power of the thera- tion of the component parts of a working peutic relationship in predicting the out- alliance has considerably influenced the de- come of psychological treatments. The im- velopment of systems for measuring the plications of these findings for practice and strength and direction of the alliance and clinical training are discussed. their effects on the outcome of psychotherapy. THEORETICAL BACKGROUND REVIEW OF STUDIES Most studies of the quality and effects of the treatment relationship are based on psy- The Alliance in Individual Psychotherapy chodynamic formulations about the nature Considerable advances have been made of client-therapist interactions. Much of the in developing measures of the treatment re- measurement technology includes variables lationship, especially in individual psycho- that were defined clinically by Zetzel (1956) therapy. The measurement technology has and Greenson (1965). Zetzel used the terms included observers', clients', and thera- "working alliance" and "therapeutic alli- pists' ratings of the treatment alliance. ance" to describe the treatment relation- One of the earliest measures of the ther- ship. She believed that, early in the therapeutic relationship, the Relationship In- apy, the client would project onto the ventory (RI), was developed by Barrett- therapist wishes that originated from the cli- Lennard (1962). Items for the RI were ent's primary relationships. According to derived from Rogers' (1957) concepts of Zetzel, the alliance would be forged if the the necessary conditions of therapy. The RI therapist used technically supportive re- consists of 16 items distributed across 4 sponses that were sensitive to these devel- dimensions: positive regard, empathic un- opmentally linked projections. Greenson dis- derstanding, unconditionality of regard, and tinguished among "transference," the "real congruence. The RI has been used exten- relationship," and the "working alliance." sively in treatment-outcome studies, and the The client's realistic reactions to the thera- consistent finding is that clients' percep- pist (real relationship) were to be differen- tions of the relationship, as measured on tiated from fantasy distortions of the rela- the RI, are related to change following a tionship (transference). Greenson also course of treatment (Gurman, 1977). The proposed that the working alliance repre- scales have been used to assess 1) clients' sented the collaborative efforts of the client perceptions of relationships longitudinally, and therapist to advance insight and change. 2) family relationships, and 3) child and This tripartite definition of the alliance has adult relationships (Barrett-Lennard, 1986). MARZIALI AND ALEXANDER 385

Developed by Orlinsky and Howard ratings of the client's involvement. Since (1975, 1986), the Therapy Session Report similar predictions could not be made from (TSR) was designed to capture clients' ex- ratings of this dimension in either the first periences of psychotherapy. It focuses on or second session, it was concluded that the four aspects of the therapeutic experience: therapists may have exerted a marked in- dialogue, exchange, feelings, and relation- fluence on the developing alliance between ship. The TRS is completed following a the first and third sessions for this change to treatment session. Orlinsky and Howard occur. In a study of brief treatment, the (1986) used responses to the questionnaire Vanderbilt group (Moras & Strupp, 1982) to characterize the interpersonal content of showed that the client's capacity for inter- the treatment relationship. They concluded personal relating could be estimated at the that the form of relatedness that occurs in time of assessment from the quality and effective treatment is one of mutual affir- duration of relationships established with mation, mutual receptivity, and sensitive col- family members and friends. Interpersonal laboration. In a later study (Saunders, relating capacity was shown to be signifi- Howard, & Orlinsky, 1989), items from the cantly associated with the quality of the al- TSR were selected to develop the Thera- liance developed with the therapist. peutic Bond Scale, which includes three di- Investigators associated with the Perm mensions: the working alliance, empathic Psychotherapy Project (Luborsky, Crits- resonance, and mutual affirmation. Analy- Christoph, Mintz, & Auerbach, 1988; Lu- ses of the psychometric properties of the borsky et al., 1980) further advanced the new scale showed that the three subscales technology for measuring the qualities of were associated with the overall quality of the treatment relationship. They developed the session and with the outcome at termi- the Penn Helping Alliance Counting Signs nation. Method (Luborsky, 1976) to quantify the The Vanderbilt psychotherapy research concept of the helping alliance. The results group developed and tested a series of mea- of their work showed that two broad types sures to assess various domains of the psy- of helping alliances could be identified. In chotherapeutic process. The Vanderbilt Psy- Type I alliances, the client perceives the chotherapy Process Scale (VPPS) was therapist as carrying the major responsibil- developed, tested, and modified by Strupp ity for advancing the helpful components of and colleagues (Strupp, Hartley, & Black- the therapy; in Type II alliances, the client wood, 1974; Gomes-Schwartz, 1978; O'Ma- perceives the treatment as a collaborative lley, Suh, & Strupp, 1983). Using audio- process during which the client works with tapes of selected therapy sessions, clinical the therapist to achieve the goals of treat- raters judged the presence and intensity of ment. The results of analyses that com- the VPPS dimensions. A factor analysis of pared a group of ten more-improved clients items in the scale yielded three subscales of with a group of ten less-improved clients process dimensions: Client Involvement, Ex- selected from the Penn Psychotherapy ploratory Processes, and Therapist-Offered Project (A'=73) showed that the more- Relationship. Analyses of associations be- improved group had a higher frequency of tween these subscales and posttreatment ef- Type II alliance "signs" than did the less- fects showed that Client Involvement in ther- improved group (Alexander & Luborsky, apy was positively associated with most 1986; Luborsky, Crits-Christoph, Mintz, & measures of outcome (Gomes-Schwartz, Auerbach, 1988). The Penn group devel- 1978). A later study (O'Malley, Suh, & oped a client and therapist self-report form Strupp, 1983) demonstrated that by the third of the Penn Alliance Scales mat was tested session of therapy, the eventual outcome in a treatment-comparison trial with drug- could be predicted on the basis of the VPPS dependent clients (Luborsky, McLellan, 386 THERAPEUTIC RELATIONSHIP

Woody, O'Brien, & Auerbach, 1985). The rating perspectives (client, therapist, and client-completed version of the question- clinical judge), significant associations be- naire was the best predictor of outcome af- tween the quality of the alliance and out- ter seven months of treatment. Further- come were evident at the third treatment more, strength of the alliance-outcome session. correlations matched or exceeded those ob- Investigators at the Langley Porter Insti- tained in other studies that used the helping tute extended the analysis of the Marziali alliance measures. alliance measure and developed the Cali- Horvath and Greenberg (1986,1989) de- fornia Therapeutic Alliance Rating System veloped a measure of the alliance, the (C ALP AS) (Marmar, Horowitz, Weiss, & Working Alliance Inventory (WAI) that Marziali, 1986; Marmar, Weiss, & Gas- represents Bordin's (1979) three dimen- ton, 1989), which has been shown to have sions of the alliance (goals, tasks, and bond). similar properties to other alliance-measure- Client and therapist versions with parallel ment systems. For example, the results of items were tested and showed positive al- several studies showed positive associa- liance-outcome associations. Tracey and tions between alliance ratings and out- Kokotovic (1989) carried out a factor- come. Also, positive contributions to the analytic study of the WAI and found that alliance were associated with clients' pre- only 12 of the 36 items were most indica- treatment interpersonal functioning (Gas- tive of the three subscale factors and that ton, Marmar, Thompson, Gallagher, 1988; these items combined to represent one over- Marmar, Weiss, & Gaston, 1989). The first riding alliance factor. Although more scale version of the CALPAS was judge-rated analysis is needed, a 12-item questionnaire and consisted of five factor-derived scales: would be easier to use and could replace Therapist Understanding, Therapist Nega- alliance measures that are considerably tive Contribution, Patient Hostile Resis- longer or are expensive to use because rat- tance, Patient Commitment, and Patient ings are obtained from a trained panel of Working Capacity. In a later study, Gaston judges. (1990) developed and tested a patient-rated With few exceptions, investigators have version of CALPAS—CALPAS-P which used alliance measures that provide only contains four scales: Patient Commitment, one perspective of the alliance: the perspec- Patient Working Capacity, Therapist Un- tive of clinical judge, therapist, or client. In derstanding and Involvement, and Working an attempt to assess the associations among Strategy Consensus. New items were de- the three perspectives of the therapeutic al- veloped to reflect an expanded theoretical liance, Marziali, Marmar, and Krupnick perspective of the alliance, and some orig- (1980) and Marziali (1984) developed an inal CALPAS items were excluded. The alliance measure that could be completed final 24-item scale was completed by pa- by the client, the therapist, and impartial tients seen in private practice. Several of clinical judges. The dimensions of the mea- the CALPAS-P subscales were related to sure parallel those used by the Vanderbilt symptomatology and problems with inti- and Penn groups. The scales were tested in macy. All scales were associated with sat- a study of time-limited psychotherapy. The isfaction with therapy. results showed that there was considerable Some investigators have explored how correspondence among the three perspec- therapist variables enhance or impede the tives of the treatment relationship, but the development of a positive treatment alli- strongest agreement was between the cli- ance. Forman and Marmar (1985) explored ents' and therapists' perceptions of the al- therapists' behavior that was associated with liance. The study corroborated the findings improvements in initially poor alliances. of other investigators; that is, from all three They analyzed the therapies of six clients MARZIALI AND ALEXANDER 387 who had initially poor alliances and found tive working alliance and more abnormal that three of them went on to have im- interpersonal behavior by the patients. proved alliances and good outcomes, but that the alliances of the other three clients The Alliance and Other did not improve and these clients had poor Forms of Treatment outcomes. Therapists' actions that occurred While much of the research on the ef- more frequently in the improved cases in- fects of the therapeutic alliance on outcome cluded 1) addressing the client's defenses, has evolved from a psychodynamic perspec- 2) addressing the client's problematic feel- tive of individual treatment, behavior ther- ings toward the therapist, and 3) linking the apists have, in the past decade, shown in- client's defensive behavior with the con- creasing interest in examining the interaction flicts in the client's feelings toward the ther- between relationship variables and behav- apist. ioral techniques. In a review of the role In another study (Henry, Schact, & played by the quality of the therapeutic re- Strupp, 1986), detailed ratings of all utter- lationship in behavior therapy, Sweet (1984) ances by therapists and clients were gener- traced the development of interest in rela- ated from a circumplex model of interper- tionship issues by behaviorists. He cited sev- sonal behavior. These researchers used eral studies (Alexander, Barton, Schaino, Benjamin's Structural Analysis of Social Be- & Parsons, 1976; Ford, 1978) showing that, havior (Benjamin, 1974) to rate 36 types contrary to expectations, relationship vari- of interpersonal behavior on two inter- ables, rather than the behavioral techniques related circumplex surfaces. They com- used, contributed significantly to the vari- pared the good- and poor-outcome cases ance in outcome. Sweet concluded that the of four therapists. The following interper- power of the therapeutic relationship in pre- sonal process variables differentiated the dicting the outcome of treatment lies in the good versus the poor outcomes: high- mutual liking, trust, and respect of client change cases were associated with the ther- and therapist. A similar review focused on apist's higher levels of "helping and pro- the importance of the client-therapist rela- tecting" and "affirming and understanding" tionship in cognitive psychotherapy (Thomp- and lower levels of "blaming and belit- son, 1989). tling." Client's "disclosing and express- There is also a growing interest in the ing" was associated with high-change cases. role of the therapeutic relationship in fam- In contrast, low-change cases had higher ily, couples, and group treatment. Al- levels of clients' "walling off and avoid- though family therapists typically do not ing." refer to a therapeutic alliance, they have In a study of similar factors, Kiesler & explored this factor in such terms as "join- Watkins (1989) examined the relationship ing with the family" (Minuchin, 1974), between the therapeutic alliance and inter- "connecting" (Davatz, 1982), initiating a personal complementarity during the early "coalitionary process" with the family phase of therapy. Following the third treat- (Sluzki, 1975), and "engaging families" ment session, 36 pairs of patients and ther- (Solomon, 1977). Similarly, marital thera- apists completed the WAI (Horvath & pists are focusing more specifically on un- Greenberg, 1986) and Kiesler's (1984) derstanding the role of the treatment bond Checklists of Interpersonal Transactions. in marital therapy (Gurman, 1982; Rutan The results showed positive associations be- & Smith, 1985). Johnson and Greenberg tween the patients' and therapists' interper- (1989) explored the relevance of the thera- sonal complementarity and their respective peutic alliance in their model of emotion- perceptions of the alliance. Of significance ally focused marital therapy. were the correlations between a less posi- Until recently, there were no tools for 388 THERAPEUTIC RELATIONSHIP assessing the quality of a family's or cou- ment sessions. The implication is that by ple's relationships with a therapist. In the the third session, the therapist's success in mid-1980s, Pinsof and Catherall (1986) de- engaging the client in the treatment process veloped preliminary versions of alliance can be judged on the basis of the quality of measures that are applicable to couples and the alliance. Although good alliances are families. The scales' psychometric proper- consistently associated with good out- ties have been explored, and studies of the comes, the specific independent and com- effects of the alliance on outcome of family bined contributions to the alliance by both treatment are in progress. the client and the therapist are unknown. In group models of treatment, the devel- There is some evidence that client factors, opment of group "cohesion" is an essential such as high levels of pathology and the factor for ensuring the continuance of the poor quality of interpersonal relationships, group and for sustaining its work. Except affect the quality of the therapeutic alli- for studies of the function of cohesion in ance. In addition, negative attitudes and be- group treatment (Yalom, 1975), little effort havior by therapists are associated with poor has been devoted to examining the effects alliances and outcomes. Similarly, when of the treatment alliance in these models of therapists fail to explore clients' communi- treatment. Budman et al. (1989) studied both cations about negative elements in the ther- cohesion and alliance in time-limited group apeutic interaction, the alliance suffers. psychotherapy. As expected, cohesion and For each measurement system reviewed, alliance were strongly related and both pre- ratings of the client's alliance behavior were dicted improved self-esteem and reduced better predictors of outcome than were rat- symptomatology. Of particular interest was ings of therapist's alliance behavior. In other the finding that cohesion measured early in words, it was the quality of the client's par- a group session (in the first 30 minutes) was ticipation in the alliance that had the greater more related to outcome than was cohesion impact on outcome. It may be that, in their measured later in a group session. current form, alliance measurement systems are not as effective in capturing the DISCUSSION therapist's contribution to the alliance. The As this review of empirical studies of the exception was the method used by Henry, therapeutic alliance demonstrates, this key Schact, and Strupp (1986)—the Structural client-therapist factor plays an important role Analysis of Social Behavior—which deter- in determining the outcome of treatment. mined therapists' as well as clients' behav- The therapeutic alliance is a potent curative ior that distinguished cases with good out- factor in all forms of treatment. This point comes from cases with poor outcomes. was emphasized by Wolfe and Goldfried Therapists who showed high levels of help- (1988) when they stated that the alliance: ing, protecting, affirming, and understanding behavior and low levels of blaming and ... is probably the quintessential integrative variable belittling behavior were in the good- because its importance does not lie within specifica- tions of one school of thought. It is now commonly outcome group. accepted by most orientations that the therapeutic relationship is of essential importance to the conduct of IMPLICATIONS psychotherapy, (p. 449) Analyses of the nature and function of Inferences about early development of the the treatment relationship have important alliance were explored in several of the re- implications for clinical practice and reviewed studies. For example, some studies search. If the "baseline" meanings of the showed that the degree of the client's in- treatment relationship are established as volvement in treatment increased consider- early as the third to fifth treatment session, ably between the first and the third treat- then it would be prudent for the therapist to MARZIALI AND ALEXANDER 389 monitor carefully the client's cues about con- supervision, both during their training and fusions and anxieties in their interactions subsequently. Perhaps more emphasis during these early sessions. When these should be placed on helping the trainees cues are detected and explored, the thera- detect relationship cues that require atten- pist demonstrates to the client a willingness tion. In a recent review of the use of man- to understand the client's unique relation- uals for training psychotherapists, several ship qualities. In tandem, the therapist will authors stressed the need to focus on factors need to monitor her or his subjective reac- in the therapists' and clients' personalities tions to the client's style of communicat- because of their effects on alliance-building ing. These self-observations help the ther- and -mending behavior (Dobson & Shaw, apist to distinguish between subjective 1988; Guest & Beutler, 1988; Strupp, But- reactions that need to be contained and sub- ler, & Rosser, 1988). The present review jective reactions that inform him or her about underlines the fact that supervision and train- the client's style of relating. Alliance- ing can no longer ignore the strong evi- mending strategies are derived from under- dence showing the alliance as a significant standing both the therapist's and client's predictor of outcome, regardless of the ori- contributions to the alliance. The results of entation of therapy. Forman and Marmar's (1985) study sug- Many trainees may lose initial training gest that some therapists are unaware of or cases because they have failed to grasp the choose to ignore problems early in the treat- client's often subtle expressions of anxiety ment relationship. In contrast, they showed and confusion about the treatment relation- that the good-outcome therapists recog- ship. Alliance measures, such as the short nized cues in the interaction that alerted form of the WAI (Horvath & Greenberg, them to conflicts in the relationship and that 1986; Tracey & Kokotovic, 1989), could they explored with their clients the mean- be used as tools for monitoring the alliance ing of these problems. Similarly, for group during the supervision of training cases. In treatment, the therapist's activity in the first supervision, alliance ratings of a treatment third of each group session may consider- session could be reviewed in tandem with ably enhance cohesive-bonding interac- an audio recording of the session. The aim tions among group members (Budman et would be to examine client-therapist inter- al., 1989). actions that explain or support the subscale Many studies have shown that clients drop scores on the inventory. Confusions, differ- out of therapy within the first three to five ences in observations, misunderstandings, sessions (Garfield, 1986). Therapists may and distortions could be discussed. The su- conclude that these clients were either not pervisor could help the therapist translate suitable candidates for psychotherapy or observations of interpersonal factors into in- were not ready to commit themselves to terventions which would communicate to this arduous process. Studies of the thera- the client that despite the intensity of the peutic alliance challenge these assumptions shared emotions, troubled transactions such because of their consistent findings of the as those occurring in the treatment relation- strong associations between the quality of ship can be explored and understood. the alliance early in therapy and outcome. The helping professions have made sig- Thus, the problems with early attrition may nificant contributions to the development be fruitfully addressed by examining the of new treatment models with a wide range quality of the therapeutic alliance and the of client populations. Yet there is a paucity relative contributions to the alliance by both of studies of the elements of new treatment the client and the therapist. processes that explain outcome. Measure- Therapists learn about the quality and ment technology for detecting the qualities functions of the treatment relationship in of the treatment relationship has advanced 390 THERAPEUTIC RELATIONSHIP considerably, and most of the measures are Forman, S.A., & Mannar, C.R. (1985). Therapist well suited to most clinical settings. Simi- actions that address initially poor therapeutic alliances in psychotherapy. American Journal of Psy- larly, measures of various outcome factors chiatry, 142, 922-926. are readily available. Thus, it is now possi- Garfield, S.L. (1986). Research on client variables in ble to design within-group process-out- psychotherapy. In S.L. Garfield & A.E. Bergin, (Eds.), Handbook of psychotherapy and behavior come studies of treatment models that have change (pp. 213-256). New York: John Wiley. been well articulated clinically, but that have Gaston, L. (1990). Reliability and criterion-related not been tested empirically. Such studies validity of the California Psychotherapy Alliance Scales—Patient Version. Paper presented at the An- could be used to explore interactions be- nual Meeting of the Society for Psychotherapy Re- tween alliance factors and strategies for in- search, Wintergreen, Washington, DC. tervention that are specific to the model, Gaston, L., Marmar, C.R., Thompson, L.W., & Gal- lagher, D. (1988). Relation of patient pretreatment especially during the early phase of treat- characteristics to the therapeutic alliance in diverse ment. If it is true that the success of therapy psychotherapies. Journal of Consulting and Clini- is supported largely by the quality of the cal Psychology, 56, 483-489. Gomes-Schwartz, B. (1978). Effective ingredients in treatment relationship, it is imperative that psychotherapy: Prediction of outcome from process this factor be monitored and measured in all variables. Journal of Consulting and Clinical Psy- forms of treatment. chology, 46, 1023-1025. Greenson, R. (1965). The working alliance and the transference neuroses. Psychoanalytic Quarterly, 34, 155-181. REFERENCES Guest, P.D., & Beutler, L.E. (1988). The impact of Alexander, J.F., Barton, G., Schaino, R.S., & Par- psychotherapy supervision on therapist orientation sons, B.V. (1976). Systems behavioral interven- and values. Journal of Consulting and Clinical Psy- tions with families of delinquents: Therapist char- chology, 56, 653-658. acteristics, family behavior and outcome. Journal Gurman, A.S. (1977). The client's perception of the of Clinical and Consulting Psychology, 44, 656- therapeutic relationship. In A.S. Gurman & A.M. 664. Razin (Eds.), Effective psychotherapy: A handbook Alexander, L., & Luborsky, L. (1986). The Penn Help- of research (pp. 503-543). New York: Pergamon ing Alliance Scales. In L. Greenberg &. W. Pinsof Press. (Eds.), The psychotherapeutic process: A research Gurman, A.S. (1982). Creating a therapeutic alliance handbook(pp.'i25-366). New York: Guilford Press. m marital therapy. In A.S. Gurman (Ed.), Ques- Barrett-Lennard, G.T. (1962). Dimensions of thera- tions and answers in the practice of family therapy pist response as causal factors in therapeutic per- (Vol. 2). New York: Brunner/Mazel. sonality change. Psychological Monographs, 76 Hartley, D., & Strupp, H.H. (1983). The therapeutic (Whole No. 562). alliance: Its relationship to outcome in brief psy- Barrett-Lennard, G.T. (1986). The Relationship In- chotherapy. In J. Masling (Ed.), Empirical studies ventory now. In L.S. Greenberg & W. Pinsof (Eds.), of psychoanalytic theories (Vol. 1, pp. 1-38) Hill- The psychotherapeutic process: A research hand- stlale. NJ: Lawrence Erlbaum. book (pp. 439-476). New York: Guilford Press. Henry, W.P., Schact, T.E., & Strupp, H.H. (1986). Benjamin, L.S. (1974). Structural analysis of social Structural analysis of social behavior: Application behavior. Psychological Review, 81, 392-425. to a study of interpersonal processes in differential Bordin, E.S. (1979). The generalizability of the psy- psychotherapeutic outcome. Journal of Consulting choanalytic concept of the working alliance. Psy- and Clinical Psychology, 54, 27-31. chotherapy: Theory, Research and Practice, 16, Horvath, A., & Greenberg, L. (1986). The develop- 252-260. ment of the Working Alliance Inventory. In L. Green- Budman, S.H., Soldz, S., Demby, A., Feldstein, M., berg & W. Pinsof (Eds.), The psychotherapeutic Springer, T., & Davis, M.S. (1989). Cohesion, al- process: A research handbook (pp. 529-556). New liance and outcome in group psychotherapy. Psy- York: Guilford Press. chiatry, 52, 339-350. Horvath, A., & Greenberg, L. (1989). Development Davatz, U. (1982). Establishing a therapeutic alliance and validation of the Working Alliance Inventory. in family systems therapy. In A.S. Gurman (Ed.), Journal of Counseling Psychology, 36, 223-233. Questions and answers in the practice of family Ivanoff, A., Blythe, B.J., & Briar, S. (1987). The therapy (Vol. 2). New York: Brunner/Mazel. empirical clinical practice debate. Social Casework, Dobson, K.S., & Shaw, B.F. (1988). The use of man- 68, 290-298. uals in cognitive therapy: Experience and issues. Johnson, S.M., & Greenberg, L.S. (1989). The ther- Journal of Consulting and Clinical Psychology, 56, apeutic alliance in marital therapy. Journal of Cog- 673-680. nitive Psychology, 3, 97-110. Ford, J.D. (1978). Therapeutic relationship in behav- Kiesler, D.J. (1984). Checklist of Interpersonal Trans- ior therapy: An empirical analysis. Journal of Con- actions. Richmond, VA: Commonwealth Univer- sulting and Clinical Psychology, 46, 1302-1314. sity. MARZIALI AND ALEXANDER 391

Kiesler, D.J., & Watkins, L.M. (1989). Interpersonal Orlinsky, D., & Howard, K. (1975). Varieties ofpsy- complementarity and the therapeutic alliance: A chotherapeutic experience. New York: Teachers Col- study of relationship in psychotherapy. Psychother- lege Press. apy, 26, 183-194. Orlinsky, D., & Howard, K. (1986). The psycholog- Lambert, M., Shapiro, D., & Bergin, A. (1986). The ical interior of psychotherapy: Explorations with the effectiveness of psychotherapy. In S. Garfield & A. therapy session reports. In L. Greenberg & W. Pin- Bergin (Eds.), The handbook of psychotherapy and sof (Eds.), Thepsychotherapeuticprocess (pp. 477- behavioral change (pp. 157-212). New York: John 500). New York: Guilford Press. Wiley. Pinsof, W., & Catherall, D.R. (1986). The integrative Luborsky, L. (1976). Helping alliance in psychother- psychotherapy alliance: Family, couples and indi- apy: The groundwork for a study of their relation- vidual therapy. Journal of Marital and Family Ther- ship to outcome. In J. Cleghom (Ed.), Successful apy, 12, 15-22. psychotherapy (pp. 92-116). New York. Brunner/ Rogers, C. (1957). The necessary and sufficient con- Mazel. ditions of therapeutic personality change. Journal Luborsky, L. (1987). Research can now affect clinical of Consulting and Clinical Psychology, 21, 95-103. practice: A happy turnaround. Clinical Psycholo- Ruttan, S.J., & Smith, J.W. (i9«5). Building thera- gists, 40, 56-60. peutic relationships with couples. Psychotherapy, Luborsky, L., Crits-Christoph, J., Mintz, J., & Auer- 22, 194-200. bach, A. (1988). Who will benefit from psychother- Saunders, S.M., Howard, K.I., & Orlinsky, D.E. apy: Predicting therapeutic outcomes. New York: (1989). The therapeutic bond scales: Psychometric Basic Books. characteristics and relationship to treatment effec- Luborsky, L., McLellan, AT., Woody, G.E.. tiveness. Psychological Assessment, 1, 323-330. O'Brien, C.P., & Auerbach, A. (1985). Therapist Sluzki, C. (1975). The coalitionary process in initiat- success and it determinants. Archives of General ing family therapy. Family Process, 14, 67-77. Psychiatry, 42, 602-611. Smith, M., & Glass, C. (1977). MeU-analysis of psy- Luborsky, L., Mintz, J., Auerbach, A., Crits- chotherapy outcome studies. American Psycholo- Christoph, P., Bachrach, H., Todd, T., Johnson, gist, 32, 752-760. M., Cohen, M., & O'Brien, C. (1980). Predicting Solomon, M,A. (1977). The staging of family treat- the Outcome of Psychotherapy: Findings of the Perm ment: An approach to developing the therapeutic Psychotherapy Project. Archives of General Psychi- alliance. Journal of Marriage and Family Counsel- atry, 37, 471-481. ing, 3, 59-66. Marmar, C.R., Horowitz, M.J., Weiss, D.S., & Mar- Strupp, H.H., Butter, S.F., & Rosser, C.L. (1988). ziali, E. (1986). The development of the Therapeu- Training in psychodynamic therapy. Journal of tic Alliance Rating System. In L. Greenberg & W. Consulting and Clinical Psychology, 56, 689-695. Pinsof (Eds.), Thepsychotherapeulicprocess: A re- Strupp, H.H., Hartley, D., ABlackwood, G. (1974). search handbook (pp. 367-390). New York: Guil- The Vanderbilt Psychotherapy Process Seal*. Un- ford Press. published manuscript, Vanderbilt University, Nash- Marmar, C.R., Weiss, D.S. & Gaston, L. (1989). ville, TN. Toward the validation of the California therapeutic Sweet, A. (1984). The therapeutic relationship in be- alliance rating system. Psychological Assessment, havior therapy. Clinical Psychology Review, 4, 253— 1, 46-52. 272. Marziali, E. (J984). Three viewpoints on the thera- Thompson, J.K. (1989). Client-therapist relationship peutic alliance: Similarities, differences, and asso- and cognitive psychotherapy: Current status and fu- ciations with psychotherapy outcome. Journal of ture directions. Journal of Cognitive Psychother- Nervous and Mental Disease, 172, 417-423. apy, 3, 83. Marziali, E., Marmar, C, & Krupnick, J. (1980). Tracey, T.J., & Kokotovic, A.M. (1989). Factor struc- Therapeutic alliance scales: Their development and ture of the Working Alliance Inventory. Psycholog- relationship to psychotherapy outcome. American ical Assessment, 1, 207-210. Journal of Psychiatry, 138, 361-364. Videka-Sherman, L. (1988). Meta analysis of re- Minuchin, S. (1974). Families and family therapy. search in social work practice in mental health. So- Cambridge, MA.: Harvard University Press. cial Work, 33, 325-337. Moras, C, & Strupp, H.H. (1982). Pre-therapy in- Wolfe, B.E., & Goldfrwd, M.R. (1988). Research on terpersonal relations, client alliance and outcome in psychotherapy integration: Recommendations and brief psychotherapy. Archives of General Psychia- conclusions from the NIMH workshop. Journal of try, 39, 405-409. Consulting and Clinical Psychology, 56, 448-451. O'Malley, S., Suh, D., & Strupp, H. (1983). The YaJom, I.D. (1975). The theory and practice of group Vanderbilt Psychotherapy Process Scale: A report psychotherapy (2nd si.). New York: Basic Books. on scale development and a process-outcome study. Zetzel, E. (1956). Current concepts of transference. Journal of Consulting and Clinical Psychology, 51, International Journal of Psychoanalysis, 37, 369- 581-586. 376.

For reprints: Elsa Marziali, Ph.D , Faculty of Social Work, University of Toronto, 246 Blow Street, West, Toronto, Ont. M5S IAJ, Canada