• unlimited access with print and download
    $ 37 00
  • read full document, no print or download, expires after 72 hours
    $ 4 99
More info
Unlimited access including download and printing, plus availability for reading and annotating in your in your Udini library.
  • Access to this article in your Udini library for 72 hours from purchase.
  • The article will not be available for download or print.
  • Upgrade to the full version of this document at a reduced price.
  • Your trial access payment is credited when purchasing the full version.
Buy
Continue searching

Clinical prediction in group psychotherapy

Dissertation
Author: Christopher Chapman
Abstract:
Prior research in individual therapy has provided evidence that therapists are poor predictors of client outcome without the aid of objective measures and often misjudge clients' perceptions of the therapeutic relationship. The focus of the current research was to conduct a similar study in a group setting. Therapists from a university counseling center and a state psychiatric hospital were recruited to test their accuracy in predicting client outcome, quality of therapeutic relationship and their own use of empirically supported group interventions. Results indicated that therapists are poor predictors of all three, providing support for the implementation of measure-based feedback systems to inform therapists about key information that may affect the effectiveness of group psychotherapy. Keywords: group psychotherapy, clinical prediction, outcome, cohesion

TABLE OF CONTENTS Introduction................................................................................................................1 Literature Review.......................................................................................................4 Factors Influencing Outcom e.........................................................................4 A New, Comprehensive Model of the Group Relationship...........................7 Defining the Therapeutic Group Relationship: The Development of the Group Questionnaire..........................................................................10 Trends in Patient-Focused Research: A Response to “Evidence-Based Practice” ............................................................................................17 The ‘Dose-Response’ Model in Tracking Therapy Outcomes......................19 Benefits of Feedback......................................................................................20 Clinical and Empirical Prediction Methods...................................................21 Can Therapists Accurately Predict Client Status and Outcome? ..................22 Providing Client Progress Feedback to Clinicians: A Review of Prior Research.............................................................................................24 Implementation of Clinical Support Tools....................................................27 Feedback in Group Psychotherapy Research.................................................29 “Quality Assurance” and Feedback in Group Psychotherapy: The CORE- R Battery............................................................................................31 Outcome Measures.........................................................................................32 Group Process Measures................................................................................33

v Benefits and Drawbacks of the Use of Measures in Group Psychotherapy..35 Statem ent of the Problem...............................................................................37 Method.......................................................................................................................38 Participants.....................................................................................................38 Instruments.....................................................................................................39 Procedures......................................................................................................42 Results........................................................................................................................44 Predicting Outcome.......................................................................................44 Predicting Group Relationship.......................................................................45 Agreement on Group-Level Interventions.....................................................47 Discussion..................................................................................................................50 Limitations and Future Directions.................................................................53 References..................................................................................................................55 Appendix I: Measures Used.......................................................................................66 Appendix II: Prediction and Feedback Forms...........................................................71

vi LIST OF TABLES Table 1: GQ Descriptive Statistics for Outpatient (UCC) and Inpatient (USH) Samples...............................................................................................................46 Table 2: Prediction on GQ Domains- University Counseling Center Setting................48 Table 3: Prediction on GQ Domains- State Hospital Setting.........................................49

vii LIST OF FIGURES Figure 1: Five Interrelated Change Processes in Group Psychotherapy..........................4 Figure 2: Client Outcomes and Therapist Predictions...................................................45

1 Clinical Prediction in Group Psychotherapy As group psychotherapy has been established as an viable and cost-effective mode of treatment (Burlingame, Fuhriman, & Mosier, 2003; Kösters, Burlingame, Nachtigall, & Strauss, 2006), debate regarding which factors contribute to therapeutic gain by group members has spurred continuing research investigating the nature of these curative processes, their relationship to one another, and their impact on psychotherapeutic outcomes. Recently, a revised CORE battery (Burlingame, Strauss, & Hwang, 2008) was developed as a response to the growing pressure mounted on clinicians to use empirically based measures to track therapeutic factors and client outcomes in their groups. The aim of the CORE-R is to augment clinical judgment by providing information regarding member selection, therapeutic group processes, and member outcome (Burlingame et al., 2008). Similar empirically-based feedback systems have been implemented for clinicians working in individual psychotherapy both nationally and internationally (Barkham, Margison, Leach, Lucock, Mellor-Clark, & Evans, 2001; Kordy, Hannover, & Richard, 2001; Lambert, Hansen, & Finch, 2001; Lueger, Howard, Martinovich, Lutz, Anderson, & Grissom, 2001;). Research examining the benefits of providing such feedback to therapists in individual therapy has shown encouraging results in regards to improving client outcome, in particular for clients who are not responding to treatment (Berking, Orth, & Lutz 2006; Harmon et al., 2006; Hawkins, Lambert, Vermeersch, Slade, & Tuttle, 2004; Lambert et al., 2002; Whipple et al., 2003). However, research examining the impact of providing similar feedback in group psychotherapy remains scant.

2 While there are ma ny hypothesized benefits to implementing assessment tools in group, such as those included with the CORE-R, many of these benefits have yet to be fully ascertained. Some therapists may question whether or not such measures are necessary to aid their clinical judgment, and therefore rely on their training and expertise to alter their interventions when necessary in psychotherapy. Prior research in individual therapy has provided evidence that therapists are poor predictors of client outcome (Breslin, Sobell, Buchan, & Cunningham, 1997; Hannan, Lambert, Harmon, Nielsen, Smart, & Shimokawa, 2005) and often misjudge clients’ perceptions of the therapeutic relationship (Burns & Auerbach, 1996). Given the research demonstrating the difficulties in accurately predicting individual client outcome and strength of the therapeutic relationship, a similar study seems warranted in a group setting, as the therapist has even less information on each individual client when compared to dyadic treatments. Thus, we expected no better and perhaps worse predictive results in group treatment. Such a study would not only examine the importance of using outcome and clinical support feedback in group, but would also represent an important new direction in group research. Given that group has lagged behind individual therapy in implementing outcome feedback, the current study would be the first to examine group leader reaction to receiving such feedback. As such, the study could provide the necessary background to engage in a RCT similar to those conducted by researchers in individual therapy using OQ-45 outcome feedback along with CSTs.

3 The current study aimed to determine whether or not the use of process and outcome measures provided new and useful information to group therapists, or whether therapists’ can accurately predict these factors. What follows is a review of the literature detailing therapeutic processes related to outcome in group psychotherapy. In addition, past studies examining the ability of therapists to accurately predict client perceptions of the therapeutic relationship and client outcome are reviewed. The chapter also focuses on the benefits of providing empirically based feedback to therapists in individual psychotherapy. The review then explores options for group therapists seeking to implement similar feedback interventions in their work, highlighting the limited research conducted regarding the usefulness of these feedback systems in group psychotherapy. The purpose of the review is to discuss how the use of assessment tools may augment clinical judgment in group psychotherapy, and the need for further research and exploration regarding the proposed benefits of using such measures. Literature Review Factors Influencing Outcome In research by Burlingame, MacKenzie, and Strauss (2004), the authors propose a model detailing factors that explain treatment outcomes in group psychotherapy, including five interrelated factors: patient characteristics, structural factors, leader characteristics, formal change theory, and small-group processes (Figure 1). Patient characteristics, such as initial level of disturbance, personality, and interpersonal style have an established importance as predictors of group process (Kivlighan & Angelone, 1992; Piper, Joyce, Rosie, & Azim, 1994) as well as outcome (Burlingame et al., 2004;

4 Yalom & Lescsz, 1995). Structural factors refer to the establishment and maintenance of group norms, such as frequency of sessions, group settings, and the size of the group. Leader characteristics include aspects of the leader’s presence in group that impact the performance of the group, including therapist warmth, empathy, and openness. These are characteristics which have been associated with group process and outcome (Hurley & Rosenberg, 1990; Mcbride, 1995). Formal Change Theories represent the diverse therapeutic orientations, ranging from cognitive behavioral, to psycho-educational, to existential, and their impact on the psychotherapy group. These formal change theories are typically used as a framework in order to direct the therapeutic activity within the group. Figure 1. Five Interrelated Change Processes in Group Psychotherapy, from Burlingame,G. M., MacKenzie, K. R., & Strauss, B. (2004).

5 Sm all Group Processes encompass areas of the group relationship with known links to therapeutic outcome (Burlingame, Fuhriman, & Johnson, 2002; Burlingame et al., 2004; Yalom & Lescsz, 1995). While the study of small group processes, their definition, and their impact on outcome comprises a voluminous body of research, this brief review will focus on several of the most well-defined and empirically validated areas in terms of their impact on group outcome: Cohesion, Working Alliance, Group Climate, and Empathy (Burlingame, Johnson, & Fuhriman, 2002; Johnson, Burlingame, Olsen, Davies, & Gleave, 2005). While there may be considerable debate as to the precise definitions of these factors, as well as their relationships (and possible areas of overlap) with one another, each has been shown to relate to member progress and outcome. Cohesion, one of the most extensively researched small group processes (Yalom & Lescsz, 2005), can be defined as the sense of togetherness or ‘we-ness’ (Yalom & Lescsz, 2005) of the group, comparable to the ‘therapeutic alliance’ described in individual therapy. However, the definition of the construct has been debated and evolved considerably across the history of group process research. Burlingame and colleagues (2002) define cohesion as the therapeutic relationship in group. As such, cohesion consists of relationships on multiple levels: member-to-group, member-to- member, member-to-leader, leader-to-group, and leader-to-leader. It also describes the sense of collaborative bonding and alliance on interpersonal levels as well as intrapersonal (group-as-whole) levels. A number of studies have linked high levels of group cohesion with therapeutic outcome (McCallum, Piper, Ogrodniczuk, & Joyce, 2002; Stokes, 1983; Tschuschke &

6 Dies, 1994), with Tschuschke and Dies (1994) declaring a ‘linear and positive’ relationship between cohesion and outcome. W hile these findings appear to validate the notion that cohesion is a crucial aspect of the group therapeutic relationship, other research presents more mixed results in regards to cohesion and its impact on outcome (Gillaspy, Wright, Campbell, Stokes, & Adinoff, 2002; Kipnes, Piper, & Joyce, 2002; Marziali, Mumoe-Blum, & McCleary, 1997). However, much of this difficulty in establishing the relationship between cohesion and outcome may be related to the diverse definitions, operationalizations, and measures used to define cohesion (Dion, 2000). Working Alliance, or the shared responsibility between group members and the group leader in focusing on and working towards treatment goals (Johnson et al., 2005), has a well-established relationship with outcome. However, the construct shares a degree of definitional overlap with cohesion, and much like cohesion, has been defined and measured in diverse ways in the body of group process research (Johnson et al., 2005). In studies in which working alliance is defined as group member alliance with the therapist in working towards treatment goals, this aspect of the group relationship has been found to be predictive of positive outcomes (Brown & O’Leary, 2001; Sexton, 1993; Strauss & Burgmeier-Lohse, 1995). Group Climate refers to the presence of a therapeutic climate that facilitates the emotional expression and self-disclosure of group members, the responsiveness of other group members to these disclosures, and the shared meaning derived from such in-group experiences (Burlingame et al., 2002). In order to benefit from group, it is vital that an atmosphere of warmth and acceptance be provided to allow group members to express and explore the meanings of their behavior and emotional expressions in a cathartic

7 manner within the group (Hurley & Rosenberg, 1990; Mcbride, 1995). While there has been less variation in terms of defining the construct of Group Climate, one of the reasons for this is the preeminence of the Group Climate Questionnaire (GCQ; MacKenzie, 1983) in group process literature. The measure has been used in a variety of studies, with the ‘Engagement’ subscale found to positively predict outcome (Johnson et al., 2005; MacKenzie, Dies, Coche, Rutan, & Stone, 1987; Ogrodniczuk & Piper, 2003) while high scores on its conflict (measuring levels of hostility within the group) and avoidance subscales appear to be negatively correlated with outcomes (Johnson et al., 2005; Ogroduniczuk & Piper, 2003; Phipps & Zastowny, 1988). Empathy, or the client’s sense of being understood by the group, has gained almost universal acceptance in regards to its therapeutic value by adherents to a variety of psychotherapeutic orientations. While proponents of these orientations may define empathy in slightly different ways, it is consistently held as a curative therapeutic factor of critical importance (Burns, 1996; Mcbride, 1995; Trad, 1993). Empathy has been connected with positive outcome in a variety of studies; in one review, Orlinsky, Grawe, and Parks (1994), examined 115 studies examining the impact of empathy on outcome and found that in 72% of the studies, the clients’ perception of empathy was positively related to outcome. In the group literature, Karterud (1988) as well as Hurley and Rosenberg (1990) have linked empathetic group leader qualities with positive outcomes. A New, Comprehensive Model of the Group Relationship While the small group processes briefly reviewed above have all been linked to therapeutic outcome for group members, little research has been done investigating the relationship between these process variables, their possible overlap, and their impact on

8 one another across member-member, member-group, and member-leader levels. In an attempt to construct an empirical definition of the group relationship and assist in mitigating the considerable confusion caused by the diverse definitions of each group relationship construct, Johnson and colleagues (2005) proposed a new model of higher- order or latent constructs to describe the group relationship. In their model, four measures were used to operationalize the latent constructs: the Group Climate Questionnaire (GCQ; MacKenzie, 1983), the cohesion scale of the Therapeutic Factors Inventory (TFI; Lese & McNair-Semands, 2000), the Working Alliance Inventory (WAI; Horvath & Greenberg, 1989), and Empathy Scale (ES; Burns & Auerbach, 1996). The TFI and the GCQ were used to target perceptions of member-group relationships, while the WAI and the ES corresponded to member-member and member-leader relationships. These measures were administered to 662 participants from 11 different counseling centers and personal growth groups at the American Group Psychotherapy Association (AGPA). Employing Exploratory Factor Analysis (EFA) to analyze the data from the composite questionnaires, these authors created an empirical description of the latent group relationship factors. This model consists of three main components of the therapeutic group relationship, each subsuming a number of other components of group process (Johnson et al., 2005). The first component, positive relational bonds, represents the individual group member’s emotional attachment and sense of affiliation with the other members of the group, including the therapist, and the group-as-a-whole. The second component, positive working relationships, represents the individual member’s collaborative engagement in working towards treatment goals with other members, and with the therapist. The third

9 component, negative relationship factors, represents aspects of the group process that m ay negatively impede the therapeutic work or impact member bonds with the other members, group leader, or group-as-a-whole (Burlingame et al., 2008). Johnson and colleagues’ new model is unique in its incorporation of three central aspects of the group therapeutic relationship that have previously been extensively researched and discussed in group process research: content, relationship roles, and quality. The main content-based divisions in group psychotherapy are conceptualized as being between alliance and cohesion, as well as work and bonding processes (Johnson et al., 2005). In past studies, alliance and cohesion have been considered the primary content differentiation in group psychotherapy (Bakali, Baldwin, & Lorentzen, 2009; Horvath & Luborsky, 1993), until Johnson and colleague’s (2005) study asserted the three-factor structure in which bonding (positive bonding relationship) and working (positive working relationship) were primary processes. Roles in group have commonly been divided into three structured relationships: member-leader, member-member, and member group (Burlingame et al., 2004; Yalom & Leszcz, 2005). Constructs of working alliance has commonly been defined as specific to the member-leader relationship, and group climate and cohesion have been tied to member-group relationships. The new model, however, analyze working and bonding processes as they operate amongst member-leader, member-member, and member-group relationships. Quality of therapeutic relationship has historically been represented by one factor with a continuum from negative to positive (Tracey & Kokotovic, 1989) or with two

10 factors representing negative and positive relationship factors, respectively. In Johnson’s study, the final model that provided the best fit for her data included two positive dimensions (positive bonding and positive working) and one negative dimension (negative relationship). Defining the Therapeutic Group Relationship: The Development of the Group Questionnaire Following Johnson and colleagues (2005) efforts at providing a unified, empirically-based definition of the group relationship, other researchers attempted to replicate these findings across five clinical settings and four countries. A study by Bormann and Strauss (2007) collected data from 67 inpatient psychodynamic groups drawn from 15 inpatient treatment centers in Germany and Switzerland. The test-of-model fit was analyzed for the whole clinical sample (N = 438) and four randomized samples to ensure the model’s robustness. Most of the fit indices revealed significant differences between the hypothesized model originally proposed by Johnson and colleagues and the empirical data in four of the five tested samples. Nevertheless, Chi-Square-Difference-Tests clearly demonstrated the predominance of the three-factor model compared to a more economical one-factor model, assuming only one general factor indicating relationship quality (therapeutic relationship). Exploratory Structural Equation Modeling (SEM) revealed a three-factor structure comparable with the original three-factor model, with only minimal modifications. While results showed that the three-factor model could not be replicated completely, the basic structure of the model was confirmed. The German sample also revealed high correlations between the four relationship factors (group climate, cohesion, alliance and empathy) and indicated

11 that the three m ajor factors (positive bonding, positive working and negative relationship) best described the complex relationships within group treatments. A study by Bakali, Baldwin, and Lorentzen (2009) further examined the factor structure put forward by Johnson and colleagues (2005) with group members in Norway attending psychoanalytic therapy groups. The Working Alliance Inventory-Short Form, the Therapeutic Factors Inventory Cohesiveness Subscale, and the Group Climate Questionnaire-Short Form were administered to 145 patients in 18 groups three times during the life of the group. They were administered in the early (sessions 3-4), middle (10-11), and late (17-18) stages of development. One particular thrust of the study focused on differentiation of roles in group (member-leader, member-member, and member-group) and how these roles relate to specific content aspects of the therapeutic relationship. The researchers used CFA to test five distinct factor structure models on their data. The first model aimed to examine whether or not the data would fit a one-factor model representing the aspect of quality in therapeutic relationships. The second model consisted of two factors, the working and bonding content dimensions. The third model was a direct reflection of Johnson and colleagues’ three-factor model used in establishing the GQ, which consists of working and bonding content dimensions, along with a factor representing the quality of therapeutic relationships in group. The fourth model consisted of two factors represented by the group relationship as distinct from the dyadic relationship (the aspect of roles), with the dyadic relationship factor confounded within the alliance content dimension. The fifth model consisted of three factors, each based on the quality of relationships in combination with the distinct relationship roles in member-

12 group and member-leader relationships. In this model the content dimension of cohesion was confounded within the member-group relationship factor and alliance was confounded within the member-leader factor. Researchers tested the models in the early, middle, and late stages of group psychotherapy. Using multilevel CFA, the researchers found that models one, two, and four were not a good fit for the data. At the early group stage, models three and five fit the data equally well, despite not meeting conventional standards for good model fit. However, the model fit was close to acceptable, and warranted further exploration in order to seek an appropriate fit. The researchers inspected the modification indices to examine if any theoretically justifiable changes could be made. This process resulted in a sixth model combining aspects of models three and five, with the primary change being modifying the WAI Bond content dimension to load on the first and second factors. The final model consisted of three factors: factor one was labelled member-leader alliance, factor two labelled positive bonding relationship, and factor three labelled negative relationship. The primary difference in this model from Johnson and colleagues’ (2005) original was that the member-leader bond was not only important to the member-leader relationship but the also to the bonding dimension of the group as a whole. This model fit exceptionally well for the early group data, CFI = 0.98, TLI = 0.97, and RMSEA = 0.04. The Chi-Square test was nonsignificant, χ 2 (38, N = 139) = 46.56, p = .16. When tested with data from the middle sessions (10-11) of therapy, the model showed adequate fit, χ 2 (38, N = 130) = 61.3, p = .01, with CFI = 0.95, TLI = 0.94, and RMSEA = 0.07. However, researchers discovered that during the middle sessions the WAI bond did not significantly load onto the second factor (positive bonding

13 relationship) and that it loaded more strongly on the first factor (member-leader alliance). Thus, the model of greatest fit during the middle sessions approximated model five as described above. For data from the later sessions of therapy, model six showed an excellent fit, with a nonsignificant chi square test, χ 2 (38, N = 130) = 42.3, p = 0.29, CFI = 0.99, TLI = 0.99, and RMSEA = 0.03. When tested with data from later sessions of therapy, the factor loadings were similar to those from the middle sessions, with WAI Bond loading significantly to the first factor but not the second. The findings suggest that the member-leader relationship and the emotional bond between patient and therapist was important to both working and bonding aspects of the alliance, as well as to positive bonding of the group as a whole in early sessions, with lesser importance in later sessions. The researchers explained that this may be due to the fact group leaders in the study were particularly active in early sessions (Bakali et al., 2009) and met individually with group members for five sessions before group began. This unique group format may have led to the divergence between their model of best fit and Johnson’s model. Despite these differences, the study supported a three-factor structure of the therapeutic group relationship, similar to the factor structure originally put forth by Johnson et al.

(2005). However, the study also indicated that the member-leader relationship in group may operate as its own relative independent therapeutic process, further highlighting the need in group psychotherapy research to examine specific relationship roles and how they relate with group therapeutic processes. A study by Krogel, Burlingame, Chapman, Renshaw, Gleave, and Beacher (2009) aimed to again test Johnson and colleagues’ factor structure of the group relationship and

14 ultimately u se it to create a measure of the group relationship with the capability of tracking group relationship factors linked to outcome. The final purpose of the study was to create a measure that is empirically based, relevant to clinicians, and easy to administer and interpret. The measure was dubbed the Group Questionnaire (GQ). The development of the GQ consisted of two steps: First, using empirical data and clinical judgment, a team of experienced group researchers and clinicians worked to pare down and select the most relevant and psychometrically sound items from Johnson’s original set of 60. Second, the GQ factor structure was tested and revised using Confirmatory Factor Analysis (CFA) and data from three populations: outpatients from a University Counseling Center (UCC), non-patient participants from the American Group Psychotherapy Association (AGPA), and inpatients from the Utah State Hospital (USH) (Krogel et al., 2009). In selecting which items to omit, empirical analyses were first conducted, followed by a thorough review of the proposed items. First, the research team reviewed Johnson’s statistical analyses of the original 60 questions used in creating the model. The researchers then identified from an empirical perspective which items from each first order factor were the strongest and which items could be dropped due to redundancy. In general, items with small factor loadings or high correlations with other items were identified for consideration of being dropped (Krogel et al., 2009). Following the empirical analysis, the researchers met and discussed the content domain of each subscale as reflected by the items it contained. A clinically relevant definition was provided to describe each scale after reviewing the items that comprised the scale. Items were considered for inclusion based on their clinical relevance to these

15 refined construct definitions and their empirical support. Using this process, item s were selected, eliminated, reworded, and combined to create the 40-item version of the Group Questionnaire (Krogel et al., 2009). Researchers administered the reduced 40-item measure to three populations in order to test the validity of Johnson’s model and explore the psychometric properties of the new measure. Participants included 486 individuals from the three group populations: outpatient university student, non-clinical, and inpatient. Because the GQ had never been used in an inpatient setting, the USH population was selected to potentially broaden the scope of the GQ and to test its viability with the inpatient population. Data were then analyzed using CFA to assess the goodness-of-fit of Johnson’s model to each of the populations separately and as a whole. Results from the CFA of the 40-item GQ provided an inadequate fit to the data Poor fit was also found when a separate CFA was performed on each of the three sample populations. However, expected relationships between first and second order factors were found by calculating correlations and regression weights among all the subscales and testing them for statistical significance using a one- tailed test. Model divisions between levels of the group relationship (member-member, member-leader, and member-group) were found to provide a good fit to the data when second order factor items were related directly and first order factors were excluded. In order to explore a better fit for the model, the 40-item GQ was further refined by removing 10 items due to small regression weights or small factor loadings with their associated first order factor. These 30 items were again subjected to the previously conducted statistical analyses. Results from the CFA of the refined GQ provided a good

16 fit to the data, replicating the good fit Johnson and colleagues (2005) found with the proposed model in her study. This suggests that the refined 30-item GQ measurement model adequately represented the relevant theoretical constructs in the samples used in Krogel’s study. The final study was conducted by Bormann and Strauss (2009) and explored the validity and structural fit of the German version of the 30-item Group Questionnaire (GQ). The study was conducted in two parts. First, structural fit was applied to a dataset from the earlier German multi-site study (Bormann & Strauss, 2007). Using LISREL, the fit indices indicated a very good structural fit (χ 2 = 629, df = 348, p < .001; CFI = 0.96; NFI = 0.94, SRMR = 0.05; RSMEA = 0.04). The reliabilities (internal consistency, Cronbachs’ Alpha) of the three GQ-subscales, positive bonding (α = 0.92), positive working (α = 0.90), and negative relationship (α = 0.77), were also acceptable. The second focus of the study evaluated the validity of the German GQ by relating it to four gold-standard process measures in use in Germany. Five-hundred inpatients from 64 groups and 8 different German hospitals took part in the study. The scale inter-correlations between the GQ-subscales and the subscales of the measures mentioned above showed all significant results and ranged from r = 0.42 to r = 0.56 (p < 0.01; Pearson correlation). The structural fit using LISREL was applied once more on the new dataset and supported the results from earlier studies. Again the fit indices indicated a good structural fit (χ 2 = 818, df = 376, p < .001; CFI = 0.97; NFI = 0.95, SRMR = 0.06; RSMEA = 0.05). Also the reliabilities of the three GQ-subscales using Cronbachs’ Alpha showed good results that are comparable with Krogel et al.’s study:

17 positive bonding (α = 0.92), positive working (α = 0.89), and negative relationship (α = 0.79). The studies described above all have endeavoured to provide a parsimonious and empirically-based definition of group therapeutic processes. The final goal of this line of research is to clarify the nature of the group relationship and to develop a comprehensive measure of the group relationship with the capability of providing clinicians useful feedback about the functioning of their groups. Trends in Patient-focused Research: A Response to “Evidence-Based Practice” It is a common reality in today’s mental health treatment climate that clinicians are asked with increasing frequency to use objective measures of process and outcome in order to demonstrate the effectiveness of their work (Burlingame et al., 2008; Lambert & Ogles, 2004). RCT’s conducted using OQ-45 feedback in concert with CST’s in individual therapy represent an effort to address this push to practice in a demonstrably empirically supported manner. The current study is the connecting step between the aforementioned individual therapy RCT’s and group treatment. Specifically, it tests the usefulness of outcome- based feedback in tandem with information regarding client perception of the therapeutic relationship in the group. The study can be seen as a preliminary step before an RCT is initiated evaluating the effects of using outcome and clinical support feedback in groups. A brief review of the RCT’s from the individual literature provides a context for the present study. Driven by the need to meet the demands of evidence-based practice, with an awareness of the limitations of efficacy research attempting to establish empirically-

Full document contains 89 pages
Abstract: Prior research in individual therapy has provided evidence that therapists are poor predictors of client outcome without the aid of objective measures and often misjudge clients' perceptions of the therapeutic relationship. The focus of the current research was to conduct a similar study in a group setting. Therapists from a university counseling center and a state psychiatric hospital were recruited to test their accuracy in predicting client outcome, quality of therapeutic relationship and their own use of empirically supported group interventions. Results indicated that therapists are poor predictors of all three, providing support for the implementation of measure-based feedback systems to inform therapists about key information that may affect the effectiveness of group psychotherapy. Keywords: group psychotherapy, clinical prediction, outcome, cohesion