Journal of Counseling Psychology The Therapist, the Client, and the Real Relationship: An Actor–Partner Interdependence Analysis of Treatment Outcome Dennis M. Kivlighan, Jr.,, Charles J. Gelso, Stacie Ain, Ann M. Hummel, and Rayna D. Markin Online First Publication, March 24, 2014. http://dx.doi.org/10.1037/cou0000012

CITATION Kivlighan, D. M., Jr.,, Gelso, C. J., Ain, S., Hummel, A. M., & Markin, R. D. (2014, March 24). The Therapist, the Client, and the Real Relationship: An Actor–Partner Interdependence Analysis of Treatment Outcome. Journal of Counseling Psychology. Advance online publication. http://dx.doi.org/10.1037/cou0000012

Journal of Counseling Psychology 2014, Vol. 61, No. 1, 000

© 2014 American Psychological Association 0022-0167/14/$12.00 DOI: 10.1037/cou0000012

BRIEF REPORT

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

The Therapist, the Client, and the Real Relationship: An Actor–Partner Interdependence Analysis of Treatment Outcome Dennis M. Kivlighan, Jr., and Charles J. Gelso

Stacie Ain

University of Maryland

George Washington University

Ann M. Hummel

Rayna D. Markin

Fleet Mental Health, Naval Base San Diego

Villanova University

The relationship between treatment progress (as rated by both clients and therapists) and real relationship (also rated by both clients and therapists) was decomposed into between-therapist and within-therapist (between-client) effects and analyzed using the actor–partner interdependence model. We reanalyzed a subset of the data, 12 therapists and 32 clients, from Gelso et al.’s (2012) study of brief, theoretically diverse outpatient treatment. Consistent with and extending previous research, clients whose therapists provided higher average levels of client-perceived real relationship across the clients treated by a given therapist had better progress ratings from both themselves and their therapists. Within each therapist’s caseload, differences between clients in client- or therapist-rated real relationship were unrelated to either client- or therapist-rated outcome. Clients whose therapists provided higher average levels of therapistperceived real relationship, across the clients treated by the therapist, had worse progress ratings from the therapists. The results provide additional evidence for the importance of between-therapist differences in therapeutic relationship qualities, both client and therapist rated. Keywords: between-therapist and within-therapist analysis, real relationship, actor–partner interdependence model

largely attributable to differences between therapists in the average quality of the therapeutic relationship that they establish. The prototypical study of the therapy relationship has clients, who are nested within therapists, rate the therapy relationship. For example, three clients of therapist X may rate her on a relationship measure as follows: Client A ⫽ 5, Client B ⫽ 6, and Client C ⫽ 7. These ratings can be decomposed into a between-therapist component, which is the average relationship rating across the three clients (M ⫽ 6), and a between-client component, which is the deviation of each client’s relationship rating from this therapist mean (⫺1 for Client A, 0 for Client B, and 1 for Client C). When clients’ ratings of the therapeutic alliance were decomposed, Baldwin et al. (2007) found that between-therapist variability, but not between-client variability, was related to pretreatment to posttreatment symptom change. Zuroff et al. (2010) also found that between-therapist variability in clients’ ratings of the Rogerian conditions of positive regard, empathy, and genuineness was related to longitudinal decreases in overall maladjustment and depressive vulnerability. However, Zuroff et al. (2010) found that between-client variability in clients’ ratings of the Rogerian conditions had a small relationship to decreasing levels of overall maladjustment. The results of these two studies show the importance of differences between therapists in determining treatment outcome. Marcus, Kashy, and Baldwin (2009) and Marcus, Kashy, Wintersteen, and Diamond (2011) used a different approach to variance partitioning. One-with-many designs were used in these stud-

Gelso and Carter (1994) defined the therapy relationship as therapists’ and clients’ mutual feelings and attitudes and how these are expressed. The working alliance, empathy, goal consensus, collaboration, countertransference management, positive regard, and congruence are relationship components related to outcome (Norcross & Lambert, 2011). Meta-analyses using these relationship variables and therapy outcome offer support for the importance of the therapy relationship (see Norcross & Lambert, 2011). Accordingly, Baldwin, Wampold, and Imel (2007) and Zuroff, Kelly, Leybman, Blatt, and Wampold (2010) attempted to better understand the association between the therapy relationship and outcome. They argue that differences in clients’ outcomes are

Dennis M. Kivlighan, Jr., Department of Counseling, Higher Education and Special Education, University of Maryland; Charles J. Gelso, Department of Psychology, University of Maryland; Stacie Ain, University Counseling Center, George Washington University; Ann M. Hummel, Fleet Mental Health, Naval Base San Diego; Rayna D. Markin, Department of Psychology, Villanova University. Correspondence concerning this article should be addressed to Dennis M. Kivlighan, Jr., Department of Counseling, Higher Education and Special Education, University of Maryland, 3214 Benjamin Building, College Park, MD 20742. E-mail: [email protected] 1

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

2

KIVLIGHAN, GELSO, AIN, HUMMEL, AND MARKIN

ies to partition the variance in working alliance ratings into (a) therapist–partner effects (parallel to the between-therapist component), (b) therapist–perceiver effects (do the clients whose therapists report a stronger working alliance report a better outcome— not examined in the previous studies), (c) client relationship effects (parallel to the between-client component), and (d) therapist relationship effects (therapists report an especially strong working alliance with one of their clients, compared with their other clients—also not examined in Baldwin et al., 2007; Zuroff et al., 2010). In addition, they decomposed client outcome into betweentherapist and between-client components. Marcus et al. (2009) found better outcomes for clients who saw their therapists as establishing stronger alliances, compared with the clients of other therapists. Like Zuroff et al. (2010) for overall maladjustment, Marcus et al. (2009) found that clients who reported a better alliance with their therapists, compared with the alliance ratings of their therapist’s other clients, had better outcomes than the outcomes of their therapist’s other clients. Neither therapist perceiver nor therapist relationship variance derived from the therapists’ ratings of the working alliance was related to client outcome. Marcus et al. (2011) also used both clients’ and therapists’ ratings of the working alliance and examined three different outcome variables (the Substance Problem Index [Dennis, 1999], self-reported days of cannabis use, and number of treatment sessions). Contrary to the findings reported above, between-therapist variance in client-rated working alliance was not related to either the self-reported days of cannabis use or the number of sessions that the client attended. However, between-therapist variance was related to decreases on the Substance Problem Index. The only effect for between-client variance pertained to reduced days of cannabis use. There was only one significant effect for between-therapist variance in therapist-rated working alliance; therapists who saw themselves as establishing stronger alliances on average had clients who reported lower scores on the Substance Problem Index. There was also one significant between-client effect: Therapists who reported a better alliance with some their clients, compared with the alliance ratings for their other clients, had clients who attended more sessions relative to their other clients. There are important consistencies and inconsistencies across these four studies. These consistencies and inconsistencies are discussed below in relation to the present study’s hypotheses. First, however, we discuss how the present research both replicates and extends the previous research. The majority of studies examined the working alliance; therefore, it was important to examine a different relationship variable. The real relationship is one component in the tripartite model of the therapeutic relationship (Gelso & Carter, 1985, 1994; Gelso & Hayes, 1998). The other relationship components are the working alliance and the transference– countertransference configuration. Gelso (2011) defined the real relationship as “the personal relationship existing between two or more persons as reflected in the degree to which each is genuine with the other and perceives the other in ways that befit the other” (pp. 12–13). The real relationship is genuine to the extent that the therapist and client relate in an authentic manner. Realism involves the therapist and client perceiving the other accurately. Research on the real relationship shows that the real relationship measured at differing points in treatment, and from the different perspectives (client and thera-

pist), is related to session-level and treatment-level outcomes (Gelso et al., 2012; Lo Coco, Gullo, Prestano, & Gelso, 2011; Marmarosh et al., 2009). The real relationship, therefore, provides a good vehicle to extend the research using variance decomposition. Recently, Gelso et al. (2012) used the actor–partner interdependence model (APIM; Kashy & Kenny, 2000; Kenny, 1996) to examine the real relationship and treatment outcome from the perspectives of both the client and the therapist. They found that client ratings of the real relationship after the first session, first quarter of treatment, and after all sessions combined related to both their own (the clients’) and their therapists’ ratings of outcome. Therapists’ ratings of the real relationship at these time points, however, did not relate to either their own or their clients’ ratings of outcome. In the present study, we reanalyze the data from Gelso et al. (2012) by decomposing both clients’ and therapists’ ratings of the real relationship into between-therapist and between-client components. This data set is ideal for extending the previous studies because the researchers collected clients’ and therapists’ ratings of the real relationship and both parties’ ratings of treatment progress.

Clients’ Perspectives on the Real Relationship Baldwin et al. (2007) and Zuroff et al. (2010) used client ratings of the working alliance or Rogerian conditions, respectively. Marcus et al. (2009) and Marcus et al. (2011) used both clients’ and therapist’s ratings of the working alliance. Clients’ ratings of outcome were used in all four studies; however, Marcus et al. (2009) also examined two behavioral measures: days of cannabis use and sessions attended. Across the four studies, the client-rated outcomes were consistently related to the between-therapist component of the client-rated relationship. However, the two behavioral measures were not related to the between-therapist component of the client-rated relationship component. As noted by Marcus et al. (2009), one reason that client relationship ratings are related to client outcome ratings is likely the common rater confound (Heppner, Wampold, & Kivlighan, 2008). One way to address the common rater confound is to use therapist outcome (progress) ratings when the client relationship variable is decomposed. On the basis of the consistent results across the four studies and Gelso et al.’s (2012) findings, we hypothesized that both clients and therapists would have higher outcome ratings when clients’ rated the real relationship with their therapists as stronger, on average (a significant and positive between-therapist effect for client-rated real relationship). This is the first study in which between-therapists effects on therapist-rated treatment progress are examined. Across the four studies reviewed above, when the clients rated the therapeutic relationship, there have been inconsistent results when between-client effects were examined. Across the seven outcomes examined, only three of the outcomes showed a relationship to the between-client aspect of the relationship. When a between-client effect was found for a particular outcome, it was consistently smaller than the between-therapist effect. Given these mixed results, we developed the following research question: What is the nature of the relation between the between-client component of the real relationship, as assessed by clients, and progress, as assessed by both clients and the therapists.

BETWEEN-THERAPIST AND BETWEEN-CLIENT DIFFERENCE

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Therapists’ Perspectives on the Real Relationship Only Marcus et al. (2009) and Marcus et al. (2011) examined therapists’ ratings of the relationship. The between-therapist component of therapists’ real relationship ratings assesses the degree to which therapists rate their real relationship compared with how other therapists rate their real relationships. In Marcus et al. (2009), there was no relationship between therapists who were high on the between-therapist component of the therapeutic alliance and client outcome. In Marcus et al. (2011), the between-therapist component of therapist-rated working alliance was only related to clients’ ratings on the Substance Problem Index. Given the mostly negative results when the between-therapist component of therapist-rated relationship was assessed, we expected that the between-therapist component of therapist-rated real relationship would be unrelated to progress assessed by both clients and therapists. Across the Marcus et al. (2009) and Marcus et al. (2011) studies, there were two significant between-client effects when therapist rated the working alliance. This suggests that because therapists saw multiple clients in the Marcus studies, and because they have seen multiple clients in their careers, they theoretically have the experience to recognize when one of their clients has a heightened capacity to form a real relationship or when they have a particularly strong real relationship with a specific client. Therefore, we hypothesized that there would be a significant positive relationship between the between-client component of the real relationship, as assessed by the therapist, and outcome, as assessed by both the clients and the therapists.

Method Participants We reanalyzed a subsample of the data reported in Gelso et al. (2012); specifically, we used data from therapists who saw at least two clients. This subsample consisted of 32 clients seen by 12 therapists. The demographic for clients was six men and 26 women; five graduate students, four seniors, seven juniors, six sophomores, six freshman, and four unspecified; European American (29); Black/ African American (one); Asian/Asian American (one), and unspecified (one). The 12 therapists were either staff psychologists or therapists (n ⫽ 6) with 1–31 years of experience, or advanced graduate students in counseling psychology (n ⫽ 6). There were two male and 10 female therapists. The racial/ethnic breakdown was 11 European Americans and one Black/African American.

Measures Real Relationship Inventory–Client version (RRI-C; Kelley, Gelso, Fuertes, Marmarosh, & Lanier, 2010). The RRI-C is a 24-item measure that assesses the strength of the real relationship from the client’s perspective. Items are rated on a Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree). Examples of items are “my therapist and I were able to be authentic in our relationship” and “my therapist’s perceptions of me were accurate.” RRI-C reliability previous estimates ranged from the mid .80s to the mid .90s (e.g., Fuertes et al., 2007; Kelley et al., 2010; Marmarosh et al., 2009). The measure’s convergent, discriminant, and construct validity have been supported by its relation in theoretically expected ways to the

3

observing ego, therapist congruence, an earlier measure of the real relationship (Eugster & Wampold, 1996), working alliance, the tendency to hide one’s true feelings and behave in ways that are expected from others, social desirability, client attachment, and clients’ attachments to their therapists (Fuertes et al., 2007; Kelley et al., 2010; Marmarosh et al., 2009). Predictive validity has been supported by the relation of the RRI-C measured early in treatment with therapy outcome (Lo Coco et al., 2011; Marmarosh et al., 2009). In the present sample, the internal consistency estimate was ␣ ⫽ .98. The Real Relationship Inventory–Therapist Form (RRI-T; Gelso et al., 2005). The RRI-T contains 24 items that assess the strength of the real relationship from the therapist’s perspective. Items are rated on a Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Examples of items include “My client shares with me the most vulnerable parts of him/herself”; “My client feels liking for the ‘real me.’” Coefficient alphas have ranged from .80 to .90 in various samples (Fuertes et al., 2007; Gelso et al., 2005; Marmarosh et al., 2009). Convergent and discriminant validity have been supported by the relation of the RRI-T to ratings of working alliance, session quality, negative transference, client insight, social desirability, and therapist and client attachment (Fuertes et al., 2007; Gelso et al., 2005; Marmarosh et al., 2009). Predictive validity of the RRI-T is supported by its relation with outcome (Marmarosh et al., 2009). For the present sample, the internal consistency estimate was ␣ ⫽ .94. Counseling Outcome Measure (COM; Gelso & Johnson, 1983). The COM consists of four items on which therapists or clients rate clients’ progress. Ratings are made on a 7-point scale, with the anchors being much worse (1), moderately worse (2), slightly worse (3), about the same (4), slightly improved (5), moderately improved (6), and much improved (7). Both therapist and client ratings have tended to be quite high (e.g., Gelso & Johnson, 1983; Gelso et al., 1997). Despite this skew, the COM has demonstrated excellent psychometric features. Internal consistency and retest reliability have been demonstrated to be satisfactory (Fuertes et al., 2007; Gelso, Kivlighan, Wine, Jones, & Friedman, 1997; Gelso & Johnson, 1983; Tracey, 1987). Coefficient alphas have been generally in the high .80s to low .90s. For the present study, coefficient alphas for therapists and clients were found to be .90 and .92, respectively. Patton, Kivlighan, and Multon (1995) found that the COM was significantly correlated with pre- to postcounseling changes in interpersonal problems and the symptoms. In addition, the COM was similar to a symptom scale in identifying high- and low-change clients in Tracey (1987). Construct validity as an outcome measure has been supported by the theoretically expected associations with predictors such as the real relationship (Fuertes et al., 2007; Gelso et al., 2012), elements of time-limited psychotherapy (Gelso & Johnson, 1983), and the interaction of client transference and insight (Gelso et al., 1997).

Procedure Therapists at the Counseling and Student Development Center of James Madison University (JMU) and at the University of Maryland’s (UM) Counseling Center were invited to participate in the research. Each therapist was told to invite the next clients he or she saw to participate in the research. Clients and therapists completed a demographic questionnaire after the first session and the real realtionship measure and additional measures not examined in this study after each session. After the final session, clients and

KIVLIGHAN, GELSO, AIN, HUMMEL, AND MARKIN

4

therapists completed the session measures and the COM. Further details can be found in Gelso et al. (2012). Therapy at both settings followed a brief therapy model. At UM, a 12-session duration limit existed, and therapy occurred on a weekly basis. At JMU, therapists were advised to work within a 12-session framework, but they had the option of continuing beyond this limit with clients of their choosing. To be as comparable as possible with the previous studies, we analyzed client and therapist ratings of the real relationship from the third treatment session.

␤2 ⫽ ␥20 ⫹ ⌫21 * (Between ⫺ Therapist Therapist ⫺ Rated RR) ⫹ ⌫22 * (Between ⫺ Therapist Client ⫺ Rated RR) ⫹ u2 ␤3 ⫽ ␥30 ⫹ u3 ␤4 ⫽ ␥40 ⫹ u4 ␤5 ⫽ ␥50 ⫹ u5

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Data Analysis

␤6 ⫽ ␥60 ⫹ u6

The data in this study are doubly nested, with clients and therapists nested within the therapeutic dyad and dyads nested within therapists. Therefore, we used a modified version of the APIM (Kashy & Kenny, 2000; Kenny, 1996). The APIM models the client’s and therapist’s outcome rating as a function of the between-therapist variance in both clients’ and therapists’ ratings of the real relationship and as a function of the between-clienttherapist variance in both clients’ and therapists’ ratings of the real relationship. The specific model examined was: Level 1 model: Y ⫽ ␤1 * (Therapist Outcome) ⫹ ␤2 * (Client Outcome) ⫹ ␤3 * (Between ⫺ Client Therapist ⫺ Rated RRActor Effect) ⫹ ␤4 * (Between ⫺ Client Client ⫺ Rated RR Actor Effect)

.

Results The average progress ratings for therapist and client, respectively, were 6.27 (SD ⫽ .62) and 6.02 (SD ⫽ .68). The average between-therapist real relationship ratings were 4.04 (SD ⫽ .45) and 2.96 (SD ⫽ .07), respectively. The average between-client real relationship ratings were 0.00 (SD ⫽ .26) and 0.00 (SD ⫽ 0.05), respectively. The correlation between therapist-rated and clientrated outcome was significant (r ⫽ .38, p ⫽ .022). Therapist-rated and client-rated real relationship were not significantly correlated (r ⫽ .26, p ⫽ .105).

Client Ratings of the Real Relationship

⫹ ␤5 * (Between ⫺ Client Therapist ⫺ Rated RR Partner Effect) ⫹ ␤6 * (Between ⫺ Client Client ⫺ Rated RR Partner Effect) ⫹ R Level 2 model: ␤1 ⫽ ␥10 ⫹ ⌫11 * (Between ⫺ Therapist therapist ⫺ rated RR) ⫹ ⌫12 * (Between ⫺ Therapist Client ⫺ Rated RR) ⫹ u1

Our first hypothesis stated that both clients and therapists would have higher progress ratings when clients rated the real relationship with their therapists as stronger, on average (a significant and positive between-therapist effect for client-rated real relationship). In our first hypothesis, it is important to test whether a client’s rating of progress is related to her or his between-therapist real relationship while controlling for her or his therapist’s betweentherapist ratings of the real relationship. The results of the APIM model are displayed in Table 1. As hypothesized, there was a significant between-therapist effect for client-rated real relationship and client-rated progress (␥22 ⫽ 0.90,

Table 1 Actor–Partner Interdependence Model for Between-Therapist and Between-Client Real Relationship, as Rated by Therapists and Clients, and Outcome, as Rated by Therapists and Clients Effect

Coefficient

SE

t(11)

p

Therapist outcome intercept ␥00 Between-therapist therapist-rated RR ␥01 Between-therapist client-rated RR ␥02 Client outcome intercept ␥10 Between-therapist therapist-rated RR ␥11 Between-therapist client-rated RR ␥12 Between-client therapist–actor RR ␥13 Between-client client–partner RR ␥14 Between-client therapist–partner RR Between-client client–partner RR

5.97 ⫺3.36 0.64 6.89 ⫺2.43 0.90 ⫺2.68 0.20 ⫺0.97 ⫺0.14

0.10 1.48 0.23 0.07 1.21 0.16 1.50 0.13 1.66 0.17

62.73 ⫺2.27 2.76 84.74 ⫺1.63 5.79 ⫺1.79 1.62 ⫺0.58 ⫺0.82

.000 .037 .014 .000 .060 .000 .091 .123 .567 .425

Effect size 5.09 0.97 3.68 1.36 4.06 0.30 1.47 0.21

Note. An actor effect is the relationship between a participant’s own RR rating and her or his own outcome rating. A partner effect is the relationship between a participant’s own RR rating and the other participant’s outcome rating. RR ⫽ real relationship.

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

BETWEEN-THERAPIST AND BETWEEN-CLIENT DIFFERENCE

SE ⫽ 0.16), t(16) ⫽ 5.79, p ⬍ .000. Therapists whose clients provided higher, on average, ratings of the real relationship had clients who rated a better therapeutic progress. Dividing the gamma by the standard deviation of the dependent variable provides an estimate of the effect size. The effect size for the relationship between between-therapist client-rated real relationship and client-rated progress is 1.36, which is considered a large effect. Also, as hypothesized, there was a significant between-therapist effect for client-rated real relationship and therapist-rated progress (␥12 ⫽ 0.64, SE ⫽ 0.23), t(16) ⫽ 2.76, p ⬍ .014. Therapists whose clients provide higher, on average, ratings of the real relationship rated their clients as having better therapeutic progress. The effect size for this relationship was 0.97, which is also considered a large effect. We also tested the relationship between the between-client component of the real relationship and progress, as assessed by both the clients and the therapists. The between-client effect for client-rated real relationship and client-rated progress was small and not significant (␥40 ⫽ 0.20, SE ⫽ 0.13), t(16) ⫽ 1.62, p ⬍ .123. Within clients seen by each therapist, differences between clients in their perceptions of the real relationship were not related to client-rated progress. The effect size for the relationship between between-client client-rated real relationship and client-rated progress is 0.30, which is considered a small effect. Also, the between-client effect for client-rated real relationship and therapist-rated progress was also small and not significant (␥60 ⫽ ⫺0.14, SE ⫽ 0.17), t(16) ⫽ ⫺0.82, p ⬍ .425. Within clients seen by each therapist, differences between clients in their perceptions of the real relationship were not related to therapistrated progress. The effect size for the relationship between between-client client-rated real relationship and therapist-rated progress is 0.21, also a small effect.

Therapists’ Ratings of the Real Relationship There was a nonsignificant negative relation for betweentherapist effect for therapist-rated real relationship and client-rated progress (␥21 ⫽ ⫺2.43, SE ⫽ 1.21), t(16) ⫽ ⫺1.63, p ⬍ .060. The effect size for the relation between between-therapist therapistrated real relationship and client-rated progress is 3.68, which is considered a very large effect. It is important to put effect size into perspective. According to the estimated slope for every one-unit change in between-therapist ratings of the real relationship, there is a 2.43 decrease in client-rated progress. This is a very large change in progress rating. However, the standard deviation for betweentherapist ratings of the real relationship is only 0.07. A one-unit change in between-therapist ratings of the real relationship would equate to a change of 14 standard deviations. In other words, the likelihood of obtaining a one-unit change in between-therapist ratings of the real relationship is extremely small. Therefore, the large effect size is a function of the very narrow band of therapist ratings of the real relationship. There was a significant negative between-therapist effect for therapist-rated real relationship and therapist-rated progress (␥11 ⫽ ⫺3.36, SE ⫽ 1.48), t(16) ⫽ ⫺2.27, p ⬍ .037. Therapists who saw themselves as having high average levels of the real relationship across the clients in their caseloads rated their clients’ progress lower. The effect size for the relationship between between-therapist therapist-rated real relationship and client-rated

5

progress is 5.09, which is a large effect. As stated above, however, this effect must be interpreted in light of the very small deviation in between-therapist ratings of the real relationship. Our hypotheses regarding therapists’ ratings of the real relationship stated that there would be a significant positive relationship between the between-client component of the real relationship and progress, as assessed by both the clients and the therapists. These hypotheses were not supported. The between-client effect for therapist-rated real relationship and client-rated progress was not significant (␥50 ⫽ ⫺0.97, SE ⫽ 1.66), t(16) ⫽ ⫺0.58, p ⬍ .567. The effect size for the relationship between between-client therapist-rated real relationship and client-rated progress is 1.47, which is considered a large effect. The between-client effect for therapist-rated real relationship and therapist-rated progress was also not significant (␥30 ⫽ ⫺2.68, SE ⫽ 1.50), t(16) ⫽ ⫺1.79, p ⬍ .091. The effect size for the relationship between between-client client-rated real relationship and therapist-rated progress is 4.06, which is also considered a large effect. According to the estimated slope for every one-unit change in between-client ratings of the real relationship, there is a 2.68 decrease in therapist-rated progress. However, the standard deviation in between-client ratings of the real relationship is only 0.05. A one-unit change in betweentherapist ratings of the real relationship would equate to 20 standard deviations. Therefore, the large effect size for between-client ratings of the real relationship is a function of the very narrow band of therapist ratings of the real relationship.

Discussion In our study, we attempted to replicate and extend Baldwin et al.’s (2007), Zuroff et al.’s (2010), and Marcus et al.’s (2009) findings that the relation between the relationship variables and outcome is best explained by between-therapist differences. As hypothesized, we found that between-therapist differences in client-rated real relationship were related to progress, as rated by both the client and the therapist. The findings for therapist-rated progress are particularly important because the relationship cannot be attributed to the common rater confound (Heppner et al., 2008). Thus, as perceived by clients, some therapists form stronger real relationships than other therapists. Those therapists with better client-perceived real relationships have clients with better progress (as rated by both clients and therapists). These significant relationships were obtained when controlling for the other partner’s perspective, which provides a more stringent test than in Baldwin et al. (2007) or Zuroff et al. (2010). We also found that between-therapist variability in therapistrated real relationship was negatively related to progress, as rated by the therapist. Between-client variability in client-rated or therapist-rated real relationship was unrelated to progress, as rated by the client or therapist. Between-therapist differences in client-rated relationship variables appears to be an important and robust finding in spite of differences in (a) the aspect of the therapeutic relationship examined, (b) therapist’s level of experience, (c) type of treatment settings, and (d) type of research design and type of dependent variable. An important focus for future research is to identify variables that predict the therapists who form better therapeutic relationships with their clients. Anderson, Ogles, Patterson, Lambert, and Vermeersch (2009) showed that a performance test of

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

6

KIVLIGHAN, GELSO, AIN, HUMMEL, AND MARKIN

therapists’ facilitative conditions could predict therapists who, on average, had better treatment progress as rated by their clients. In addition, Laska, Smith, Wislocki, Minami, and Wampold (2013) showed that an experienced supervisor, unaware of client outcomes, could reliably identify more effective therapists. Effective therapists were viewed as able to effectively deal with client avoidance, having a flexible interpersonal style, and able to foster a strong therapeutic alliance. If more and less effective therapists can be reliably identified, is it possible to help the less effective therapist become more effective? This is an important question for future research. Are the studies in which a significant relationship between the between-therapist relationship component, as rated by the client, and outcome is found tapping either (a) a general therapist ability of therapists to form a therapeutic relationship or (b) a specific relationship ability? This question can only be answered by research in which multiple aspects of the therapeutic relationship in the same study are examined simultaneously. This study shows that who rates the real relationship is important. Between-therapist variance in therapist-rated real relationship had the opposite effect when compared with betweentherapist variance in client-rated real relationship. The therapists who rated themselves as generally forming stronger real relationships compared with their peers tended to have poorer therapistrated progress. We suspect that these poorer performing therapists were the least able to accurately assess their ability to form real relationships. Kruger and Dunning (1999) showed that poorer performers, across a variety of domains, consistently overestimated how they compared with their peers. It would be interesting to see whether feedback interventions or supervision could help therapists to more accurately calibrate their real relationship assessments to their performance. Across all of the studies, there are only a couple of instances in which between-client relationship effects were related to treatment outcome or progress. These consistent null findings serve to emphasize the importance of the therapist’s compared with the client’s contribution to the relationship. Like the Marcus et al. (2009) and Marcus et al. (2011) studies, we examined in the present study between-therapist and between-client variance using the therapist’s rating of the relationship. We thought that therapists would be in a better position to recognize when one of their clients had a heightened capacity to form a real relationship or when they had a particularly strong real relationship with a specific client. If this recognition did occur, it was not related to either client-rated or therapist-rated progress. None of the studies examining betweentherapist and between-client effects have examined possible crosslevel interactions involving between-therapist and between-client effects. It may be that between-client differences in the therapeutic relationship are important for therapists who are especially good or especially bad at forming these relationships. Two different strategies have been used to partition the variance in client and therapist relationship ratings. In Baldwin et al. (2007), Zuroff et al. (2010), and the present study, average relationship ratings across the therapists’ caseloads were the between-therapist component, and client deviations from these averages defined the between-client component. Marcus et al. (2009) and Marcus et al. (2011) used the one with many (OWM) method to partition the variance in client and therapist relationship ratings into therapist perceiver, client partner, client relationship, and therapist relation-

ship components. One of the advantages of the OWM analysis is the ability to examine reciprocity in client and therapist relationship ratings. An issue with the OWM method is that the variance in the outcome measure must be partitioned into between-therapist and between-client components. For example, if between-therapist variance in a client outcome rating is 10%, then the therapist perceiver (between-therapist) component of the relationship can only predict a maximum of 10% of the total variance in outcome. In the method used in this study, the between-therapist component can predict the total variance in progress. The results of the present study should be placed in the context of the study’s strengths and limitations. The present study had fewer therapists (N ⫽ 12) than did the Baldwin et al. (2007), Zuroff et al. (2010), and Marcus et al. (2009) studies, but a similar number of therapist as in the Marcus et al. (2011) study (N ⫽ 14). The smaller number of therapists in the present study decreased the power to detect between-therapist and between-client effects. The lower power may have contributed to the lack of findings for some of the between-client effects. For therapist ratings, there was little variability in their ratings of either the between-therapists or the betweenclients components of the real relationship. Perhaps the therapist version of the RRI is not as sensitive to relationship differences as the client version. However, in other studies in which the therapist version was used, the standard deviations for therapist ratings were substantially (approximately 5 times) larger than in the present study (e.g., Marmarosh et al., 2009). Future research should address why therapists may not differentiate between themselves or their clients in terms of the real relationship. We examined relatively brief therapies provided by therapists from various theoretical orientations; therefore, the results may not generalize to longer term treatment or therapies of one particular theoretical orientation. In terms of type of treatment delivered, Zuroff et al. (2010) showed that between-therapist effects were consistent across cognitive-behavior therapy, interpersonal therapy. Future research, however, may wish to examine the association between therapist- and client-rated real relationship, on the one hand, and observer rated progress, on the other. What can psychotherapy researchers and clinicians learn from studies decomposing relationship variables into between-therapist and between-client components? First and foremost, the therapist’s ability to form high-quality therapeutic relationships is central in the outcome of therapy. Therefore, it is important for researchers to be able to identify therapists who have this capacity and to create training that is focused on the therapist capacity to develop therapeutic relationships. In addition, from this study’s findings, there appears to be a group of therapists who overestimate their ability to form therapeutic relationships. These therapists may lack the metacognitive capacity or emotional insight to accurately assess their abilities. Therefore, research is needed on ways to increase this type of insight and metacognitive ability. The relation between the client’s perception of the real relationship and the client’s self-rated progress adds to existing research on the relation between the therapeutic relationship and progress. However, further research would be needed to determine whether the therapeutic relationship is only facilitative of outcomes such as symptom reduction or quality-of-life improvement, or whether the therapy relationship is a strong enough indicator of client change and psychological health to be considered a dimension of client progress.

BETWEEN-THERAPIST AND BETWEEN-CLIENT DIFFERENCE

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

References Anderson, T., Ogles, B. M., Patterson, C. L., Lambert, M. J., & Vermeersch, D. A. (2009). Therapist effects: Facilitative interpersonal skills as a predictor of therapist success. Journal of Clinical Psychology, 65, 755–768. doi:10.1002/jclp.20583 Baldwin, S. A., Wampold, B. E., & Imel, Z. E. (2007). Untangling the alliance– outcome correlation: Exploring the relative importance of therapist and patient variability in the alliance. Journal of Consulting and Clinical Psychology, 75, 842– 852. doi:10.1037/0022-006X.75.6.842 Dennis, M. L. (1999). Global Appraisal of Individual Needs (GAIN): Administration guide for the GAIN and related measures (Version 1299). Bloomington, IL: Chestnut Health Systems. Eugster, S. L., & Wampold, B. E. (1996). Systematic effects of participant role on evaluation of the psychotherapy session. Journal of Consulting and Clinical Psychology, 64, 1020 –1028. doi:10.1037/0022-006X.64.5 .1020 Fuertes, J. N., Mislowack, A., Brown, S., Gur-Arie, S., Wilkinson, S., & Gelso, C. J. (2007). Correlates of the real relationship in psychotherapy: A study of dyads. Psychotherapy Research, 17, 423– 430. doi:10.1080/ 10503300600789189 Gelso, C. J. (2011). The real relationship in psychotherapy: The hidden foundation of change. Washington, DC: American Psychological Association. doi:10.1037/12349-000 Gelso, C. J., & Carter, J. A. (1985). The relationship in counseling and psychotherapy: Components, consequences, and theoretical antecedents. The Counseling Psychologist, 13, 155–243. doi:10.1177/ 0011000085132001 Gelso, C. J., & Carter, J. A. (1994). Components of the psychotherapy relationship: Their interaction and unfolding during treatment. Journal of Counseling Psychology, 41, 296 –306. doi:10.1037/0022-0167.41.3 .296 Gelso, C. J., & Hayes, J. A. (1998). The psychotherapy relationship: Theory, research, and practice. New York, NY: Wiley. Gelso, C. J., & Johnson, D. H. (1983). Explorations in time-limited counseling and psychotherapy. New York, NY: Teachers College Press. Gelso, C. J., Kelley, F. A., Fuertes, J. N., Marmarosh, C., Holmes, S. E., Costa, C., & Hancock, G. R. (2005). Measuring the real relationship in psychotherapy: Initial validation of the Therapist Form. Journal of Counseling Psychology, 52, 640 – 649. doi:10.1037/0022-0167.52.4.640 Gelso, C. J., Kivlighan, D. M., Jr., Busa-Knepp, J., Spiegel, E. B., Ain, S., Hummel, A. M., . . . Markin, R. D. (2012). The unfolding of the real relationship and the outcome of brief psychotherapy. Journal of Counseling Psychology, 59, 495–506. doi:10.1037/a0029838 Gelso, C. J., Kivlighan, D. M., Jr., Wine, B., Jones, A., & Friedman, S. C. (1997). Transference, insight, and the course of time-limited therapy. Journal of Counseling Psychology, 44, 209 –217. doi:10.1037/00220167.44.2.209 Heppner, P. P., Wampold, B. E., & Kivlighan, D. M., Jr. (2008). Research design in counseling (3rd ed.). Belmont, CA: Thomson Brooks/Cole. Kashy, D. A., & Kenny, D. A. (2000). The analysis of data from dyads and

7

groups. In H. T. Reis, C. M. Judd, H. T. Reis, & C. M. Judd (Eds.), Handbook of research methods in social and personality psychology (pp. 451– 477). New York, NY: Cambridge University Press. Kelley, F. A., Gelso, C. J., Fuertes, J. N., Marmarosh, C., & Lanier, S. (2010). The Real Relationship Inventory: Development and psychometric investigation of the client form. Psychotherapy: Theory, Research, Practice, Training, 47, 540 –553. doi:10.1037/a0022082 Kenny, D. A. (1996). Models of non-independence in dyadic research. Journal of Social and Personal Relationships, 13, 279 –294. doi: 10.1177/0265407596132007 Kruger, J., & Dunning, D. (1999). Unskilled and unaware of it: How difficulties in recognizing one’s own incompetence lead to inflated self-assessments. Journal of Personality and Social Psychology, 77, 1121–1134. doi:10.1037/0022-3514.77.6.1121 Laska, K. M., Smith, T. L., Wislocki, A. P., Minami, T., & Wampold, B. E. (2013). Uniformity of evidence-based treatments in practice? Therapist effects in the delivery of cognitive processing therapy for PTSD. Journal of Counseling Psychology, 60, 31– 41. doi:10.1037/a0031294 Lo Coco, G., Gullo, S., Prestano, C., & Gelso, C. (2011). Relation of the real relationship and the working alliance to the outcome of brief psychotherapy. Psychotherapy, 48, 359 –367. doi:10.1037/a0022426 Marcus, D. K., Kashy, D. A., & Baldwin, S. A. (2009). Studying psychotherapy using the one-with-many design: The therapeutic alliance as an exemplar. Journal of Counseling Psychology, 56, 537–548. doi:10.1037/ a0017291 Marcus, D. K., Kashy, D. A., Wintersteen, M. B., & Diamond, G. S. (2011). The therapeutic alliance in adolescent substance abuse treatment: A one-with-many analysis. Journal of Counseling Psychology, 58, 449 – 455. doi:10.1037/a0023196 Marmarosh, C. L., Gelso, C. J., Markin, R. D., Majors, R., Mallery, C., & Choi, J. (2009). The real relationship in psychotherapy: Relationships to adult attachment, working alliance, negative transference, and therapy outcome. Journal of Counseling Psychology, 56, 337–350. doi:10.1037/ a0015169 Norcross, J. C., & Lambert, M. J. (2011). Psychotherapy relationships that work II. Psychotherapy, 48, 4 – 8. doi:10.1037/a0022180 Patton, M. J., Kivlighan, D. M., Jr., & Multon, K. D. (1995). Missouri psychoanalytic counseling research project–first findings. Unpublished manuscript, University of Missouri—Columbia. Tracey, T. J. (1987). Stage differences in the stage dependencies of topic initiation and topic following behavior. Journal of Counseling Psychology, 34, 123–131. doi:10.1037/0022-0167.34.2.123 Zuroff, D. C., Kelly, A. C., Leybman, M. J., Blatt, S. J., & Wampold, B. E. (2010). Between-therapist and within-therapist differences in the quality of the therapeutic relationship: Effects on maladjustment and self-critical perfectionism. Journal of Clinical Psychology, 66, 681– 697.

Received August 26, 2013 Revision received November 7, 2013 Accepted November 22, 2013 䡲

The therapist, the client, and the real relationship: an actor-partner interdependence analysis of treatment outcome.

The relationship between treatment progress (as rated by both clients and therapists) and real relationship (also rated by both clients and therapists...
129KB Sizes 1 Downloads 3 Views