A meta-analytic examination of client-therapist perspectives of the working alliance

Using 53 studies, comprising 52 separate data sets, published in refereed journals from 1985 through 2006, the authors conducted meta-analyses of the correlation and mean difference between client therapist alliance ratings. Client and therapist alliance ratings were moderately correlated (/r .36, SD .00); clients’ ratings were higher than ratings by their therapists (/d .63, SD .42). Client disturbance was a significant moderator of client therapist alliance rating discrepancies; clients with milder disturbances or with substance abuse problems tended to have larger rating discrepancies with their therapists than clients with more severe disturbances or moderate disturbance without substance abuse. Several researchers have investigated client and therapist agreement on what is happening in therapy and found that convergent perspectives are associated with better outcomes (Cummings, Hallberg, Slemon, & Martin, 1992; Cummings, Martin, Hallberg, & Slemon, 1992; Kivlighan & Arthur, 2000; Reis & Brown, 1999). The client therapist working alliance (also known as therapeutic alliance, working relationship, and helping alliance) is common to all psychotherapeutic interventions (Gelso & Carter, 1985). Although theorists differ somewhat in their conceptualizations of the alliance (Bordin, 1979; Hausner, 2000; Hentschel, 2005), most emphasize client therapist collaboration and consensus on the goals and tasks of therapy (Horvath & Bedi, 2002) as well an emotional bond between client and therapist (Martin, Garske, & Davis, 2000). Despite the importance of client therapist collaboration, Fitzpatrick, Iwakabe, and Stalikas (2005) indicated that therapist and client perspectives of the alliance do not always agree. Authors of numerous studies (Bachelor, 1991; Bachelor & Salame, 2000; Cecero, Fenton, Nich, Frankforter, & Carroll, 2001; Fitzpatrick et al., 2005; Hatcher, Barends, Hansell, & Gutfreund, 1995; Hilsenroth, Peters, & Ackerman, 2004; Mallinckrodt, 1991, 1993; Mallinckrodt & Nelson, 1991; Ogrodniczuk, Piper, Joyce, & McCallum, 2000; Tichenor & Hill, 1989) have observed that clients and therapists appear to view the alliance differently, with clients generally giving higher alliance ratings. Some studies (Casey, Oei, & Newcombe, 2005; Kivlighan & Shaughnessy, 1995; Mallinckrodt & Nelson, 1991) also have reported relatively low correlations between client and therapist alliance ratings. Several factors appear to influence the alliance, and some or all may relate to client therapist divergence in perspective. It is our goal to use meta-analyses to examine client therapist alliance ratings as well as factors that influence the alliance that might serve as possible moderators of these rating. One of these factors is client disturbance. In a summary of 11 studies, Horvath (1991), as cited in Constantino, Castonguay, & Schut, 2002) reported that clients who have difficulty maintaining social relationships were more likely to have difficulty forming a working alliance. More severely disturbed clients generally have relationship problems, and several studies (Gaston, Thompson, Gallager, Cournoyer, & Gagnon, 1998; Gunderson, Najavits, Leonhard, Sullivan, & Sabo, 1997; Hersoug, Hoglend, Monsen, & Havik, 2001; Lingiardi, Filippucci, & Baiocco, 2005; Zuroff et al., 2000) indicated that more severe client disturbance is related to lower alliances ratings. In contrast to the often reported pattern of client alliance ratings being higher than ratings by therapists, severely disturbed clients in studies by Gehrs and Goering (1994) and Correspondence concerning this article should be addressed to Georgiana Shick Tryon, The Graduate Center (CUNY), PhD Program in Educational Psychology, 365 Fifth Avenue, New York, NY 10016. E-mail: [email protected] Psychotherapy Research, November 2007; 17(6): 629 642 ISSN 1050-3307 print/ISSN 1468-4381 online # 2007 Society for Psychotherapy Research DOI: 10.1080/10503300701320611 D ow nl oa de d B y: [T ry on , G eo rg ia na S hi ck ] A t: 16 :1 2 17 O ct ob er 2 00 7 Gunderson et al. (1997) frequently rated the alliance lower than did their therapists. Others (Joyce & Piper, 1998; Paivio & Bahr, 1998), however, have not found a relationship between client disturbance and alliance ratings. Within the larger category of degree of client disturbance, client diagnosis could be a factor that influences alliance ratings. Most client participants in alliance-rating studies, however, have several diagnoses that usually are listed by study authors, but generally alliance data are not presented or analyzed according to diagnoses. Some studies, however, use more homogeneous client samples. For example, several studies have used clients who were being treated for drug or alcohol problems (Barber et al., 1999; Calsyn, Klinkenberg, Morse, & Lemming, 2006; Cecero et al., 2001; Connors, Carroll, DiClemente, Longabaugh, & Donovan, 1997; Fenton, Cecero, Nich, Frankforter, & Carroll, 2001; Luborsky et al., 1996; Meier & Donmall, 2006; Meier, Donmall, Barrowclough, McElduff, & Heller, 2005; Meier, Donmall, McElduff, Barrowclough, & Heller, 2006; Petry & Bickel, 1999). Some of these authors have commented about finding large alliance rating discrepancies, with clients with substance abuse problems viewing the alliance much more positively than their therapists. Thus, some specific client diagnoses may impact alliance ratings. Another factor that may impact client therapist perspectives on the alliance is the experience level of the therapist. Experienced therapists have had both more opportunities to practice psychotherapy and a wider exposure to different types of clients than trainees, so their perceptions of the alliance may differ from that of trainees, and clients may also rate them differently than they do trainees. Hersoug et al. (2001) and Mallinckrodt and Nelson (1991) found that alliance ratings by both clients and therapists differed depending on therapist experience. Dunkle and Friedlander (1996), however, did not find such a relationship. Length of therapy is another factor that may moderate client therapist alliance rating differences. It may be that rating convergence is greater in longer term therapy because it covers a longer period of time, during which client and therapist could presumably develop similar perspectives. There have been alliance studies comparing longer and shorter term therapies (e.g., Stiles, Agnew-Davis, Hardy, Barkham, & Shapiro, 1998), but none have compared client therapist alliance perspective convergence for shorter relative to longer therapies. Differences in client therapist alliance ratings may also depend on the scales used to assess the alliance. Several alliance-rating instruments have both client and therapist versions. These alliance measures are based on various theoretical conceptualizations of the alliance and generally have been found to correlate highly with each other (Cecero et al., 2001; Fenton et al., 2001; Hatcher & Barends, 1996; Stiles et al., 2002: Tichenor & Hill, 1989), confirming the common elements of the alliance. Although the various alliance instruments are similar, they are not identical, and it is possible that client therapist convergence could be influenced by the alliance measure used. Finally, the type of therapeutic treatment that clients receive could moderate client therapist alliance ratings. The alliance is an important component of all types of treatment, and meta-analyses have shown that the relationship between alliance and treatment outcome is not moderated by type of treatment (Horvath & Bedi, 2002; Martin et al., 2000). Although the alliance is common to all therapies, different theoretical orientations emphasize different aspects of the alliance (Raue, Castonguay, & Goldfried, 1993), and there is evidence that the theoretical orientation of the alliance rater may influence alliance ratings (Raue, Putterman, Goldfried, & Wolitzky, 1995). Thus, the type of treatment may impact therapists’ alliance ratings and play a role in client therapist alliance rating differences. Although convergence of client and therapist perspectives on what is happening in psychotherapy is desirable (Kivlighan & Arthur, 2000), the literature that we reviewed suggests that, in regard to the alliance, client therapist perspective divergence may be the rule rather than the exception. To what extent do clients’ and therapists’ perceptions of the alliance converge? Are client therapist alliance perspectives moderated by factors such as client disturbance, therapist experience, length of therapy, alliance instrument, and treatment type? Over the past 21 years, several studies have assessed the alliance from both client and therapist perspectives. The aim of this study is to use meta-analyses to investigate the relationship between client and therapist alliance ratings by assessing (a) the correlation between client and therapist ratings of the alliance and (b) the mean differences in client therapist alliance ratings along with their possible moderating factors.