You are on page 1of 457

Psychotherapy Relationships That Work

This page intentionally left blank

That Work
Second Edition

Edited by
John C. Norcross

Published in the United States of America by Oxford University Press, Inc.,
198 Madison Avenue, New York, NY, 10016
United States of America

Oxford University Press, Inc., publishes works that further Oxford University’s
objective of excellence in research, scholarship, and education.
Oxford is a registered trade mark of Oxford University Press
in the UK and in certain other countries

© Oxford University Press, Inc. 2011

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, without the prior permission in writing of Oxford
University Press, Inc., or as expressly permitted by law, by licence, or under terms agreed with the
appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope
of the above should be sent to the Rights Department, Oxford University Press, Inc., at the
address above

You must not circulate this work in any other form and you must impose this same condition on
any acquirer

Library of Congress Cataloging-in-Publication Data

Psychotherapy relationships that work : evidence-based

responsiveness / edited by John C. Norcross. — 2nd ed.
p. ; cm.
Includes bibliographical references.
ISBN 978-0-19-973720-8 (alk. paper)
1. Psychotherapist and patient. 2. Evidence-based psychotherapy.
I. Norcross, John C., 1957-
[DNLM: 1. Professional-Patient Relations—Meta-Analysis.
2. Psychotherapy—methods—Meta-Analysis. 3. Evidence-Based
Practice—Meta-Analysis. WM 420]
RC480.8.P78 2011


1 3 5 7 9 10 8 6 4 2
Typeset in Adobe Garamond Pro
Printed on acid-free pape
Printed in the United States of America
Dedicated to

Emma and Owen

Daily reminders of the healing power of nurturing relationships

Arnold A. Lazarus, Ph.D.

Lifelong champion of adapting psychotherapy to the individual patient
This page intentionally left blank

A cordial welcome to the second edition of a single professional association (Division

Psychotherapy Relationships That Work. This of Psychotherapy), but this second edition
volume seeks, like its predecessor, to iden- was sponsorship by both the American
tify effective elements of the psychotherapy Psychological Association (APA) Division
relationship and to determine effective of Clinical Psychology and the APA
methods of adapting or tailoring that rela- Division of Psychotherapy. Second, we
tionship to the individual patient. That is, retitled the focus evidence-based psychother-
we summarize the empirical research apy relationships instead of empirically sup-
and clinical practice on what works ported (therapy) relationships to parallel the
in general as well as what works in particu- contemporary movement to the newer ter-
lar. This dual focus has been characterized minology. This title change, in addition,
as “two books in one”: one book on rela- properly emphasizes the confluence of the
tionship elements and one book on adapta- best research, clinical expertise, and patient
tion methods. characteristics in a quality treatment rela-
My hope in this book, as with the first tionship. Third, we expanded the breadth
edition, is to advance a rapprochement of coverage. New reviews were commis-
between the warring factions in the culture sioned on the alliance with children and
wars of psychotherapy and to demonstrate adolescents, the alliance in couple and
that the best available research clearly dem- family therapy, collection of real-time
onstrates the healing qualities of the ther- feedback from clients, patient preferences,
apy relationship. The first edition brought culture, and attachment style. Fourth, we
renewed and corrective attention to the decided to insist on meta-analyses for all
substantial research behind the therapy research reviews. These original meta-anal-
relationship and, in the words of one yses enable direct estimates of the magni-
reviewer (Psychotherapy Research, 2003, tude of association and the ability to search
p. 532), “will convince most psychothera- for moderators. Unfortunately, that also
pists of the rightful place of ESRs (empiri- meant that several relationship elements
cally supported relationships) alongside and adaptation methods in the first edition
ESTs in the treatments they provide.” Note (self-disclosure, transference interpreta-
the desired emphasis on “alongside” treat- tions, anaclitic vs. introjective styles,
ments, not “instead of ” or “better than.” assimilation of problematic experiences)
were excluded due to their insufficient
Changes in the New Edition number of studies. Fifth, we improved the
The aims of this edition of Psychotherapy process for determining whether a particu-
Relationships That Work remain the same as lar relationship element—say, the alliance
its predecessor, but its sponsorship, meth- or empathy—could be classified as
odology, and scope differ. First, the inaugu- demonstrably effective, probably effective,
ral edition of the book was sponsored by or promising but insufficient research

vi i
to judge. We constituted expert panels to evidence-based practices or treatment
establish a consensus on the evidentiary guidelines. We hope our work will inform
strength of the relationship elements and and balance any efforts to focus exclusively
adaptation methods. Experts indepen- on techniques or treatments to the neglect
dently reviewed and rated the meta-analy- of the humans involved in the enterprise.
ses on several objective criteria, thus adding Our third priority was insurance carriers
a modicum of rigor and consensus to the and accreditation organizations, many of
process, which was admittedly less so in the which have unintentionally devalued
first edition of the book. the person of the therapist and the
The net result is a compilation of two centrality of the relationship by virtue of
dozen, cutting-edge meta-analyses devoted reimbursement decisions. Although sup-
to what works in the therapy relationship portive of the recent thrust toward science
and what works in adapting that relation- informing practice, we must remind all
ship to the individual client and his/ parties to the therapy relationship that
her singular situation. This new edition, healing cannot be replaced with treating,
appearing 10 years after the first incarna- caring cannot be supplanted by managing.
tion, presents a slightly slimmer book offer- Finally, Psychotherapy Relationships That
ing more practical, bulleted information Work is intended for psychotherapy
on clinical practice at the end of each researchers seeking a central resource on
chapter. the empirical status of the multiple, inter-
dependent qualities of the therapy
Probable Audiences relationship.
One of our earliest considerations in plan-
ning the first edition of the book concerned Organization of the Book
the intended audiences. Each of psycho- The opening chapter introduces the book
therapy’s stakeholders—patients, practitio- by outlining the purpose and history of the
ners, researchers, trainers, students, interdivisional Task Force and its relation
organizations, insurance companies, and to previous efforts to identify evidence-
policymakers—expressed different prefer- based practices in psychotherapy. This
ences for the content and length of the chapter also presents the key limitations of
volume. our work.
We prepared the book for multiple audi- The heart of the book is composed of
ences but in a definite order of priority. research reviews on the therapist’s relational
First came clinical practitioners and train- contributions and recommended therapeu-
ees of diverse theoretical orientations and tic practices predicated on that research.
professional disciplines. They need to Section II—Effective Elements of the
address urgent pragmatic questions: What Therapy Relationship: What Works in
do we know from the empirical research General—features eleven chapters on rela-
about cultivating and maintaining the tionship elements primarily provided by
therapeutic relationship? What are the the psychotherapist. Chapters 2–5 report
research-supported means of adapting that on broader, more inclusive relationship ele-
relationship to the patient beyond his/her ments. The therapy alliance and group
diagnosis? Our second priority was accorded cohesion are composed, in fact, of multiple
to the mental health disciplines themselves, elements. Chapters 6–9 feature more
specifically those committees, task forces, specific elements of the therapy relation-
and organizations promulgating lists of ship, and Chapters 10–12 review specific
viii pre fac e
therapist behaviors that promote the rela- Report the effect size as weighted r
tionship and favorable treatment results. (in Section II) or d (in Section III).
Section III—Tailoring the Therapy Relation- • Moderators. Present the results of the
ship to the Individual Patient: What Works moderator analyses on the association
in Particular—features eight chapters on between the relationship element and
adaptation methods. They feature probably treatment outcome.
and demonstrably effective means of tailor- If available in the studies, examine the
ing psychotherapy to the entire person possible moderating effects of
beyond diagnosis alone. (1) rater perspective (assessed by therapist,
The final section of the book consists of patient, or external raters),
a single chapter. It presents the Task Force (2) therapist variables, (3) patient factors,
conclusions, including a list of evidence- (4) different measures,
based relationship elements and adaptation (5) time of assessment (when in the course
methods, and our recommendations, of therapy), and (6) type
divided into general, practice, training, of psychotherapy/theoretical orientation.
research, and policy recommendations. • Patient Contribution. The meta-
analyses pertain largely to the
Chapter Guidelines psychotherapist’s contribution to the
With the exception of the bookends relationship; by contrast, this section will
(Chapters 1 and 21), all chapters use the address the patient’s contribution to that
same section headings and adhere to a con- relationship and
sistent structure, as follows: the distinctive perspective he/she brings to
the interaction.
• Introduction (untitled). Introduce the
• Limitations of the Research. Point to
relationship element or the adaptation
the major limitations of both the meta-
method and its historical context.
analysis and the available studies.
• Definitions and Measures. Define in
• Therapeutic Practices. Emphasize what
theoretically neutral language
works. Bullet the practice implications
the relationship element or adaptation
from the foregoing research, primarily in
method. Identify any highly similar
terms of the therapist’s contribution and
or equivalent constructs from
secondarily in terms of the patient’s
diverse theoretical traditions.
Review the popular measures used
in the research and included in the These research reviews are based on the
ensuing meta-analysis. results of empirical research linking the
• Clinical Example. Provide several relationship element or adaptation method
concrete examples of the relationship to psychotherapy outcome. Outcome was
behavior being reviewed. Portions of inclusively defined but consisted largely of
psychotherapy transcripts are encouraged. distal posttreatment outcomes. Authors
• Meta-Analytic Review. Compile all were asked to specify the outcome criterion
available empirical studies linking the when a particular study did not employ
relationship behavior to treatment a typical end-of-treatment measure of
outcome in the English language. Use the symptom or functioning. Indeed, the type
Meta-Analysis Reporting Standards of outcome measure was frequently ana-
(MARS) as a general guide for the lyzed as a possible moderator of the overall
information included in the chapter. effect size.

p re fac e ix
Acknowledgments sharing their expertise. Dr. Bruce Wampold
Psychotherapy Relationships That Work would expertly reviewed each meta-analysis and
not have been possible without a decade of provided valuable guidance on the entire
organizational and individual support. On project. Members of the expert consensus
the organizational front, the board of direc- panels critiqued each meta-analysis and
tors of the APA Division of Psychotherapy rated the evidentiary strength of the results;
and the APA Division of Clinical Psychology I appreciate the generosity of Drs. Guillermo
commissioned and supported the Task Bernal, Franz Caspar, Louis Castonguay,
Force. In particular, I am indebted to the Charles Gelso, Mark Hilsenroth, Michael
presidents of the respective divisions: Drs. Lambert, and Bruce Wampold. The Steering
Jeffrey Barnett, Nadine Kaslow, and Jeffrey Committee of the first Task Force assisted
Magnavita of the psychotherapy division, in canvassing the literature, defining the
and Drs. Marsha Linehan, Irving Weiner, parameters of the project, selecting the
and Marvin Goldfried of the clinical divi- contributors, and writing the initial con-
sion. At Oxford University Press, Joan clusions. I am grateful to them all: Steven
Bossert shepherded both books through J. Ackerman, Lorna Smith Benjamin,
the publishing process and recognized early Larry E. Beutler, Charles J. Gelso, Marvin
on that they would compliment Oxford’s R. Goldfried, Clara E. Hill, Michael
landmark Treatments That Work. This J. Lambert, David E. Orlinsky, and Jackson
second edition has been improved by the P. Rainer. Last but never least, my
OUP book team of Sarah Harrington, Jodi immediate family—Nancy, Jonathon, and
Nardi, and Tony Orrantia. Rebecca—tolerated my absences, preoccu-
On the individual front, many people pations, and irritabilities associated with
modeled and manifested the ideal thera- editing this book with a combination of
peutic relationship throughout the course empathy and patience that would do any
of the project. The authors of the respective seasoned psychotherapist proud.
chapters, of course, were indispensible John C. Norcross, PhD
in generating the research reviews and in Clarks Summit, Pennsylvania

x pre fac e

Jennifer Alonso, B.S. AC Del Re, M.A.

Department of Psychology, Department of Counseling Psychology,
Brigham Young University University of Wisconsin–Madison
Rebecca M. Ametrano, B.A. Gary M. Diamond, Ph.D.
Department of Psychology, Department of Psychology,
University of Massachusetts Amherst Ben-Gurion University of the Negev
Diane B. Arnkoff, Ph.D. Erin M. Doolin, M.Ed.
Department of Psychology, Department of Counseling Psychology,
Catholic University of America University of Wisconsin–Madison
Sara B. Austin, B.S. Robert Elliott, Ph.D.
Department of Psychology, University of School of Psychological Sciences and
Wisconsin–Madison Health, University of Strathclyde
Guillermo Bernal, Ph.D. William D. Ellison, M.S.
Institute for Psychological Research, Department of Psychology,
University of Puerto Rico Pennsylvania State University
Samantha L. Bernecker, B.S. Valentín Escudero, Ph.D.
Department of Psychology, Departamento de Psicología,
Pennsylvania State University Universidad de A Coruña
Larry E. Beutler, Ph.D. Catherine Eubanks-Carter, Ph.D.
Pacific Graduate School of Psychology, Ferkauf Graduate School of Psychology,
Palo Alto University Yeshiva University
Kathy Blau, M.S. Barry A. Farber, Ph.D.
Pacific Graduate School of Psychology, Department of Counseling and Clinical
Palo Alto University Psychology, Teachers College
Arthur C. Bohart, Ph.D. Columbia University
Department of Psychology, California Christoph Flückiger, Ph.D.
State University–Dominguez Hills and Department of Clinical Psychology and
Graduate College of Psychology and Psychotherapy, University of Bern
Humanistic Studies, Saybrook University Myrna L. Friedlander, Ph.D.
Gary M. Burlingame, Ph.D. Department of Educational and
Department of Psychology, Counseling Psychology, University at
Brigham Young University Albany/State University of New York
Jennifer L. Callahan, Ph.D. Charles J. Gelso, Ph.D.
Department of Psychology, Department of Psychology,
University of North Texas University of Maryland-College Park
Michael J. Constantino, Ph.D. Carol R. Glass, Ph.D.
Department of Psychology, Department of Psychology,
University of Massachusetts-Amherst Catholic University of America
Don E. Davis, M.S. Leslie S. Greenberg, Ph.D.
Department of Psychology, Department of Psychology,
Virginia Commonwealth University York University

T. Mark Harwood, Ph.D. Michael A. McDaniel, Ph.D.
Private Practice Department of Management,
Chicago, Illinois Virginia Commonwealth University
Jeffrey A. Hayes, Ph.D. Aaron Michelson, M.S.
Counseling Psychology Program, Pacific Graduate School of Psychology,
Pennsylvania State University Palo Alto University
Laurie Heatherington, Ph.D. J. Christopher Muran, Ph.D.
Department of Psychology, Derner Institute of Advanced
Williams College Psychological Studies,
John Holman, M.S. Adelphi University
Pacific Graduate School of Psychology, John C. Norcross, Ph.D.
Palo Alto University Department of Psychology,
Joshua N. Hook, Ph.D. University of Scranton
Department of Psychology, James O. Prochaska, Ph.D.
University of North Texas Cancer Prevention Research Consortium,
Adam O. Horvath, Ed.D. University of Rhode Island
Faculty of Education & Department of Melanie M. Domenech Rodríguez, Ph.D.
Psychology, Simon Fraser University Department of Psychology,
Ann M. Hummel, M.S. Utah State University
Department of Psychology, Jeremy D. Safran, Ph.D.
University of Maryland-College Park Department of Psychology,
Marc S. Karver, Ph.D. New School for Social Research
Department of Psychology, Lori N. Scott, M.S.
University of South Florida Department of Psychology,
Satoko Kimpara, Ph.D. Pennsylvania State University
Pacific Graduate School of Psychology, Kenichi Shimokawa, Ph.D.
Asian American Community Involvement Family Institute,
(AACI) and Palo Alto University Northwestern University
Marjorie H. Klein, Ph.D. Stephen R. Shirk, Ph.D.
Department of Psychiatry, Department of Psychology,
University of Wisconsin–Madison University of Denver
Gregory G. Kolden, Ph.D. JuliAnna Z. Smith, M.A.
Department of Psychiatry and Center for Research on Families,
Psychology, University of University of Massachusetts Amherst
Wisconsin–Madison Timothy B. Smith, Ph.D.
Paul M. Krebs, Ph.D. Department of Counseling
Department of General Internal Psychology and Special Education,
Medicine, New York University Brigham Young University
Medical Center Xiaoxia Song, Ph.D.
Michael J. Lambert, Ph.D. Department of Psychology,
Department of Psychology, Ohio University
Brigham Young University Joshua K. Swift, Ph.D.
Kenneth N. Levy, Ph.D. Department of Psychology,
Department of Psychology, University of Alaska Anchorage
Pennsylvania State University Dianne Symonds, Ph.D.
Debra Theobald McClendon, Ph.D. Faculty of Community and
Department of Psychology, Health Studies, Kwantlen
Brigham Young University Polytechnic University

xii co n t r i bu to r s
Georgiana Shick Tryon, Ph.D. Chia-Chiang Wang, M.Ed.
Ph.D. Program in Educational Department of Rehabilitation
Psychology, The Graduate Center, Psychology and Special Education,
City University of New York University of Wisconsin–Madison
David Verdirame, M.S. Jeanne C. Watson, Ph.D.
Pacific Graduate School of Department of Adult Education,
Psychology, Palo Alto University University of Toronto
Barbara M. Vollmer, Ph.D. Greta Winograd, Ph.D.
Department of Counseling Psychology Department,
Psychology, University of State University of
Denver New York-New Paltz
Bruce E. Wampold, Ph.D. Everett L. Worthington, Jr., Ph.D.
Department of Counseling Department of Psychology,
Psychology, University of Virginia Commonwealth
Wisconsin–Madison University

co n t r i bu to r s xiii
This page intentionally left blank

Part One • Introduction

1. Evidence-Based Therapy Relationships 3
John C. Norcross and Michael J. Lambert

Part Two • Effective Elements of the Therapy Relationship:

What Works in General
2. Alliance in Individual Psychotherapy 25
Adam O. Horvath, A. C. Del Re, Christopher Flückiger, and Dianne Symonds
3. Alliance in Child and Adolescent Psychotherapy 70
Stephen R. Shirk and Marc S. Karver
4. Alliance in Couple and Family Therapy 92
Myrna L. Friedlander, Valentín Escudero, Laurie Heatherington, and
Gary M. Diamond
5. Cohesion in Group Therapy 110
Gary M. Burlingame, Debra Theobald McClendon, and Jennifer Alonso
6. Empathy 132
Robert Elliott, Arthur C. Bohart, Jeanne C. Watson, and
Leslie S. Greenberg
7. Goal Consensus and Collaboration 153
Georgiana Shick Tryon and Greta Winograd
8. Positive Regard and Affirmation 168
Barry A. Farber and Erin M. Doolin
9. Congruence/Genuineness 187
Gregory G. Kolden, Marjorie H. Klein, Chia-Chiang Wang, and
Sara B. Austin
10. Collecting Client Feedback 203
Michael J. Lambert and Kenichi Shimokawa
11. Repairing Alliance Ruptures 224
Jeremy D. Safran, J. Christopher Muran, and Catherine Eubanks-Carter
12. Managing Countertransference 239
Jeffrey A. Hayes, Charles J. Gelso, and Ann M. Hummel

Part Three • Tailoring the Therapy Relationship to the Individual
Patient: What Works in Particular
13. Reactance/Resistance Level 261
Larry E. Beutler, T. Mark Harwood, Aaron Michelson, Xiaoxia Song,
and John Holman
14. Stages of Change 279
John C. Norcross, Paul M. Krebs, and James O. Prochaska
15. Preferences 301
Joshua K. Swift, Jennifer L. Callahan, and Barbara M. Vollmer
16. Culture 316
Timothy B. Smith, Melanie Domenech Rodríguez, and Guillermo Bernal
17. Coping Style 336
Larry E. Beutler, T. Mark Harwood, Satoko Kimpara, David Verdirame, and
Kathy Blau
18. Expectations 354
Michael J. Constantino, Carol R. Glass, Diane B. Arnkoff,
Rebecca M. Ametrano, and JuliAnna Z. Smith
19. Attachment Style 377
Kenneth N. Levy, William D. Ellison, Lori N. Scott, and
Samantha L. Bernecker
20. Religion and Spirituality 402
Everett L. Worthington, Jr., Joshua N. Hook, Don E. Davis, and
Michael A. McDaniel

Part Four • Conclusions and Guidelines

21. Evidence-Based Therapy Relationships: Research Conclusions and
Clinical Practices 423
John C. Norcross and Bruce E. Wampold

Index 431

xvi ta b l e o f co n t e n ts
This page intentionally left blank

1 Evidence-Based Therapy Relationships

John C. Norcross and Michael J. Lambert

The culture wars in psychotherapy dramati- controlled/clinical trial), and the scientific-
cally pit the treatment method against medical model? Or do you belong to
the therapy relationship. Do treatments the side of the therapy relationship, the
cure disorders or do relationships heal effectiveness and process-outcome studies,
people? Which is the most accurate vision and the relational-contextual model? Such
for practicing, researching, and teaching polarizations not only impede psychothera-
psychotherapy? pists from working together but also hinder
Like most dichotomies, this one is mis- attempts to provide the most efficacious psy-
leading and unproductive on multiple chological services to our patients.
counts. For starters, the patient’s contribu- We hoped that a balanced perspective
tion to psychotherapy outcome is vastly would be achieved by the adoption of an
greater than that of either the particular inclusive, neutral definition of evidence-
treatment method or the therapy relation- based practice. The American Psychological
ship (Lambert, 1992). The empirical evidence Association (2006, p. 273) did endorse just
should keep us mindful and a bit humble such a definition: “Evidence-based practice
about our collective tendency toward ther- in psychology (EBPP) is the integration of
apist centricity (Bohart & Tallman, 1999). the best available research with clinical
For another, decades of psychotherapy expertise in the context of patient charac-
research consistently attest that the patient, teristics, culture, and preferences.” However,
the therapist, their relationship, the treat- even that definition has been comman-
ment method, and the context all contrib- deered by the rival camps as polarizing
ute to treatment success (and failure). We devices. On the one side, some erroneously
should be looking at all of these determi- equate EBP solely with the best available
nants and their optimal combinations research and particularly the results of
(Norcross, Beutler, & Levant, 2006). RCTs on treatment methods, while on the
But perhaps the most pernicious and other side, some mistakenly exaggerate the
insidious consequence of the false dichot- primacy of clinical or relational expertise
omy of treatment versus relationship has while neglecting research support.
been its polarizing effect on the discipline. Within this polarizing context, in 1999,
Rival camps have developed, and countless the American Psychological Association
critiques have been published on each side (APA) Division of Psychotherapy com-
of the culture war. Are you on the side of the missioned a task force to identify, opera-
treatment method, the RCT (randomized tionalize, and disseminate information on

empirically supported therapy relation- the therapy relationship and to determine
ships. That task force summarized its find- effective methods of adapting or tailoring
ings and detailed its recommendations in therapy to the individual patient on the
the first edition of this book (Norcross, basis of his/her (transdiagnostic) character-
2002). In 2009, the Division of Psycho- istics. In other words, we were interested in
therapy along with the Division of Clinical both what works in general and what works
Psychology commissioned a second task for particular patients. This twin focus has
force on evidence-based therapy relation- been characterized as “two books in one”:
ships to update the research base and one book on relationship elements and one
clinical practices on the psychotherapist– book on adaptation methods under the
patient relationship. This second edition, same cover.
appearing 10 years after its predecessor, does For the purposes of our work, we again
just that. adopted Gelso and Carter’s (1985, 1994)
Our hope now, as then, is to advance a operational definition of the therapy rela-
rapprochement between the warring fac- tionship: The relationship is the feelings
tions and to demonstrate that the best and attitudes that therapist and client have
available research clearly supports the heal- toward one another, and the manner in
ing qualities of the therapy relationship and which these are expressed. This definition is
the beneficial value of adapting that rela- quite general, and the phrase “the manner
tionship to patient characteristics beyond in which it is expressed” potentially opens
diagnosis. The bulk of the book summa- the relationship to include everything
rizes the best available research and clinical under the therapeutic sun (see Gelso &
practices on numerous elements of the Hayes, 1998, for an extended discussion).
therapy relationship and on several meth- Nonetheless, it serves as a concise, consen-
ods of treatment adaptation. In doing so, sual, theoretically neutral, and sufficiently
our grander goal is to repair some of the precise definition.
damage incurred by the culture wars in We acknowledge the deep synergy
psychotherapy and to promote integration between treatment methods and the thera-
between science and practice. peutic relationship. They constantly shape
In this chapter, we begin by tracing the and inform each other. Both clinical expe-
purpose and processes of the interdivi- rience and research evidence (e.g., Rector,
sional Task Force. We explicate the need for Zuroff, & Segal, 1999; Barber et al., 2006)
identifying evidence-based elements of the point to a complex, reciprocal interaction
therapy relationship and means of match- between the interpersonal relationship and
ing or adapting treatment to the individual. the instrumental methods. Consider this
In a tentative way, we offer two models to finding from a large collaborative study:
account for psychotherapy outcome as a For patients with a strong therapeutic alli-
function of various therapeutic factors (e.g., ance, adherence to the treatment manual
patient, relationship, technique). The latter was irrelevant for treatment outcome, but
part of the chapter features the limitations for patients with a weak alliance, a moder-
of the Task Force’s work and responds to ate level of therapist adherence was associ-
frequently asked questions. ated with the best outcome (Barber et al.,
2006). The relationship does not exist apart
The Interdivisional Task Force from what the therapist does in terms of
The dual purposes of the interdivisional Task method, and we cannot imagine any treat-
Force were to identify effective elements of ment methods that would not have some

4 i n t ro d u c t i o n
relational impact. Put differently, treat- promulgation of effective psychotherapy. It
ment methods are relational acts (Safran & does so by expanding or enlarging the typi-
Muran, 2000). cal focus of evidence-based practice to ther-
For historical and research convenience, apy relationships. Focusing on one area—
the field has distinguished between rela- in this case, the therapy relationship—may
tionships and techniques. Words like unfortunately convey the impression that
“relating” and “interpersonal behavior” are this is the only area of importance. We
used to describe how therapists and clients review the scientific literature on the
behave toward each other. By contrast, therapy relationship and provide clinical
terms like “technique” or “intervention” are recommendations based on that literature
used to describe what is done by the thera- without, we trust, degrading the simultane-
pist. In research and theory, we often treat ous contributions of the treatments, patients,
the how and the what—the relationship or therapists to outcome. Indeed, we wish
and the intervention, the interpersonal and that more psychotherapists would acknowl-
the instrumental—as separate categories. edge the inseparable context and practical
In reality, of course, what one does and interdependence of the relationship and the
how one does it are complementary and treatment. That can prove a crucial step in
inseparable. To remove the interpersonal reducing the polarizing strife of the culture
from the instrumental may be acceptable wars and in improving the effectiveness of
in research, but it is a fatal flaw when psychotherapy (Lambert, 2010).
the aim is to extrapolate research results An immediate challenge to the Task
to clinical practice (see Orlinsky, 2000; Force was to establish the inclusion and
2005 special issue of Psychotherapy on the exclusion criteria for the elements of the
interplay of techniques and therapeutic therapy relationship. We readily agreed that
relationship). the traditional features of the therapeu-
In other words, the value of a treatment tic relationship—the alliance in individual
method is inextricably bound to the rela- therapy and cohesion in group therapy, for
tional context in which it is applied. Hans example—and the Rogerian facilitative
Strupp, one of our first research mentors, conditions would constitute core elements.
offered an analogy to illustrate the insepa- We further agreed that discrete, relatively
rability of these constituent elements. nonrelational techniques were not part of
Suppose you want your teenager to clean our purview, but that a few relational meth-
his or her room. Two methods for achiev- ods would be included. Therapy methods
ing this are to establish clear standards were considered for inclusion if their con-
and to impose consequences. A reasonable tent, goal, and context were inextricably
approach, but the effectiveness of these interwoven into the emergent therapy rela-
two evidence-based methods will vary on tionship. We settled on three relationship
whether the relationship between you and behaviors (collecting real-time client feed-
the teenager is characterized by warmth back, repairing alliance ruptures, and man-
and mutual respect or by anger and mis- aging countertransference) because these
trust. This is not to say that the methods methods are deeply embedded in the inter-
are useless, merely that how well they work personal character of the relationship itself.
depends upon the context in which they But which relational behaviors to include
are used (Norcross, 2010). and which to exclude under the rubric of
The work of the Task Force applies psy- the therapy relationship bedeviled us, as it
chological science to the identification and has the field.

n o rc ro s s , l a m b e rt 5
How does one divide the indivisible rela- Their respective ratings of demonstrably
tionship? For example, is support similar effective, probably effective, or promising
enough to positive regard or validation to be but insufficient research to judge were then
considered in the same meta-analysis, or is combined to render a consensus. These
it distinct enough to deserve a separate conclusions are presented in the last chap-
research review? We struggled on how finely ter of this book.
to slice the therapy relationship. As David The deliberations of the Steering
Orlinsky opined in one of his e-mail mes- Committee and the expert panels were not
sages, “It’s okay to slice bologna that thin, easy or unanimous. Democracy is messy
but I doubt that it can be meaningfully and inefficient; science is even slower and
done to the relationship.” We agreed, as a painstaking. We debated and, in most
group, to place the research on support in instances, voted on our decisions. We relied
the positive regard chapter, but we under- on expert opinion, professional consensus,
stand that some practitioners may under- and most importantly, reviews of the
standably take exception to collapsing these empirical evidence. But these were all
relationship elements. As a rule, we opted human decisions—open to cavil, conten-
to divide the research reviews into smaller tion, and future revision.
chunks so that the research conclusions
were more specific and the practice impli- Therapy Relationship
cations more concrete. Recent years have witnessed the controver-
In our deliberations, several members of sial compilation of practice guidelines and
the Steering Committee advanced a favor- evidence-based treatments in mental health.
ite analogy: the therapy relationship is like In the United States and other countries,
a diamond, a diamond composed of mul- the introduction of such guidelines has
tiple, interconnected facets. The diamond provoked practice modifications, training
is a complex, reciprocal, and multidimen- refinements, and organizational conflicts.
sional entity. The Task Force endeavored to Insurance carriers and government policy-
separate and examine many of these facets. makers are increasingly turning to such
Once these decisions were finalized, we guidelines to determine which psychother-
commissioned original meta-analyses on apies to approve and fund. Indeed, along
the relationship elements and the adap- with the negative influence of managed
tation methods. The chapters and the care, there is probably issue no more cen-
meta-analyses therein were reviewed and tral to clinicians than the evolution of evi-
subsequently underwent at least one revi- dence-based practice in psychotherapy
sion. Once revised, two consensus panels (Barlow, 2000).
(each composed of five experts) were estab- All of the efforts to promulgate evidence-
lished to review the evidentiary strength based psychotherapies have been noble in
of the relationship element or adaptation intent and timely in distribution. They
method according to the following criteria: are praiseworthy efforts to distill scien-
number of empirical studies, consistency tific research into clinical applications
of empirical results, independence of sup- and to guide practice and training. They
portive studies, magnitude of association wisely demonstrate that, in a climate of
between the relationship element and out- accountability, psychotherapy stands up to
come, evidence for causal link between empirical scrutiny with the best of health
relationship element and outcome, and the care interventions. And within psychol-
ecological or external validity of research. ogy, these have proactively counterbalanced
6 i n t ro d u c t i o n
documents that accord primacy to biomed- Person of the Therapist
ical treatments for mental disorders and Most practice guidelines and evidence-
largely ignore the outcome data for psycho- based practice compilations depict dis-
logical therapies. On many accounts, then, embodied psychotherapists performing
the extant EBP efforts have addressed the procedures on DSM disorders. This stands
realpolitik of the socioeconomic situation in marked contrast to the clinician’s experi-
(Messer, 2001; Nathan, 1998). ence of psychotherapy as an intensely inter-
At the same time, many practitioners personal and deeply emotional experience.
and researchers have found these recent Although efficacy research has gone to
efforts to codify evidence-based treatments considerable lengths to eliminate the indi-
seriously incomplete. While scientifically vidual therapist as a variable that might
laudable in their intent, these efforts have account for patient improvement, the ines-
largely ignored the therapy relationship capable fact is that it’s simply not possible
and the person of the therapist. If one were to mask the person and the contribution of
to read previous efforts literally, disembod- the therapist (Orlinsky & Howard, 1977).
ied therapists apply manualized inter- The curative contribution of the person
ventions to discrete DSM disorders. Not of the therapist is, arguably, as empirically
only is the language offensive on clinical validated as manualized treatments or
grounds to some practitioners, but the psychotherapy methods (Duncan, Miller,
research evidence is weak for validating Wampold, & Hubble, 2010).
treatment methods in isolation from the Multiple and converging sources of
therapy relationship and the individual evidence indicate that the person of the
patient. psychotherapist is inextricably intertwined
Suppose we asked a neutral scientific with the outcome of psychotherapy. A large,
panel from outside the field to review the naturalistic study estimated the outcomes
corpus of psychotherapy research to deter- attributable to 581 psychotherapists treat-
mine what is the most powerful phenome- ing 6,146 patients in a managed care set-
non we should be studying, practicing, and ting. About 5% of the outcome variation
teaching. Henry (1998, p. 128) concludes was due to therapist effects and 0% due
that the panel to specific treatment methods (Wampold
& Brown, 2005). Quantitative reviews of
would find the answer obvious, and therapist effects in psychotherapy outcome
empirically validated. As a general trend studies show consistent and robust effects—
across studies, the largest chunk of probably 5% to 9% of psychotherapy
outcome variance not attributable to outcome (Crits-Christoph et al., 1991). In
preexisting patient characteristics involves reviewing the research, Wampold (2001,
individual therapist differences and the p. 200) concluded that “a preponderance
emergent therapeutic relationship between of evidence indicates that there are large
patient and therapist, regardless of therapist effects . . . and that the effects
technique or school of therapy. This is the greatly exceed treatment effects.”
main thrust of three decades of empirical Two controlled studies examining thera-
research. pist variables in the outcomes of cognitive-
behavioral therapy are instructive (Huppert
What’s missing, in short, are the person et al., 2001; Project MATCH Research
of the therapist and elements of the thera- Group, 1998). In the Multicenter Collab-
peutic relationship. orative Study for the Treatment of Panic

n o rc ro s s , l a m b e rt 7
Disorder, considerable care was taken to average at the center was about 8%. In a
standardize the treatment, the therapist, related study of many of the same thera-
and the patients in order to increase the pists (Anderson, Ogles, Patterson, Lambert,
experimental rigor of the study and in order & Vermeersch, 2009), the strongest predic-
to minimize therapist effects. The treat- tor of patient outcome was these therapists’
ment was manualized and structured, the interpersonal skills.
therapists were identically trained and
monitored for adherence, and the patients Relationship Elements
rigorously evaluated and relatively uniform. A second omission from most evidence-
Nonetheless, the therapists significantly based practice guidelines has been the
differed in the magnitude of change among decision to validate only the efficacy
caseloads. Effect sizes for therapist impact of treatments or technical interventions,
on outcome measures ranged from 0% as opposed to the therapy relationship or
to 18%. In the similarly controlled multi- therapist interpersonal skills. This decision
site study on alcohol abuse conducted both reflects and reinforces the ongoing
by Project MATCH, the therapists were movement toward high-quality compara-
carefully selected, trained, supervised, and tive effectiveness research (CER) on brand-
monitored in their respective treatment name psychotherapies. “This trend of putting
approaches. Although there were few out- all of the eggs in the “technique” basket
come differences among the treatments, began in the late 1970s and is now reach-
over 6% of the outcome variance (1%–12% ing the peak of influence” (Bergin, 1997,
range) was due to therapists. Despite impres- p. 83).
sive attempts to experimentally render indi- Both clinical experience and research
vidual practitioners as controlled variables, findings underscore that the therapy rela-
it is simply not possible to mask the person tionship accounts for as much of the
and the contribution of the therapist. outcome variance as particular treatment
Further evidence comes from naturalis- methods (Lambert & Barley, 2002), espe-
tic studies of clinical practice rather than cially after the effects of researcher allegiance
research settings where attempts are made to treatment are accounted for (Luborsky
to reduce individual therapist’s contribu- et al., 1999). An early and influential review
tion to patient outcomes. Okiishi, Lambert, by Bergin and Lambert (1978, p. 180)
Nielsen, and Ogles (2003) examined the anticipated the contemporary research con-
outcomes of clients seen by 56 therapists sensus: “The largest variation in therapy
practicing a variety of treatment methods. outcome is accounted for by pre-existing
Despite the fact that the psychothera- client factors, such as motivation for
pists had similarly disturbed clients, there change, and the like. Therapist personal
were dramatic differences in client outcome factors account for the second largest pro-
as a function of seeing a top-rated thera- portion of change, with technique variables
pist or one at the bottom. On average, cli- coming in a distant third.”
ents seeing a top-rated therapist achieved Even those practice guidelines enjoin-
reliable improvement, while those clients ing practitioners to attend to the therapy
seen by bottom-ranked therapists were relationship do not provide specific, evi-
unchanged or slightly worse off after treat- dence-based means of doing so. The
ment. Client deterioration for the low per- APA Template for Developing Guidelines
formers included one therapist who had (Task Force on Psychological Intervention
21% of his/her clients deteriorate while the Guidelines, 1995, pp. 5–6), for example,

8 i n t ro d u c t i o n
sagely recognizes that factors common to Paul’s (1967) iconic question: “What treat-
all therapies, “such as the clinician’s ability ment, by whom, is most effective for this
to form a therapeutic alliance or to gener- individual with that specific problem, and
ate a mutual framework for change, are under which set of circumstances?” Every
powerful determinants of success across psychotherapist recognizes that what works
interventions” but only vaguely addresses for one person may not work for another;
how research protocols or individual prac- we seek “different strokes for different
titioners should do so. For another exam- folks.”
ple, the scholarly and comprehensive review To many, the means of such matching
on treatment choice from Great Britain was to tailor the psychotherapy to the
(Department of Health, 2001) devotes a patient’s disorder or presenting problem—
single paragraph to the therapeutic rela- that is, to find the best treatment for a
tionship. Its recommended principle is particular disorder. The research suggests
that “Effectiveness of all types of therapy that it is certainly useful for select disorders;
depends on the patient and the therapist some psychotherapies make better mar-
forming a good working relationship” riages with some mental health disorders
(p. 35), but no evidence-based guidance (e.g., Barlow, 2007; Nathan & Gorman,
is offered on which therapist behaviors 2007; Roth & Fonagy, 2004). Indeed, the
contribute to that relationship. Likewise, overwhelming majority of randomized
although most treatment manuals mention clinical trials in psychotherapy compare the
the importance of the therapy relationship, efficacy of specific treatments for specific
few specify which therapist qualities or in- disorders (Lambert, 2011).
session behaviors lead to a curative However, only matching psychotherapy
relationship. to a disorder is incomplete and not always
All of this is to say that extant lists of effective (Wampold, 2001). Particularly
EBPs and best practices in mental health absent from much of the research has been
give short shrift—some would say lip ser- the person of the patient, beyond his/her
vice—to the person of the therapist and disorder. As Sir William Osler, father of
the emergent therapeutic relationship. The modern medicine, said: “It is sometimes
vast majority of current attempts are thus much more important to know what sort
seriously incomplete and potentially mis- of a patient has a disease than what sort of
leading, both on clinical and empirical disease a patient has.”
grounds. Most practice guidelines and evidence-
based compilations unintentionally reduce
Treatment Adaptation our clients to a static diagnosis or prob-
Since the earliest days of modern psycho- lem. The impressive American Psychiatric
therapy, practitioners have realized that Association Practice Guidelines for the
treatment should be tailored to the indi- Treatment of Psychiatric Disorders (2006), to
viduality of the patient and the singularity take one prominent example, is organized
of his/her context. As early as 1919, Freud exclusively around diagnoses. Virtually all
introduced psychoanalytic psychotherapy practice guidelines are directed toward cat-
as an alternative to classical analysis on the egorical disorders. DSM diagnoses have
recognition that the more rarified approach ruled the evidence-based roost to date.
lacked universal applicability (Wolitzky, This choice flies in the face of clini-
2011). The mandate for individualizing cal practice and research findings that a
psychotherapy was embodied in Gordon categorical, nonpsychotic Axis I diagnosis

n o rc ro s s , l a m b e rt 9
exercises only a modest impact on treat- Research studies problematically collapse
ment outcome (Beutler, 2000). While the numerous patients under a single diagnosis.
research indicates that certain psychothera- It is a false and, at times, misleading pre-
pies make better marriages for certain supposition in randomized clinical trials
disorders, psychological therapies will be that the patient sample is homogenous.
increasingly matched to people, not simply Perhaps the patients are diagnostically
diagnoses. homogeneous, but nondiagnostic variabil-
The process of creating the optimal ity is the rule, as every clinician also knows.
match in psychotherapy has been accorded It is precisely the unique individual and the
multiple names: adaptation, responsive- singular context that many psychothera-
ness, attunement, matchmaking, custom- pists attempt to treat.
izing, prescriptionism, treatment selection, Moreover, most practice and EBP guide-
specificity factor, differential therapeutics, lines do little for those psychotherapists
tailoring, treatment fit, and individualizing. whose patients and theoretical conceptual-
By whatever name, the goal is to enhance izations do not fall into discrete disorders
treatment effectiveness by tailoring it to the (Messer, 2001). Consider the client who
individual and his/her singular situation. seeks more joy in his/her life, but who
In other words, psychotherapists endeavor does not meet diagnostic criteria for any
to create a new therapy for each patient. disorder, whose psychotherapy stretches
This position can be easily misunder- beyond 20 sessions, and whose treatment
stood as an authority figure therapist pre- objectives are not easily specified in mea-
scribing a specific form of psychotherapy surable, symptom-based outcomes. Current
for a passive client. Far from it: the goal is evidence-based compilations have little to
for an empathic therapist to arrange for contribute to this kind of treatment (see
an optimal relationship collaboratively O’Donohue, Buchanan, & Fisher, 2000,
with an active client on the basis of the cli- for general characteristics of ESTs).
ent’s personality, culture, and preferences. The upshot of these considerations is
If a client frequently resists, for example, that a truly evidence-based psychotherapy
then the therapist considers whether he or will necessarily consider the person of the
she is pushing something that the client psychotherapist, the therapy relationship,
finds incompatible (preferences), or the and means to adapt or tailor that relation-
client is not ready to make those changes ship to the individual patient—in addi-
(stage of change) or is uncomfortable with tion to diagnosis. Otherwise, evidence-based
a directive style (reactance). practice will prove clinically incomplete as
As every clinician knows, different types well as scientifically suspect.
of patients respond more effectively to dif-
ferent types of treatments and relationships. Effect Sizes
Clinicians strive to offer or select a therapy The second edition of this book endeav-
that accords with the patient’s personal char- ors to systemically appraise the empirical
acteristics, proclivities, and worldviews—in research performed on elements of the
addition to diagnosis. Any differential effec- therapy relationship and means of treatment
tiveness of different therapies may well prove adaptation in order to identify what works.
to be a function of cross-diagnostic patient The subsequent chapters feature original
characteristics, such as patient preferences, meta-analyses on the link between the rela-
coping styles, stages of change, personality tionship elements (Section II) and adapta-
dimensions, and culture. tion methods (Section III) to psychotherapy

10 i n t ro d u c t i o n
outcome. Insisting on meta-analyses for mean r of .30. As shown in Table 1.1, this
all these chapters enables direct estimates is a medium effect size. That translates into
of the magnitude of association in the form happier and healthier clients: patients with
of effect sizes. And conducting these meta- empathic therapists tend to progress more
analytic tests with random effects models in treatment and experience a higher prob-
permits generalization to studies outside ability of eventual improvement.
the samples, although the random effects Consider another example, this one
model is slightly less powerful than the involving the effectiveness of tailoring
fixed effect model (Rosenthal, 1995). therapy. The authors of Chapter 16 con-
The meta-analyses in Section II of the ducted a meta-analysis on 65 experimental
book all employed the weighted r. This and quasi-experimental studies, involving
decision improved the consistency among 8,620 patients, which evaluated the impact
the meta-analyses, enhanced their inter- of culturally adapted treatments versus
pretability among the readers (square r for traditional (nonadapted) treatments. The
the amount of variance accounted for), and resultant d of .46 favored those clients
enabled direct comparisons of the meta- receiving a culturally adapted therapy. As
analytic results to one another as well as to seen in Table 1.1, this effect size also repre-
d (the ES typically used when comparing sents a medium, beneficial effect; incor-
the relative effects of two treatments). In all porating clients’ culture into treatment
of these analyses, the larger the magnitude typically enhances the effectiveness of
of r, the higher the probability of patient psychotherapy.
success in psychotherapy. By convention Given the large number of factors contrib-
(Cohen, 1988), an r of .10 in the behav- uting to such success, and the inherent com-
ioral sciences is considered a small effect, plexity of psychotherapy, we do not expect
.30 a medium effect, and .50 a large effect. large, overpowering effects of any one of its
The meta-analyses presented in Section III facets. Instead, we expect to find a number of
of the book, by contrast, employed the helpful facets. And that is exactly what we
weighted d. That is the common indica- find in the following chapters—beneficial,
tor of a difference between two treatments medium-sized effects of several elements of
or conditions: in this case, the difference the complex therapy relationship.
between the conventional or unadapted
therapy and the adapted therapy. In all of Accounting for Psychotherapy
these analyses, the larger the value of d, Outcome
the higher the effectiveness of the specific What, then, accounts for psychotherapy
adaptation or tailoring. By convention success (and failure)? This question repre-
(Cohen, 1988), a d of .30 in the behavioral sents an understandable desire for clarity
sciences is considered a small effect, .50 a and guidance, but we answer with trepi-
medium effect, and .80 a large effect. dation. Our collective ability to answer in
Table 1.1 presents several concrete ways meaningful ways is limited by the huge
to interpret r and d in health care. For variation in methodological designs, theo-
example, the authors of Chapter 6 con- retical orientations, treatment settings, and
ducted a meta-analysis of 57 studies that patient presentations. Of the dozens of vari-
investigated the link between therapist ables that contribute to patient outcome,
empathy and patient success at the end of only a few can be included in any given
treatment. Their meta-analysis, involving a study. How can we divide the indivisible
total of 3,599 clients, found a weighted complexity of psychotherapy outcome?

n o rc ro s s , l a m b e rt 11
Table 1.1 Interpretation of Effect Size (ES) Statistics
Cohen’s Type of Percentile of Success rate of Number
d r Benchmark effect treated patientsa treated patientsb needed to treatc
1.00 Beneficial 84 72% 2.2
.90 Beneficial 82 70% 2.4
.80 .50 Large Beneficial 79 69% 2.7
.70 Beneficial 76 66% 3.0
.60 Beneficial 73 64% 3.5
.50 .30 Medium Beneficial 69 62% 4.1
.40 Beneficial 66 60% 5.1
.30 Beneficial 62 57% 6.7
.20 .10 Small Beneficial 58 55% 10.0
.10 No effect 54 52% 20.0
.00 0 No effect 50 50%
−.10 No effect 46 48%
–.20 –.10 Detrimental 42 45%
–.30 Detrimental 38 43%
Sources: Adapted from Cohen (1988); Norcross, Hogan, & Koocher (2008); and Wampold (2001)
Each ES can be conceptualized as reflecting a corresponding percentile value: in this case, the percentile standing of the average treated
patient after psychotherapy relative to untreated patients.
Each ES can also be translated into a success rate of treated patients relative to untreated patients; a d of .70, for example, would translate into
approximately 66% of patients being treated successfully compared with 50% of untreated patients.
Number needed to treat (NNT) refers to the number of patients who need to receive the experimental treatment vis-à-vis the comparison to
achieve one success. An effect size of .70 approximates an NNT of 3: three patients need to receive psychotherapy to achieve a success relative
to untreated patients (Wampold, 2001).

Nonetheless, psychotherapy research has based on decades of research, but not for-
made tremendous strides in clarifying the mally derived from meta-analytic methods
question and addressing the uncertainty. (see Lambert & Barley, 2002, for details).
Thus, we tentatively advance two models The patient’s extratherapeutic change—
that account for psychotherapy outcome, self-change, spontaneous remission, social
averaging across thousands of outcome support, fortuitous events—accounts for
studies and hundreds of meta-analyses, and roughly 40% of success. Common factors,
acknowledging that this matter has been variables found in most therapies regard-
vigorously debated for over six decades. We less of theoretical orientation, probably
implore readers to consider the following account for another 30%. The therapy
percentages as crude estimates, not as exact relationship represents the sine qua non
numbers. of common factors, along with client and
The first model estimates the percentage therapist factors. Technique factors, explain-
of explained psychotherapy outcome vari- ing approximately 15% of the variance, are
ance as a function of therapeutic factors. those treatment methods fairly specific
This comparative importance of each of to the prescribed therapy, such as biofeed-
these factors is summarized in Figure 1.1. back, transference interpretations, desensi-
The percentages presented in Figure 1.1 are tization, or two-chair work. Finally, playing

12 i n t ro d u c t i o n
Individual Other factors
therapist 3%
Expectancy Treatment 7%
(placebo effect) method
15% 8%

Common factors
Therapy variance
Extratherapeutic relationship 40%
change 12%

Fig. 1.1 % of Improvement in Psychotherapy Patients Fig. 1.2 % of Total Psychotherapy Outcome Variance
as a Function of Therapeutic Factors. Atrributable to Therapeutic Factors.

an important role is expectancy—the pla- How to improve psychotherapy outcome?

cebo effect, the client’s knowledge that he/ Follow the evidence; follow what contrib-
she is being treated and his/her conviction utes to psychotherapy outcome. Begin by
in the treatment rationale and methods. leveraging the patient’s resources and self-
These four broad factors account for the healing capacities; emphasize the therapy
explained outcome variance. relationship and so-called common factors;
The second model begins with the unex- employ research-supported treatment meth-
plained variance in psychotherapy outcome, ods; select interpersonally skilled and clini-
which necessarily decreases the amount of cally motivated practitioners; and adapt all
variance attributable to the therapeutic fac- of them to the patient’s characteristics, per-
tors. As summarized in Figure 1.2, psycho- sonality, and worldviews. This, not simply
therapy research cannot explain all of the matching a treatment method to a particu-
variation in psychotherapy success. To be lar disorder, will maximize success.
sure, some of this is attributable to measure- The differences between the two models
ment error and fallible research methods, help explain the rampant confusion in
but some is also attributable to the complex- the field regarding the relative percentages
ity of human behavior. Thereafter, we esti- accounted for by relationships and tech-
mate that the patient (including severity of niques. The first model (Figure 1.1) presents
disorder) accounts for approximately 30% only the explained variance and separates
of the total variance, the therapy relation- common factors and specific factors, whereas
ship for 12%, the specific treatment method the second model (Figure 1.2) presents the
for 8%, and the therapist for 7% (when not total variance and assigns common factors
confounded with treatment effects). In this to each of the constituent elements. Hence,
model, we assume that common factors are it is essential to inquire whether the percent-
spread across the therapeutic factors—some ages attributable to particular therapeutic
pertain to the patient, some to the therapy factors are based on total or explained vari-
method, some to the treatment method, ance and how common factors are concep-
and some to the therapist him/herself. tualized in a particular model.

n o rc ro s s , l a m b e rt 13
Despite the differing percentages, both Force may have failed to make necessary
models converge mightily on several take- distinctions.
home points. One: patients contribute the Another lacuna in the Task Force work is
lion’s share of psychotherapy success (and that we may have neglected, relatively
failure). Simply consider the probable speaking, the productive contribution of
outcome of psychotherapy with an adjust- the client to the therapy relationship. We
ment disorder in a healthy person in the decided not to commission a separate chap-
action stage versus a chronically mentally ter on the client’s contributions; instead, we
ill person presenting in precontemplation/ asked the authors of each chapter to address
denial. Two: the therapeutic relationship them. We encouraged authors to pay atten-
generally accounts for as much psychother- tion to the chain of events among the ther-
apy success as the treatment method. Three: apist’s contributions, the patient processes,
particular treatment methods do matter in and eventual treatment outcomes. This, we
some cases, especially with severe anxiety hoped, would maintain the focus on what
disorders treated via systematic exposure is effective in patient change. Further, all of
(Lambert & Ogles, 2004). Four: Adapting the chapters in Section III examine patient
or customizing therapy to the patient contributions directly in terms of specific
enhances the effectiveness of psychotherapy patient characteristics. Nonetheless, by
probably by innervating multiple path- omitting separate chapters on the client, we
ways—the patient, the relationship, the may be understandably accused of an omis-
method, and the expectancy. Fifth: psycho- sion akin to the error of leaving the rela-
therapists need to consider multiple factors tionship out at the expense of method. This
and their optimal combinations, not only book may be “therapist centric” in minimiz-
one or two of their favorites. ing the client’s relational contribution and
self-healing processes.
Limitations of the Task Force Another prominent limitation across
A single task force can accomplish only so these research reviews is the difficulty of
much work and cover only so much con- establishing causal connections between the
tent. As such, we wish to acknowledge sev- relationship behavior and treatment out-
eral necessary omissions and unfortunate come. The only meta-analysis in Section II
truncations in our work. that contains randomized clinical trials
The products of the Task Force proba- (RCTs) capable of demonstrating a causal
bly suffer from content overlap. We may effect is collecting client feedback. (Note
have cut the “diamond” of the therapy that most of the meta-analyses in Section
relationship too thin at times, leading to III were conducted on RCTs and are capa-
a profusion of highly related and possi- ble of causal conclusions.) Causal infer-
bly redundant constructs. Goal consensus, ences are always difficult to make concerning
for example, correlates highly with parts process variables, such as the therapy rela-
of the therapeutic alliance, but these are tionship. Does the relationship cause
reviewed in separate chapters. Collecting improvement or simply reflect it? The inter-
client feedback and repairing alliance rup- pretation problems of correlational studies
tures, for another example, may represent (third variables, reverse causation) render
different sides of the same therapist behav- such studies less convincing than RCTs.
ior, but these too are covered in separate It is methodologically difficult to meet the
meta-analyses. Thus, to some, the content three conditions to make a causal claim:
may appear swollen; to others, the Task nonspuriousness, covariation between the
14 i n t ro d u c t i o n
process variable and the outcome measure, collaborative, and supportive therapist
and temporal precedence of the process versus a nonempathic, authoritarian, disre-
variable (Feeley, DeRubeis, & Gelfand, spectful, and unsupportive therapist.
1999). We still need to determine whether A final interesting drawback to the pres-
and when the therapeutic relationship is a ent work, and psychotherapy research as a
mediator, moderator, or mechanism of whole, is the paucity of attention paid to
change in psychotherapy (Kazdin, 2007). the disorder-specific and treatment-specific
At the same time as we acknowledge this nature of the therapy relationship. It is pre-
central limitation, let’s remain mindful of mature to aggregate the research on how
several considerations. First, the establish- the patient’s primary disorder or the type
ment of temporal ordering is essential for of treatment impacts the therapy relation-
causal inference, but it is not sufficient. ship, but there are early links. For exam-
In showing that these facets of a therapy ple, in the National Institute on Drug
relationship precede positive treatment Abuse Collaborative Cocaine Treatment
outcome, we can certainly state that the Study, higher levels of the working alliance
therapy relationship is, at a minimum, were associated with increased retention in
an important predictor and antecedent of supportive-expressive therapy, but in cog-
that outcome. Second, within these reality nitive therapy, higher levels of alliance were
constraints, dozens of lagged correlational, associated with decreased retention (Barber
unconfounded regression, structural equa- et al., 2001). In the treatment of severe
tion, and growth curve studies suggest that anxiety disorders, the specific treatments
the therapy relationship probably causally seem to exert a larger effect than the ther-
contributes to outcome (e.g., Barber et al., apy relationship, but in depression, the
2000). For example, using growth curve relationship appears more powerful. The
analyses and controlling for prior improve- therapeutic alliance in the NIMH Treat-
ment and eight prognostically relevant client ment of Depression Collaborative Research
characteristics, Klein and colleagues (2003) Program, in both psychotherapy and phar-
found that the early alliance significantly macotherapy, emerged as the leading force
predicted later improvement in 367 chron- in reducing a patient’s depression (Krupnick
ically depressed clients. Although we need et al., 1996). The therapeutic relationship
to continue to parse out the causal linkages, probably exhibits more impact in some dis-
the therapy relationship has probably been orders and in some therapies than others
shown to exercise a causal association to (Beckner, Vella, Howard, & Mohr, 2007).
outcome. Third, some of the most precious As with research on specific treatments,
behaviors in life are incapable on ethical it may no longer suffice to ask “Does the
grounds of random assignment and experi- relationship work?” but “How does the
mental manipulation. Take parental love as relationship work for this disorder and this
an exemplar. Not a single randomized clini- treatment?”
cal trial has ever been conducted to conclu-
sively determine the causal benefit of a Frequently Asked Questions
parental love on children’s functioning, The interdivisional Task Force on Evidence-
yet virtually all humans aspire to it and Based Therapy Relationships has generated
practice it. Nor can we envision an institu- considerable enthusiasm in the professional
tional review board (IRB) ever approving community, but it has also provoked mis-
a grant proposal to randomize patients in understandings and reservations. Here we
a psychotherapy study to an empathic, address frequently asked questions (FAQs)

n o rc ro s s , l a m b e rt 15
about the Task Force’s objectives and convey the impression that it is the only
results. area of importance. This is certainly not
♦ What is the relationship of this task force our intention. Relationship factors are
to the Division 12 Task Force on Research- important, and we need to review the sci-
Supported Treatments (now the standing entific literature and provide clinical and
Committee on Science and Practice)? training recommendations based upon that
Questions abound regarding the con- literature. This can be done without trivial-
nection of the task forces, probably because izing or degrading the effects of specific
they are both associated with the same treatments.
division of the American Psychological ♦ Isn’t your report just warmed-over Carl
Association. Organizationally, the Task Rogers?
Forces are separate creatures. Their respec- No. While Rogers’ (1957) facilitative
tive foci obviously diverge: one looking at conditions are represented in this book,
therapist contributions to the relationship they comprise only about 15% of the
and patient responsiveness, the other look- research critically reviewed. More funda-
ing at treatment methods for specific disor- mentally, we have moved beyond a limited
ders. However, both task forces share the and invariant set of necessary relationship
same book publisher (Oxford University conditions. Monolithic theories of change
Press) and overarching goals (to identify and one-size-fits-all therapy relationships
and promulgate evidence-based practices). are out; adapting the therapy to the unique
♦ Are you saying that treatment methods patient is in.
are immaterial to psychotherapy outcome? ♦ An interpersonal view of psychotherapy
Absolutely not. The empirical research seems at odds with what managed care and
shows that both the therapy relationship bean counters ask of me in my clinical prac-
and the treatment method make frequent tice. How do you reconcile these?
contributions to treatment outcome. It It is true that a dominant image of modern
remains a matter of judgment and method- psychotherapy, among both researchers
ology on how much each contributes, but and reimbursers, is as a mental health treat-
there is virtual unanimity that both the ment. This “treatment” or “medical” model
relationship and the method (insofar as inclines people to define process in terms of
we can separate them) “work.” Looking method, therapists as providers trained in
at either treatment methods or therapy rela- the application of techniques, treatment in
tionships alone is incomplete. We encour- terms of number of contact hours, patients
age practitioners and researchers to look at as embodiments of psychiatric disorders,
multiple determinants of outcome, partic- and outcome as the end result of a treat-
ularly client contributions. ment episode (Orlinsky, 1989).
♦ But are you not exaggerating the effects It is also true that the Task Force mem-
of relationship factors and/or minimizing bers believe this model to be restricted and
the effects of treatments in order to set up the inaccurate. The psychotherapy enterprise is
importance of your work? far more complex and interactive than the
We think not and hope not. With linear “Treatment operates on patients to
the guidance of Task Force members and produce effects” (Bohart & Tallman, 1999).
external consultants, we have tried to avoid We would prefer a broader, integrative
dichotomies and polarizations. Focusing model that incorporates the relational and
on one area—the psychotherapy relation- educational features of psychotherapy, one
ship—in this volume may unfortunately that recognizes both the interpersonal and
16 i n t ro d u c t i o n
instrumental components of psychother- capable of more malleability and “mood
apy, one that appreciates the bidirectional transcendence” than might be expected. In
process of therapy, and one in which the Gurman’s (1973) research, for example,
therapist and patient cocreate an optimal expert therapists appeared to be less handi-
process and outcome. capped by their own “bad moods” than
♦ Won’t these results contribute fur- were their less skilled peers. From the
ther to deprofessionalizing psychotherapy? literature on the cognitive psychology of
Aren’t you unwittingly supporting efforts expertise (Schacht, 1991), experienced
to have any warm, empathic person perform psychotherapists are disciplined improvisa-
psychotherapy? tionalists who have stronger self-regulating
Perhaps some will misuse our conclu- skills and more flexible repertoires than
sions in this way, but that is neither our novices. The research on the therapist’s
intent nor commensurate with our meta- level of experience suggests that experience
analyses. It trivializes psychotherapy to begets heightened attention to the client
characterize it as simply “a good relation- (less self-preoccupation), an innovative
ship with a caring person.” The research perspective, and in general, more endorse-
shows that an effective psychotherapist is ment of an “eclectic” orientation predi-
one who employs specific methods, who cated on client need (Auerbach & Johnson,
offers strong relationships, and who cus- 1977). Indeed, several research studies (see
tomizes both treatment methods and rela- Beutler, Machado, & Neufeldt, 1994) have
tionship stances to the individual person demonstrated that therapists can consis-
and condition. That requires considerable tently use different treatment models in a
training and experience, the antithesis of discriminative fashion.
“anyone can do psychotherapy.” Thus, our clinical experience and the
♦ Are psychotherapists really able to adapt modest amount of research attest that sea-
their relational style to fit the proclivities and soned practitioners can shift back and forth
personalities of their patients? Where is the among different relationship styles for a
evidence we can do this? given case. Whether inexperienced psycho-
Relational flexibility conjures up many therapists can do so is still unanswered.
concerns, but two of particular import in And we caution therapists to be careful that
this question: the limits of human capacity the blending of stances and strategies never
and the possibility of capricious posturing deteriorates into playacting or capricious
(Norcross & Beutler, 1997). Although posturing.
the psychotherapist can, with training ♦ What should we do if we are unable or
and experience, learn to relate in a number unwilling to adapt our therapy to the patient
of different ways, there are limits to our in the manner that research indicates is likely
human capacity to modify relationship to enhance psychotherapy outcome?
stances. It may be difficult to change inter- Five avenues spring to mind. First,
action styles from client to client and ses- address the matter forthrightly with the
sion to session, assuming one is both aware patient as part of the evolving therapeutic
and in control of one’s styles of relating contract and the creation of respective
(Lazarus, 1993) tasks, in much the same way one would
Can psychotherapists authentically differ with patients requesting a form of therapy
from their preferred or habitual style of or a type of medication that research has
relating? There is some research supporting indicated would fit particularly well in their
this assertion. Experienced therapists are case but which is not in your repertoire.

n o rc ro s s , l a m b e rt 17
Second, treatment decisions are the result therapy relationship and the particular
of multiple, interacting, and recursive means of adapting it to individual patients.
considerations on the part of the patient, It is premature to proffer the last word, but
the therapist, and the context. A single it is time to codify and disseminate what
evidence-based guideline should be seri- we do know. We look forward to regular
ously considered, but only as one of many updates on our research conclusions and
determinants of treatment itself. Third, practice recommendations.
formal tracking of patient functioning ♦ So, are you saying that the therapy rela-
during the course of psychotherapy pro- tionship (in addition to the treatment method)
vides a systematic way of assessing the is crucial to outcome, that it can be improved
consequences of treatment as it unfolds. by certain therapist contributions, and that it
Determine if, in this particular case, the can be effectively tailored to the individual
treatment is helping. Fourth, an alternative patient?
to the one-therapist-fits-most-patients per- Precisely. And this book, on the basis
spective is practice limits. Without a will- of the empirical research, suggests impor-
ingness and ability to engage in a range of tant directions for practitioners, trainers,
interpersonal stances, the therapist may researchers, and policymakers.
limit his or her practice to clients who fit
that practice. Psychotherapists need not Concluding Reflections
offer all services to all patients. Finally, con- The future of psychotherapy portends the
sider a judicious referral to a colleague who integration of science and service, of the
can offer the relationship stance (or treat- instrumental and the interpersonal, of
ment method or medication) indicated for the technical and the relational in the tradi-
a particular patient. tion of evidence-based practice (Norcross,
♦ Are the Task Force’s conclusions and rec- Freedheim, & VandenBos, 2011). Evidence-
ommendations intended as practice standards? based therapy relationships align with this
No. These are research-based conclu- future and embody a crucial part of
sions that can lead, inform, and guide prac- evidence-based practice, when properly
titioners toward evidence-based therapy conceptualized. We can imagine few prac-
relationships and responsiveness to patient tices in all of psychotherapy that can confi-
needs. They are not legal, ethical, or profes- dently boast that they integrate “the best
sional mandates. available research with clinical expertise in
♦ Well, don’t these represent the offi- the context of patient characteristics, culture,
cial positions of Division 12 (Clinical), and preferences” (American Psychological
Division 29 (Psychotherapy), or the American Association Task Force on Evidence-Based
Psychological Association? Practice, 2006) as well as the relational
No. No. No. behaviors and treatment adaptations pre-
♦ Isn’t it premature to launch a set of sented in this book. We are reminded daily
research-based conclusions on the therapy that research can guide how we create, culti-
relationship and patient matching? vate, and customize that powerful human
Science is not a set of answers; science is relationship.
a series of processes and steps by which we Moreover, we fervently hope this book
arrive closer and closer to elusive answers. will indirectly serve another master: to heal
A vast amount of sophisticated research the damage incurred by the culture wars in
over the past five decades has been con- psychotherapy. If our Task Force is even a
ducted on both the general elements of the little bit successful, then the pervasive gap
18 i n t ro d u c t i o n
between the science and practice commu- Barlow, D. H. (Ed.). (2007). Clinical handbook of
nities will be narrowed, and the insidious psychological disorders: A step-by-step treatment
manual (4th ed.). New York: Guilford.
dichotomy between the therapy relation-
Beckner, V., Vella, L., Howard, I., & Mohr, D. C.
ship and the treatment method will be (2007). Alliance in two telephone-administered
lessened. Phrased more positively, psycho- treatments: Relationship with depression and
therapists from all camps and communities health outcomes. Journal of Consulting and
will increasingly collaborate, and our patients Clinical Psychology, 75, 508–512.
will benefit from the most efficacious treat- Bergin, A. E. (1997). Neglect of the therapist and the
human dimensions of change: A commentary.
ments and relationships available.
Clinical Psychology: Science and Practice, 4, 83–89.
Bergin, A. E., & Lambert, M. J. (1978). The
References evaluation of outcomes in psychotherapy. In
American Psychiatric Association. (2006). Practice S. L. Garfield & A. E. Bergin (Eds.), Handbook
guidelines for the treatment of psychiatric dis- of psychotherapy and behavior change (pp.
orders: Compendium 2006. Washington, DC: 139–189). New York: Wiley.
Author. Beutler, L. E. (2000). David and Goliath: When
American Psychological Association Task Force on empirical and clinical standards of practice meet.
Evidence-Based Practice. (2006). Evidence-based American Psychologist, 55, 997–1007.
practice in psychology. American Psychologist, 61, Beutler, L.E., Machado, P. P. P., & Neufeldt, S.A.
271–285. (1994). Therapist variables. In A. E. Bergin &
Anderson, T., Ogles, B. M., Patterson, C. L., S. L. Garfield (Eds.), Handbook of psychotherapy
Lambert, M. J., & Vermeersch, D. A. (2009). and behavior change (4th ed., pp. 229–269).
Therapist effects: Facilitative interpersonal skills New York: Wiley.
as a predictor of therapist success. Journal of Bohart, A. C., & Tallman, K. (1999). How clients
Clinical Psychology, 65, 755–768. make therapy work: The process of active self-
Auerbach, A. H., & Johnson, M. (1977). In healing. Washington, DC: American Psycho-
A. S. Gurman & A. M. Razin (Eds.), Effective logical Association.
psychotherapy: A handbook of research. New York: Cohen, J. (1988). Statistical power analysis for
Pergamon. the behavioral sciences (2nd ed.). Hillsdale, NJ:
Barber, J. P., Connolly, M. B., Crits-Christoph, P., Erlbaum.
Gladis, L., & Siqueland, L. (2000). Alliance Crits-Christoph, P., Baranackie, K., Kurcias, J. S.,
predicts patients’ outcomes beyond in-treatment Beck, A. T., Carroll, K., Perry, K., et al. (1991).
change in symptoms. Journal of Consulting and Meta-analysis of therapist effects in psycho-
Clinical Psychology, 68, 1027–1032. therapy outcome studies. Psychotherapy Research,
Barber, J. P., Gallop, R., Crits-Christoph, P., Frank, A., 1, 281–91.
Thase, M. E., Weiss, R. D., et al. (2006). The Department of Health. (2001). Treatment choice in
role of therapist adherence, therapist compe- psychological therapies and counseling. London:
tence, and alliance in predicting outcome of Department of Health Publications.
individual drug counseling: Results from the Duncan, B. L., Miller, S. D., Wampold, B. E., &
National Institute on Drug Abuse Collaborative Hubble, M. A. (Eds.). (2010). The heart and soul
Cocaine Treatment Study. Psychotherapy Research, of change (2nd ed.). Washington, DC: American
16, 229–240. Psychological Association.
Barber, J. P., Luborsky, L., Gallop, R., Feeley, M., DeRubeis, R. J., & Gelfand, L. A.
Crits-Christoph, P., Frank, A., Weiss, R. D., (1999). The temporal relation of adherence and
et al. (2001). Therapeutic alliance as a predictor alliance to symptom change in cognitive therapy
of outcome and retention in the National for depression. Journal of Consulting and Clinical
Institute on Drug Abuse Collaborative Cocaine Psychology, 67, 578–582.
Treatment Study. Journal of Consulting & Clinical Gelso, C. J. (Ed.). (2005). The interplay of tech-
Psychology, 69, 119–124. niques and the therapeutic relationship in psych-
Barlow, D. H. (2000). Evidence-based practice: otherapy. Psychotherapy, 42(4), whole.
A world view. Clinical Psychology: Science and Gelso, C. J., & Carter, J. A. (1985). The relationship
Practice, 7, 241–242. in counseling and psychotherapy: Components,

n o rc ro s s , l a m b e rt 19
consequences, and theoretical antecedents. The Lambert, M. J., & Barley, D. E. (2002). Research
Counseling Psychologist, 13, 155–243. summary on the therapeutic relationship and
Gelso, C. J., & Carter, J. A. (1994). Components psychotherapy outcome. In J. C. Norcross (Ed.),
of the psychotherapy relationship: Their interac- Psychotherapy relationships that work (pp. 17–32).
tion and unfolding during treatment. Journal of New York: Oxford University Press.
Counseling Psychology, 41, 296–306. Lambert, M. J., & Ogles, B. M. (2004). The
Gelso, C. J., & Hayes, J. A. (1998). The psycho- efficacy and effectiveness of psychotherapy. In
therapy research: Theory, research, and practice. M. J. Lambert (Ed.), Begin and Garfield’s hand-
New York: Wiley. book of psychotherapy and behavior change (5th
Gurman, A. S. (1973). Effects of therapist and ed., pp. 139 – 193). New York: Wiley.
patient mood on the therapeutic functioning of Lazarus, A. A. (1993). Tailoring the therapeutic
high- and low-facilitative therapists. Journal of relationship, or being an authentic chameleon.
Consulting and Clinical Psychology, 40, 48–58. Psychotherapy, 30, 404–407.
Henry, W. P. (1998). Science, politics, and the Luborsky, L., Diguer, L., Seligman, D. A.,
politics of science: The use and misuse of empir- Rosenthal, R., Krause, E.D., Johnson, S., et al.
ically validated treatment research. Psychotherapy (1999). The researcher’s own therapy allegiances:
Research, 8, 126–140. A “wild card” in comparisons of treatment
Huppert, J. D., Bufka, L. F., Barlow, D. H., efficacy. Clinical Psychology: Science and Practice,
Gorman, J. M., Shear, M. K., & Woods, S. W. 6, 95–106.
(2001). Therapists, therapist variables, and cog- Messer, S. B. (2001). Empirically supported treat-
nitive–behavioral therapy outcome in a multi- ments: What’s a nonbehaviorist to do?. In
center trial for panic disorder. Journal of B. D. Slife, R. N. Williams, & S. H. Barlow
Consulting and Clinical Psychology, 69, 747–755. (Eds.), Critical issues in psychotherapy. Thousand
Kazdin, A. E. (2007). Mediators and mechanism Oaks, CA: Sage.
of change in psychotherapy research. In Nathan, P. E. (1998). Practice guidelines: Not yet
S. Nolen-Hoeksema, T. D. Cannon, & ideal. American Psychologist, 53, 290–299.
T. Widiger (Eds.), Annual Review of Clinical Nathan, P. E., & Gorman, J. M. (Eds.). (2007).
Psychology, 3, 1–27. A guide to treatments that work (3rd ed.).
Klein, D. N., Schwartz, J. E., Santiago, N. J., New York: Oxford University Press.
Vivian, D., Vocisano, C., Castonguay, L. G., Norcross, J. C. (Ed.). (2002). Psychotherapy
et al. (2003). Therapeutic alliance in depression relationships that work: Therapist contributions
treatment: Controlling for prior change and and responsiveness to patient needs. New York:
patient characteristics. Journal of Consulting and Oxford University Press.
Clinical Psychology, 71, 997–1006. Norcross, J. C. (2010). The therapeutic relationship.
Krupnick, J. L., Sotsky, S. M., Simmens, S., In B. L. Duncan, S. D. Miller, B. E. Wampold,
Moyer, J., Elkin, I., Watkins, J., et al. (1996). & M. A. Hubble (Eds.), Heart & soul of
The role of the therapeutic alliance in psycho- change in psychotherapy (2nd ed.). Washington,
therapy and pharmacotherapy. Journal of DC: American Psychological Association.
Consulting and Clinical Psychology, 64, 532–539. Norcross, J. C., & Beutler, L. E. (1997). Determining
Lambert, M. J. (1992). Psychotherapy outcome the therapeutic relationship of choice in
research: Implications for integrative and eclectic brief therapy. In J. N. Butcher (Ed.), Personality
therapists. In J. C. Norcross and M. R. Goldfried assessment in managed health care: A practi-
(Eds.), Handbook of psychotherapy integration tioner’s guide. New York: Oxford University
(pp. 94–129). New York: Basic Books. Press.
Lambert, M. J. (2010). Prevention of treatment Norcross, J. C., Beutler, L. E., & Levant, R. F.
failure: The use of measuring, monitoring, & (Eds.). (2006). Evidence-based practices in
feedback in clinical practice. Washington, DC: mental health: Debate and dialogue on the funda-
American Psychological Association. mental questions. Washington, DC: American
Lambert, M. J. (2011). Psychotherapy research Psychological Association.
and its achievements. In J. C. Norcross, Norcross, J. C., Hogan, T. P., & Koocher, G. P.
G. R. VandenBos, & D. K. Freedheim (Eds.), (2008). Clinician’s guide to evidence-based prac-
History of psychotherapy (2nd ed.). Washington, tices: Mental health and the addictions. New York:
DC: American Psychological Association. Oxford University Press.

20 i n t ro d u c t i o n
Norcross, J. C., Freedheim, D. K., & Rector, N. A., Zuroff, D. C., & Segal, Z. V. (1999).
VandenBos, G. R. (2011). Into the future: Cognitive change and the therapeutic alliance:
Retrospect and prospect in psychotherapy. The role of technical and nontechnical factors in
In J. C. Norcross, G. R. VandenBos, & cognitive therapy. Psychotherapy, 36, 320–328.
D. K. Freedheim (Eds.). (2011). History of Rogers, C. R. (1957). The necessary and sufficient
psychotherapy (2nd ed.). Washington, DC: conditions of therapeutic personality change.
American Psychological Association. Journal of Consulting Psychology, 22, 95–103.
O’Donohue, W., Buchanan, J. A., & Fisher, J. E. Rosenthal, R. (1995). Writing meta-analytic
(2000). Characteristics of empirically supported reviews. Psychological Bulletin, 118, 183–192.
treatments. Journal of Psychotherapy Practice and Roth, A., & Fonagy, P. (2004). What works for
Research, 9, 69–74. whom? A critical review of psychotherapy research
Okiishi J., Lambert, M. J., Nielsen, S. L., & (2nd ed.). New York: Guilford.
Ogles, B. M. (2003). In search of supershrink: Safran, J. D., & Muran, J. C. (2000). Negotiating
Using patient outcome to identify effective the therapeutic alliance. New York: Guilford.
and ineffective therapists. Clinical Psychology and Schacht, T. E. (1991). Can psychotherapy educa-
Psychotherapy, 10, 361–373. tion advance psychotherapy integration? A view
Orlinsky, D. E. (1989). Researchers’ images of from the cognitive psychology of expertise.
psychotherapy: Their origins and influence Journal of Psychotherapy Integration, 1, 305–320.
on research. Clinical Psychology Review, 9, Task Force on Psychological Intervention Guidelines.
413–441. (1995). Template for developing guidelines:
Orlinsky, D. E. (2000, August). Therapist interper- Interventions for mental disorders and psychosocial
sonal behaviors that have consistently shown posi- aspects of physical disorders. Washington, DC:
tive correlations with outcome. Paper presented at American Psychological Association.
the 108th annual convention of the American Wampold, B. E. (2001). The great psychother-
Psychological Association, Washington, DC. apy debate: Models, methods, and findings.
Orlinsky, D., & Howard, K. E. (1977). The Mahwah, NJ: Lawrence Erlbaum.
therapist’s experience of psychotherapy. In Wampold, B. E., & Brown, G. S. (2005). Estimating
A. S. Gurman & A. M. Razin (Eds.), Effective variability in outcomes attributable to therapists:
psychotherapy: A handbook of research. New York: A naturalistic study of outcomes in managed
Pergamon. care. Journal of Consulting and Clinical Psychology,
Paul, G. L. (1967). Strategy of outcome research in 73, 914–923.
psychotherapy. Journal of Consulting Psychology, Wolitzky, D. L. (2011). Psychoanalytic theo-
31, 109–118. ries of psychotherapy. In J. C. Norcross,
Project MATCH Research Group. (1998). Therapist G. R. VandenBos, & D. K. Freedheim (Eds.),
effects in three treatments for alcohol problems. History of psychotherapy (2nd ed.). Washington,
Psychotherapy Research, 8, 455–474. DC: American Psychological Association.

n o rc ro s s , l a m b e rt 21
This page intentionally left blank
Effective Elements
of the Therapy
Relationship: What
Works in General
This page intentionally left blank

2 Alliance in Individual Psychotherapy

Adam O. Horvath, AC Del Re, Christoph Flückiger, and Dianne Symonds

Since our last review of the literature was the work of Carl Rogers and his col-
in 2002, research on the alliance in psy- leagues. By applying rigorous empirical
chotherapy has continued to flourish. By methods to the examination of person-
searching the electronic databases at the centered treatment, they not only proved
end of 2000, we located just over 2,000 ref- that the therapy process can be explored
erences using the keywords alliance, helping beyond anecdotal records, but also moved
alliance, working alliance, and therapeutic the concept of the therapeutic relationship
alliance. The same search in early 2010 to the center of the healing process. Rogers
yielded over 7,000 items. The growing attrac- and colleagues generated an important
tion of the alliance concept appears to be body of literature exploring the interper-
the result of a number of related sources: sonal interior of psychotherapy (Rogers,
One reason is the convergence of evi- Gendlin, Kiesler, & Truax, 1967).
dence, staring in the ’70s, that different A third important precursor can be traced
psychotherapies typically produce similar back to the 1930s: A growing curiosity and
beneficial effect for clients (e.g., Luborsky, interest in the integration of diverse theo-
Singer, & Luborsky, 1975; Smith & Glass, ries of psychotherapies (Frank & Frank,
1977; Stiles, Shapiro, & Elliot, 1986). 1991; Rosenzweig, 1936). The desire to
Although the “Dodo bird interpretation” reconcile some conflicting therapeutic
(All have won and all must have their methods and their underlying theories
prizes. . .) of these meta-analyses of psy- eventually led to the founding of the Society
chotherapy effectiveness has proven some- for the Exploration of Psychotherapy
what controversial (Chambless, 2002), Integration (SEPI) in 1983. On the prac-
most therapists and researchers alike have tice side, psychotherapists in North America
accepted the notion that a large part of started to reject the strict boundaries of
what is helpful for clients receiving psycho- classical theories and became increasingly
therapy is shared across diverse treatments. interested in utilizing a variety of effective
The quality of the therapeutic relationship methods irrespective of their “school;” the
in general, and the alliance in particular, field was moving from theoretical monism
are obvious “common factors” shared by to an eclectic pragmatism. The value of
most if not all psychotherapies. aspects of therapist–client relatedness (e.g.,
Another important precursor of the alli- alliance) found ready acceptance among
ance concept, and a pioneering force in the those committed to psychotherapy integra-
development of therapy process research, tion (Goldfried, 1980).

But perhaps the most potent force respon- (1910/1913). His basic premise was that
sible for the sustained growth of interest in all relationships were transference based
the alliance was the consistent finding of a (Freud, 1912/1958). Early in his writings,
moderate but robust relationship between he struggled with the question of what keeps
the alliance and treatment outcome across a the analysand in therapy in the face of the
broad spectrum of treatments in a variety of psyche’s unconscious fear and rejection
client/problem contexts (Horvath & Bedi, of exploring repressed material. His first
2002; Horvath & Symonds, 1991; Martin, formulation suggested that he thought that
Ganske, & Davis, 2000). there was an “analyst” within the patient
In this chapter, we reexamine the empir- supporting the healing journey (Freud,
ical evidence linking the alliance to out- 1912/1958). Later he speculated about the
come in individual psychotherapy with reality-based collaboration between therapist
adults. But the relation between alliance and client, a conjoint effort to conquer the
and therapy is only the first level of interest. client’s pain. He also referred to this process
Beyond the strength of the overall alliance– as the unobjectionable or positive transfer-
outcome link, it was our intent to use the ence (Freud, 1913/1940). Both the wisdom
accumulated data to examine the role of of recognizing the client’s attachment to
several potential moderators and mediators the therapist, and his ambiguity about the
that impact this relationship, with particu- status of this attachment (reality based and
lar attention to issues that help us better conscious versus transferential and uncon-
understand the way alliance and treatment scious) has echoed throughout the evolution
results are linked. of the concept.
The term ego alliance was coined by
Definitions and Measures Sterba (1934), who conceptualized it as
The term alliance (also therapeutic alliance, part of the client’s ego-observing process
working alliance, and helping alliance) as it is that alternated with the experiencing (trans-
used in the research literature, can refer to ferential) process. Zetzel (1956) used the
a number of related constructs; at this term therapeutic alliance to refer to the
time we do not have a single, consensually patient’s ability to use the healthy part of
accepted definition of the concept (Horvath her/his ego to link up or join with the
& Luborsky, 1993; Saketopoulou, 1999). analyst to accomplish the therapeutic tasks.
While there are important shared aspects in Greenson (1965, 1967) made a distinction
the way researchers use the construct in the between the working alliance, the client’s
literature (e.g., Bordin, 1980, 1989; Gaston ability to align with the tasks of analysis,
et al., 1995; Hatcher & Barends, 2006; and the therapeutic alliance, which refers to
Horvath & Luborsky, 1993), there are also the capacity of therapist and client to form
nontrivial differences among authors about a personal bond.
the precise meaning of the term (Psycho- During the 1970s efforts were made to
therapy, 43(3), whole). The best way to grasp extrapolate and extend the concept of the
the complexity of the current status of this alliance from its psychodynamic roots to
concept is by briefly reviewing its history. encompass components of the relational
elements of all helping endeavors: Luborsky
Definitions (1976) proposed an extension of Zetzel’s
The concept of the alliance (though not (1956) and Stone’s (1961) concept. He sug-
the term itself ) originated with Freud gested that the alliance between therapist

26 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
and client developed in two phases: The build and ebb in the normal course of
first phase, Type I alliance, involved the events, and that the repair of these stresses
client’s belief in the therapist as a potent in the alliance would constitute the core
source of help, and the therapist providing task of any helping relationship.
a warm, supporting, and caring relation- The most distinguishing feature of the
ship. This level of alliance results in a secure modern pantheoretical reconceptualization
holding relationship within which the work of the alliance is its emphasis on collabora-
of the therapy can begin. The second phase, tion and consensus (Bordin, 1980; Hatcher,
Type II alliance, involved the client’s invest- Barends, Hansel, & Gutfreund, 1995;
ment and faith in the therapeutic process Luborsky, 1976). In contrast to previous
itself, a commitment to the core concepts conceptualizations that emphasized either
undergirding the therapy (e.g., nature of the therapist’s contributions to the relation-
the problem, value of the exploratory pro- ship (i.e., Rogers & Wood, 1974) or the
cess), as well as a willing investment of her unconscious distortions of the relation
or himself, to share the ownership for the between therapist and client, the revised
therapy process. While Luborsky’s (1976, alliance theory emphasized the active col-
1994) assumptions about the therapy pro- laboration between the participants.
cess itself were grounded in psychody- An equally significant consequence of
namic theory, his description of the alliance the way the alliance concept was reintro-
as a therapeutic process was quite general. duced is that there were two different voices
Luborsky and his team also pioneered theorizing about the concept. Each want-
an alliance assessment method for raters, ing to separate the idea from its long his-
using transcripts or audio recordings, to tory within the psychodynamic framework
count signs of in-session events indicative and operationalize the concept in a way
of the presence of either type of alliance. in which it would be compatible with most,
Bordin (1975, 1976, 1989, 1994) pro- if not all, theoretical approaches. But nei-
posed a somewhat different pan-theoretical ther of these theorists (Bordin or Luborsky)
alliance concept he called the working alli- offered a precise definition of how this
ance. His concepts of the alliance were new conceptualization of the alliance
based on Greenson’s’ (1965) ideas as a start- related to (or was different from) other
ing point but departed from the psychody- concepts that are parts of the therapeutic
namic premises even more clearly than relationship. This theoretical ambiguity
Luborsky did. For Bordin, the alliance was created a void which was filled by a number
centrally the achievement of collaborative of alliance measures developed in para-
stance in therapy and was built on three llel between 1978 and 1986. What we
components: agreements on the therapeu- know about the alliance and its relation
tic goals, consensus on the tasks that make to outcome and other therapy variables
up therapy, and a bond between the client has been gleaned from studies that, in prac-
and the therapist. He predicted that differ- tice, define the alliance by the instru-
ent therapies would place different demands ment used to measure it. In this sense, the
on the relationship, thus the “profile” of the instrumentation defines the construct.
ideal working alliance would be different In the following section we review the alli-
across theoretical orientations. Bordin also ance instruments and discuss the differ-
proposed that as therapy progresses, the ences and similarities of their undergirding
strength of the working alliance would conceptualizations.

ho rvat h , re , f lü c k i g e r, s y m o n d s 27
Measures the numbers were similar to those reported
In this chapter we refer to the alliance in for the core measures. In Table 2.1, each
the singular. However, in the database of instrument is identified using the label
201 studies we have assembled for this or identification the authors provided.
meta-analysis, over 30 different alliance However, in the moderator analyses we dis-
measures were used, not counting different cuss later in this chapter, the less often used
versions of the same instrument. Similar to measures (n of use ≤ 3) were merged into one
previous reports, the four core measures: category: “Other.” In this “Other” category
California Psychotherapy Alliance Scale are: some newer alliance measures with rela-
(CALPAS, Gaston & Marmar, 1994), tively few administrations, measures devel-
Helping Alliance Questionnaires (HAq, oped for the specific investigation, and
Alexander & Luborsky, 1987), Vanderbilt instruments originally developed for rela-
Psychotherapy Process Scale (VPPS, tionship constructs other than the alliance.
O’Mally, Suh & Stupp, 1983), and Work- Adding to the diversity of measures is the
ing Alliance Inventory (WAI, Horvath & fact that, over time, the four core instru-
Greenberg, 1986) accounted for approxi- ments have evolved as well and currently
mately two thirds of the data. In examina- exist in a number of different forms (e.g.,
tions of the shared factor structure of the short versions, observer versions, versions
WAI, CALPAS, and HAq, the concept of specific to context and/or application, trans-
“confident collaborative relationship” was lations). The relation of these modified
the central common theme (Hatcher et al., instruments to the original is not always
1995; Hatcher & Barends, 1996). Each of well documented. As we noted, the diver-
these four instruments has been in use for sity in the definition of the alliance
over 20 years and has demonstrated an via the use of a variety of assessment mea-
acceptable levels of internal consistency. sures has become an important issue. The
The methods of reporting reliability of consequences of these differences will be
measures were somewhat inconsistent, but discussed in the section evaluating the inter-
we estimated that clients’ and therapists’ pretation of this corpus of research.
rating of the alliance using these core mea-
sures were in the range of 0.81–0.87
Clinical Examples
(Cronbach’s alpha). Rated (observer) mea-
The alliance represents an emergent quality
sures tended to report interrater reliability
of partnership and mutual collaboration
indexes of similar values. However, the
between therapist and client. As such, it is
shared variance, even among these well-
not the outcome of a particular intervention;
established measures, has been shown to be
its development can take different forms and
less than 50% (Horvath, 2009).
may be achieved almost instantly or nurtured
Fifty four of the research reports in
over a longer period of time depending on
our data set used less-well-validated instru-
the kind of therapy and the stage of treatment
ments or assessment procedures; the rela-
(Bordin, 1994). The following is an excerpt
tion of most of these measures to the core
from an early session that illustrates the
instruments, or to each other, are not well
challenges of negotiating the clients’ whole-
documented, and sometimes nonexistent.
hearted participation in the therapy process:
Relatively little data are available with
respect to their psychometric properties, Client (C): Well aren’t you going to ask
but when this information was provided, me what this reminds me of?

28 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Table 2.1 Research Reports included in the Meta Analysis
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Adler (1988) 12 Various C.T WAI, E, L TC, SCL/BSI, RSE, C, T 0.28 44
Allen et al. ∗ Inpatient T ITAS E, L, A Overall Outcome, T 0.54 37
(1985)a GAS, Outcome
Allen et al. ∗ Inpatient T ITAS E, L Premature C 0.54 37
(1986)a Termination
Andreoli 6 Crisis T ITAS E Overall Outcome, T 0.57 16
et al. (1993)b intervention Interpersonal
Specific Outcomes
Ankuta 6 Crisis T ITAS E Overall Outcome T 0.02 44
(1993) intervention
Arnow et al. 20 CBT C WAI-S E Premature O 0.10 681
(2003) Termination
Baldwin ∗ Various C WAI M OQ-45 C 0.24 331
Barber et al. 20 Various C, T HAq-II, E, M SCL/BSI, Addiction C 0.10 121
(2006) CALPAS Severity Index, BDI
Barber et al. 40 Various C, O HAq-II, E, M SCL/BSI, Addiction C 0.13 83
(1999)a CALPAS Severity Index, BDI
Barber et al. ∗ Dynamic C CALPAS E, M, L BDI C 0.37 83
Barber et al. 40 Various C CALPAS E, M Addiction Severity C 0.08 265
(2001)a Index
Barber et al. 36 Dynamic C HAq, E Addiction Severity C 0.10 89
(2008) CALPAS Index
Barkham 12 Interpersonal O CALPAS E Overall Outcome C 0.41 12
et al. (1993)
Bassler et al. 14 w Various C HAq E Overall Outcome C 0.16 237
Bethea et al. 8 CBT C, T, HAq II E, M, Drug Use, C, O 0.21 25
(2008) O L, A Functioning
Adherence, Pain
Bieschke 7 Various C WAI L Change in Distress C 0.38 90
et al. (1995)
Biscoglio ∗ CBT C, T WAI-S E GAS, IIP, SCL/BSI, C, O 0.21 32
(2005) TC
Botella ∗ Various C WAI-S E Premature O 0.16 190
(2008) Termination
(Continued )

Table 2.1 Continued
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Bredel et al. ∗ Various C NSI E Satisfaction C, O 0.44 78
Broome 46 Drug Counseling, C 3-item M Premature C 0.11 167
(1996)b Methodone NSI Termination
Brotman 16 Various C, O WAI, E HRSD O 0.31 51
(2004) HA(r)
Burns et al. 12 w Rehabilitation C WAI-S E Cardiac Depression C 0.12 79
(2007) Scale, Diet Progress,
Exercise and Diet
General Health
Busseri et al. ∗ Eclectic C, T WAI E, M SCL/BSI, TC C, T 0.36 54
Busseri et al. 8 Eclectic C, T WAI E PTQ, TC, SCL C, T 0.35 50
Card (1991) 6 Cognitive- O CALPAS E, M, L STAI, BDI, HRSD, C, O 0.07 55
behavior SCL/BSI
Castonguay 15 Cognitive, O WAI M BDI, HRSD, GAS C, O 0.57 30
et al. (1996) Medication
Chilly 16 Interpersonal C WAI E BDI C 0.52 9
Cislo (1998) 10 Various C HAq A Session Impact C 0.30 47
Clarkin et al. ∗ Inpatient O ITAS A GAS O 0.39 96
(1987) Psychiatric Unit
Cloitre et al. 16 Various C WAI-S E Premature C, O 0.27 30
(2004) Termination, PTSD
Coleman ∗ Eclectic C WAI-S C SCL/BSI, SWLS 0.12 102
Connors 12 w Various C, T WAI E DpD, Abstinence C 0.11 579
et al. (1997)a
Constantino 19 CBT, C HAq E, M Purge Frequency 0.29 75
et al. (2005)b Interpersonal
Crits- 54 Dynamic O HAq(cs) E Composite C, T, O 0.39 43
Cristoph Outcome, Residual
et al. (1988) Gain
Davis et al. 26 w CBT O WAI-S A PANSS, WBI 0.43 26
de Roten 4 Dynamic C HAq M, E, SCL/BSI, C 0.45 70
et al. (2004) A Evaluation
Questionnaire, SAS
(Continued )

Table 2.1 Continued
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Dearing 12 CBT C WAI E DpD, Abstinence, C 0.29 208
et al. (2005) Drinking Related
Satisfaction with
Deu et al. 10 w Interpersonal C HAq E Depressive O 0.18 17
(2009) Symptoms
Dorsch et al. ∗ Various C HAq II E ACQ, BDI, BSQ, C 0.61 30
(2002) Clinical
Improvement, SCL/
Dundon ∗ Various C, T WAI E Abstinence, Sessions O 0.08 194
et al. (2008) Attended
Dunn et al. 18 w CBT C CALPAS E PANSS O -0.11 29
Eaton et al. ∗ Various O TARS A Overall Outcome, C, T 0.00 40
(1988) SCL/BSI
Emmerling ∗ Eclectic C WAI-S E GHQ C 0.42 56
et al. (2009)
Fakhoury ∗ Various T HA E Rehospitalization O 0.14 223
et al. (2007)
Feeley 12 Cognitive O HAr A BDI C 0.40 25
Ferleger 41 Dynamic O CALPAS E SCL/BSI, TC, C 0.09 40
(1993) Social Adjustment
Florsheim 90–100 Various C, O WAI E, M Drug Use, Teachers C, O 0.22 78
et al. (2000) days (residential and Youths Report
program) Form, Recidivism
Flückiger ∗ CBT C BPSR E, M, L SCL/BSI, IIP, C, T 0.58 47
et al. (2005) Satisfaction,
Improvement, Goal
Forbes et al. NS Counseling C WAI-S E PTSD symptoms C 0.10 84
Forman 6 Rehabilitation C, T WAI M, L Global Outcome C, T 0.48 29
Frank et al. 56 Various T ITAS M Premature C, O 0.32 46
(1990) Termination,
Specific Symptoms,
Overall Outcome,
Symptom Severity,
Social Relations
(Continued )

Table 2.1 Continued
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Freitas ∗ Therapeutic C, T WAI E Lengths of T, O 0.00 80
(2001) Community Treatment,
Fries et al. 25 w Various C BPSR A PANSS C, O 0.32 30
Gaiton 24 CBT T, O CALPAS E Composite C, O 0.14 38
(2004) Outcome
Gallop et al. 10 Inpatient Eating C, T WAI E Premature C 0.16 31
(1994) Disorders Unit Termination
Gaston et al. 18 Various C, T CALPAS E, M, L BDI, HRSD C 0.21 18
(1991)a, b
Gaston et al. 18 Dynamic C, T CALPAS A Depression-Anxiety, C 0.15 32
(1994)a, b Interpersonal
Behavior Scale
Gaston et al. 18 Various O CALPAS A BDI, HRSD C 0.34 88
Geider ∗ Experiential O CALPAS A Global Outcome C 0.48 10
Geiser et al. ∗ Various C HAq II E ACQ, BDI, BSQ, C, T 0.55 231
(2002) GAF
Gerstley 48 Various C, T HAq E Addictive Severity O 0.36 30
et al. (1989) Index
Godfrey 6w CBT O OAS E Chronic Fatique C 0.10 71
et al. (2007)
Gomes- 18 Various O VPPS A Overall Ratings, C, T, O 0.46 35
Swartz MMPI
(1978)a Maladjustment, TC
Greenberg 6 Gestalt C WAI E Scale of Indecision, C, T 0.62 31
et al. (1982) STAI, TC
Greenberg 32 Experiential C WAI A SCL/BSI, IIP, C 0.14 32
et al. (2002) Intrex, TC
Grob et al. 19w Inpatient O ITAS E, M, L Overall T 0.41 60
(1989) Improvement
Gunderson ∗ Various C, T HAq E, M, L SCL/BSI, SAS, C 0.22 28
et al. (1997) GAS
Gunther 15 Various O CALPAS E, L SCL/BSI C 0.25 41
Gutfreund 29 Various O CALPAS A SCL/BSI, Dynamic C 0.16 46
(1992) Outcome
(Continued )

Table 2.1 Continued
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Hansson 4w Inpatient C ITAS E, L SCL/BSI, CPRS, C 0.19 106
et al. (1992) DTES, TC
Hardy et al. 16 w CBT C CALPAS A BDI C 0.71 24
Hartley et al. 18 Various O VTAS A Composite Gain C, T, O 0.27 28
(1983) Scale
Hartmann 12 Dynamic O CS E, M, L SCL/BSI, IIP C 0.46 10
Hatcher 51 Dynamic C CALPAS Various Improvement to C 0.10 230
et al. (1996) Date
Hawley et al. 16 Various O VTAS A HRSD O 0.27 162
Hayes et al. NS CBT C, O WAI Severity Rating O 0.26 18
Hays (1994) 6 Various C, T WAI E Global Outcome, C, T 0.30 29
Personal Growth,
Relations with
∗ Mother–infant O WAI E Growth C 0.35 58
Hervé et al.
(2008) Consultation
Hilliard et al. 25 Dynamic C,T,O SASB M Interject-best/worst, C, T, O 0.21 64
(2000) Intrex SCL/BSI, Global
Hopkins 12 Various C, T WAI E SCL/BSI C 0.25 15
Hopkins 30 Case T WAI-S C MCAS T 0.24 28
et al. (2006) Management
Horowitz 12 Dynamic O TARS A SCL/BSI, PCS C, O 0.11 52
et al. (1984)
Horvath 10 Various C, T WAI E PTQ C, T 0.49 29
Howard 16 Various C WAI E BDI, HRSD, IIP C, O 0.57 47
Howard 16 CBT C WAI M BDI C, 0.67 19
et al. (2006)
Huber et al. ∗ Various C, T TRS E BDI, Contentment, C, T 0.28 275
(2003) Premature
Ilgen et al. ∗ Alcohol and drug C, T WAI E Alcohol Abstincence O 0.11 785
(2006a)a Abstinence Self-Efficacy, DpD
(Continued )

Table 2.1 Continued
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Ilgen et al. ∗ Alcohol and Drug C, T WAI E Alcohol Abstincence O 0.11 785
(2006b)a Abstinence Self-Efficacy, DpD
Irelan (2004) ∗ Various C WAI E Premature O 0.35 40
Jacob (2003) 13 Various C WAI E OQ-45, Panic C 0.16 80
Janecke 38 w Various C HAq E IIP, Satisfaction, C 0.00 50
(2003) Symptom Reduction
Johansson ∗ Various C, T HAq II E SCL/BSI, IIP O 0.23 122
et al. (2006)
Joyce et al. 20 Dynamic C, T NSI A General Symptoms, C, T, O 0.29 64
(1998) Individual
Joyce et al. 18 Various C, T AAS A Improvement, C, T, O 0.27 144
(2003)a Severity of
Jumes 28 w Inpatient, C WAI E BPRS, GAS O 0.28 121
(1995) Medication
Kabuth et al. ∗ Hospital O HAq E, L Social O 0.41 33
(2005) Development,
Karver et al. 12 w CBT C, T, WAI-S, E CES-D C 0.12 12
(2008)a O AOCS
Katz (1999) 5 Dynamic C WAI-S E Premature O 0.03 100
Kech (2008) 16 IPT C NSI A Depression C 0.56 20
Kelly et al. ∗ Various C, T WAI-S A SCL/BSI C 0.28 83
Kivlighan 12 Various C, T WAI E, M, L Interpersonal C 0.17 21
et al. (1995) Problems
Kivlighan 4 Various C WAI E IIP, BIC O 0.55 38
et al. (2000)
Klee et al. 29 Various O TARS E SCL/BSI, Global C 0.23 32
(1990) Outcome
Klein et al. 12 CBT T WAI-S E, M HRSD O 0.31 367
(Continued )

Table 2.1 Continued
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Knaevelsrud 5 w CBT C WAI-S L SCL/BSI, IES C 0.48 41
et al. (2007)
Kokotovic 4 Various C, T WAI E Premature C 0.13 105
et al. (1990) Termination
Kolden 4 Dynamic C TBS E Mental Health C 0.30 60
(1996) Index
Konzag et al. 12 w Various C, T HAq E SCL/BSI C 0.21 225
Kramer et al. ∗ Various C HAq A SCL/BSI C 0.25 50
Kramer et al. ∗ Various C HAq A SCL/BSI C 0.80 50
Krupnick 16 Various O VTAS E, A Global Outcome C, O 0.46 206
et al. (1994)a
Krupnick 16 Various O VTAS E, A HRSD, BDI O 0.46 206
et al. (1996)
Kukla et al. ∗ Vocational C WASc A Job Tenure, O -0.18 91
(2009) Program Working Duration
Lansford 12 Dynamic O AWR A Global Outcome C, T, O 0.89 6
Lieberman ∗ Acute Inpatient C, T ITGA, EH E Symptom C 0.30 63
et al. (1992) Improvement, GAS,
Defense Style, RSE
Liebler et al. ∗ Various C BPSR M SCL/BSI C 0.07 87
Loneck et al. ∗ Intake Interview O VPPS E Referral O 0.23 39
(2002) Appointment
Luborsky 52 Dynamic O HAq(cs), E, L, A Rated Benefits, C, T, O 0.54 20
et al. (1983) HAq(r) Residual Gain,
Luborsky 0.79 77
et al. (1985)a
Mallinckrodt 12 Various C, T WAI E Global Outcome C, T 0.63 40
Mallinckrodt 15 Brief C WAI E SCL/BSI, Social C 0.54 34
(1996) Interpersonal Support, BDI
(Continued )

Table 2.1 Continued
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Marmar 18 Various C, T CALPAS E BDI C 0.18 18
et al.
Marmar 12 Dynamic O CALTARS A Patterns of C 0.39 52
et al. Individual
(1989b) Change
Scores, SCL
Marmarosh ∗ Various C, T WAI-S E SCL/BSI C 0.30 48
et al. (2009)
Marshel 50 Dynamic C HAq, E Premature C -0.06 101
(1986) TARS, Termination
Marziali 12 Dynamic O TARS A Composite C, O 0.35 10
et al. (1981) Outcome
Marziali 20 Dynamic C, T, TARS A Behavioral C, T, O 0.24 42
(1984) O Symptom Index,
SAS, Global
Marziali 30 Dynamic C TAS† E, L SAS, Objective C 0.79 17
et al. (1999)b Behavior Index,
McNeil 12 Various C, T, AQ A General Symptoms O 0.22 99
(2006) O
McLeod ∗ Various O TPOCS A Trait and Stait C 0.50 22
et al. (2005) Anxiety
Meier et al. ∗ Alcohol and C, T WAI-S E Premature O 0.01 187
(2006a)a Drug Termination
Meier et al. ∗ Alcohol and Drug C, O WAI-S E Premature O 0.01 187
(2006b)b Abstinence Termination
Meyer et al. 16 Various O VTAS E HRSD, BDI C, O 0.49 151
Missirlian 16 Experiential C WAI-S E, M, L SCL/BSI, BDI, IIP, C 0.37 32
et al. (2005) RSE
Mohl et al. ∗ Various C HAq E Premature C 0.30 80
(1991) Termination
Moleiro 20 Alcohol and C STS, A BDI, Composite C, O 0.48 186
(2003)a Drug TPRS Outcome
(Continued )

Table 2.1 Continued
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Morgan et al. 52 Dynamic O HAr E, A Composite C, T, O 0.59 20
(1982)a Outcome, Rated
Moseley 14 Various C WAI E State-Trait Anxiety, C 0.28 25
(1983) Self-Concept, TC,
Multon et al. 7 Career C WAI-S E SCL/GSI, C 0.14 42
(2001) Counseling Instability
Muran et al. 20 Cognitive C CALPAS A SCL/BSI, C, T 0.38 37
(1995) Interpersonal
Problems, GAS,
TC, Overall
Muran et al. 30 w Various C, T WAI-S E Premature O 0.38 99
(2009) Termination,
O’Malley ∗ Various O VPPS E Overall Outcome, C, T, O 0.55 38
et al. (1983)a TC
Ogrodniczuk 20 Interpretive, C, T NSI A General Symptoms, C, T, O 0.35 67
et al. (2000) Supportive Individual
Pantalon 19 w CBT C IVRS A Abstinence, O 0.46 16
et al. (2004) Premature
Pavio et al. 12 Experiential C WAI E, L SCL/BSI, C 0.24 33
(1998) SASB Introject,
Piper et al. 19 Dynamic C, T AAS A Composite C, T, O 0.52 64
(1991) Outcome
Piper et al. 19 Dynamic C, T AAS A State-Trait C, T 0.54 30
(1995) Anxiety, BDI,
SCL/BSI, Overall
Piper et al. 20 Dynamic C, T NSI A Composite C, O 0.10 144
(2004)a Outcome
Pos (2007) 18 Experiential C WAI M SCL/BSI, BDI C 0.34 74
Pos et al. 18 Experiential C WAI M SCL/BSI, BDI C 0.34 74
(Continued )

Table 2.1 Continued
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Priebe et al. 20 Case management C BAS E Hospitalization O 0.28 58
(1993) months Index, Work Axis,
Prigatano 6 Neuropsychology T NAS L Productivity O 0.40 35
et al. (1994) months Rehabilitation
Pugh (1991) 12 Various C, T WAI E SCL/BSI, TC C, T 0.18 55
Pyne (1991) 6 Various T, O HAq(r), A Global Outcome, C, T, O 0.34 29
VPPS Premature
Ramnerö 16 CBT T WAI-S M Outcome O -0.06 59
et al. (2007) Composite
Reiner ∗ Dynamic C TBS E Overall Outcome O 0.40 82
Reis et al. 16 Dynamic C WAI E HRSD O 0.07 58
Riley (1992) 8 Various C, T WAI, E, L SCL/BSI, TC, GAS C, T, O 0.17 61
Rogers et al. ∗ Case C, T WAI-S E Depressive O 0.27 64
(2008) Management Symptoms
Rounsaville 14 Interpersonal O VPPS E Schedule for O, C 0.25 35
et al. (1987)a Affective
Disorders, SAS,
Patient Self-
Rudolf et al. ∗ Dynamic C, T TRS E, L Composite C, T 0.44 238
(1993) Outcome
Safran et al. 20 Cognitive C WAI, E SCL/BSI, C, T 0.53 22
Success, TC
Sammet ∗ Various C HAq A SCL/BSI, IIP C 0.16 213
et al. (2004)
Samstag 30 Various C, T WAI-S A SCL/BSI, IIP C 0.55 48
et al. (2008)
Santiago 12 CBT C WAI-S E HRSD O 0.22 324
et al. (2005)
Saunders 26 Dynamic C TSR E Mental Health C 0.16 114
(2000)a Index
Saunders 26 Dynamic C TBS E Session Quality, C, O 0.20 113
et al. (1989)a Termination
(Continued )

Table 2.1 Continued
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Schauenburg 11 Dynamic C, T HAq L SCL/BSI, Severity C 0.23 284
et al. (2005) Rating
Schleussner ∗ Dynamic C HAq E Satisfaction C 0.13 57
Schönberger 14 Rehabilitation C, T WAI-S E EBIQ C, T 0.14 59
et al.
Schönberger 14 Rehabilitation C, T WAI-S† E EBIQ C, T 0.31 103
et al.
Schönberger 14 Rehabilitation C, T WAI-S E Composite C, T 0.14 59
et al. Outcome
Schönberger 14 Rehabilitation C, T WAI-S E Cognitive C, T 0.14 104
et al. (2007)a Functioning
Sexton 10 Various C WAI-S E BOPS, Beck C, O 0.40 27
(1996) Anxiety Scale,
BSO, Zung,
Global Problem
Sherer et al. ∗ Rehabilitation C, O, CALPAS, E Premature O 0.18 56
(2007) T NAS Termination,
Shirk et al. 12 w CBT C, T, AOCS A BDI, Depressive C, T 0.25 50
(2008)a O Symptoms
Solomon 2 years Case C, T WAI L Quality of Life, C, O 0.28 82
et al. (1995) Management Compliance,
Satisfaction with
Treatment, other
Sonnenberg 11 Inpatient C, T ITAS E SCL/BSI C 0.03 63
Spinhoven ∗ Various C, T WAI E Symptom Status O 0.25 70
et al. (2007)
Stevens et al. 30 30 C WAI E, M, L Outcome C, T 0.37 44
(2007) Composite
Stiles et al. 12 Various C ARM A SCL/BSI, BDI, IIP, C 0.25 76
(2004) SAS, RSE
(Continued )

Table 2.1 Continued
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Strauser et al. ∗ Mental C WAS A Employment C 0.41 97
(2004)b Retardation Prospects, Job
Strauss ∗ CBT C CALPAS A WISPI C 0.41 25
Strauss et al. ∗ CBT C CALPAS A WISPI, SCID-II, C, O 0.45 30
(2006) BDI
Svartberg 20 Dynamic C FAI M SCL/BSI, DAS C 0.38 11
et al. (1994)
Tichenor 16 Various C, T, WAI, A SCL/BSI, Self C, T, O 0.16 8
(1989) O CALPAS, Concept, TC,
Trepka et al. 16 CBT C CALPAS, A BDI C 0.50 30
(2004) ARM
Tryon et al. 19 Various C, T HAq M Premature C 0.20 74
(1990) Termination
Tryon et al. 13 Various C, T WAI-S E Premature C 0.25 86
(1993) Termination
Tryon et al. 10 Various C, T WAI-S E Premature C 0.25 71
(1995) Termination
Tunis et al. 180 Methadone C CALPAS E, M, Premature C 0.34 20
(1995) days Detox. L, A Termination,
Opioid Use, HIV
Risk Behavior
Van et al. 16 Various C HAq M Depressive C 0.24 62
(2008) Symptoms
Vogel et al. 12 CBT C HAq M Y-BOCS O 0.36 37
Vronmans 8 Narrative C WAI E, M, L BDI, OQ-45 C 0.48 34
(2007) Therapy
Wettersten 12 Various C WAI A SCL/BSI, C 0.27 32
(2000) Satisfaction
Wettersten 12 Various C WAI A SCL/BSI, C 0.27 32
et al. (2005)b Satisfaction
Wilson et al. 19 Various C HRQ E Frequency of C 0.00 154
(2002) Vomiting
Windholtz 16 Dynamic O VPPS M SCL/BSI, C, T, O 0.20 38
et al. (1988) Overall Change,
(Continued )

Table 2.1 Continued
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Yeomans 230 Dynamic O CALPAS E Premature C 0.05 20
et al. (1994) Termination
Zuroff et al. 16 Various O VTAS L DAS, C 0.10 149
(2000) Maladjustment
Zuroff et al. 16 CBT O BLRI E Maladjustment C, O 0.18 48
(2006) Composite
Raters: C = client, T = therapist, O = other/observer
Time E = early, M = middle, L = late, A = averaged alliance
RG = residual gain score
Alliance AAS = Alberta Alliance Scale
Measures: AE = Active Engagement
AOCS = Alliance Observation Coding System
AQ = Alliance Questions
ARM = Agnew Relationship Measure
AWR = Alliance Weakenings and Repairs
BAS = Berlin Alliance Scale
BLRI = Barrett-Lennard Relationship Inventory
BPSR = Bern Post Session Report
CALPAS = California Psychotherapy Alliance Scale
CALTARS = California Therapeutic Alliance Rating Scale
CIS = Client Involvement Scale
CS = Coordination Scale
EH = Patient expectation of helpfulness
FAI = Facilitative Alliance Inventory
HA(r) = Penn Helping Alliance Scale - Rated
HAq = Helping Alliance Questionnaire - Self-Rated
HA(cs) = Helping Alliance Counting Signs
HRQ = Helping Relationship Questionnaire
ITAS = Various Inpatient Therapeutic Alliance Scales
ITGA = Inpatient Task and Goal Agreement
IVRS = Interpersonal Variables Rating Scale
NAS = Neuropsychology Alliance Scale, Prigatano Alliance Scale
NSI = Non Standard Instrument (Measure developed for the specific research project)
OAS = Observer Alliance Scale
SASB = Structural Analysis of Social Behavior
STS TPRS = Systematic Treatment Selection Therapy Process Rating Scale
TARS = Therapeutic Alliance Rating Scale
TBS = Therapeutic Bond Scale
TRS = Therapeutic Relationship Scale
VTAS = Vanderbilt Therapeutic Alliance Rating Scale,
VPPS = Vanderbilt Psychotherapy Process Scale,
WAI = Working Alliance Inventory,
WAI-S = Working Alliance Inventory - Short version
WASu = Working Alliance Survey
WASc = Working Alliance Scale
(Continued )

Table 2.1 Continued
Outcome ACQ = Agoraphobic Cognitions Questionnaire
measures: BDI = Beck Depression Inventory
BIC = Battery of Interpersonal Capabilities
BOPS = Brief Outpatient Psychopathology Scale
BPRS = Brief Psychiatric Rating Scale
BSQ = Body Sensation Questionnaire
CES-D = Center of Epidemiologic Studies Depression Scale
CPRS = Comprehensive Psychopathological Rating Scale
DAS = Dysfunctional Attitudes Scale
DpD = Drinking per Day
DTES = Drug Taking Evaluation Scale
EBIQ = European Brain Injury Questionnaire
GAS = Global Assessment Scale
GHQ = General Health Questionnaire
HRSA = Hamilton Rating Scale for Anxiety
HRSD = Hamilton Rating Scale for Depression,
IES = Impact of Event Scale
MCAS = Multnomah Community Ability Scale
PANSS = Positive and Negative Syndrome Scale
PICS = Pattern of Individual Change Scores
PTQ = Post Therapy Questionnaire
RSE = Rosenberg Self-Esteem Index
SCL/BSI = Symptom Checklist 90, Brief Symptom Inventory
SEQ = Session Evaluation Questionnaire
STAI = State-Trait Anxiety Inventory
SWLS = Satisfaction with Life Scale
TC = Target Complaints
WBI = Working Behavior Inventory
WISPI = Wisconsin Personality Disorder Inventory
Y-BOCS = Yale-Brown Obsessive-Compulsive Scale
Zung = Zung’s Self-Rating of Depression
Treatment: ∗
= various length
w = weeks of treatment

Partial Scale
Studies that share data with other report(s).
Studies that contributed multiple independent alliance–outcome relations.

Therapist (T): You think I should? recognize it, and perhaps prevent a
C: You do; always. re-play of the same old grooves . . .
T: Because we agreed that looking at [pause 10 sec]
connection between past relationship T: I said that “we agreed” that this
patterns and how you and [name] are is the way to go, but I get the
getting on is . . . sense that you may not be
C: {voice over} Yes, unfinished convinced that’s so . . . it is
business . . . And all that. such a good idea.
T: It may be that there is a pattern C: Look, I mean . . . you are the
here, which would be useful to therapist and I keep fucking up with
explore and understand better. my “old lady.” So I guess I better
Once we understand it, we can start thinking & talking about these

42 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
patterns . . . I wish there was a pill therapy—he drops his previous agenda
or electric shock therapy to . . ., it and demonstrates his commitment to find
would be faster. a way of working collaboratively with his
T: Maybe we better take a step back. patient. Clients frequently have a mixture
I am a therapist, but I can’t give you of hopes and worries about discussing long
a pill or shock you to fix you. And suppressed feelings and memories of deep
looking for these unfinished patterns significance. The therapist’s challenge in
don’t seem to make an awful lot building the alliance is to recognize, legiti-
of sense to you . . . right now. But mize, and work through these issues and
I hear you are willing to be a “good engage the client in a joint exploration of
client” these obstacles.
C: But this not what it is about, about The following excerpt provides another
me being good, I mean, right? brief example of such a process:
T: What would you say if you were not
a “good patient”? Would you rebel? C: “[topic discussed last week]” . . . was
C: I guess I might . . . It’s crazy you interesting . . . But sometimes I can’t
know, before I got married I was a remember what I talked about from
pretty wild dog . . . long hair, one week to the next.
motorcycles, some pretty crazy stuff. T: . . . I think we ended up last talking
T: So, what happened? Where did the about how difficult it is to imagine
“crazy you” go? What did you do how things would be different.
with him? C: <overlap> I sometimes wonder . . .
C: Married, good job, slick house, nice what do therapists do after the
kids, you know . . . session? I mean . . . Do you walk
T: You think I might meet this around the block to forget all this
character? He seems to have been craziness . . .? Do you go home and
shut up but not forgotten . . . He dream about it?
might have something interesting T: Hmm, I . . .
to say . . . C: [overlap] I mean, it is not like having
C: I might be a little afraid of my old a discussion with a friend; though
self . . . But [with different voice]: goodness knows, I sometimes forget
Doc, I’m trash, my old man was about those too. I think to myself,
trash, but he put his money in good does he (T) need to hear all this?
booze; not in psychiatrists’ pockets! How often did I tell you that stuff?
T: He did not have much faith in this I read that Freud sometimes napped
therapy business behind the couch . . . Not, mind
C: Yeah, Of course you should not let you, that I think you nap! But
him write the cheque for the session; sometimes you look tired <Laughs>.
it would for sure bounce . . . [both Oh, don’t mind; this was a useful
laugh] session. <looks at the clock> Are we
done? <Stands up>
In the above excerpt the therapist T: So I guess sometimes you wonder
starts off defending his “modus operandi,” “what is it in it for him (T)”?
but when he becomes aware of the client’s C: I knew you’d say that!
ambivalent feelings about dealing with T: Well . . . I am not “really a friend.”
the past—and possibly about being in It is a strange thing to pour one’s
ho rvat h , re , f lü c k i g e r, s y m o n d s 43
heart out to someone and then in a way I have not been able to talk
wonder: Did it mean anything to about it before. Last week, I
him? What am I to him? mean . . . But kind of pulled back
C: Yeah, I guess . . . That’s therapy, for and felt mixed up when we
you! <stand up again as to go> started . . .I don’t like risking myself
T: Not sure if you want to talk about much do I? . . . Hmm, I guess I went
this or go? to the right school: “The hit and run
C: Well it is late . . . academy of motherly love” . . . I am
T: Interesting that this came up to-day. so tired of it [pause] . . . I think I am
And . . . kind of left hanging . . . making the connection . . . [pause]
between us. We got someplace today.
C: You mean “Hit & Run” . . .? when
It is important to note that clients, espe-
I don’t . . . get . . . something . . .
cially in the beginning of treatment, may
[I want] I don’t wait for an answer.
appear to be hostile, rejecting, or fearful of
T: There was something you
treatment or the therapist. The therapist’s
wanted . . . from me . . .?
ability to respond with acceptance and an
C: Doesn’t take a rocket scientist to
openness to discuss these challenges is an
figure out . . . When you where
important asset in establishing the alliance.
asking “does it (therapy) work for
There is some research evidence to show
you” {reference to last week’s
that therapists who respond with their own
discussion} I thought here it
negativity to client’s hostile remarks will
comes . . .
likely damage the alliance (Henry, 1994).
T: You mean I’ll quit on you?
The last excerpt offers a brief illustration
C: I know you would not do that.
of what the concept that we psychologists
I know you wouldn’t. But, I mean,
call alliance feels like from the client’s
we are talking about this all this
time, and I think . . . I talk about
it to others too {relates an incident C: Yeah, I am more comfortable
of talking about his marriage working with you . . . After finishing
to a colleague} Now I know she with Dr. “K” I was not too sure
{colleague} feels sorry for me, but of about getting into therapy again for
course this doesn’t help either. But two years. My previous therapist—
that’s different. Kind of . . .it’s not I went to him for about a year—he
sympathy I need, but sometimes was great at listening . . . I mean he
feels . . . <voice goes shallow, eyes had a good reputation and I think
moist> he was older than you. He must
T: You want from me . . . how I as have heard of these things before.
person feel . . . about . . . But I thought he was afraid that if
C: <Change of expression; sarcastic> he told me something I’ll do it like
Good fucking time to bring it up! a robot or something. I mean,
T: Does this; like this . . . remind . . . I know these are my decisions and
C: You mean do I do this Hit & Run I got to get my own answers and
with B (wife), yeah. I’ve been sometimes you tell me that I’m
thinking about that. Kind of stupid trying to get around busting my
but interesting; I felt we were own ass by getting you to tell me
really . . . I was telling you something what you think . . .
44 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Th: I . . . account for correlations of the outcome
C: [Chuckle]. It’s OK, you do it quite measures within studies. In addition, we
nicely. But I can tell. [Pause] But extracted some alliance–outcome relations
you respect an honest question and not available previously, and adjusted for
seem to try to work with me the variations in the number of participants
way I want to, not always out of used to calculate alliance–outcome rela-
the book . . . tions within studies. As a result, both the
I mean the other day, last week I mean, ES and sample size (k) associated with some
I was . . . I just could not let go of studies in this report are not identical to the
that anger. I guess I was not very well values reported in previous meta-analyses.
behaved here, as a patient I mean. To locate research with data on the rela-
But it was important for me to hear tion between alliance and outcome from
when you said “you will not let go 2000 to 2009, we searched the PsycINFO
without taking a piece of me.” Then database using the same search parame-
we talked . . . I talked like a normal ters as the Horvath and Bedi (2002) meta-
client. But I needed to get a foot into analysis published in the previous edition
you and hear “ouch” for me to look of this book. In addition, we had access to
at what is happening. I needed your a list of e-mail address for persons with
“ouch” to see into me, and not a whom the first author corresponded on the
finger from on high. subject of the alliance; these individuals
were invited to identify studies meeting the
These brief excerpts were selected to selection criteria.
illustrate how different therapy contexts The criteria for inclusion in this report
draw on diverse therapist resources, and were: (1) the study author referred to the
also the fact that the concept of the alli- therapy process variable as alliance (includ-
ance unites the notions of interventions ing variants of the term), (2) the research
and the development of the relationship in was based on clinical as opposed to analog
therapy. Alliance is built by doing the work data, (3) five or more patients participated
of therapy collaboratively. in the study, and (4) the data reported was
such that we could extract or estimate
Meta-Analytic Review a value indicating the relation between
Sources of Data alliance and outcome.
To locate research published between 1973 In reviewing the retrieved material, we
and 2000, we relied on the three previously discovered that there is a growing literature
published meta-analyses (Horvath & Bedi, linking alliance to the effectiveness of med-
2002; Horvath & Symonds, 1991; Martin, ical interventions as well as a variety of
Garske, & Davis, 2000). However, most social and even legal services. However,
of the previously published effect sizes it was decided that this literature was out-
(ESs) were recomputed for this analysis, side the scope of this report. We chose
using a more detailed coding system to take to focus more narrowly on the relation
account of added features and to better between the quality of the alliance and
identify interdependencies in the underly- outcome in the context of psychological
ing data when more than one research treatments. Alliance research conducted
report shared the same client sample. We on couples and family therapy and alliance
also applied more sensitive statistical analy- research on children were also excluded as
ses to the previously published data to these topics are covered by other chapters
ho rvat h , re , f lü c k i g e r, s y m o n d s 45
in this volume. However, treatments for appeared overwhelmingly (153) in peer-
substance abuse as well as psychological reviewed journals, with some (5) in book
problems that involve psychoactive medi- chapters, while 43 items came from unpub-
cations are included. lished (mostly dissertations) sources. The
In contrast to previous meta-analyses, later represent a significant increase
we attempted to include research published in the proportion of unpublished research
in languages other than English. Our liter- in the current data compared to previous
ature search was extended to material pub- meta-analyses. In total, the data captures
lished in Italian, German, and French. information based on over 14,000 treat-
A search was conducted of the German ments. (In Table 2.1, we provide ESs
language database (PSYNDEX) using the associated with each manuscript, but the
same inclusion criteria as for the English aggregated effect sizes, and all of the cal-
language searches. One hundred and culations presented below, were adjusted
fifty- two German abstracts were retrieved. for shared (nonindependent) data and
Of these, 17 manuscripts contained usable are based on the 190 independent effects
alliance–outcome data and were included sizes.)
in the analysis. For the French and Italian The number of eligible studies included
literature, we searched in PsycINFO with in this chapter is roughly double the size of
the additional keywords French OR Francais the data that were available in the previous
OR Italian OR Italiano. We accessed the chapter. The growth in the literature over
search platforms EBSCO (USA) and OVID the past decade means not only that there
(Europe). Of the 87 French articles located, are more studies available for analysis, but
73 manuscripts were written in English also that there is a significant increase in
and published in English journals; of the the types of therapies, treatment contexts,
remaining 14 items, 2 had usable alliance– client problems, and research designs cap-
outcome data; these are included in the tured by the current analysis.
analyses. Twenty-six Italian manuscripts Even with an effort to include non-
were located; of these 14 were published in English publications, the geographic distri-
English journals, and none of the Italian- bution of research in our data is strongly
only papers had usable data. In total, 19 biased: 153 manuscripts came from North
research reports unavailable in English were America (134 USA, 19 Canada), 45 from
included in the analysis. Europe (22 from German-speaking coun-
The 201 research reports included in the tries, 10 from Scandinavia, 8 from UK,
meta-analysis are listed in Table 2.1. Thirty- and 8 from other countries in Europe),
nine of these manuscripts were based on a and three research reports came from
shared data; that is, two or more reports pro- Australia. Notwithstanding these limita-
vided alliance–outcome information derived tions, it is reasonable to claim that the data
from a common pool of clients. Thus, some we present closely mirrors the universe of
of these reported effect sizes were not inde- alliance research, since it appears that most
pendent. In addition, 10 research publica- foreign-speaking researchers who do this
tions listed in the table reported multiple kind of work publish in English language
alliance–outcome relations based on two or journals.
more independent samples.
The data on which our analysis is based Methods of Analysis
includes both published (158) and unpub- For our numerical estimates, we used the
lished (53) research. The published research random-effects model. The reasons for this

46 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
were twofold. First, given the broad range of correlation (which has a negative skew),
of applications, research designs, and mea- and the Fisher’s z transformation results in
surement approaches within our data, an approximately normal distribution.
we could not assume the existence of an In cases where the primary study reported
underlying, homogeneous, singular, alliance– more than one level of a categorical vari-
outcome index of alliance–outcome rela- able (e.g., both clients’ and therapists’
tions. By using a fixed-effects model, we alliance scores), dependencies at the mod-
would “. . . assume homogeneity of under- erator level were accounted for by randomly
lying treatment effects across studies [and selecting one within-study level per study.
this] may lead to substantial understate- This procedure allowed for a fully indepen-
ment of uncertainty” (National Research dent analysis at the moderator level. This
Council, 1992, p. 187). Second, the random- random selection procedure provided a
effects model, apart from requiring fewer safeguard from violating the assumption of
assumptions, yields a more conservative independence in testing differences among
estimate and hence leads to safer, more levels of moderators; however, using this
trustworthy, conclusions (Cooper, Hedges, procedure also reduced the sample size and
& Valentine, 2009; Hunter & Schmidt, thus the power of the analysis. All proce-
2004). A random-effects model assumes dures for this meta-analysis were conducted
that the studies analyzed are selected from using the MAc (Del Re & Hoyt, 2010) and
a population of studies and thus the results RcmdrPlugin.MAc (Del Re, 2010) meta-
are generalizable to the larger universe of analysis packages for the R statistical
studies. software program (R Development Core
In many studies, there were a number Team, 2009).
of different outcome measures and hence
multiple effect sizes were reported. In order Results
to account for the dependencies among The aggregate effect size, for the 190 inde-
outcome measures, due to multiple within- pendent alliance–outcome relations repre-
study ESs, we employed Hunter & Schmidt’s senting over 14,000 treatments was r =
(2004) aggregation procedures to obtain 0.275. The 95% confidence interval of
one correlation effect size per study. These this aggregated ES ranged from 0.249 to
procedures take into account the correla- 0.301. The aggregated value is adjusted for
tion among within-study outcome mea- sample size, as well as the intercorrela-
sures and thus yield a more precise estimate tion among outcome measures. The magni-
of the population parameter. In cases where tude of the relationship in the current
the primary studies did not provide actual meta-analysis is a little larger but similar to
correlations among outcome measures, the values reported in previous research
the estimate of between-outcome measure (Horvath & Bedi, 2002, r = 0.21, k = 100;
correlation was set to 0.50 (Wampold, Horvath & Symonds, 1991, r = 0.26, k = 26;
1997). Martin, Garske, & Davis, 2000, r = 0.22,
When conducting categorical and con- k = 79). The median effect size of ESs of the
tinuous moderator analyses, all correlations current data set is 0.28 (not adjusted for
were transformed to a Fisher’s z (Fisher, sample size), suggesting that the group of
1924) and then transformed back to r effect sizes we collected are not strongly
for interpretive purposes. The correlation skewed. The overall effect size of 0.275 is
coefficient is known to be nonnormally statistically significant at p < 0.0001 level,
distributed, particularly with high values indicating a moderate but highly reliable
ho rvat h , re , f lü c k i g e r, s y m o n d s 47
relation between alliance and psychother- hidden in dusty file drawers to generate an
apy outcome. aggregate ES that was no longer statistically
This effect size of 0.275 was esti- significant.
mated based on studies located using elec- Another way to explore the question of
tronic databases. Therefore, this estimate whether there is a sampling bias effecting
is potentially vulnerable to the file drawer the data is by inspecting the funnel plot of
bias (Sutton, 2009): the possibility that the the collection of ES in our set. A funnel
research literature we accessed represents a plot is a diagram of standard error on the
biased sample, as there might be a number vertical axis as a function of effect size on
of studies with smaller or null ESs languish- the horizontal axis. In the presence of bias,
ing in file drawers and unlisted in databases we would expect the plot to show a higher
(possibly rejected by journals because they concentration of studies on one side of the
report nonsignificant results). The conse- mean than the other. Typically, smaller
quence of such a scenario can be evaluated sample size studies (having larger standard
by computing the fail-safe N. This is the errors) are more likely to be published
number of studies with ES = 0 that would if they have larger-than-average effects. In
make the aggregate ES in the database sta- the absence of publication bias, we would
tistically nonsignificant (p > 0.05). We have expect the studies to be distributed rela-
calculated the fail-safe value (Rosenthal, tively symmetrically around the aggregated
1979): there would have to be over one ES. The funnel plot in Figure 2.1 does not
thousand ES = 0 (null) additional studies indicate a strongly biased set of data, but

Publication bias


Standard error



−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5

Fisher’s z

Fig. 2.1 Funnel plot of the ESs in the meta analysis.

48 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
neither is it perfectly symmetrical about the relation, we would expect the reported
vertical axes. research results to cluster around a popula-
We investigated two possible sources tion parameter with deviations from
of systemic bias in the distribution of the true value due only to random errors.
ESs: date of publication and study sample We computed the I 2 statistic, which pro-
size. There was a small and statistically vides an estimate of the percentage of
nonsignificant negative time trend observed variance of ESs over and above the amount
( p = 0.082). Over time (1972–2009) research- of variability that can be accounted for by
ers were reporting slightly decreasing ESs. random (chance) variation. The I 2 of 0.56
This makes intuitive sense because recent we obtained indicates that the variance in
studies use more sophisticated methods our data is approximately 56% greater than
for controlling for pre-therapy effects that one would expect if all the studies were
might impact the strength of the alliance. measuring the same relation. This finding,
There are also more studies published in and of itself, is not surprising; research-
recently involving client populations with ers assessed alliance at different points
more severe psychological problems. Both of therapy, in a variety of therapy con-
of these factors would likely exert a down- texts, using therapists, clients, or observers
ward pressure on the correlation between for their evaluations. In addition, outcomes
alliance and outcome. were measured from a variety of perspec-
More surprisingly, we found a signifi- tives, sometimes immediately after treat-
cant relation between sample size and ments, at other times at follow-up points.
ES (r = –.25 p <0.01). The best fitting Heterogeneity in the data encouraged us to
regression line for this puzzling association investigate the possible moderators effect-
is quadratic, nonlinear; the studies with ing the alliance outcome correlations.
sample sizes between 100 to 200 appear
to report lower ESs compared to studies Moderators and Mediators
with both smaller and larger sample size. Alliance Measures
This effect may be an artifact of some sort In our meta-analysis, over 30 different alli-
but will require further investigation. ance instruments were employed, but only
In sum, the overall relation between the four core instruments (CALPAS, HAq,
alliance and outcome in individual psycho- VPPS, WAI) were used in three or more
therapy is robust, is not effected by the file studies. (The HAq family is composed of
drawer problem, and accounts for approxi- two quite different instruments: the original
mately 7.5% of the variance in treatment [1983] versions coded as HAq, and the
outcomes. [1996] revision coded as HAqII). Therefore,
for the current study, we compared the
Variability of Effect Sizes aggregated ES of each of the four core instru-
There is a great deal of variability in the ments plus a residual category called “other.”
alliance–outcome relations across the 190 The box and whisker plot (Figure 2.2) dis-
ESs in the current data set. Similar to what plays the ESs associated with these measures
we found in the previous meta-analysis (range r = 0.23–0.39). The differences
(Horvath & Bedi, 2002), the group of among them were not significant (Q = 1.851,
alliance–outcome relations in this data set df = 5). However, it should be noted that
are not homogenous (Q = 498.42, df = 189, within the four core instruments, only the ES
p < 0.00001). If all the alliance–outcome associated with the CALPAS and the VPPS
research in our data were sampling the same were homogeneous. Effect sizes reported
ho rvat h , re , f lü c k i g e r, s y m o n d s 49


Effect size



CALPAS (k = 24) HAQ (k = 25) HAQ II (k = 5) VPPS (k = 5) WAI (k = 78) OTHER (k = 52)

Alliance measure

Fig. 2.2 Box-and-Whisker plot of Effect Sizes (ES) associated with different alliance measures.

within each of the other measures were more only, if multiple assessments were available
variable than one would expect from chance in a study, we chose the earliest assessment
or random error alone. A likely reason for available for the computations. Figure 2.3
the homogeneity of the the CALPAS and shows the results of this analysis graphi-
the VPPS measures are available in fewer cally. As one would anticipate, the relation
versions than the WAI and the HAq. between alliance and outcome grew in
magnitude as the alliance and outcome
Time-of-Alliance Assessment became closer in time. The omnibus Q sta-
The time-of-alliance assessment was tistic for the overall contrast among these
grouped into four categories: Early: Sessions time categories was highly significant
1–5; Mid: after the fifth, but four or more (Q 17.42, df = 3, p <0.001), but the post-
sessions before end-of-treatment; Late: hoc pairwise multiple comparisons were
within three sessions of end-of-treatment; not statistically significant (p > 0.05), due
and Averaged a combination of assessment to the large within-category heterogeneity.
points. A number of researchers provided
information on multiple assessment times Sources-of-Alliance Assessment
within the same study. In most of our anal- The alliance can be rated from three per-
yses, one ES was randomly selected if mul- spectives: Clients, therapists, and observers.
tiple ESs were available in order to ensure Client and observer ratings are similar
independence of the data. However, for [Clients r = 0.282 (k = 109); Observer
reasons of clinical relevance, in this analysis r = 0.295 (k = 47)], and both of these

50 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p

Effect size




Early (k = 113) Mid (k = 33) Late (k = 36) Average (k = 52)

Session alliance rated

Fig. 2.3 Box-and-Whisker plot of Effect Sizes between alliance and outcome measured at different
phases of treatment.

perspectives of the alliance provided better Outcome Measures

prediction of therapy outcome than thera- As was the case with the alliance mea-
pist evaluations (r = 0.196, k = 40). These sures, a broad range of therapy outcome
findings are consistent with previous measures were utilized in the studies in
research (Horvath & Bedi 2002; Horvath our meta-analysis. Of the over 35 out-
& Symonds, 1991). However, the differ- come assessments, only three measures —
ences among these categories were not Beck Depression Inventory (Beck et al.,
statistically significant (Qb = 5.16, df = 2, 1961, BDI), SCL (Derogatis, Lippman, &
p = 0.076). An examination of the distribu- Covi, 1973), and premature termination or
tion of ESs within these categories indi- “dropout”— were utilized in sufficient fre-
cated that the variability of the ESs in the quency (5 ESs or more) to permit analysis.
client and therapist ratings was over 50% As a result, only a subset of 60 ESs that uti-
greater than expected by chance. That is, lized these three outcome measures could
the ESs in the set were heterogeneous, likely be included in this analysis. Table 2.2
due to the variety of measuring instru- displays the results of this analysis. The
ments used. Such a high degree of variabil- Q statistics for this set of moderators is
ity within these categories made it less significant (Q 10.98, df = 2, p = 0.004).
likely that the differences between the Post-hoc analysis indicated a statistically
categories of raters would reach statistical significant difference between studies using
significance. dropout (or premature termination) and

ho rvat h , re , f lü c k i g e r, s y m o n d s 51
Table 2.2 Relation between the Alliance and Outcome Measures
K ES 95% CI (Lower) 95% CI (Upper) p

Dropout 15 0.164 0.062 0.262 .001

SCL 28 0.276 0.195 0.353 .000
BDI 17 0.409 0.304 0.505 .000

BDI as an outcome measure. In consider- the alliance–outcome relation among them

ing these results, it is helpful to keep in were not statistically significant (Qb = 4.85,
mind that the BDI is most often used as a df = 3, p = 0.183)*. These results support
symptom-specific outcome measure for the claim that the alliance is a pantheoreti-
clients receiving treatment for depression. cal factor in diverse types of treatments.
The relation between alliance and therapy However, it should be noted that only 93
outcome for the treatment for depression out of the total of 190 ES in the data set
tends to be relatively high. Dropout as a could be fitted into the four categories used
treatment outcome was almost exclusively for this analysis.
utilized in studies of clients with substance
abuse problems. While unilateral termina- Raters of Outcome Data
tion represents, in one sense, “hard” out- Similar to the measurement of alliance,
come data, the substance abuse treatments researchers used outcome evaluations
included in our data were highly varied, obtained from clients, therapists, indepen-
and clients in these treatments were often dent observers, or some combination of
volatile and multidisordered. This being these sources. In the data set, 109 ESs
the case, individuals might have termi- were based on clients, 47 ESs on observers,
nated therapy for a variety of reasons, apart 12 ESs on therapists, and 22 ESs were gen-
from lack of progress in treatment. The erated by other sources (e.g., dropouts,
observed differences among these studies days of sobriety, rehospitalization). The
are consistent with this hypothesis; the 15 difference among the alliance–outcome ES
ESs within the dropout group were highly obtained by these disparate raters was statis-
variable (p = 0.37). tically significant (Q = 8.34. df = 3, p < 0.05),
but the post-hoc test of pairwise contrast of
Types of Treatments differences was not statistically significant.
Bordin (1976, 1994) argued that the alli- Again, this result is likely due to the large
ance is a significant factor in all types of variability within these rater categories.
helping relationships. We explored the evi-
dence for this claim by contrasting the Halo Effect
averaged effect sizes associated with CBT, We examined the question of whether
IPT, psychodynamic, and substance abuse ESs were inflated when the alliance and
treatments. The ESs for each type of treat- outcome information came from the
ment, reflecting the strength of the associa- same (e.g., client rates both alliance and
tion between alliance and outcome, were outcome) or different sources (e.g., client
highly significant (p <0.001). But an analy- rated alliance and observer rated outcome).
sis of the contrast between the treatment
categories indicated that the differences in *Qb = between-group statistic

52 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Table 2.3 Alliance outcome correlations disaggregated by raters of assessment.
Alliance rater Outcome rater
Client Therapist Observer Other
Client 0.30 (98) 0.27 (14) 0.21 (40) 0.34 (13)
Therapist 0.18 (31) 0.29 (20) 0.13 (20) 0.30 (12)
Observer 0.24 (28) 0.39 (10) 0.27 (17) 0.58 (7)
Other 0.16 (9) N/A 0.16 (2) 0.22 (8)
Note: Numbers in parentheses (k); Diagonal values represent data generated when the alliance and outcome were rated by the same source.
Data in this table are not independent; some studies provided data from more than one source.

Table 2.3 shows that ESs from studies in is that the categories we coded were not
which the same raters completed both the pure factors, but variables that interact
outcome, and alliance measures were indeed with one another. To better understand
higher on average than those coming from the influence of these variables on the rela-
different rater categories, but the difference tion between alliance and outcome, one
between these values fell slightly below needs to examine the complexities of these
the critical level for statistical significance variables acting together.
(p = 0.079). It is notable that the difference Modeling the full complement of poten-
between same source versus independent tial moderator interactions is statistically
source ESs has increased progressively unmanageable even with a large data set
since this effect was first analyzed in 1991 such as the one we collected. Not all the
using 27 data points (Horvath & Symonds, levels of these categorical variables intersect,
1991). Keeping in mind this apparent and there are computational difficulties
trend, the possibility that same-source eval- because many of the joint values are not
uations might be inflated may be a concern independent (i.e., are based on the same
of clinical significance in the future. data). Taking these limits into consider-
ation, a random effects multipredictor
Publication Sources meta-regression was computed to explore
The lion’s share of the ESs in our data was the joint impact of a subset of the clini-
published in refereed journals (153), fol- cally most interesting categorical modera-
lowed by unpublished studies (43), and 5 tors: The alliance raters (client, therapist,
extracted from books. The effect sizes associ- observer), the alliance measurements (using
ated with these sources were r = 0.287, only the core instruments WAI, HAq, HAq
0.237, and 0.399, respectively. The differ- II, VPPS, and CALPAS), and the three
ences among these ESs were not significant. major outcome indexes (BDI, SCL, and
dropout). The effects of the year of publica-
Interaction among Moderators tion and sample size were controlled in this
We examined a number of categorical analysis.
variables that potentially moderate the Because of the restrictions in the data,
alliance–outcome relation, but within most only 54 ESs could be entered into this
of these moderators, the range of ESs was analysis. However, almost half of the total
quite broad; most levels of moderator variance among these alliance–outcome rela-
categories were themselves heterogeneous. tions were explained by the individual and
One possible reason for such heterogeneity joint effects of these variables (R 2 = 0.46).

ho rvat h , re , f lü c k i g e r, s y m o n d s 53
It was particularly interesting that the The results of these research studies defies
addition of the joint effect of the alliance easy summary. Multiple studies found some
measure × alliance rater alone contributed support for the existence of quadratic alli-
an R 2 change of 0.23. While the results ance patterns and their relation to posi-
of this analysis cannot be generalized to tive outcome in short-term therapy (e.g.,
the alliance–outcome literature, since only Horvath & Marx, 1991; Kivlighan &
about 25% of the studies could be used for Shaughnessy, 2000; Man, 1973; Miller
these calculations, the findings strongly et al., 1983). But several other studies (e.g.,
suggest that the abundant heterogeneity in Piper et al., 2004; Stevens et al., 2007; Stiles
the research findings is due in large part to et al., 2004) were unable to replicate their
the range of methods used to measure the results and confirm this hypothesis. There
alliance combined with the variety of means is some support for the prediction that
used to assess outcome. linear and increasing levels of the quality
The cost of inclusivity in defining both of alliance over the length of treatment
the outcome and especially the process are associated with better outcome than
(alliance) variable in research praxis is the flat-linear and decreasing-linear patterns
difficulty in arriving at a focused conclu- (de Roten et al., 2004; Kramer et al., 2009;
sion. An exemplar of the how this broad Piper et al., 2004). However, the associa-
conceptualization of variables both enriches tion of linear-increasing pattern with out-
the alliance research literature and at the come is problematic: some researchers
same time creates challenges in generat- analyzed data that was collected close to
ing convergence toward clinically useful the end of therapy; these “late” alliance
conclusion, we will briefly review one of measures are difficult to disentangle from
the most dynamic strands of the current outcome (Horvath & Luborsky, 1993).
research agenda: the investigation of the Kivlighan and Shaughnessy (2000) have
dynamics of the alliance over the span of also identified “U” and reverse or inverted
treatment. “U” patterns in brief therapy. However,
there are number of reports confirming the
Patterns of Alliance over Time existence of these patterns but the number
The patterns of growth and development of published reports unable to confirm the
of the alliance over the course of psycho- pattern (e.g., Kramer et al., 2009; Piper
therapy associated with a good outcome et al., 2004; Stiles et al., 2004; Stevens
have been of continued interest to research- et al., 2007) were about equally divided.
ers. Bordin (1985, 1989) suggested that The diversity among these findings is
the repairs of stresses or tears in the alliance almost certainly related to a number of
will make an important contribution research design and measurment issues. The
toward therapeutic gains. Gelso and Carter different length of therapies is one: research-
(1994) predicted that a rise in early alliance ers have examined treatments ranging
followed by a decline in the quality of the from as few as 6 to over 30 sessions. Some
alliance, followed again by an increase, studies explored the change in the level
would be associated with positive outcomes and the shape of alliance patterns over time
in therapy. There have been a number of independently (Stiles, et al., 2004; Strauss,
research projects aimed at investigating 2001), but others choose not to make those
these and other predictions in order to distinctions (e.g., Kivlighan & Shaughnessy,
document the relation between the various 2000). In addition, a number different
alliance patterns and outcome. alliance measures and a variety of statistical

54 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
approaches, have been used by different (i.e., growth or growth-decay-and-growth
researchers, (e.g., regression models, HLM, pattern) may be most beneficial is not yet
and cluster analysis), and there is no consis- resolved.
tency across studies in terms of the length
of time in the course of psychotherapy over Limitations of the Research
which patterns were examined. As well, This chapter is based on a numerical
researchers used different criteria for the synthesis of the research results. While
identification of growth patterns and curve our team has made a sustained effort to
shape. As a result, it is not yet clear if seek out all the available research on
the lack of convergence in this literature alliance–outcome relation, no meta-analysis
signals that the hypothesized good alliance is comprehensive, and this one is no excep-
development patterns are local to specific tion. At the very least, by the time the chap-
contexts (e.g., type of treatment or client ter goes to press, there will be, no doubt, a
problems), and thus cannot be generalized, number of new studies available.
or that the broad diversity of research meth- A more significant challenge is the
ods obscure a yet-to-be discovered general “apples and oranges” problem (Hunter &
pattern of good alliance development. Schmidt, 1990, p. 521). In including all
There are a couple of hypotheses, how- research in which the authors refer to the
ever, with promise of convergence: The process variable as alliance, we might have
most consistent finding appears to be the collected and summarized a number of
proposition that some fluctuation, that is, different kinds of things. This is a serious
change over time, particularly in the mid- concern, especially in light of the fact that
phase of therapy, is associated with posi- the ESs in this data set are quite diverse.
tive outcome when contrasted with a linear A practical response to this conceptual
or stable alliance pattern. It is also safe to problem is to assert that this chapter reports
conclude that no single pattern of alliance the results of alliance–outcome relation as
development or growth has been consis- it is researched.
tently documented as better or more pre- There are also some technical constraints
dictive of good outcome than alternative to the analyses we reported. We chose to
shapes, across different kinds of therapies use independent data whenever possible.
in different lengths of treatments. There To achieve this, on many occasions we
is mounting evidence that in treatments needed to randomly discard some data (ES)
where the quality of alliance is steadily in order to make sure that only one result
declining over time, the outcome is usually from a particular data pool was used in each
poor (Stiles, 2004). There is also conver- analysis. As a result, we lost power to detect
gence that some variability in the quality differences in a number of analyses. In the
of the alliance is likely an indication of long run, the use of independent data is
superior outcome compared to a situation statistically well justified, but the resulting
where the alliance is level and stable, so constraints put on the computations are
long as the overall quality of the alliance also important to consider.
does not decline over time (Safran, 1993;
Safran, Crocker, McMain, & Murray, Therapeutic Practices
1990; Safran & Muran, 1996, 2000; Strauss The positive relation between the quality of
et al., 2006). But, the amount of varia- the alliance and treatment outcomes for
tion that is optimal for outcome, or indeed many different types of psychotherapies is
the period over which the fluctuation confirmed in this meta-analysis. The question

ho rvat h , re , f lü c k i g e r, s y m o n d s 55
of whether alliance contributes to outcome through some media (e.g., Internet,
beyond early therapy gains (e.g., Feeley, telephone). Different forms of therapy
DeRubeis, & Gelfand, 1999) has also call on diverse relational resources and
been largely resolved: a number of studies different levels of intimacy and intensity.
that controlled for this factor found that The therapist and client must find the
alliance is predictive of outcome above level of collaboration suited to achieve
and beyond early gains (e.g., Barber, et al., the work of therapy—even if they do not
2001; Brotman, 2004; Constantino, have face-to-face contact.
Arnow, Blasey & Agras, 2005; Gaston, • In the early phases of therapy,
et al., 1991; Klein, et al., 2003; Strauss, modulating the methods of therapy (tasks)
et al., 2006). While the overall ES of to suit the specific client’s needs,
r = 0.275 accounts for a relatively modest expectations, and capacities is important
proportion of the total variance in treat- in building the alliance. Clients are often
ment outcome, the magnitude of this cor- naïve in their expectation of what therapy
relation makes it one of the strongest and entails, how they have to participate in the
most robust predictors of treatment success process, and unaware of the links between
that research has been able to document what is happening moment to moment
(Wampold, 2001). In the following sec- during the session and the changes they
tion, we condense some of the most salient desire. Bridging the client’s expectations
points for practicing therapists. and what the therapist believes to be the
most appropriate interventions is an
• The alliance is not the same as the important and delicate task. Alliance
therapeutic relationship. The relationship emerges, in part, as a result of the smooth
is made of several interlocking elements coordination of these elements.
(empathy, responsiveness, creating a safe • Therapists need to closely monitor
secure environment, etc.) The alliance is the client’s perspective on the alliance
one way of conceptualizing what has been throughout treatment. It is frequently the
achieved by the appropriate use of these case that therapists’ and clients’
elements. perceptions of the alliance, particularly
• The fostering of the alliance is not early in treatment, do not converge.
separate from the interventions therapists Misjudging the client’s experience of the
implement to help their clients; it is alliance (i.e., believing that it is in good
influenced by, and is an essential and shape when the client does not share this
inseparable part of, everything that perception) could render therapeutic
happens in therapy. In this sense, the interventions less effective.
therapist does not “build alliance” but • The strength of the alliance, within
rather does the work of treatment in such or between sessions, often fluctuates in
a way that the process forges an alliance response to a variety of in-therapy factors,
with the client. such as the therapist challenging clients
• The development of a “good enough” to grapple with difficult conflicts,
alliance early in therapy is vital for therapy misunderstandings, and transference.
success. The sense of collaboration creates These “normal” variations —as long as
a working space, room to introduce new they are attended to and resolved— are
ways of addressing the clients concerns. associated with good treatment outcomes.
• The alliance matters in all forms of • Therapists’ non-defensive response to
therapy, including treatments mediated client negativity or hostility is critical for

56 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p

maintaining a good alliance. Therapists Barber, J. P., Connolly, M. B., Crits-Christoph, P.,
ought to neither internalize nor to ignore Gladis, L., & Siqueland, L. (2000). Alliance
predicts patients’ outcome beyond in-treatment
client’s negative responses.
change in symptoms. Journal of Consulting and
• Clients presenting with high Clinical Psychology, 68(6), 1027–1032.
interpersonal anxiety or with personality ∗
Barber, J. P., Gallop, R., Crits-Christoph, P.,
disorders tend to be particularly Barrett, M. S., Klostermann, S., McCarthy, K. S.,
challenging in terms of alliance et al. (2008). The role of the alliance and tech-
development and maintenance. niques in predicting outcome of supportive-
expressive dynamic therapy for cocaine
dependence. Psychoanalytic Psychology, 25(3),
References ∗
Barber, J. P., Gallop, R., Crits-Christoph, P.,
Asterisk indicates works used in meta- Frank, A., Thase, M. E., Weiss, R. D., et al.
analysis. (2006). The role of therapist adherence, therapist

Adler, J. V. (1988). A study of the working alliance in competence, and alliance in predicting outcome
psychotherapy. Unpublished doctoral dissertation, of individual drug counseling: Results from the
University of British Columbia. National Institute Drug Abuse Collaborative
Alexander, L. B., & Luborsky, L. (1987). The Penn Cocaine Treatment Study. Psychotherapy Research,
Helping Alliance Scales. In L. S. Greenberg 16(2), 229–240.

& W. M. Pinsoff (Eds.), The psychotherapeutic Barber, J. P., Luborsky, L., Crits-Cristoph, P.,
process: A research handbook (pp. 325–356). Thase, M. E., Weiss, R., Frank, A., et al. (1999).
New-York: Guilford. Therapeutic alliance as predictor of outcome in

Allen, J. G., Tarnoff, G., & Coyne, L. (1985). treatment of cocaine dependence. Psychotherapy
Therapeutic alliance and long-term hospital Research, 9, 54–73.

treatment outcome. Comprehensive Psychiatry, Barber, J. P., Luborsky, L., Gallop, R.,
26, 187–194. Crits-Christoph, P., Frank, A., Weiss, R. D.,

Allen, J. G., Tarnoff, G., Coyne, L., Spohn, H. E., et al. (2001). Therapeutic alliance as a predic-
Buskirk, J. R., & Keller, M. W. (1986). An inno- tor of outcome and retention in the National
vative approach to assessing outcome of long- Institute on Drug Abuse Collaborative Cocaine
term psychiatric hospitalization. Hospital and Treatment Study. Journal of Consulting and
Community Psychiatry, 37, 376–380. Clinical Psychology, 69(1), 119–124.
∗ ∗
Andreoli, A., Frances, A., Gex-Mabry, M., Barkham, M., Agnew, R. M., & Culverwell, A.
Aapro, N., Gerin, P., & Dazord, A. (1993). Crisis (1993). The California Psychotherapy Alliance
intervention in depressed patients with and Scales: A pilot study of dimensions and ele-
without DSM-II-R personality disorders. Journal ments. British Journal of Medical Psychology, 66,
of Nervous and Mental Disease, 181, 732–737. 157–165.
∗ ∗
Ankuta, G. Y. (1993). Initial and later therapeutic Bassler, M., Potratz, B., & Krauthauser, H. (1995).
alliance and psychotherapy outcome. Unpublished Der “Helping Alliance Questionnaire” (HAQ) von
doctoral dissertation, Michigan State University, Luborsky [The Helping Alliance Questionnaire
East Lansing. (HAQ) by Luborsky]. Psychotherapeut, 40(1),

Arnow, B. A., Manber, R., Blasey, C., Blalock, J. A., 23–32.
Rothbaum, B. O., Thase, M. E., et al. (2003). Beck, A. T., Ward, C. H., Mendelson, M., Mock, J.,
Therapeutic reactance as a predictor of out- & Erbaugh, J. (1961). An inventory for measur-
come in the treatment of chronic depression. ing depression. Archives of General Psychiatry, 4,
Journal of Consulting and Clinical Psychology, 71, 561–571.

1025–1035. Bethea, A. R., Acosta, M. C., & Haller, D. L.

Baldwin, S. A., Wampold, B. E., & Imel, Z. E. (2008). Patient versus therapist alliance: whose
(2007). Untangling the alliance-outcome corre- perception matters? Journal of Substance Abuse
lation: Exploring the relative importance of Treatment, 35(2), 174–183.

therapist and patient variability in the alliance. Bieschke, K. J., Bowman, G. D., Hopkins, M.,
Journal of Consulting and Clinical Psychology, Levine, H., & McFadden. (1995). Improve-
75(6), 842–852. ment and satisfaction in short-term therapy at

ho rvat h , re , f lü c k i g e r, s y m o n d s 57
a university counseling center. Journal of College ∗
Busseri, M. A., & Tyler, J. D. (2003).
Student Development, 36, 553–559. Interchangeability of the Working Alliance

Biscoglio, R. L. (2005). Patient and therapist per- Inventory and Working Alliance Inventory,
sonality, therapeutic alliance, and overall outcome Short Form. Psychological Assessment, 15(2),
in Brief Relational Therapy. Unpublished doctoral 193–197.
dissertation, City University, New York, NY. ∗
Busseri, M. A., & Tyler, J. D. (2004). Client-
Bordin, E. S. (1975, September). The working alli- therapist agreement on target problems, working
ance: Basis for a general theory of psychotherapy. alliance, and counselling outcome. Psychotherapy
Paper presented at the Society for Psychotherapy Research, 14(1), 77–88.
Research, Washington, DC. ∗
Card, C. A. (1991). Therapeutic alliance, comple-
Bordin, E. S. (1976). The generalizability of the mentarity, and distress during stress prevention
psychoanalytic concept of the working alliance. training after HIV testing. Unpublished doctoral
Psychotherapy: Theory, Research, and Practice, 16, dissertation, New School for Social Research,
252–260. New York.
Bordin, E. S. (1980, June). Of human bonds that ∗
Castonguay, L. G., Goldfried, M. R., Wiser, S.,
bind or free. Paper presented at the Society for Raue, P. J., & Hayes, A. M. (1996). Predicting
Psychotherapy Research, Pacific Grove, CA. the effect of cognitive therapy for depression:
Bordin, E. S. (1989, June). Building therapeutic A study of unique and common factors. Journal
alliances: The base for integration. Paper presented of Clinical and Consulting Psychology, 64,
at the Society for Psychotherapy Research, 497–504.
Berkley, CA. Chambless, D. L. (2002). Beware the dodo bird:
Bordin, E. S. (1994). Theory and research on the The dangers of overgeneralization. Clinical
therapeutic working alliance: New directions. In Psychology: Science and Practice, 9(1), 13–16.
A. O. Horvath & L. S. Greenberg (Eds.), The ∗
Chilly, C. (2004). The effects of the interpersonal
working alliance: Theory, research, and practice. framework of intervention on therapeutic compli-
(pp. 13–37). New York: Wiley. ance, alliance, and outcome. Unpublished doc-

Botella, L., Corbella, S., Belles, L., Pacheco, M., toral dissertation, York University, Toronto,
Maria Gomez, A., Herrero, O., et al. (2008). Canada.
Predictors of therapeutic outcome and process. ∗
Cislo, D. A. (1988). Client perceptions of the thera-
Psychotherapy Research, 18(5), 535–542. peutic alliance and session outcome. Unpublished

Bredel, S., Brunner, R., Haffner, J., & Resch, F. doctoral dissertation, University of Toledo,
(2004) Behandlungserfolg, Behandlungserleben Toledo, OH.
und Behandlungszufriedenheit aus der Sicht von ∗
Clarkin, J. F., & Crilly, J. L. (1987). Therapeutic
Patienten, Eltern und Therapeuten - Ergebnisse alliance and hospital treatment outcome. Hospital
einer evaluativen Studie aus der stationären and Community Psychiatry, 38, 871–875.
Kinder- und Jugendpsychiatrie [Treatment suc- ∗
Cloitre, M., Chase Stovall-McClough, K.,
cess, experience of treatment, and treatment sat- Miranda, R., & Chemtob, C. M. (2004).
isfaction from the viewpoint of patients, parents, Therapeutic alliance, negative mood regulation,
and therapists]. Praxis der Kinderpsychologie und and treatment outcome in child abuse-related
Kinderpsychiatrie. 53(4), 256–276. posttraumatic stress disorder. Journal of Consulting

Broome, K.M. (1996). Antisocial personality and and Clinical Psychology, 72(3), 411–416.
drug abuse treatment process. Unpublished doc- ∗
Coleman, D. (2006). Client personality, working
toral dissertation. Texas Christian University, alliance and outcome: A pilot study. Social Work
Fort Worth. in Mental Health, 4(4), 83–98.
∗ ∗
Brotman, M. A. (2004). Therapeutic alliance and Connors, G. J., Carroll, K. M., DiClemente, C. C.,
adherence in cognitive therapy for depression. Longabaugh, R., & Donovan, D. M. (1997).
Unpublished doctoral dissertation, University of The therapeutic alliance and its relationship to
Pennsylvania, Philadelphia. alcoholism treatment participation and outcome.

Burns, J. W., & Evon, D. (2007). Common and Journal of Consulting and Clinical Psychology,
specific process factors in cardiac rehabilitation: 65(4), 588–598.
Independent and interactive effects of the work- ∗
Constantino, M. J., Arnow, B. A., Blasey, C., &
ing alliance and self-efficacy. Health Psychology, Agras, W. S. (2005). The association between
26(6), 684–692. patient characteristics and the therapeutic alliance

58 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
in cognitive-behavioral and interpersonal ther- alcohol dependence. Drug & Alcohol Dependence,
apy for bulimia nervosa. Journal of Consulting 95(3), 230–236.

and Clinical Psychology, 73(2), 203–211. Dunn, H., Morrison, A. P., & Bentall, R. P. (2006).
Cooper, H., Hedges, L.V., & Valentine, J. C. The relationship between patient suitability,
(2009). The handbook of research synthesis and therapeutic alliance, homework compliance
meta-analysis (2nd edition). New York, Russell and outcome in cognitive therapy for psycho-
Sage Foundation. sis. Clinical Psychology & Psychotherapy, 13(3),

Crits-Cristoph, P., Cooper, A., & Luborsky, L. 145–152.

(1988). The accuracy of therapists’ interpreta- Eaton, T. T., Abeles, N., & Gutfreund, M. J.
tions and the outcome of dynamic psycho- (1988). Therapeutic alliance and outcome:
therapy. Archives of General Psychiatry, 56, Impact of treatment length and pretreatment
490–495. symptomology. Psychotherapy: Theory, Research,

Davis, L. W., & Lysaker, P. H. (2007). Therapeutic Practice and Training, 25, 536–542.

alliance and improvements in work performance Emmerling, M. E., & Whelton, W. J. (2009).
over time in patients with schizophrenia. The Stages of change and the working alliance in
Journal of Nervous and Mental Disease, 195(4), psychotherapy. Psychotherapy Research, 19(6),
353–357. 687–698.
∗ ∗
Dearing, R. L., Barrick, C., Dermen, K. H., & Fakhoury, W. K. H., White, I., Priebe, S., & PLAO
Walitzer, K. S. (2005). Indicators of client Study Group (2007). Be good to your patient:
engagement: Influences on alcohol treatment How the therapeutic relationship in the treat-
satisfaction and outcomes. Psychology of Addictive ment of patients admitted to assertive outreach
Behaviors, 19(1), 71–78. affects rehospitalization. Journal of Nervous and
Del Re., A. C. (2010). RcmdrPlugin.MAc: Meta- Mental Disease, 195, 789–791.

Analysis with Correlations (MAc) Rcmdr Plug-in. Feeley, W. M. (1993). Treatment components of
R package version 1.0.5. http://CRAN.Rproject. cognitive therapy for major depression: The good,
org/package=RcmdrPlugin.MAc the bad, and the inert. Unpublished doctoral
Del Re., A. C., & Hoyt, W. T. (2010). MAc: Meta- dissertation. University of Pennsylvania,
Analysis with Correlations. R package version Philadelphia.
1.0.5. Feeley, M., DeRubeis, R. J., & Gelfand, L. A.

de Roten, Y., Fischer, M., Drapeau, M., Beretta, V., (1999). The temporal relation of adherence and
Kramer, U., Favre, N., et al. (2004). Is one assess- alliance to symptom change in cognitive therapy
ment enough? Patterns of helping alliance devel- for depression. Journal of Consulting and
opment and outcome. Clinical Psychology & Clinical Psychology, 67(4), 578–582.

Psychotherapy, 11(5), 324–331. Ferleger, N. A. (1993). The relationship of depen-

Deu, A., Heimbrock, F., & Röhrle B. (2009). Der dency and self-criticism to alliance, complementarity
Zusammenhang zwischen Merkmalen Sozialer and outcome in short-term dynamic psychotherapy.
Netzwerke und der therapeutischen Beziehung. Unpublished doctoral dissertation, Fordham
Eine kombinierte Prozess-Outcome-Studie [The University, Bronx, NY.
relation between social network attributes Fisher, R. A. (1924) On a distribution yielding
and the therapeutic relationship - A combined the error functions of several well known statis-
process-outcome study]. In Röhrle, B., & tics. Proceedings of the International Congress of
Laireiter, A. R. (Eds.). Soziale Unterstützung Mathematics, Toronto, 2, 805–813.

und Psychotherapie (pp. 227–253). Tübingen, Florsheim, P., Shotorbani, S., Guest-Warnick, G.,
Germany: dgvt Verlag. Barratt, T., & Hwang, W. (2000). Role of the

Dorsch, B., & Joraschky, P. (2002). Katamnestische working alliance in treatment of delinquent
Untersuchung an Patienten mit einer Angststörung boys in community-based programs. Journal of
nach stationärer Therapie [Follow-up study of Clinical Child Psychology, 29, 94–107.

patients with anxiety disorders after inpatient Flückiger, C., Regli, D., & Grawe, K. (2005).
therapy]. Lage, Germany: Verlag Hans Jacobs. Allgemeine Psychotherapie im Gruppensetting:

Dundon, W. D., Pettinati, H. M., Lynch, K. G., Das Zusammenspiel von gruppen- und inter-
Xie, H., Varillo, K. M., Makadon, C., et al. ventionsspezifischen Wirkfaktoren [General psy-
(2008). The therapeutic alliance in medical- chotherapy in the group setting: Interaction
based interventions impacts outcome in treating between group- and intervention-specific factors].

ho rvat h , re , f lü c k i g e r, s y m o n d s 59
Gruppenpsychotherapie und Gruppendynamik, International. Section B: Physical Sciences and
41(3), 306–322. Engineering, 65(09), 4828.
∗ ∗
Forbes, D., Parslow, R., Creamer, M., Allen, N., Gallop, R., Kennedy, S. H., & Stern, S. D.
McHugh, T., & Hopwood, M. (2008). (1994). Therapeutic alliance on an inpatient
Mechanisms of anger and treatment outcome in unit for eating disorders. International Journal
combat veterans with posttraumatic stress disor- of Eating Disorders, 16, 405–410.
der. Journal of Traumatic Stress, 21(2), 142–149. Gaston, L., Goldfried, M. R., Greenberg, L. S.,

Forman, N.W. (1990). The nature of trait empathy Horvath, A. O., Raue, P. J., & Watson, J.
in clients with chronic pain and their counselors (1995). The therapeutic alliance in psychody-
and its impact on the development of the working namic, cognitive-behavioral and experiential
alliance and outcome. Unpublished doctoral dis- therapies. Journal of Psychotherapy Integration,
sertation, Ohio State University, Columbus. 15, 1–26.

Frank, A. F., & Gunderson, J. G. (1990). The role Gaston, L., & Marmar, C. (1994). The California
of the therapeutic alliance in the treatment of Psychotherapy Alliance Scales. In A. O. Horvath
schizophrenia. Archives of General Psychiatry, 47, & L. S. Greenberg (Eds.), The working alliance:
228–236. Theory, research and practice . New York: Wiley.

Frank, J. D., & Frank, J. B. (1991). Persuasion and Gaston, L., Marmar, C. R., Thompson, L. W., &
healing: A comparative study of psychotherapy (3rd Gallager, D. (1991). Alliance prediction of out-
ed.). Baltimore: Johns Hopkins University Press. come beyond in-treatment symptomatic change as

Freitas, T. T. (2001). The relationship between psychotherapy progresses. Psychotherapy Research,
neurocognitive impairment, working alliance, and 1, 104–112.

length of stay in a therapeutic community. Gaston, L., Piper, W. E., Debbane, E. G.,
Unpublished doctoral dissertation, Norfolk State Bienvenu, J. P., & Garant, J. (1994). Alliance
University, Virginia. and technique for predicting outcome in
Freud, S. (1912/1958). The dynamics of trans- short- and long-term analytic psychotherapy.
ference [Zur Dynamik der Übertragung]. In Psychotherapy Research, 4, 121–135.

J. Strachey (Ed. & Trans.), The standard edition of Gaston, L., Thompson, L., Gallager, D.,
the complete psychological works of Sigmund Freud Cournoyer, L. G., & Gagnon, R. (1998).
(Vol. 12, pp. 99–108). London: Hogarth Press. Alliance, technique and their interactions in
Freud, S. (1913). On the beginning of treatment: predicting outcome of behavioral, cognitive and
Further recommendations on the technique of brief dynamic therapy. Psychotherapy Research, 8,
psychoanalysis [Zur Einleitung der Behandlung– 190–209.

Weitere Ratschläge zur Technik der Psycho- Geider, F. J. (1997). Die Messung therapeutischer
analyse]. In J. Strachey (Ed.), Standard edition of Allianz: Dimensionierung und Konstruktva-
the complete psychological works of Sigmund Freud lidierung einer deutschen Version der California
(Vol. 12, pp. 122–144). London: Hogarth. Psychotherapeutic Alliance Scales (CALPAS) [The
Freud, S. (1913/1940). The technique of psycho- measurement of therapeutic alliance: Dimension-
analysis. In J. Strachey (Ed.), Standard edition of ality and construct validation of a German ver-
the complete psychological works of Sigmund Freud sion of the California Psychotherapeutic Alliance
(Vol. 23, pp. 172–182). London: Hogarth. Scales (CALPAS)]. Unpublished doctoral disser-

Fries, A., Pfammatter, M., Andres, K., & tation, University of Ruprecht-Karls-University,
Brenner, H. D. (2003). Wirksamkeit und Heidelberg, Germany.

Prozessmerkmale einer psychoedukativen und Geiser, F., Bassler, M., Bents, H., Carls, W.,
bewältigungsorientierten Gruppentherapie für Joraschky, P., Michelitsch, B., et al. (2002).
schizophren und schizoaffektiv Erkrankte Bewertung des Therapieerfolgs durch Patienten mit
[Outcome and process characteristics of a psy- Angststörungen nach stationärer Psychotherapie
choeducational and coping-oriented group [Patient’s subjective assessment of therapy out-
therapy for patients with schizophrenia and come after inpatient psychotherapy for anxiety
schizoaffective disorder]. Verhaltenstherapie, disorders.] Der Nervenarzt, 73(1), 59–64.
13(4), 237–243. Gelso, C. J., & Carter, J. A. (1994). Components

Gaiton, L. R. (2004). Investigation of therapeutic of the psychotherapy relationship: Their interac-
alliance in a treatment study with substance- tion and unfolding during treatment. Journal of
abusing women with Ptsd. Dissertation Abstracts Counseling Psychology, 41, 296–306.

60 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
∗ in short-term inpatient psychiatric care.
Gerstley, L., McLellan, A.T., Alterman, A.I.,
Woody, G.E., Luborsky, L., & Prout, M. (1989). Scandinavian Journal of Social Medicine, 20,
Ability to form an alliance with the therapist: A 45–50.

possible marker of prognosis for patients with Hardy, G. E., Cahill, J., Shapiro, D. A., Barkham,
antisocial personality disorder. American Journal M., Rees, A., & Macaskill, N. (2001). Client
of Psychiatry, 146, 508–512. interpersonal and cognitive styles as predictors of
∗ response to time-limited cognitive therapy for
Godfrey, E., Chalder, T., Ridsdale, L., Seed, P.,
& Ogden, J. (2007). Investigating the ‘active depression. Journal of Consulting and Clinical
ingredients’ of cognitive behaviour therapy and Psychology, 69(5), 841–845.

counselling for patients with chronic fatigue in Hartley, D. E., & Strupp, H. H. (1983). The
primary care: Developing a new process mea- therapeutic alliance: Its relationship to out-
sure to assess treatment fidelity and predict come in brief psychotherapy. In J. Masling
outcome. British Journal of Clinical Psychology, (Ed.), Empirical studies in analytic theories. (Vol. 1,
46(3), 253–272. pp. 1–37). Hillside, NJ: Erlbaum.

Goldfried, M. R. (1980). Toward the delinea- Hartmann, H. A. (2001). Clients’ contribution
tion of therapeutic change principles. American to the alliance and the role it plays in outcome.
Psychologist, 35, 991–999. Unpublished doctoral dissertation, Adelphi
∗ University. New York, NY.
Gomes-Schwartz, B. (1978). Effective ingredients

in psychotherapy: Prediction of outcome from Hatcher, R. L., & Barends, A. W. (1996). Patient’s
process variables. Journal of Consulting and view of the alliance in psychotherapy: Exploratory
Clinical Psychology, 46, 1023–1035. factor analysis of three alliance measures. Journal

Greenberg, L. S., & Malcolm, W. (2002). Resolving of Clinical and Consulting Psychology, 64,
unfinished business: Relating process to outcome. 1326–1336.
Journal of Consulting and Clinical Psychology, Hatcher, R. L., & Barends, A. W. (2006). How a
70(2), 406–416. return to theory could help alliance research.
∗ Psychotherapy: Theory, Research, Practice, Training,
Greenberg, L. S., & Webster, M. C. (1982).
Resolving decisional conflict by Gestalt two- 43(3), 292–299.
chair dialogue: Relating process to outcome. Hatcher, R. L., Barends, A., Hansell, J., &
Journal of Counseling Psychology, 29, 468–477. Gutfreund, M. J. (1995). Patient’s and therapist’s
Greenson, R. R. (1965). The working alliance shared and unique views of the therapeutic
and the transference neuroses. Psychoanalysis alliance: An investigation using confirmatory
Quarterly, 34, 155–181. factor analysis in a nested design. Psychoanalysis
Greenson, R. R. (1967). Technique and practice of Quarterly, 63, 636–643.

psychoanalysis. New York: International University Hawley, L. L., & Moon-Ho Ringo, H. (2006). The
Press. relationship of perfectionism, depression, and
∗ therapeutic alliance during treatment for
Grob, M. C., & Eisen, S. V. (1989). Most likely to
succeed: Correlates of good versus poor hospital depression: Latent difference score analysis.
outcome in young adult inpatients. Psychiatric Journal of Consulting & Clinical Psychology,
Hospital, 20, 23–30. 74(5), 930–942.
∗ ∗
Gunderson, J. G., Najavits, L. M., Leonhard, C., Hayes, S. A., Hope, D. A., VanDyke, M. M., &
Sullivan, C. N., & Sabo, A. N. (1997). Ontogeny Heimberg, R. G. (2007). Working alliance for cli-
of the therapeutic alliance in borderline patients. ents with social anxiety disorder: Relationship with
Psychotherapy Research, 7, 301–309. session helpfulness and within-session habituation.
∗ Cognitive Behaviour Therapy, 36(1), 34–42.
Gunther, G. J. (1992). Therapeutic alliance, patient

object relations and outcome in psychotherapy. Hays, V.L. (1994). The effects of the therapeutic
Unpublished doctoral dissertation, Michigan alliance and social support on therapy outcome and
State University, East Lansing. mental health of women. Unpublished doctoral
∗ dissertation, University of Wisconsin, Madison.
Gutfreund, M. J. (1992). Therapist interventions:
Their relation to therapeutic alliance and outcome in Henry, W. P., & Strupp, H. H. (1994). The
dynamic psychotherapy. Unpublished doctoral dis- therapeutic alliance as interpersonal process.
sertation, Michigan State University, East Lansing. In A. O. Horvath & L. S. Greenberg (Eds.), The
∗ working alliance: Theory, research and practice.
Hansson, L., & Berglund (1992). Stability of ther-
apeutic alliance and its relationship to outcome New York: Wiley.

ho rvat h , re , f lü c k i g e r, s y m o n d s 61
∗ Horvath, A. O., & Luborsky, L. (1993). The role
Hervé, M.-J., White-Koning, M., Paradis-
Guennou, M., Guiraud, A., Picot, M.-C., of the therapeutic alliance in psychotherapy.
Grandjean, H., et al. (2008). Adaptation d’une Journal of Consulting and Clinical Psychology, 61,
echelle d’alliance therapeutique au contexte des 561–573.
consultations mere-nourrisson. Etude prelimi- Horvath, A. O., & Marx, R. W. (1991). The devel-
naire [Adaptation of a therapeutic alliance scale opment and decay of the working alliance during
in mother-infant consultations. A preliminary time-limited counselling. Canadian Journal of
study]. Devenir, 20(1), 65–85. Counselling, 24, 240–259.
∗ Horvath, A. O., & Symonds, B. D. (1991).
Hilliard, R. B., Henry, W. P., & Strupp, H. H.
(2000). An interpersonal model of psychother- Relation between working alliance and out-
apy: Linking patient and therapist developmental come in psychotherapy: A meta-analysis. Journal
history, therapeutic process, and types of of Counseling Psychology, 38, 139–149.

outcome. Journal of Consulting and Clinical Howard, I. (2003). Interpersonal factors in relation-
Psychology, 68(1), 125–133. ship to therapy outcome and therapeutic alli-
∗ ance in a cohort of multiple sclerosis patients.
Hopkins, M., & Ramsundar, N. (2006). Which fac-
tors predict case management services and how do Unpublished doctoral dissertation, Alliant
these services relate to client outcomes? Psychiatric International University, San Francisco.

Rehabilitation Journal, 29(3), 219–222. Howard, I., Turner, R., Olkin, R., & Mohr, D. C.
∗ (2006). Therapeutic alliance mediates the rela-
Hopkins, W. E. (1988). The effects of conceptual
level matching on the working alliance and out- tionship between interpersonal problems and
come in time-limited counseling. Unpublished depression outcome in a cohort of multiple scle-
doctoral dissertation, University of Maryland, rosis patients. Journal of Clinical Psychology,
College Park. 62(9), 1197–1204.
∗ ∗
Horowitz, M. J., Marmar, C., Weiss, D. S., Huber, D., Brandl, T., Henrich, G., & von Rad, M.
DeWitt, K. N., & Rosenbaum, R. (1984). Brief (2003). Interaktive Beziehungsmuster und ihre
psychotherapy of bereavement reactions: The Bedeutung für psychotherapeutische Verän-
relationship of process to outcome. Archives of derungen [Satisfaction and compliance after a
General Psychiatry, 41, 438–448. psychosomatic-psychotherapeutic consultation
∗ session]. Zeitschrift für Klinische Psychologie und
Horvath, A. O. (1981). An exploratory study of the
working alliance: Its measurement and relationship Psychotherapie, 32(2), 85–93.
to outcome. Unpublished doctoral dissertation, Hunter, J. E., & Smith, F. L. (1990). Methods of
Vancouver, Canada. Meta-analysis: Correcting error and bias in research
Horvath, A. O. (2009). Conceptual and methodolog- findings. London: Sage.
ical challenges in alliance research: Is it time for a Hunter, J. E., & Schmidt, F. L. (2004). Methods of
change. Paper presented at the European Society meta-analysis (2nd edition). Thousand Oaks,
for Psychotherapy Research Conference. Bolzano, CA: Sage.

Italy. Ilgen, M. A., McKellar, J., Moos, R., & Finney, J. W.
Horvath, A. O., & Bedi, R. P. (2002). The alliance. (2006a). Therapeutic alliance and the relationship
In J. C. Norcross (Ed.), Psychotherapy relation- between motivation and treatment outcomes in
ships that work: Therapist contributions responsive- patients with alcohol use disorder. Journal of
ness to patients (pp. 37–70). New York: Oxford Substance Abuse Treatment, 31(2), 157–162.

University Press. Ilgen, M., Tiet, Q., Finney, J., & Moos, R. H.
Horvath, A. O., Gaston, L., & Luborsky, L. (2006b). Self-efficacy, therapeutic alliance, and
(1993). The role of alliance in psychotherapy. alcohol-use disorder treatment outcomes. Journal
In N. E. Miller, L. Luborsky, J. P. Barber, & of Studies on Alcohol, 67(3), 465.

J. P. Docherty (Eds.), Psychodynamic treatment Irelan, C. S. (2004). The therapeutic alliance, resolu-
research: A handbook for clinical practice. tion of alliance ruptures and termination outcome.
(pp. 247–274). New York: Basic Books. Central Michigan University, Michigan.

Horvath, A. O., & Greenberg, L. S. (1986). Jacob, K. L. (2003). Toward understanding change
Development of the working alliance Inventory. in psychotherapy: A closer look at the relation-
In L. S. Greenberg & W. M. Pinsof (Eds.), The ship between client-therapist match and treat-
psychotherapeutic process: A research handbook ment outcome. Unpublished doctoral dissertation,
(pp. 529–556). New York: Guilford. Clark University, Worchester, MA.

62 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
∗ outpatient therapy. Psychotherapy: Theory,
Janecke, N. (2003). Bedeutung der Motivation und
therapeutischen Beziehung in der Psychotherapie Research, Practice, Training, 46(2), 193–202.

[The importance of motivation and therapeutic Kivilighan, D. M., & Shaugnessy, P. (1995).
alliance in psychotherapy]. Unpublished doc- Analysis of the development of the working alli-
toral dissertation, University of Innsbruck, ance using hierarchical linear modeling. Journal
Innsbruck, Austria. of Counseling Psychology, 42, 338–349.
∗ ∗
Johansson, H. K., & Eklund, M. (2006). Helping Kivlighan, D. M, Jr., & Shaughnessy, P. (2000).
alliance and early dropout from psychiatric out- Patterns of working alliance development:
patient care: The influence of patient factors. A typology of client’s working alliance ratings.
Social Psychiatry and Psychiatric Epidemiology, Journal of Counseling Psychology, 47(3), 362–371.

41(2), 140–147. Klee, M. R., Abeles, N., & Mullere, R. T. (1990).
∗ Therapeutic alliance: Early indicators, course,
Joyce, A. S., & Piper, W. E. (1998). Expectancy,
the therapeutic alliance, and treatment out- and outcome. Psychotherapy, 27, 166–174.

come in short-term individual psychotherapy. Klein, D. N., Schwartz, J. E., Santiago, N. J.,
Journal of Psychotherapy Practice and Research, 7, Vivian, D., Vocisano, C., Castonguay, L. G.,
236–248. et al. (2003). Therapeutic alliance in depres-
∗ sion treatment: Controlling for prior change and
Joyce, A. S., Ogrodniczuk, J. S., Piper, W. E., &
McCallum, M. (2003). The alliance as mediator patient characteristics. Journal of Consulting and
of expectancy effects in short-term individual Clinical Psychology, 71(6), 997–1006.

therapy. Journal of Consulting and Clinical Knaevelsrud, C., & Maercker, A. (2007). Internet-
Psychology, 71(4), 672–679. based treatment for PTSD reduces distress and
∗ facilitates the development of a strong thera-
Jumes, M. T. (1995). The developmental course of
the inpatient working alliance in state psychiatric peutic alliance: A randomized controlled clinical
hospital clients. Unpublished doctoral disserta- trial. BMC Psychiatry, 7, 13–10.

tion, University of Texas Southwestern Medical Kokotovic, A. M., & Tracey, T. J. (1990). Working
Center, Dallas. alliance in the early phase of counseling. Journal of
∗ Counseling Psychology, 37, 16–21.
Kabuth, B., De Tychey, C., & Vidailhet, C. (2005).

Alliance therapeutique avec les mères et evolu- Kolden, G. G. (1996). Change in early sessions
tion clinique des enfants d’un hôpital de of dynamic therapy: Universal processes and
jour [Mothers’ therapeutic alliance and children the generic model of psychotherapy. Journal
outcome in a day hospital]. Annales Médico of Clinical and Consulting Psychology, 64,
Psychologiques, 163(6), 486–492. 489–496.
∗ ∗
Karver, M., Shirk, S., Handelsman, J. B., Fields, S., Konzag, T., Bademer-Greulich, U., Bahreke, U.,
Crisp, H., Gudmundsen, G., et al. (2008). & Fikentscher, E. (2004). Therapeutische
Relationship processes in youth psychotherapy: Beziehung und Therapieerfolg bei der station-
Measuring alliance, alliance-building behaviors, ären Psychotherapie von Persönlichkeitsstörungen
and client involvement. Journal of Emotional & [Therapeutic relationship and outcome in psy-
Behavioral Disorders, 16(1), 15–28. chotherapy for personality disorders]. Zeitschrift
∗ für Psychosomatische Medizin und Psychotherapie,
Katz, J. (1999). Self-handicapping, the working alli-
ance, interpersonal tendencies and the prediction 50(4), 394–405.

of drop-out. Unpublished doctoral dissertation, Kramer, U., de Roten, Y., Beretta, V., Michel, L.,
New School for Social Research, New York. & Despland, J.-N. (2009). Alliance patterns
∗ over the course of short-term dynamic psycho-
Kech, S. (2008). Einflussfaktoren auf den
Behandlungserfolg der Interpersonellen Psychother- therapy: The shape of productive relationships.
apie bei stationären Depressionspatienten: Analyse Psychotherapy Research, 19(6), 699–706.

der Wirkmechanismen [Determinats of success Kramer, U., de Roten, Y., Beretta, V., Michel, L.,
of interpersonal psychotherapy with depressive & Despland, J.-N. (2008). Patient’s and
inpatients: Analysis of effective factors.] Published therapist’s views of early alliance building in
doctoral dissertation, University of Freiburg, dynamic psychotherapy: Patterns and relation
Freiburg, Germany. http://www.freidok.uni- to outcome. Journal of Counseling Psychology, 55(1), 89–95.
∗ Kramer, U., de Roten, Y., Beretta, V., Michel, L., &
Kelly, A. E., & Yuan, K.-H. (2009). Clients’
secret keeping and the working alliance in adult Despland, J.-N. (2009). Alliance patterns over

ho rvat h , re , f lü c k i g e r, s y m o n d s 63
the course of short-term dynamic psychother- alliance methods for predicting outcomes of psy-
apy: The shape of productive relationships. chotherapy: A counting signs vs. a global rating
Psychotherapy Research, 19(6), 669–706. method. Journal of Nervous and Mental Disease

Krupnick, J. L., Elkin, I., Collins, J., Simmens, S., 171, 480–491.

Sotsky, S. M., Pilkonis, A., & Watkings, J. T. Luborsky, L., McLellan, A. T., Woody, G. E.,
(1994). Therapeutic alliance and clinical out- O’Brien, C. P., & Auerbach, A. (1985). Therapist
come in the NIMH Treatment of Depression success and its determinants. Archives of General
Collaborattive Research Program: Preliminary Psychiatry, 42, 602–611.
findings. Psychotherapy, 31, 28–35. Luborsky, L., Singer, B., & Luborsky, L. (1975).

Krupnick, J. L., Sotsky, S. M., Simmens, A., Comparative studies of psychotherapies; “Is
Moyer, J., Elkin, I., Watkins, J., et al. (1996). it true that everybody has won and all must
The role of the alliance in psychotherapy and have prizes”? Archives of General Psychiatry, 32,
pharmacotherapy outcome: Findings in the 995–1008.

National Institute of Mental Health treatment Mallinckrodt, B. (1993). Session impact, work-
of depression collaborative research program. ing alliance, and treatment outcome in brief
Journal of Clinical and Consulting Psychology, 64, counseling. Journal of Counseling Psychology, 40,
532–539. 25–32.
∗ ∗
Kukla, M., & Bond, G. R. (2009). The working Mallinckrodt, B. (1996). Change in working
alliance and employment outcomes for people alliance, social support and psychological symp-
with severe mental illness enrolled in voca- toms in brief therapy. Journal of Counseling
tional programs. Rehabilitation Psychology, 54(2), Psychology, 43, 448–455.

157–163. Marmar, C. R., Gaston, L., Gallager, D., &

Lansford, E. (1986). Weakenings and repairs of the Thompson, L. W. (1989b). Therapeutic alli-
working alliance in short-term psychotherapy. ance and outcome in behavioral, cognitive, and
Professional Psychology: Research and Practice, 17, brief dynamic psychotherapy in late-life depres-
364–366. sion. Journal of Nervous and Mental Disease 177,

Lieberman, P. B., von Rehn, S., Dickie, E., & 464–472.

Elliott, B. (1992). Therapeutic effects of brief Marmar, C., Weiss, D. S., & Gaston, L. (1989a).
hospitalization: The role of a therapeutic alli- Toward the validation of the California
ance. Journal of Psychotherapy Practice & Research, Therapeutic Alliance Rating System. Psychological
1(1), 56–63. Assessment: A Journal of Consulting and Clinical

Liebler, A., Biella, M., Schulz, W., & Hartmann, U. Psychology, 1, 46–52.

(2004). Differenzielle Therapieverläufe und Marmarosh, C. L., Gelso, C. J., Markin, R. D.,
Behandlungserfolg in der stationären Grup- Majors, R., Mallery, C., & Choi, J. (2009). The
penpsychotherapie [Differential courses of treat- real relationship in psychotherapy: Relationships
ment and their relation to outcome in inpatient to adult attachments, working alliance, transfer-
group psychotherapy]. Gruppenpsychotherapie und ence, and therapy outcome. Journal of Counseling
Gruppendynamik. 40(1), 1–21. Psychology, 56(3), 337–350.
∗ ∗
Loncek, B., Banks, S., Way, B., & Bonaparte, E. Marshel, R. L. (1986). An exploration of the
(2002). An empirical model of therapeutic pro- relationship between the therapeutic alliance
cess for psychiatric emergency room clients with and premature termination. Unpublished doc-
dual disorders. Social Work Research, 26(3), 132. toral dissertation, Northwestern University,
Luborsky, L. (1976). Helping alliances in psycho- Evanston, IL.
therapy. In J. L. Cleghhorn (Ed.), Successful psy- Martin, D. J., Garske, J. P., & Davis, K. M. (2000).
chotherapy (pp. 92–116). New York: Brunner/ Relation of the therapeutic alliance with out-
Mazel. come and other variables: A meta analytic review.
Luborsky, L. (1994). Therapeutic alliance measures Journal of Clinical and Consulting Psychology, 68,
as predictors of future benefits of psychotherapy. 438–450.

In A. O. Horvath & L. S. Greenberg (Eds.), The Marziali, E. (1984). Three viewpoints on the
working alliance: Theory, research and practice. therapeutic alliance scales: Similarities, differ-
New York: Wiley. ences and associations with psychotherapy out-

Luborsky, L., Crits-Cristoph, P., Alexander, L., come. Journal of Nervous and Mental Disease
Margolis, M., & Cohen, M. (1983). Two helping 172, 417–423.

64 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
∗ ∗
Marziali, E., Marmar, C., & Krupnick, J. (1981). Moleiro, C. M. D. M. (2003). Change in therapy:
Therapeutic alliance scales: Development and A model of the effects of systematic treatment
relationship to psychotherapy outcome. American selection variables and quality of alliance on indi-
Journal of Psychiatry, 138, 361–364. vidual growth. Unpublished doctoral disserta-

Marziali, E., Munroe-Blum, H., & McClery, L. tion, University of California, Santa Barbara.

(1999). The effects of the therapeutic alliance Morgan, R., Luborsky, L., Crits-Cristoph, P.,
on the outcomes of individual and group psy- Curtis, H., & Solomon, J. (1982). Predicting
chotherapy with borderline personality disorder. the outcomes of psychotherapy by the Penn
Psychotherapy Research, 9, 452–467. Helping Alliance Rating Method. Archives of

McLeod, B. D., & Weisz, J. R. (2005). The Therapy General Psychiatry, 39, 397–402.

Process Observational Coding System-Alliance Moseley, D. (1983). The therapeutic relationship and
Scale: Measure characteristics and prediction its association with outcome. Unpublished thesis,
of outcome in usual clinical practice. Journal University of British Columbia, Vancouver,
of Consulting and Clinical Psychology, 73(2), Canada.

323–333. Multon, K. D., Heppner, M. J., Gsybers, N. C.,

McNeil, D. C., (2006). Patient, therapist, and Zook, C., & Ellis-Kalton, C. A. (2001).
observer perspectives on cohesion and alliance Client psychological distress: An important
and their relationship to outcome in psychody- factor in career counseling. The Career
namic group psychotherapy for persons experienc- Development Quarterly, 49, 324–335.

ing complicated grief. Unpublished doctoral Muran, J. C., Gorman, B. S., Safran, J. D.,
dissertation, University of Alberta, Edmonton, Twining, L, Samstag, I. W., & Winston, A.
Canada. (1995). Linking in-session change to overall

Meier, P. S., & Donmall, M. C. (2006). Differences outcome in short term cognitive therapy.
in client and therapist views of the working Journal of Consulting and Clinical Psychology,
alliance in drug treatment. Journal of Substance 63, 651–657.

Use, 11(1), 73–80. Muran, J. C., Safran, J. D., Gorman, B. S.,

Meier, P. S., Donmall, M. C., McElduff, P., Samstag, L. W., Eubanks-Carter, C., &
Barrowclough, C., & Heller, R. F. (2006). The Winston, A. (2009). The relationship of early
role of the early therapeutic alliance in predict- alliance ruptures and their resolution to pro-
ing drug treatment dropout. Drug & Alcohol cess and outcome in three time-limited psycho-
Dependence, 83(1), 57–64. therapies for personality disorders. Psychotherapy:

Meyer, B., Pilkonis, P. A., Krupnick, J. L., Theory, Research, Practice, Training, 46(2),
Egan, M. K., Simmens, S. J., & Sotsky, S. M. 233–248.
(2002). Treatment expectancies, patient alli- National Research Council (1992). Combining
ance and outcome: Further analyses from the information: Statistical issues and opportunities
National Institute of Mental Health Treatment for research. Washington DC: National Academy
of Depression Collaborative Research Program. Press.

Journal of Consulting and Clinical Psychology, Ogrodniczuk, J. S., Piper, W. E., Joyce, A. S., &
70(4), 1051–1055. McCallum, M. (2000). Different perspectives of
Miller, N. E., Luborsky, L., Barber, J. P., & the therapeutic alliance and therapist technique in
Docherty, J. P. (1993). Psychodynamic treat- 2 forms of dynamically oriented psychotherapy.
ment research: A handbook for clinical practice. Canadian Journal of Psychiatry, 45, 452–458.

New York, NY US: Basic Books. O’Malley, S. S., Suh, C. S., & Strupp, H. H.

Missirlian, T. M., Toukmanian, S. G., Warwar, S. H., (1983). The Vanderbilt Psychotherapy Process
& Greenberg, L. S. (2005). Emotional arousal, Scale: A report on the scale development and a
client perceptual processing, and the working process-outcome study. Journal of Clinical and
alliance in experiential psychotherapy for depres- Consulting Psychology, 51, 581–586.

sion. Journal of Consulting and Clinical Psychology, Pantalon, M. V., Chawarski, M. C., Falcioni, J.,
73(5), 861–871. Pakes, J., & Schottenfeld, R. S. (2004). Linking

Mohl, P. C., Martinez, D., Ticknor, C., Huang, M., process and outcome in the community rein-
& Cordell, L. (1991). Early dropouts from forcement approach for treating cocaine depen-
psychotherapy. Journal of Nervous and Mental dence: A preliminary report. American Journal of
Disease, 179(8), 478–481. Drug & Alcohol Abuse, 30(2), 353–367.

ho rvat h , re , f lü c k i g e r, s y m o n d s 65
∗ ∗
Pavio, S. C., & Bahr, L. B. (1998). Interpersonal Ramnerö, J., & Öst, L.-G. (2007). Therapists’ and
problems, working alliance, and outcome in clients’ perception of each other and working
short-term experiential therapy. Psychotherapy alliance in the behavioral treatment of panic dis-
Research, 8, 392–406. order and agoraphobia. Psychotherapy Research,

Piper, W. E., Azim, H. F. A., Joyce, A. S., 17(3), 328–337.

MacCallum, M., Nixon, G. W. H., & Segal, P. S. Reiner, P. A. (1987). The development of the thera-
(1991). Quality of object relations vs. interper- peutic alliance. Unpublished doctoral disserta-
sonal functioning as predictor of therapeutic alli- tion, University of North Carolina, Chapel Hill.

ance and psychotherapy outcome. Journal of Reis, S., & Grenyer, B. F. S. (2004). Fearful attach-
Nervous and Mental Disease 179, 432–438. ment, working alliance and treatment response

Piper, W. E., Boroto, D. R., Joyce, A. S., for individuals with major depression. Clinical
McCallum, M., & Azim, H. F. A. (1995). Pattern Psychology & Psychotherapy, 11(6), 414–424.

of alliance and outcome in short-term individual Riley, J. D. (1992). The impact of pre-therapy and
psychotherapy. Psychotherapy, 32, 639–647. extra-therapy role preparation on the development

Piper, W. E., Ogrodniczuk, J. S., & Joyce, A. S. of therapeutic alliance and outcome. Unpublished
(2004). Quality of object relations as a modera- doctoral dissertation, Pacific Graduate School of
tor of the relationship between pattern of Psychology, Palo Alto, CA.
alliance and outcome in short-term individual Rogers, C. R., Gendlin, G. T., Kiesler, D. V., &
psychotherapy. Journal of Personality Assessment, Truax, L. B. (1967). The therapeutic relationship
83(3), 345–356. and its impact: A study of psychotherapy with schizo-

Pos, A. E., & B. (2007). Experiential treatment phrenics. Madison: University of Wisconsin Press.
for depression: A test of the experiential theory of Rogers, C. R., & Wood, J. K. (1974). Client-
change, differential effectiveness, and predictors of centered theory: Carl R. Rogers. . In A. Burton
maintenance of gains. Unpublished dissertation, (Ed.), Operational theories of personality
York University. Toronto, Canada. (pp. 211–258.). New York: Bruner/Mazel.
∗ ∗
Pos, A. E., Greenberg, L. S., & Warwar, S. H. Rogers, N., Lubman, D. I., & Allen, N. B. (2008).
(2009). Testing a model of change in the Therapeutic alliance and change in psychiatric
experiential treatment of depression. Journal symptoms in adolescents and young adults
of Consulting and Clinical Psychology, 77(6), receiving drug treatment. Journal of Substance
1055–1066. Use, 13(5), 325–339.

Priebe, S., & Gruyters, T. (1993). The role of Rosenthal, R. (1979). The file drawer problem
helping alliance in psychiatric community care: and tolerance for null results. Psychological
A prospective study. Journal of Nervous and Bulletin, 86, 638–41.
Mental Disease, 181, 552–557. Rosenzweig, S. (1936). Some implicit common

Prigatano, G. P., Klonoff, P. S., O’Brien, K. P., factors in diverse methods of psychotherapy.
Altman, I. M., Amin, K., Chiapello, D., American Journal of Orthopsychiatry, 6, 412–415.

Shepherd, J., Cunningham, M., & Mora, M. Rounsaville, B. J., Chevron, E. S., Prusoff, B. A.,
(1994). Productivity after neuropsychologically Elkin, I., Imber, S., Sotsky, S., (1987). The
oriented milieu rehabilitation. Journal of Head relation between specific and general dimensions
Trauma Rehabilitation, 9, 91–102. of the psychotherapy process in interpersonal

Pugh, M. A. (1991). The working alliance, prether- psychotherapy of depression. Journal of Clinical
apy interpersonal relationships and therapeutic out- and Consulting Psychology, 55, 379–384.

come. Unpublished doctoral dissertation, Virginia Rudolf, G., & Manz, R. (1993). Zur prognos-
Commonwealth University, Richmond. tischen Bedeutung der therapeutischen Arbeits-

Pyne, S. C. (1991). Therapist’s use of countertransfer- beziehung aus der Perspektive von Patienten
ence in short-term psychotherapy: A focus on rescu- und Therapeuten [The prognostic relevance of
ing and counterhostility. Unpublished doctoral a therapeutic working alliance as seen by patients
dissertation, University of Colorado, Boulder. and therapists]. Psychotherapie, Psychosomatik,
R Development Core Team (2009). R: A language Medizinische Psychologie. 43(6), 193–199.
and environment for statistical computing. Safran, J. D. (1993). The therapeutic alliance
R Foundation for Statistical Computing, rupture as a transtheoretical phenomenon:
Vienna, Austria. ISBN 3-900051-07-0, http:// Definitional and conceptual issues. Journal of Psychotherapy Integration, 3(1), 33–49.

66 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Safran, J. D., Crocker, P., McMain, S., & Murray, P. Langzeitstudie zur Psychotherapieforschung
(1990). The therapeutic alliance rapture as a [Effectiveness of individual psychological psy-
therapy event for emperical investigation. Psycho- choanalytic psychotherapy - A longitudinal study
therapy: Theory, Research, Practice and Training, on psychotherapy research]. Zeitschrift für indi-
27, 154–165. vidualpsychologie, 30(1), 51–77.
Safran, J. D., & Muran, C. (2000). Negotiating the Schönberger, M., Humle, F., & Teasdale, T. W.
therapeutic alliance: A relational treatment guide. (2006a). Subjective outcome of brain injury
New York: Guilford. rehabilitation in relation to the therapeutic work-

Safran, J. D., & Wallner, L. K. (1991). The relative ing alliance, client compliance and awareness.
predictive validity of two therapeutic alliance Brain Injury, 20(12), 1271–1282.

measures in cognitive therapy. Psychological Schönberger, M., Humle, F., & Teasdale, T. W.
Assessment, 3, 188–195. (2006b). The development of the therapeutic
Saketopoulou, A. (1999). The psychotherapeu- working alliance, patients’ awareness and their
tic alliance in psychodynamic psychotherapy: compliance during the process of brain injury
Theoretical conceptualizations and research rehabilitation. Brain Injury, 20(4), 445–454.

findings. Psychotherapy, 36, 329–342. Schönberger, M., Humle, F., & Teasdale, T. W.

Sammet, V. I., Staats, H., & Schauenburg, H. (2007). The relationship between clients’ cogni-
(2004). Beziehungserleben und Therapieergebnis tive functioning and the therapeutic working
in stationärer Psychotherapie [Relationship expe- alliance in post-acute brain injury rehabilitation.
riences and therapeutic outcome in inpatient Brain Injury, 21(8), 825–836.

psychotherapy]. Zeitschrift für Psychosomatische Schönberger, M., Humle, F., Zeeman, P., &
Medizin und Psychotherapie, 50(4), 376–393. Teasdale, T. (2006c). Working alliance and

Samstag, L. W., Muran, J. C., Wachtel, P. L., patient compliance in brain injury rehabilitation
Slade, A., Safran, J. D., & Winston, A. (2008). and their relation to psychosocial outcome.
Evaluating negative process: A comparison of Neuropsychological Rehabilitation, 16(3), 298–314.

working alliance, interpersonal behavior, and Sexton, H. (1996). Process, life events, and symp-
narrative coherency among three psychother- tomatic change in brief eclectic psychotherapy.
apy outcome conditions. American Journal of Journal of Consulting and Clinical Psychology, 64,
Psychotherapy, 62(2), 165–194. 1358–1365.
∗ ∗
Santiago, N. J., Klein, D. N., Vivian, D., Sherer, M., Evans, C. C., Leverenz, J., Stouter, J.,
Arnow, B. A., Blalock, J. A., Kocis, J. H., et al. Irby, J. W., Jae Eun, L., et al. (2007). Therapeutic
(2005). The therapeutic alliance and CBASP- alliance in post-acute brain injury rehabilitation:
specific skill acquisition in the treatment of Predictors of strength of alliance and impact
chronic depression. Cognitive Therapy and of alliance on outcome. Brain Injury, 21(7),
Research, 29, 803–817. 663–672.
∗ ∗
Saunders, S. M. (2000). Examining the relationship Shirk, S. R., Gudmundsen, G., Kaplinski, H. C.,
between the therapeutic bond and the phases & McMakin, D. L. (2008). Alliance and out-
of treatment outcome. Psychotherapy: Theory, come in cognitive-behavioral therapy for adoles-
Research, Practice, Training, 37(3), 206–218. cent depression. Journal of Clinical Child &

Saunders, S. M., Howard, K. I., & Orlinsky, D. E. Adolescent Psychology, 37(3), 631–639.
(1989). The therapeutic bond scales: Psychometric Smith, M. L., & Glass, G. V. (1977). Meta-analysis
characteristics and relationship to treatment effec- of psychotherapy outcome studies. American
tiveness. Psychological Assessment, 1, 323–330. Psychologist, 32, 752–760.
∗ ∗
Schauenburg, H., Dinger, U., & Starck, M. (2005). Solomon, P., Draine, J., & Delaney, M. A. (1995).
Zur Bedeutung der Einzeltherapeuten für das The working alliance and consumer case man-
Therapieergebnis in der stationären Psycho- agement. Journal of Mental Health Adminstration,
therapie - eine Pilotstudie [The significance of 22, 126–134.

individual psychotherapists for psychothera- Sonnenberg, R. T. (1996). An examination of
peutic outcomes in inpatient psychotherapy. the outcome of a brief psychiatric hospitalization
A pilot study]. Psychotherapie, Psychosomatik, with particular reference to the therapeutic alliance
Medizinische Psychologie, 55(7), 339–346. and select patient characteristics, Unpublished

Schleussner, D. (2005). Wirksamkeit individualpsy- doctoral dissertation, University of Nebraska,
chologisch-psychoanalytischer Psychotherapie - Eine Lincoln.

ho rvat h , re , f lü c k i g e r, s y m o n d s 67
∗ ∗
Spinhoven, P., Giesen-Bloo, J., van Dyck, R., Tichenor, V. (1989). Working alliance: A measure
Kooiman, K., & Arntz, A. (2007). The therapeu- comparison. Unpublished doctoral dissertation,
tic alliance in schema-focused therapy and University of Maryland, Chapel Hill, MD.
transference- focused psychotherapy for border- Tichenor, V., & Hill, C. E. (1989). A comparison of
line personality disorder. Journal of Consulting & six measures of working alliance. Psychotherapy:
Clinical Psychology, 75(1), 104–115. Theory, Research, and Practice, 26, 195–199.

Sterba, R. F. (1934). The fate of the ego in analytic Trepka, C., Rees, A., Shapiro, D. A., Hardy, G. E.,
therapy. International Journal of Psychoanalysis, & Barkham, M. (2004). Therapist competence
115, 117–126. and outcome of cognitive therapy for depression.

Stevens, C. L., Muran, J. C., Safran, J. D., Cognitive Therapy & Research, 28(2), 143–157.

Gorman, B. S., & Winston, A. (2007). Levels Tryon, G., & Kane, A. S. (1990). The helping alli-
and patterns of the therapeutic alliance in brief ance and premature termination. Canadian
psychotherapy. American Journal of Psychotherapy, Psychologist, 3, 223–238.

61(2), 109–129. Tryon, G. S., & Kane, A. S. (1993). Relationship

Stiles, W. B., Glick, M. J., Osatuke, K., Hardy, G. E., of working alliance to mutual and unilateral
Shapiro, D. A., Agnew-Davies, R., et al. (2004). termination. Journal of Counseling Psychology,
Patterns of alliance development and the 40, 33–36.

rupture-repair hypothesis: Are productive rela- Tryon, G. S., & Kane, A. S. (1995). Client involve-
tionships U-shaped or V-shaped? Journal of ment, working alliance and type of psycho-
Counseling Psychology, 51(1), 81–92. therapy termination. Psychotherapy Research, 5,
Stiles, W. B., Shapiro, D., & Elliot, R. (1986). 189–198.

Are all psychotherapies equivalent? American Tunis, S. L., Deluchi, K. L., Schwartz, K.,
Psychologist, 41, 165–180. Banys, P., & Sees, K. L. (1995). The relationship
Stone, L. (1961). The Psychoanalytic Situation: of counselor and peer alliance to drug use and
An Examination of Its Development and Essential HIV risk behaviors in a six-month methadone
nature. New York: International Universities detoxification program. Addictive Behaviors, 20,
Press. 395–405.
∗ ∗
Strauser, D. R., Lustig, D. C., & Donnell, C. Van, H. L., Hendriksen, M., Schoevers, R. A.,
(2004). The relationship between working Peen, J., Abraham, R. A., & Dekker, J. (2008).
alliance and therapeutic outcomes for individu- Predictive value of object relations for therapeu-
als with mild mental retardation. Rehabilitation tic alliance and outcome in psychotherapy for
Counseling Bulletin, 47(4), 215–223. depression: An exploratory study. Journal of

Strauss, J. L. (2001). Patterns and impact of the Nervous & Mental Disease, 196(9), 655–662.

alliance and affective arousal in psychotherapy: Vogel, P. A., Hansen, B., Stiles, T. C., &
An application to cognitive therapy for avoidant Gotestam, K. G. (2006). Treatment motivation,
and obsessive-compulsive personality disorders. treatment expectancy, and helping alliance as
Unpublished doctoral dissertation, University of predictors of outcome in cognitive behavioral
Miami, Coral Gables, FL. treatment of OCD. Journal of Behavior Therapy

Strauss, J. L., Hayes, A. M., Johnson, S. L., and Experimental Psychiatry, 37(3), 247–255.

Newman, C. F., Brown, G. K., Barber, J. P., et al. Vronmans, L. P. (2007). Process and outcome of
(2006). Early alliance, alliance ruptures, and symp- narrative therapy for major depressive disorder in
tom change in a nonrandomized trial of cognitive adults: Narrative reflexivity, working alliance and
therapy for avoidant and obsessive-compulsive improved symptom and interpersonal outcomes.
personality disorders. Journal of Consulting and Unpublished doctoral dissertation, Queensland
Clinical Psychology, 74(2), 337–345. University of Technology, Brisbane, Australia.
Sutton, A. J. (2009). Publication bias. In H. Cooper, Wampold, B. E. (2001). The great psychotherapy
L. V. Hedges, & J. C. Valentine (Eds.), The hand- debate: Models, methods, and findings. Mahwah,
book of research synthesis and meta-analysis (2nd NJ: Erlbaum.
ed.). New York: Russell Sage Foundation. Wampold, B. E., Mondin, G. W., Moody, M.,

Svartberg, M., & Stiles, T. C. (1994). Therapeutic Stich, F., Benson, K., & Ahn, H. (1997). A meta-
alliance, therapist competence and client change analysis of outcome studies comparing bona fide
in short-term anxiety-provoking psychotherapy. psychotherapies: Empiricially, ‘all must have
Psychotherapy Research, 4, 20–33. prizes.’ Psychological Bulletin, 122(3), 203–215.

68 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
∗ ∗
Wettersten, K. B. (2000). Solution-focused brief Yeomans, F. E., Gutfreund, J., Selzer, M. A.,
therapy, the working alliance, and outcome: A com- Clarkin, J., Hull, J. W., & Smith, T. E. (1994).
parative analysis. Unpublished doctoral disserta- Factors related to drop-outs by borderline
tion, University of Kansas, Lawrence. patients: Treatment contract and therapeutic

Wettersten, K. B., Lichtenberg, J. W., & alliance. Journal of Psychotherapy Practice and
Mallinckrodt, B. (2005). Associations between Research, 3, 16–24.
working alliance and outcome in solution- Zetzel, E. R. (1956). Current concepts of transfer-
focused brief therapy and brief interper- ence. International Journal of Psychoanalysis, 37,
sonal therapy. Psychotherapy Research, 15(1–2), 369–376.

35–43. Zuroff, D. C., & Blatt, S. J. (2006). The therapeutic

Wilson, G. T., Fairburn, C. C., Agras, W. S., relationship in the brief treatment of depres-
Walsh, B. T., & Kraemer, H. (2002). Cognitive- sion: Contributions to clinical improvement and
behavioral therapy for bulimia nervosa: Time enhanced adaptive capacities. Journal of Consult-
course and mechanisms of change. Journal ing and Clinical Psychology, 74(1), 130–140.

of Consulting and Clinical Psychology, 70(2), Zuroff, D. C., Blatt, S. J., Sotsky, S. M.,
267–274. Krupnick, J. L., Martin, D. J., Sanislow, C. A.,

Windholtz, M. J., & Silbershatz, G. (1988). et al. (2000). Relation of therapeutic alliance and
Vanderbilt psychotherapy process scale: A repli- perfectionism to outcome in brief outpatient
cation with adult outpatients. Journal of Clinical treatment for depression. Journal of Clinical &
& Consulting Psychology, 56, 56–60. Consulting Psychology, 68, 114–124.

ho rvat h , re , f lü c k i g e r, s y m o n d s 69
Alliance in Child and
3 Adolescent Psychotherapy

Stephen R. Shirk and Marc S. Karver

The therapeutic alliance has a long history in engaging children and adolescents in a
the child and adolescent psychotherapy liter- working relationship is a major challenge
ature dating to the work of Anna Freud for those who treat young clients.
(1946). In contrast, research on the alliance in Relationship processes, including the alli-
youth treatment is relatively new. In their ance, have been neglected in the child and
2003 meta-analysis of relationship predictors adolescent literature for other reasons. In
of child and adolescent treatment outcomes, contrast to the adult literature, where treat-
Shirk and Karver (2003) identified only one ment equivalence has prompted the search
study that met the inclusion criteria used in for common factors, the development of spe-
adult alliance meta-analyses. Fortunately, the cific treatment methods has remained the
last decade has produced a substantial increase focal point of youth therapy research. This
in research on the alliance in child and ado- difference in focus is not just another exam-
lescent treatment, but the total number of ple of child research lagging behind its adult
studies still pales in comparison to the adult counterpart but, rather, reflects the absence
literature (see Chapter 2). of a “Dodo Bird verdict” for treatment equiv-
The discrepancy between the adult and alence in youth outcomes. Broad-band meta-
youth alliance research is not surprising. In analyses of youth treatment outcomes
general, research on child and adolescent indicate that behavioral treatments tend to
therapy has lagged behind its adult coun- produce significantly better results than non-
terpart in many areas. Yet, developmen- behavioral therapies across many childhood
tal differences between children and adults disorders (Weisz, Weiss, Alicke, & Klotz,
make the alliance especially important in 1987; Weisz, Weiss, Han, & Granger, 1995),
youth treatment. Children and adolescents a finding that holds up even after controlling
rarely refer themselves for treatment (Shirk for differences in methodological quality
& Saiz, 1992). Often young clients fail to (Weiss & Weisz, 1995). Although this per-
acknowledge the existence of psychologi- spective has its critics (Miller, Wampold, &
cal problems, and when they do, they attri- Varhely, 2008; Spielmans, Gatlin, & McFall,
bute their cause to environmental factors. 2010), the focus on specific treatment proce-
Children, like persons with severe mental dures has dislocated research on common
illness, lack both awareness of their prob- factors in the child and adolescent literature.
lems and interest in self-exploration that Despite this trend, research on the alliance in
facilitate involvement in therapy (Wright, youth therapy has expanded in recent years
Everett, & Roisman, 1986). Consequently, (Shirk & Karver, 2006; Zack, Castonguay,

& Boswell, 2007), partially in response to parenting strategies (Kazdin & Wassell,
growing recognition of within-treatment 2000). Although the content of therapeutic
variability in outcomes. work varies across treatments, the association
The aims of this chapter are threefold. between alliance and outcome is presumed
First, we review definitions, measures, and to be mediated through involvement in treat-
clinical examples of the alliance in the ment tasks.
child and adolescent literature. Of particu- In contrast to this perspective, play ther-
lar importance are developmental issues apists have long emphasized the curative
that distinguish youth and adult alliances. nature of the therapy relationship (Axline,
Second, we provide a meta-analytic review 1947). In this tradition, the child’s experi-
of alliance–outcome associations in child ence of the therapist as supportive, attuned,
and adolescent therapy. Third, we summa- and nonjudgmental was essential for thera-
rize the research on client factors and peutic change (Shirk & Russell, 1996;
therapist strategies that facilitate alliance Wright, Everett, & Roisman, 1986). Drawing
formation with children and adolescents. on the work of Rogers (1957), therapy was
not conceptualized as treatment, as some-
Definitions and Measures thing you do to the child, but rather as
Two views of the therapeutic relationship an opportunity for growth. The relational
were prominent in the early history of child conditions of empathy, genuineness, and
therapy. Anna Freud (1946) observed that an positive regard are posited as the active
“affectionate attachment” between child and ingredients of therapy. The development
therapist is a “prerequisite for all later work” and maintenance of an emotional bond
in child therapy (p. 31). In this early state- facilitates emotional and behavioral change.
ment, we find an enduring distinction in the Associations between bond and outcome
alliance literature, the distinction between in this tradition are direct rather than medi-
bond and work, between the emotional rela- ated through therapeutic work.
tionship and the collaborative relationship Common to the foregoing perspectives
(Estrada & Russell, 1999; Shirk & Saiz, is an emphasis on an emotional connec-
1992). Of equal importance, the link between tion between child and therapist. Emotional
bond and collaboration is framed function- bond, then, appears to be a core compo-
ally; the emotional bond enables the child to nent of alliance with children. This view has
work purposefully on the tasks of therapy. taken root in recent approaches to assessing
The bond itself is not posited as curative, but the alliance in child and adolescent ther-
rather as a catalyst for promoting therapeutic apy (e.g., Shirk, Gudmundsen, Kaplinski,
work. Interestingly, this view is revived in & McMakin, 2008; Shirk & Russell, 1996;
later cognitive-behavioral formulations of Shirk & Saiz, 1992).
the therapy relationship. The alliance serves In contrast, some have criticized this
specific technical procedures and can facili- perspective for failing to acknowledge the
tate child involvement in tasks ranging from social contractual features of the therapeutic
exposure to homework completion (Kendall, alliance. “Traditional theories of child and
Comer, Marker, Creed, & Puliafico, 2009; adolescent psychotherapy appear to have
Shirk & Karver, 2006). In the area of parent overly focused on the bond as necessary and
management training, where parents rather sufficient for change” (DiGiuseppe, Linscott,
than children are the focus of treatment, the & Jilton, 1996, p. 87). From this per-
alliance is hypothesized to improve parent spective, the central component of alli-
attendance and adherence to adaptive ance, especially with older children and

s h i rk , k a rve r 71
adolescents, consists of agreements regard- adults, the relationship with a therapist
ing treatment goals and the methods for fills a need for such connection. Indeed,
accomplishing them. The fact that youth this type of relationship might be quite
are typically referred by others makes the therapeutic for relationship-deprived chil-
establishment of agreements both difficult dren. However, other children may anchor
and essential for treatment collaboration. their positive feelings for their therapists
Given the press toward autonomy in ado- on features not typically regarded as thera-
lescence, this issue takes on added impor- peutic, for example, how fun, stimulating,
tance. At present, however, it is not clear or rewarding the therapist might be. In
if goal agreement is equally relevant for such cases it is unclear if the “bond” reflects
younger child clients as with adolescents. an experience of the therapist as an “ally,”
As these clinical perspectives suggest, there or as a valued playmate. In fact, A. Freud
are important parallels between adult and (1946) distinguished this type of relation-
youth models of alliance. Consistent with ship from the alliance. In the latter, the
Bordin’s (1979) pantheoretical model, three therapeutic bond is based on experiencing
facets of alliance—emotional bond, task the therapist as someone who can be counted
collaboration (work), and agreements (goal on for help with emotional or behavioral
consensus)—are prominent in the youth lit- problems. This is a rather tall order for many
erature. Although it is tempting to view this children and adolescents, and possibly for
convergence as evidence for configural invari- some adults. But it draws attention to the
ance in the alliance construct across age potential developmental differences and the
groups, at least two studies have failed to multiplicity of meanings in the emotional
fully support the three-factor model with bond in child versus adult therapy.
youth. These studies produced a single-factor A second developmental issue concerns
solution, thus suggesting that features of the the task dimension of the alliance. In the
alliance may be less differentiated at younger adult literature, tasks are framed in terms of
ages (DiGiuseppe et al., 1996; Faw, Hogue, agreements about the content and methods
Johnson, Diamond, & Liddle, 2005). of therapy; in essence, whether there is
Of equal importance, a number of devel- consensus between client and therapist on
opmental issues contribute to differences the substance of therapy (Bordin, 1979).
across youth and adults in the nature of Such judgments may exceed the cognitive
bond, task, and goal dimensions of alliance. capacities of many child and adolescent
Consider first the therapy bond. Anna clients. For example, studies of children’s
Freud (1946) noted some time ago that understanding of therapy have shown
a child’s relationship with the therapist important developmental progressions in
could arise from a number of sources, not their recognition of therapy processes and
all of them developmentally equivalent. parameters (e.g., Shirk & Russell, 1998, for
For many children, the relationship with a a review). For example, children’s causal rea-
therapist is an opportunity to obtain grati- soning may limit their ability to understand
fications not available in other contexts. As links between specific therapy tasks and
A. Freud observed, “if no one at home plays subsequent therapy goals (Shirk, 1988).
games with the child, for example, he might Perhaps it should not be surprising that
like to come to treatment because there research finds little agreement between child
a grown-up pays attention to him” (Sandler, and therapist ratings of task collaboration,
Kennedy, & Tyson, 1980, p.47). For chil- but greater convergence for therapy bond
dren who lack sustaining relationships with (Shirk & Saiz, 1992). Such developmental

72 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
concerns have prompted some investigators therapists are faced with establishing and
to suggest that task collaboration with chil- maintaining an alliance with the youth and
dren is best assessed through observation his or her parent(s). Most research on alli-
(Karver et al., 2008; Shirk & Karver, 2006). ance–outcome relations with children
In essence, the task dimension of alliance in and adolescents has focused on the youth–
youth therapy, especially with children, therapist relationship. A notable exception
should be operationalized as observed is in the area of parent management train-
participation in therapy tasks and not as ing where parents are the primary focus of
agreements about such tasks. child therapy. It is possible that alliances
Developmental issues complicate the goal with parents and youth may relate to differ-
dimension of the alliance as well. An impor- ent sets of outcomes. For example, Hawley
tant difference between adult and youth and Weisz (2005) found that parent, but
therapy is the involvement of other family not youth, alliance predicted better therapy
members aside from the identified client. participation. Youth alliance, but not parent
Minimally, parents or guardians are involved alliance, predicted symptom change. These
in transportation to and payment of therapy. findings suggest that a strong alliance with
Quite often, however, parents are more parents is important for treatment continu-
actively involved as informants about client ation, whereas the youth alliance may be
functioning, collateral participants, or even more critical for treatment outcomes.
as therapeutic collaborators who help with At present, a unified definition of alliance
treatment implementation outside sessions. has yet to emerge in the youth literature. In
Consequently, the therapist is faced with fact, our review of the research literature
multiple sets of goals, and often the goals of reveals 10 different alliance measures for
parents and youth diverge. Agreement on children and adolescents. No study has
goals, then, is complicated by whose goals examined the concurrent validity of the most
are considered. A study of clinic-referred frequently used measures. Thus, it is not
children (Hawley & Weisz, 2003) examined clear if different measures with similar names
therapist, child, and parent agreement about are assessing the same facets of the alliance.
the most important problems to be addressed The two most frequently used patient
in therapy. Amazingly, more than 75% of and therapist report instruments in youth
child, parent, and therapist triads began research are the Working Alliance Inventory
treatment without agreement on even one (WAI; Horvath & Greenberg, 1989) and
target problem. Nearly half failed to agree the Therapeutic Alliance Scale for Children
on one broad problem domain such as (TASC; Shirk & Saiz, 1992). The WAI has
aggression versus depression. It is interesting been used primarily with adolescents
to note that therapists agreed with parents and the TASC with children and young
more often than with children. Such evi- adolescents. Although the WAI, originally
dence suggests that agreement on goals may developed for adult therapy, has been
mean something quite different in youth modified for use with adolescents (Linscott,
therapy than in adult therapy At present, it DiGuiseppe, & Jilton, 1993), the original
is not clear if agreement between parent and or short version has been employed
therapist or child and therapist is a better most frequently. The WAI measures the
predictor of treatment outcome. quality of the therapeutic relationship across
A related issue involves the presence of three subscales: bonds, tasks, and goals. The
multiple alliances in youth treatment even final item pool for the measure was gener-
when it is child focused. Unlike with adults, ated on the basis of content analysis of

s h i rk , k a rve r 73
Bordin’s (1979) model of working alliance. or task categories and consistently sorted
Expert raters evaluated items for goodness items were retained. The resulting coding
of fit with the working alliance construct. system includes eight bond items and six
The TASC was developed specifically task collaboration items. Interrater reliabil-
for child therapy and also was based on ity has been shown to be good across items.
Bordin’s model (1979). Two dimensions Bond and task dimensions are highly
are assessed: bond between child and thera- correlated, consistent with what has been
pist, and level of task collaboration. Unlike found with youth self-reports of alliance
the WAI, task collaboration does not refer dimensions, suggesting that alliance may
to agreements on tasks, but to ratings be a unitary construct in youth therapy.
of actual collaboration on tasks such as
“talking about feelings” and “trying to solve Clinical Examples
problems.” The therapist version of the The following verbal interactions derived
TASC involves ratings of the child’s bond from a composite of cases reflect different
and task involvement rather than the thera- features of the therapeutic alliance with
pist’s own. Although items on the bond young clients. The first example illustrates a
subscale remain constant, items on the task strong emotional bond between a young
collaboration scale vary with type of treat- adolescent and her therapist:
ment in order to be consistent with CBT Therapist: So, what is it like when
(cognitive-behavioral therapy) or psycho- you’re feeling really down?
dynamic tasks. The subscales show good Client: I get like I don’t want to talk to
internal consistency and relatively high anyone. I’m like get away, leave me
levels of stability over a 4- to 7-week period alone. My dad asks me how I’m doing
(Shirk et al., 2008). Although therapist and I just say nothing or walk away.
and child agreement on bond ratings are Therapist: You just want some space.
medium to strong, agreement is substan- You don’t want to be pushed.
tially lower for task collaboration. Client: Exactly.
A number of observational measures Therapist: In here, I’m going to ask you
have appeared in the youth literature but a lot about how you are feeling. If you
none have become the “gold standard.” feel like I’m pushing you, is it possible
One measure that was developed specifi- you will not want to talk with me.
cally for child and adolescent therapy and Client: I don’t think that’ll happen
that has been used in more than one study because you’re not in my face.
is the Therapy Process Observation Coding Talking gets my stress out. When
System–Alliance Scale (McLeod & Weisz, I’m in a bad mood on the day of
2005). This observation scale took as its our meetings, I look forward to our
starting point the distinction between talking . . . it helps keep me going
bond and task collaboration found in factor because I know you get me.
analyses of child and therapist reports of
alliance (Shirk & Saiz, 1992) and factor In the next example, the goal dimension
analyses of process codes (Estrada & Russell, of the alliance is prominent. Here the
1999). Items from a broad range of mea- therapist explores the adolescent’s goals
sures that mapped onto the bond and task for therapy.
dimensions were initially included in the Therapist: I know what your parents are
item pool, and redundant items reduced. hoping for from our therapy, but
Expert raters then sorted items into bond what are your goals?

74 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Client: I want to stop worrying that I’ll Therapist: So using what we’ve worked
say something wrong, so I won’t just on might have a payoff?
feel all stuck. Client: Yeah, like what we practiced in
Therapist: It sounds like you’d like to here.
shift your focus away from all the
It should be noted that other features
things that make you worry and feel
of interactions can reflect the condition of
the alliance. For example, a child who is
Client: Yeah, when I think about what
unresponsive to therapist questions or who
could happen, I become so nervous
is only willing to talk about topics unrelated
I just avoid everyone. I want to go
to problems or issues that prompted ther-
walk up to the ins (popular girls) and
apy demonstrates behaviorally low levels of
just be right there talking and not all
collaboration. Similarly, many children will
what if.
actively participate in games and unstruc-
Therapist: So, if we could change how
tured play but will avoid talking about con-
much you worry and think about all
cerns or practicing relevant skills in session.
the negatives that would be a good
Though such children appear to like their
therapist, it is not evident that the therapist
Client: Definitely, I’m tired of worrying
is viewed as someone who could help with
all the time.
emotional or behavioral problems.
In the final example, an older child talks
with his therapist about dealing with anger.
Prior Reviews
The client’s statements reflect the collabora-
In their 2003 meta-analysis of associations
tive aspect of the alliance:
between relationship variables and treatment
Client: I feel better since we last talked. outcomes, Shirk and Karver found only 23
That stuff we worked on was pretty studies with quantifiable relationships pub-
helpful. lished over the previous 27 years. The major-
Therapist: That’s cool. Great. What ity of these studies did not assess alliance per
did you do? se but evaluated various dimensions of rela-
Client: Like . . . I forgot what it is tionship quality such as therapist warmth,
called . . . like . . . I controlled my therapeutic climate, or treatment participa-
temper . . . when I got angry . . . tion. Only ninestudies examined the alli-
I was like OK like take a deep ance, and of these nine only one evaluated
breath . . . then I walked away. the alliance prospectively in individual ther-
Therapist: Great. It helped bring your apy. Overall, Shirk and Karver (2003) found
anger down. that relationship variables are related to
Client: mmhmm. youth treatment outcomes with a weighted
Therapist: You made a good decision. mean correlation of 0.20, slightly smaller
Some people get angry and are than, but similar to, estimates from the adult
like, hey, I’m right, I’m not backing literature (see Chapter 2).
down. Although Shirk and Karver (2003) did
Client: If I get up in their face when not report results for alliance studies alone,
I’m mad, I end up losing anyway. a reanalysis of the data indicates that the
Therapist: Losing anyway? weighted mean alliance–outcome association
Client: Yeah, I pay for it later. Get in in the sample of nine studies was r = 0.25.
trouble and stuff. It should be noted, however, that this

s h i rk , k a rve r 75
estimate includes both prospective and child and adolescent literature. First, consis-
concurrent measurement of alliance and tent with earlier meta-analyses, we expected
outcome, and the assessment of alliance in alliance to predict outcomes and hypothe-
individual and family therapy. sized that this association might be moder-
In a subsequent meta-analysis, Karver, ated by timing of alliance assessment. Prior
Handelsman, Fields, & Bickman (2006) research on relationship processes in child
specifically examined alliance–outcome rela- and adolescent treatment has been criticized
tions for child and adolescent clients. These for the inclusion of studies involving the
estimates did not include parent and family concurrent measurement of alliance and
alliance data. Karver et al. (2006) identified outcome (Shirk & Karver, 2003). It has been
10 studies that assessed youth alliance in suggested that concurrent measurement
relation to outcome. Correlations varied inflates estimates of association. In order to
widely across studies and ranged from 0.05 address this possibility, we examined associa-
to 0.49 with a weighted mean correlation of tions by timing of alliance measurement.
0.21. Like the original meta-analysis by Shirk Second, some evidence suggests that parent
and Karver (2003), these results indicate a alliance may be more strongly associated
moderate association between alliance and with treatment continuation and attendance
outcome in child and adolescent therapy. than treatment outcome. (Hawley & Weisz,
2005), thus we examined this possibility by
Meta-Analytic Review comparing correlations by type of alliance
In recent years, there has been significant (parent vs. youth). Third, consistent with
growth in the number of studies evaluating prior results (Shirk & Karver, 2003), we did
alliance–outcome relations in child and not expect a difference in the strength of
adolescent therapy. In order to provide an alliance–outcome relations by treatment
estimate of this association based on a larger type; instead, we expected comparable asso-
sample of studies than previously reported, ciations across behavioral and nonbehavioral
we conducted a meta-analysis of current therapies. We did, however, predict differ-
evidence on alliance–outcome relation- ences in alliance–outcome relations as a
ships in child and adolescent therapy. Our function of type of problem treated, as pre-
meta-analysis is restricted to child- and vious research (Shirk & Karver, 2003) found
adolescent-focused treatments and does somewhat stronger alliance–outcome rela-
not include studies of family therapy (which tions for youth with externalizing (disrup-
are reviewed in Chapter 4). Consistent with tive) problems than with internalizing
prior youth meta-analyses, we included (emotional) problems. Therefore, we exam-
both prospective and concurrent assess- ined outcomes by type of problem. Finally,
ments of alliance and outcome because we did not expect age-related differences in
of the limited number of studies in the alliance–outcome relations. Comparisons
literature. However, we provide separate were made between studies with child sam-
estimates of association for each design. ples and those with adolescent samples.
Because parents are often included in child
and adolescent therapy, and in fact, they Alliance–Outcome Studies
may be the focus of behavioral treatments, To identify applicable studies that mea-
we also provide an estimate of association sured the relationship between alliance and
between parent alliance and outcome. outcome, a three- pronged approach was
Our analyses were guided by a number of used. First, prior reviews of the alliance-
hypotheses based on prior findings in the to-outcome relationship were examined for

76 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
qualifying manuscripts (Karver et al., 2006; Calculation of Effect Sizes
Shirk & Karver, 2003). Citations of these Because most studies reported results (alli-
articles were then examined as a means to ance-to-outcome relationships) as corre-
identify additional manuscripts. Second, lations, the product-moment correlation
the PsycINFO database was searched from coefficient r was used as our effect size
2004 forward to identify articles that have estimate. All results (typically product-
been published since the last major meta- moment correlation coefficient rs) in each
analytic review of the therapeutic alliance study were converted to Fisher’s Z in order
in child and adolescent therapy. Finally, to normalize the r distribution (Hedges &
Google Scholar was used to search for Olkin, 1985). For all studies it was possi-
studies that may have been missed and ble to compute effect sizes, thus no effects
for unpublished manuscripts. For both were imputed as zero. In most studies, more
searches, child and adolescent were used in than one alliance–outcome relationship
conjunction with the terms alliance or rela- was reported. In order to correct for bias
tionship and therapeutic or therapy. due to correlated effects within studies and
To be included in the current meta- an unequal number of associations reported
analysis, studies had to meet the following in different studies, we averaged (simple
criteria: (1) the study had to include a spe- mean) the Fisher’s Z’s for each study. In
cific measure explicitly described in the order to calculate a more precise estimate
manuscript as an alliance measure; (2) the of the overall relationship between alliance
alliance had to be related to some indicator and outcome, we weighted the average
or measure of posttreatment outcome and effect size (Z) for each study by the number
not another process variable; (3) the study of participants in the study. We weighted
had to be of individual or group mental each effect size so that the final estimate of
health treatment delivered to a youth under the alliance-to-outcome relationship prop-
age 18 or a parent; (4) the study could not erly accounts for the fact that more precise
be an analog study; (5) the study needed to estimates should be given more weight in
be available in English; (6) the study must the aggregate. The weighted effect sizes for
have included at least 10 participants; and each study were aggregated, and then this
(7) if the study did not directly report a sum was divided by the sum of the weights
correlation between alliance and outcome, (number of participants per study minus 3)
enough information had to be available in for each study, resulting in an estimate of
the manuscript to calculate the effect size. the overall alliance-to-outcome relationship.
The resulting sample consisted of 29 stud- This weighted effect size Z was then con-
ies with 2,202 youth clients and 892 parents. verted back to the product-moment correla-
Studies were coded for type of alliance (youth tion coefficient r. We then analyzed effect
vs. parent), timing of alliance measurement size estimates by the type of alliance, timing
(early- or middle- during first two thirds of of alliance measurement, type of therapy,
therapy sessions, late- during the last third of type of problem treated, and child age.
therapy sessions, at termination, or posttreat-
ment, or combined), type of treatment (cog- Results of the Meta-Analysis
nitive-behavioral/behavioral, nonbehavioral, The 29 studies that met inclusion criteria are
or mixed), type of problem (internalizing, displayed in Table 3.1. The sample includes
externalizing, mixed, or substance abuse), 26 published studies and 3 doctoral disser-
and mean age of youth (child = less than 12; tations. Twenty-eight studies evaluated the
adolescent = ages 12–18). child or adolescent alliance, and 10 studies

s h i rk , k a rve r 77
Table 3.1 Reviewed Studies, Alliance Measures, Classifications, and Effect Sizes
Study N Alliance measure Classifications Wt. mean r
Adler (1998) 92 Parent Evaluation Questionnaire Both ages 0.24
Mixed problems
Auerbach et al. (2008) 39 Working Alliance Inventory - Short Adolescent 0.12
Substance Abuse
Champion (1998) 19 Child Behavior in Therapy Scale Child 0.18
Mixed problems
Chiu et al. (2009) 34 Therapy Process Observation Child 0.21
System - Alliance Internalizing
Colson et al. (1991) 69 Therapeutic Alliance Difficulty Scale Adolescent 0.28
Mixed problems
Creed & Kendall (2005) 68 Therapeutic Alliance Scale for Children Adolescent 0.30
Darchuck (2007) 40 Working Alliance Inventory - Short Adolescent 0.25
Substance Abuse
Diamond et al. (2006) 353 Working Alliance Inventory - Short Adolescent 0.20
Substance abuse
Eltz et al. (1995) 38 Penn Helping Alliance Questionnaire Adolescent 0.32
Mixed problems
Florsheim et al. (2000) 78 Working Alliance Inventory Adolescent 0.12
Gavin et al. (1999) 60 Treatment Alliance Scale Adolescent 0.03
Mixed problems
Green (1996) 25 Family Engagement Scale Child 0.58
Empathy and Understanding Scale Mixed problems
Green 20 Family Engagement Scale Adolescent −0.04
Empathy and Understanding Scale Mixed problems
Handwerk et al. (2008) 71 Working Relationship Scale Adolescent 0.25
Mixed problems
Hawley & Garland (2008) 78 Working Alliance Inventory - Short Adolescent 0.29
Mixed problems
(Continued )

Table 3.1. Continued
Study N Alliance measure Classifications Wt. mean r
Hawley & Weisz (2005) 81 Therapeutic Alliance Scale for Children Both 0.13
Mixed problems
Hintikka (2006) 45 Working Alliance Inventory Adolescent 0.07
Mixed problems
Hogue et al. (2006) 56 Vanderbilt Therapeutic Alliance Scale Adolescent −0.02
Substance abuse
Holmqvist et al. (2007) 59 Penn Helping Alliance Questionnaire Adolescent 0.13
Karver et al. (2008) 23 Alliance Observation Coding System Adolescent 0.08
Working Alliance Inventory Internalizing
Kaufman et al. (2005) 45 Working Alliance Inventory - Short Adolescent 0.00
Mixed problems
Kazdin et al. (2006) 310 Working Alliance Inventory Child 0.29
Therapeutic Alliance Scale for Children Externalizing
Kazdin et al. (2005) 185 Working Alliance Inventory Child 0.21
Kazdin & Whitley (2006) 218 Working Alliance Inventory Child 0.30
Therapeutic Alliance Scale for Children Externalizing
McLeod & Weisz (2005) 22 Therapy Process Observation Child 0.25
System - Alliance Internalizing
Shirk et al. (2008) 50 Therapeutic Alliance Scale Adolescent 0.26
for Adolescents Internalizing
Smith et al. (2008) 55 Penn Helping Alliance Questionnaire Adolescents 0.36
Mixed problems
Tetzlaff et al. (2005) 434 Working Alliance Inventory - Short Adolescent 0.24
Substance abuse
Zaitsoff et al. (8) 36 Penn Helping Alliance Questionnaire Adolescent 0.48
Eating Disorders

examined the parent alliance. In terms of an rw = 0.38 (CI = +/− 0.24). As predicted,
timing of alliance measurement, 15 studies studies that included a measure of alliance
assessed the alliance early or in the middle of later in treatment, either as part of an aver-
treatment, 8 studies measured alliance late or age, change score, or slope, or simply as a
posttreatment, and 8 studies assessed alliance late measure of alliance, yielded larger effects
over time as an average and 2 assessed alliance ( p < 0.05) than studies that measured alli-
as a slope or change score. In terms of types of ance early in treatment (See Figure 3.1).
treatment, 20 studies involved nonbehavioral With regard to client characteristics, age
therapy, 14 involved behavioral or cognitive- and type of problem were examined as
behavioral therapy, and 4 involved mixed potential moderators. Studies with child
therapies, usually the inclusion of family ther- samples (under age 12) yielded significantly
apy with individual therapy. Eleven studies larger weighted mean correlations (r = 0.27;
included child samples, 12 included only CI = +/− 0.08) than studies with adolescent
adolescents, and 5 studies included both chil- samples (r = 0.17; CI = +/− 0.05). A number
dren and adolescents. One study could not of studies included both children and ado-
be classified. Finally, 9 studies focused on lescents and produced a mean correlation of
internalizing problems, 7 on externalizing 0.12 (CI= +/− 0.10), also significantly dif-
problems, 5 on substance abuse, 6 on mixed ferent from the child estimate ( p < 0.05).
problems, and 1 on eating disorders. As Previous research (e.g, Shirk & Karver,
shown in Table 3.1, weighted correlations 2003) did not find age-related differences in
averaged 0.19 with a confidence interval of alliance–outcome associations, and this find-
+/− 0.04 (range = −.09 to 0.59). This esti- ing is somewhat surprising given the clinical
mate is very similar to earlier results for the focus on alliance difficulties with adolescents
association between relationship variables (Castro-Blanco & Karver, 2010).
and outcomes (rw = 0.20), but slightly lower Effect sizes did not differ across types
than the estimate based on studies that mea- of treatment. The weighted mean correla-
sured alliance and outcome (rw = 0.25). The tions for behavioral/cognitive behavioral,
current meta-analysis includes over three nonbehavioral, and mixed therapies were
times as many alliance studies and excluded 0.18 (CI = +/− 0.05), 0.19 (CI = +/− 0.06),
studies of family therapy. and 0.20 (CI = +/− 0.08), respectively.
These results are consistent with earlier
Mediators and Moderators findings reported by Shirk and Karver
A number of variables were expected to (2003). Nor was there a difference in the
moderate the strength of association between weighted mean correlation for therapist–
alliance and outcome. One methodological parent and therapist–youth alliance. Both
variable was timing of alliance assessment. It associations averaged rw=19.
was hypothesized that concurrent measure- Consistent with earlier findings (Shirk &
ments would show stronger associations Karver, 2003), strength of alliance–outcome
than prospective designs. Studies that relations varied as a function of type of treated
assessed alliance early produced an rw = 0.15 problem. The weighted mean correlation for
(CI = +/− 0.06), those that measured it late externalizing problems was 0.24 (CI = +/−
or posttreatment resulted in an rw = 0.24 0.07). In contrast, internalizing problems,
(CI = +/− 0.09), those that used an average substance abuse, and mixed problems showed
resulted in an rw = 0.21 (CI = +/− 0.07), weighted mean correlations of 0.17 (CI =
and the two that used a slope or a change +/− 0.11), 0.14 (CI = +/− 0.07), and 0.20
score across early and late alliances yielded (CI = +/− .07), respectively. One study

80 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
evaluated youth with eating disorders and outcome in child and adolescent therapy.
and produced a mean correlation of 0.53 Estimates of alliance–outcome relations in the
(CI = +/− 0.34). The strength of alliance– child and adolescent literature may be slightly
outcome relations differed between studies inflated by the inclusion of studies that only
with externalizing samples and those with measure alliance late in treatment. However,
substance abuse samples ( p < 0.05). Other studies that assessed alliance with methods
differences were not reliable. The difference that included early measures of alliance—-
in alliance–outcome effect size for youth with through averaged scores, slopes, or change
externalizing versus internalizing problems scores—produced effects that were slightly
was in the same direction as previous research, larger than the overall estimate for all studies.
but it did not attain statistical significance. Thus, results support the view that the thera-
Despite theoretical models that posit a peutic alliance is an important predictor of
mediated relation between alliance and outcome in youth therapy.
outcome, no study examined mediation The strength of alliance–outcome rela-
through involvement in specific treatment tions did not vary with type of treatment.
tasks. Only one study (Karver et al., 2008) The alliance thus appears to be important
demonstrated a strong link between early for therapies that vary widely in terms
alliance and later task involvement. of specific treatment procedures, including
therapies that focus on teaching contingency
Summary of Meta-Analytic Findings management to parents. The alliance is an
Results from our meta-analysis indicate a important component of broad classes of
small-to-medium association between alliance child and adolescent therapy, and it appears


Effect size




Early (k = 113) Mid (k = 33) Late (k = 36) Average (k = 52)

Session alliance rated

Fig. 3.1 Alliance effect sizes and confidence intervals by time of measurement

s h i rk , k a rve r 81
to contribute to outcomes in treatments as therapy outcomes? One way of thinking
different as manual-guided CBT and non- about this issue is to use a common metric
directive, play therapy. It is noteworthy and to benchmark effects from comparative
that alliance also is predictive of outcome outcome studies with the results of the cur-
in parent-focused therapies. rent meta-analysis. The mean correlation
The strength of association between alli- obtained in this study, rw = 0.19, converts
ance and outcome varied as a function of to an effect size, d, of approximately 0.39.
several variables. Our results showed a slightly How does this compare with effect sizes
stronger association between alliance and obtained by comparing two or more youth
outcome among children compared with treatments involving different procedures?
adolescents. A prior meta-analysis did not Two recent estimates are relevant. In a
find such a difference (Shirk & Karver, comparison of evidence-based treatments
2003). It is possible that age-related dif- (EBTs) with usual-care (UC) therapy, Weisz,
ferences in alliance–outcome associations Jensen-Dose, and Hawley (2006) found
are confounded with the typical problems EBTs to be superior to UC with an effect size
treated in these two age groups. Specifically, of 0.25 for direct comparisons of psychother-
child studies often include samples of disrup- apies. In an effort to estimate the impact of
tive children and adolescent studies include specific treatment methods on outcome,
substance abuse samples. Given that correla- Miller et al. (2008) conducted a meta-
tions were stronger for externalizing samples analysis of youth comparative outcome stud-
than substance abuse samples, the age differ- ies in which two or more bona fide treatments
ences in alliance–outcome associations could were evaluated. Although they found some
be due to age differences in problem types. evidence for method effects, the total out-
It is not clear why stronger associations come variance explained by treatment
between alliance and outcome are found method was 0.037, or when converted to a
with externalizing youth, as they were in correlation, an r of 0.19, identical to the
our earlier meta-analysis (Shirk & Karver, mean correlation obtained in our current
2003). We have speculated that the chal- meta-analysis. Thus, when benchmarked to
lenge of engaging oppositional and disrup- findings from comparisons of treatment
tive youth increases variability in alliance, methods, the contribution of alliance to out-
and thus increases the possibility of larger come appears similar to the contribution of
correlations than those obtained in samples specific methods. Of course, it is possible that
with a more attenuated range of alliance differences in specific methods are more criti-
scores. It is also possible that alliance plays cal for some disorders than others, for exam-
an especially critical role in the treatment ple, obsessive-compulsive disorder versus
of disruptive problems, possibly by facili- depression, just as alliance appears to be more
tating the internalization of skills or an strongly related to outcome for externalizing
empathic attitude toward others. compared to internalizing disorders.
The question of how important the A critical question, then, is the relation
alliance is to youth outcomes deserves between the alliance and specific treatment
some comment. When viewed in terms of factors in child and adolescent therapy.
explained variance, the estimate of less than Are these factors complementary and do
4% of total outcome variance seems rather they contribute to outcome in an additive
small. However, an important question is manner, or might alliance actually account
how does this compare with the contribu- for differences in outcome currently
tion of specific treatment methods to youth attributed to specific methods? Only one

82 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
published study in the child and adolescent One exception is motivation for change.
literature has attempted to evaluate the latter Consistent with the adult literature (see
possibility. Kaufman and colleagues (2005) Chapter 2), results generally show that
examined the alliance as a potential media- youth with greater motivation, more prob-
tor of treatment effects obtained in a com- lem recognition, and more stated reasons
parison of group CBT and a Life Skills for changing their behavior form more
group for adjudicated, depressed adoles- positive alliances (Christensen & Skogstad,
cents. These investigators found more posi- 2009; Estrada & Russell, 1999; Fitzpatrick
tive alliances in the group CBT condition & Irannejad, 2008), though one study
than in the Life Skills condition, but this failed to find these associations (Garner,
difference in alliance did not account Godley, & Funk, 2008).
for significant variance in depression out- Within the youth clinical literature, ado-
comes. Unfortunately, few comparative lescence is viewed as one of the most difficult
outcome studies have assessed the alliance periods for alliance formation (Castro-
in the youth literature; consequently, the Blanco & Karver, 2010; Meeks, 1971). The
relative contribution of alliance and specific developmental press toward autonomy, the
factors cannot be readily addressed. increasing centrality of peer relationships,
and growing doubts about adults’ capacity
Client and Therapist Contributions for understanding youth experiences can
A small but growing number of studies contribute to alliance difficulties. One might
have examined predictors of the therapeu- expect, then, relatively clear evidence for
tic alliance in child and adolescent therapy. more positive alliances among children than
Emerging evidence points to a number of adolescents. Yet, the limited evidence is
client characteristics and therapist behav- mixed with two studies showing more posi-
iors associated with alliance formation. tive alliances among older youth (Garner,
Godley, & Funk, 2008; Gavin, Wamboldt,
Pretreatment Predictors of Child Sorokin, Levy, & Wamboldt, 1999), and
and Adolescent Alliance two studies reporting younger children
Numerous clinical accounts have high- forming better alliances (Creed & Kendall,
lighted some of the potential challenges to 2005; DeVet, Kim, Charlot-Swilly, & Ireys,
engaging children and adolescents in a 2003). It is certainly possible that adoles-
working alliance (Castro-Blanco & Karver, cents’ greater capacity for understanding
2010; A. Freud, 1946; Meeks, 1971). therapy rationale and tasks might contribute
Among the obstacles to alliance formation to better alliances among older than younger
are the limited problem recognition and youth. However, the paucity of research evi-
acknowledgement, the tendency to attri- dence on this issue pales in comparison with
bute problems to extermal sources, low practice-based observations of alliance diffi-
motivation for change, absence of self-re- culties with adolescents. Given that most
ferral (or the presence of coaxed or coerced treatment studies focus on either children or
referral), and a lack of understanding of the adolescents, direct comparisons of alliance
therapy processes. Although some of these processes are likely to remain limited.
factors could be indicative of psychopathol- Two other pretreatment factors have
ogy, most are simply a consequence of attracted some attention in the research
developmental level. Surprisingly few stud- literature: type and severity of psychopathol-
ies have examined direct links between these ogy and interpersonal functioning. Research
developmental factors and youth alliance. on symptom severity and alliance formation

s h i rk , k a rve r 83
has produced mixed results. Three studies Some studies have taken a more indirect
found no relationship between initial symp- approach to looking at the contribution of
tom severity and subsequent alliance pretreatment relationship factors to alliance
(Bickman et al., 2004; DeVet et al., 2003; formation. The idea behind the indirect
Eltz et al., 1995), one found a negative rela- approach is that if youth have poor family
tionship between overall level of maladaptive relationships or low levels of social support,
functioning and alliance (Green, Kroll, Imrie, these factors might indicate poor inter-
Frances, Begum et al., 2001), and two actu- personal skills or poor prior experiences of
ally showed a positive relationship between healthy relationships. Consistent with this
initial severity and alliance (Christensen & perspective, a number of studies have found
Skogstad, 2009; Shirk et al., 2008). It is likely social support—both youth and parent
that greater clarity will be attained by exam- reported—to be related to youth and/or
ining specific symptoms (e.g., depressive parent alliances (DeVet et al., 2003; Fields
symptoms) and alliance rather than overall et al., 2010; Garner, Godley, & Funk,
symptom severity. For example, clinical 2008; Hawley & Garland, 2008; Kazdin
accounts have highlighted the unique chal- & Whitley, 2006). As these results suggest,
lenges of engaging youth with high levels of alliance formation is a social process, and
oppositional and rule-breaking behavior youth relational experiences and compe-
(Gallager, Kurtz, & Blackwell, 2010). High tencies appear to impact the alliance.
levels of defiance, distrust of adult authority, Finally, it should be noted that gender and
and externalization of problems have been race have been considered in a small number
found to make alliance formation especially of studies. The evidence is split on the role of
difficult with this group. (Bickman et al., gender; two studies showed no association
2004; Garcia & Weisz, 2002) between gender and alliance (Creed &
In contrast to symptom severity, inter- Kendall, 2005; Fitzpatrick & Irannejad,
personal variables appear associated with 2008), but three showed females to rate the
youth alliance. A recurrent finding in the alliance more positively than male youth
adult literature is that quality of past (Christensen & Skogstad, 2009; Eltz, Shirk,
and present relationships predict alliance & Sarlin, 1995; Wintersteen, Mensinger, &
quality (e.g., Hersoug, Monsen, Havik, Diamond, 2005). One study showed the
& Hoglend, 2002; Mallinckrodt, Coble, & opposite with males rating the alliance more
Gantt, 1995). One study (Eltz, Shirk, & positively than females (Hawke, Hennen &
Sarlin, 1995) found that youth with more Gallione, 2005). Given that the majority of
interpersonal problems, but not greater child and adolescent therapists are female, it
overall problem severity, had more alliance would be useful to know if gender matching
difficulties, and another (Fields et al., 2010) has an impact on alliance, and if so, for whom.
found social competence to be related to The evidence with regard to race effects is
alliance. In the former study, a history extremely limited. Hawke and colleagues
of child maltreatment predicted early alli- (2005) found Hispanic and African-American
ance difficulties even after controlling for youth to have stronger youth–therapist
problem severity. Perhaps it should not be alliances than European-American youth,
surprising that youth with interpersonal but Wintersteen et al. (2005) found no
trauma histories, especially abuse in the race effects. The impact of therapist–youth
context of the family, would be cautious matching on race has not been examined.
forming a close relationship with an adult The reality is that we know very little about
caregiver (therapist). whether gender, race, or matching on these

84 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
variables is related to the alliance in youth patronizing, predicted weaker therapist
therapy. reported alliance at Session 7.
Two findings seem especially important.
Therapist Strategies for Strengthening First, therapist collaborative behavior,
the Youth Alliance including the establishment of treatment
Despite the importance of the alliance in goals with the child, may be critical for alli-
youth therapy, research on therapist behav- ance formation. Recognition and valida-
iors that contribute to alliance formation tion of the child’s perspective on this critical
has only recently emerged. With the excep- issue may help differentiate the therapist
tion of research on family therapy (Diamond, from parents, and help the child view the
Liddle, Hogue, & Dakof, 1999), all of therapist as an ally. Second, therapists need
these studies have been conducted with to balance their approach between setting a
children and adolescents in individual CBT. collaborative tone without being either too
Although generalizations to other forms of formal or overly familiar with the child.
child therapy would be premature, some Similarly, therapists need to balance their
patterns have emerged across these initial focus on problems with the maintenance of
studies and deserve consideration. rapport. Too little focus on problems can
Two studies address therapist strate- amount to supporting avoidance, and too
gies with children. Creed and Kendall (2005) great a focus can undermine alliance.
examined a set of therapist behaviors hypoth- Therapists must be mindful of the child’s
esized to promote or interfere with alliance level of tolerance and work to gradually
formation during the first three sessions of facilitate the child’s ability to talk about anxi-
CBT with anxious children. Child and ther- ety. Therapist flexibility about the pace of
apist reports were used to assess the alliance treatment may be critical for alliance
at Sessions 3 and 7. Child-reported alliance formation.
at Session 3 was positively associated with Chu and Kendall (2009) evaluated thera-
therapist collaboration strategies, including pist flexibility as a strategy for promot-
presenting therapy as a team effort, building ing client involvement in CBT for child
a sense of togetherness by using words like anxiety disorders. In this context, flexibility
“we,” “us,” and “let’s,” and by helping the is conceptualized as treatment adapta-
child set goals for therapy. In contrast, two tion occurring within the parameters of
therapist verbal behaviors predicted a weaker treatment fidelity: this is a way of indivi-
alliance at Session 3. Not surprisingly, “push- dualizing manual-guided therapy. In fact,
ing the child to talk” about anxiety and anx- therapist flexibility was significantly associ-
ious situations was negatively associated with ated with increases in child involvement
early child alliance. Similarly, therapist efforts over the course of therapy. Later child
to “emphasize common ground,” that is, involvement predicted treatment improve-
therapists’ comments like “Me, too!” in ment and lower levels of impairment. Results
response to children’s statements about inter- from this study, in conjunction with find-
ests and activities were predictive of weaker ings reported by Creed and Kendall (2005),
alliances. For therapist-reported alliance, suggest that therapists who provide CBT in
none of the therapist behaviors predicted a flexible manner, who are less didactic, less
alliance scores at Session 3, but collaborative pressing, and better able to integrate client
strategies predicted better alliances at needs within the treatment protocol are
Session 7. Talking to the child in an overly more likely to facilitate better alliances and
formal manner, that is, being too didactic or greater treatment involvement.

s h i rk , k a rve r 85
In addition to the foregoing research on focus in the first session. After initially social-
child anxiety, Shirk and colleagues have izing the adolescent to treatment, therapists
published a series of studies on engagement who reported more positive alliances later
of depressed adolescents. Their work has turned their attention to responding to ado-
focused on the evaluation of three clusters lescent concerns while providing some hope
of therapist engagement strategies—motiva- for change. In a complementary manner,
tional strategies that focus on goal setting therapists who opened and closed the first
and mobilizing clients’ intention to change, session with a focus on treatment socializa-
socialization strategies that focus on clarify- tion were more likely to have adolescents
ing roles and tasks in therapy, and experien- report positive alliances than therapists who
tial strategies that focus on eliciting the did not follow this pattern.
client’s experience and the provision of Finally, Jungbluth and Shirk (2009)
support. In addition, alliance-impeding uncovered an interesting pattern in their
behaviors (therapist lapses) were examined, analysis of first-session engagement behav-
including therapists’ failure to acknowledge ior and subsequent adolescent involvement
client emotion, therapists misunderstanding in CBT for depression. Therapists who
client’s statements, and therapist criticism. provided less structure in the first session
In their first study, Karver et al. (2008) were more likely to have adolescents who
examined engagement strategies in relation were highly involved in CBT tasks in later
to early alliance in a small sample of adoles- sessions than therapists who initiated ther-
cents who had attempted suicide and were apy with high structure. Low structure is
treated with either problem-solving therapy not equivalent to therapist inactivity but,
or nondirective, supportive therapy. Results rather, indicates greater exploration of ado-
showed that therapist lapses were the most lescents’ experiences and motives as well as
robust predictor of subsequent alliance greater provision of support.
across both conditions. Failure to respond Although the foregoing results are prom-
to expressed emotion was one of the most ising, they also must be considered prelimi-
characteristic problems. nary. At present only five studies have been
In a second study, Russell, Shirk, and published that address the contribution
Jungbluth (2008) examined the same set of of therapist behaviors and strategies to alli-
engagement strategies in a school-based ance formation and therapy involvement.
trial of CBT for adolescent depression. All of these studies have been conducted in
A unique feature of this study was the pre- the context of CBT, and only two child-
diction of subsequent alliance from tempo- hood disorders, anxiety and depression,
ral patterns in therapist behavior rather than have been considered. Nevertheless, some
from overall frequency of behavior. Reliable initial patterns can be discerned.
temporal patterns in therapist engagement Results with children suggest that “push-
behavior were identified. Therapists who ing” child clients, especially in a more
steadily increased their focus on being formal, didactic manner, is counterproduc-
responsive and remoralizing the adolescent tive for alliance formation. With adoles-
after the first 10 minutes of their sessions, cents there is evidence that greater attention
but who then dampened the rate of increase to the teen’s experience, especially in a
over the rest of the session, reported way that acknowledges the adolescent’s
more positive alliances two sessions later. perspective and expressions of emotion, is
These results suggest that more positive alli- associated with more positive alliance and
ances are associated with a shift in therapist involvement. Taken together, these results

86 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
suggest that a less directive and less task- First, too many studies still rely on out-
focused approach to therapy is critical at come assessments from a single source. As a
the start of treatment. Subsequent alliance result, alliance–outcome relations can be
and involvement appear to benefit from inflated by shared source variance. Second,
therapists taking the time to attend to their studies that include multiple outcome mea-
client’s experiences and life stories. Efforts to sures from multiple sources often fail to
engage the client by pushing or praising are distinguish among primary and secondary
contraindicated. This pattern is somewhat outcomes. Although it makes sense at this
inconsistent with the emphasis on psycho- early stage of research to explore a wide
education as a method of treatment social- range of outcomes, too few associations are
ization and suggests that client-centered based on specific hypotheses. It is possible
strategies at the start of therapy may be that the alliance as an interpersonal con-
more effective for alliance formation. struct will predict interpersonal outcomes
like changes in relational schema, social
Limitations of the Research engagement, or support-seeking behaviors.
As indicated in our review of definitions Third, research must evaluate the tempo-
and measures, the field has yet to reach ral relationship between alliance and out-
consensus about the crucial features of the comes. Research has revealed early gains in
therapeutic alliance with youth. We were child and adolescent therapy, yet no study
not surprised to find a large number of alli- has examined early alliance in relation to
ance measures in a relatively small number early symptom changes. This last issue is
of studies. One way to advance the field particularly important, as no study in the
would be to conduct a study in which mul- youth literature has ruled out the possibility
tiple measures of alliance are administered that alliance is actually predicted from early
in order to derive core underlying dimen- improvement. Designs that account for pos-
sions. Such an empirical approach could sible symptom changes prior to the measure
anchor future development of the alliance of alliance are clearly needed. And finally,
construct with youth. no study in the youth literature has evalu-
Our meta-analysis revealed some impor- ated the contribution of alliance to outcome
tant progress in research on alliance– while controlling for client adherence to
outcome relations with children and specific treatment tasks. Alliance may very
adolescents. First, the number of studies well predict involvement in specific therapy
has more than tripled since the first meta- components (Karver et al., 2008), but too
analysis of relationship predictors of child few studies assess alliance in relation to other
outcomes. Many new studies do not mea- important process variables.
sure alliance and outcome concurrently, as In conclusion, the alliance has a long
was found in the earlier meta-analysis, but history in the child and adolescent litera-
examine associations prospectively. And ture. Recent research progress on alliance–
there is clearly a trend toward assessing alli- outcome relations indicates that this
ance at multiple points in treatment and long-standing interest is clearly justified.
evaluating alliance trajectories in relation Alliance is a predictor of youth therapy
to outcomes. Finally, a growing number outcomes and may very well be an essential
of studies assess alliance from multiple ingredient that makes diverse child and
perspectives—client, therapist, parent, and adolescent therapies work. Future research
observer. Despite this progress, areas for on the relative contribution of alliance and
improvement remain. specific factors to youth outcomes, as well

s h i rk , k a rve r 87
as the contribution of alliance to child and • Although it can be tempting to try to
adolescent utilization of specific treatment connect with young clients by finding
procedures, will surely clarify its clinical shared interests and activities, initial
importance. evidence suggests that such efforts may be
counterproductive. Alliance formation
Therapeutic Practices appears to be better served by emphasizing
• Alliances with both youth and their the collaborative nature of therapy.
parents are predictive of treatment • Attending to client’s experiences and
outcomes. Consequently, psychotherapists acknowledging their expressed emotion are
need to attend to the development of crucial for alliance formation, especially
multiple alliances, not just to the alliance with adolescents. Providing an opportunity
with the youth. A solid alliance with the for client-directed interaction at the start
parent may be particularly important for of psychotherapy appears to set the stage
treatment continuation. for subsequent treatment involvement.
• Parents and youth often have divergent
views about treatment goals. Formation of
a therapeutic alliance with both youth and
parent requires the therapist to attend to An asterisk (∗) indicates studies included in
multiple perspectives and to develop a the meta-analysis.

Adler, A. G. (1998). The alliance between child
treatment plan that accommodates both therapist and parent: How it predicts treatment
youth and parent perspectives. outcome and reflects parent characteristics.
• The maintenance of a positive alliance Unpublished dissertation.

over time predicts successful outcomes with Auerbach, S., May, J., Stevens, M., & Kiesler, D.
youth. Therapists are advised to monitor (2008). The interactive role of working alli-
ance and counselor-client interpersonal behav-
alliance over the course of treatment. iors in adolescent substance abuse treatment.
Alliance formation is not simply an early International Journal of Clinical and Health
treatment task, it is a recurrent task. Psychology, 8(3), 617–629.
• Youth are likely to have a limited Axline, V. (1947). Play therapy. Boston: Houghton
understanding of therapy. Initial results Mifflin.
suggest that early alliance formation with Bickman, L., Andrade, A., Lambert, E., Doucette, A.,
Sapyta, J., Boyd A., et al. (2004). Youth thera-
youth requires the therapist to balance peutic alliance intensive treatment settings. The
active listening to the youth with Journal of Behavioral Health Services and Research.
providing an explicit framework for 31(2), 134–148.
understanding therapy processes (roles, Bordin, E. S. (1979). The generalizability of the
tasks, relevance). Overemphasizing the psychoanalytic concept of the working alliance.
latter to the exclusion of the former Psychotherapy: Theory, Research, and Practice, 16,
appears to interfere with alliance Castro-Blanco, D., & Karver, M. (2010). The elusive
formation, at least with adolescents. alliance: Treatment engagement strategies with with
• “Pushing” youth to talk about high-risk adolescents. Washington, DC: APA Books.

emotionally sensitive material appears to Champion, K. M. P. (1998). The change process
undermine alliance, at least among in therapeutic programs for children. (ProQuest
younger clients. Therapists can “scaffold” Information & Learning). Dissertation Abstracts
International: Section B: The Sciences and
young clients’ emotional disclosures by Engineering, 59, 90.
initially focusing on emotional experiences ∗
Chiu, A. W., McLeod, B. D., Har, K., & Wood, J. J.
of low intensity and gradually attending (2009). Child-therapist alliance and clinical out-
to emotions of higher intensity. comes in cognitive behavioral therapy for child

88 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
anxiety disorders. Journal of Child Psychology and Estrada, A., & Russell, R. (1999). The development
Psychiatry, 50, 751–758. of the child psychotherapy process scales (CPPS).
Chu, B., & Kendall, P. (2009). Therapist respon- Psychotherapy Research, 9(2), 154–166.
siveness to child engagement: Flexibility within Faw, L., Hogue, A., Johnson, S., Diamond, G. M.,
manual-based CBT for anxious youth. Journal & Liddle, H. A. (2005). The Adolescent
of Clinical Psychology, 65(7), 736–754. Therapeutic Alliance Scale (ATAS): Initial psy-
Christensen, M., & Skogstad, R. (2009). What chometrics and prediction of outcome in family-
predicts quality of the therapeutic alliance in a based substance abuse prevention counseling.
cognitive behavioral treatment for children with Psychotherapy Research, 15(1–2), 141–154.
anxiety disorders? Therapeutic alliance measured Fields, S., Handelsman, J., Karver, M. S., &
from the patient, therapist and observer perspec- Bickman, L. (2010). Youth and parent predictors
tive. Unpublished manuscript. Det Psykologiske of the therapeutic alliance. Manuscript submitted.
Facultet. Fitzpatrick, M., & Irannejad, S. (2008). Adolescent

Colson, D.B., Cornsweet, C., Murphy, T., readiness for change and the working alliance in
O’Malley, F., Hyland, P.S., McParland, M., counseling. Journal of Counseling & Development.
Coyne, L. (1991). Perceived treatment difficulty 86(4), 438–445.

and the therapeutic alliance in an adolescent Florsheim, P., Shotorbani, S., Guest-Warnick, G.,
psychiatric hospital, American Journal of Ortho- Barratt, T., & Hwang, W. (2000). Role of the
psychiatry, 61, 221–229. working alliance in the treatment of delinquent

Creed, T., & Kendall, P. (2005). Therapist alliance- boys in community-based programs. Journal of
building behavior within a cognitive-behavioral Clinical Child Psychology, 29, 94–10.
treatment for anxiety in youth. Journal of Consult- Freud, A. (1946). The psychoanalytic treatment of chil-
ing and Clinical Psychology, 73(3), 498–505. dren. New York: International Universities Press.

Darchuk, A. J. (2007). The role of the therapeutic Garcia, J., & Weisz, J. (2002). When youth mental
alliance and its relationship to treatment out- health care stops: Therapeutic relationship prob-
come and client motivation in an adolescent lems and other reasons for ending youth outpa-
substance abuse treatment setting. (ProQuest tient treatment. Journal of Consulting and Clinical
Information & Learning). Dissertation Abstracts Psychology, 70(2), 439–443.
International: Section B: The Sciences and Garner, B., Godley, S., & Funk, R. (2008).
Engineering, 68, 5, 195. Predictors of early therapeutic alliance among
DeVet, K., Kim, Y., Charlot-Swilley, D., & Ireys H. adolescents in substance abuse treatment. Journal
(2003). The therapeutic relationship in child of Psychoactive Drugs, 40(1) 55–65.

therapy: Perspectives of children and mothers. Gavin, L. A., Wamboldt, M. Z., Sorokin, N.,
The Journal of Clinical Child & Adolescent Levy, S. Y., & Wamboldt, F. S. (1999). Treatment
Psychology, 32 (2), 277–283. alliance and its association with family func-
Diamond, G., Liddle, H., Hogue, A., & Dakof, G. tioning, adherence, and medical outcome in
(1999). Alliance-building interventions with adolescents with severe, chronic asthma. Journal
adolescents in family therapy: A process study. of Pediatric Psychology, 24(4), 355–365.

Psychotherapy: Theory, Research, Practice, Training, Green, J.M. (1996) Engagement and empathy:
36(4), 355–368. A pilot study of the therapeutic alliance in out-

Diamond, G.S., Liddle, H.A., Wintersteen, M.B., patient child psychiatry, Child Psychology and
Dennis, M.L., Godley, S.H., & Tims, F. (2006). Psychiatry Review, 1, 130–138.

Early therapeutic alliance as a predictor of treat- Green, J., Kroll, L., Imrie, D., Frances, F.,
ment outcome for adolescent cannabis users in Begum, K., Harrison, L., et al. (2001). Health gain
outpstient treatment. The American Journal of and outcome predictors during inpatient and
Addictions, 15, 26–33. related day treatment in child and adolescent
DiGiuseppe, R., Linscott, J., & Jilton, R. (1996), psychiatry. Journal of the American Academy of
Developing the therapeutic alliance in child- Child & Adolescent Psychiatry, 40(3), 325–332.

adolescent psychotherapy. Applied and Preventive Handwerk, M., Smith, G., Thompson, R.,
Psychology, 5, 85–100. Spellman, D., & Daly, D. (2008). Psychotropic

Eltz, M., Shirk, S., & Sarlin, N. (1995). Alliance medication utilization at a group-home residen-
formation and treatment outcome among mal- tial facility for children and adolescents. Journal
treated adolescents. Child Abuse & Neglect, 19(4), of Child and Adolescent Psychopharmacology,
419–431. 18(5), 517–525.

s h i rk , k a rve r 89
Hawke, J., Hennen, J., & Gallione, P. (2005). Karver, M., Handelsman, J., Fields, S., &
Correlates of therapeutic involvement among Bickman, L. (2006). Meta-analysis of relation-
adolescents in residential drug treatment. The ship variables in youth and family therapy:
American Journal of Drug and Alcohol Abuse, Evidence for different relationship variables in
31(1), 163–177. the child and adolescent treatment literature.

Hawley, K., & Garland, A. (2008). Working Clinical Psychology Review, 26, 50–65.

alliance in adolescent outpatient therapy: Youth, Karver, M., Shirk, S., Handelsman, J. B., Fields, S.,
parent and therapist reports and associations Crisp, H., Gudmundsen, G., et al. (2008).
with therapy outcomes. Child and Youth Care Relationship processes in youth psychotherapy:
Forum, 37(2), 59–74. Measuring alliance, alliance-building behaviors,
Hawley, K., & Weisz, J. (2003). Child, parent and and client involvement. Journal of Emotional and
therapist (dis)agreement on target problems in Behavioral Disorders, 16(1), 15–28.

outpatient therapy: The therapist’s dilemma and Kaufman, N., Rohde, P., Seeley, J., Clarke, G.,
its implications. Journal of Consulting and & Stice, E. (2005). Potential mediators of
Clinical Psychology, 71(1), 62–70. cognitive-behavioral therapy for adolescents

Hawley, K. M., & Weisz, J. R. (2005). Youth versus with comorbid major depression and conduct
parent working alliance in usual clinical care: disorder. Journal of Consulting and Clinical
Distinctive association with retention, satisfac- Psychology, 73(1), 38–46.

tion, and treatment. Journal of Clinical Child and Kazdin, A. E., Marciano, P. L., & Whitley, M. K.
Adolescent Psychology, 34, 117–128. (2005). Therapeutic alliance in cognitive-
Hedges, L. V., & Olkin, I. (1985). Statistical behavioral treatment of children referred for
methods for meta-analysis. New York: Academic oppositional, aggressive, and antisocial behavior.
Press. Journal of Consulting and Clinical Psychology, 73,
Hersoug, A., Monsen, J., Havik, O., & Hoglen, P. 726–730.
(2002). Quality of early working alliance in psy- Kazdin, A., & Wassell, G. (2000). Predictors of
chotherapy: Diagnoses, relationship and intra- barriers to treatment and therapeutic change in
psychic variables as predictors. Psychotherapy and outpatient therapy for antisocial children and
Psychosomatics, 71(1), 18–27. their families. Mental Health Services Research, 2,

Hintikka, U., Laukkanen, E., Marttunen, M., & 27–40.

Lehtonen, J. (2006). Good working alliance Kazdin, A. E., & Whitley, M. K. (2006).
and psychotherapy are associated with positive Pretreatment social relations, therapeutic alli-
changes in cognitive performance among ado- ance, and improvements in parenting prac-
lescent psychiatric inpatients. Bulletin of the tices in parent management training. Journal
Menninger Clinic, 70, 316–335. of Consulting and Clinical Psychology, 74(2),

Hogue, A., Dauber, S., Stambaugh, L., Cecero, C., 346–355.

& Liddle, H. (2006). Early therapeutic alli- Kazdin, A. E., Whitley, M., & Marciano, P. L.
ance and treatment outcome in individual and (2006). Child-therapist and parent-therapist
family therapy for adolescent behavior problems. alliance and therapeutic change in the treatment
Journal of Consulting and Clinical Psychology, of children referred for oppositional, aggressive,
74(1) 121–129. and antisocial behavior. Journal of Child

Holmqvist, R., Hill, T., & Lang, A. (2007). Psychology and Psychiatry, 47, 436–445.
Treatment alliance in residential treatment of Kendall, P., Comer, J., Marker, C., Creed, T., Puliafico,
criminal adolescents. Child and Youth Care A., Hughes, A., et al. (2009). In-session exposure
Forum, 36(4), 163–178. tasks and therapeutic alliance across the treatment
Horvath, A. O., & Greenberg, L. S. (1989). of childhood anxiety disorders. Journal of Consulting
Development and validation of the Working and Clinical Psychology, 77(3), 517–525.
Alliance Inventory. Journal of Counseling Linscott, J., DiGiuseppe, R., & Jilton, R. (1993).
Psychology, 36(2), 223–233. A measure of TA in adolescent psychotherapy. Poster
Jungbluth, N., & Shirk, S. (2009). Therapist presented at the 101st Annual Convention of the
strategies for building involvement in cognitive– American Psychological Association, Toronto,
behavioral therapy for adolescent depression. Canada.
Journal of Consulting and Clinical Psychology, Mallinckrodt, B., Coble, H., & Gantt, D. (1995).
77(6), 1179–1184. Working alliance, attachment memories, and

90 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
social competencies of women in brief therapy. ∗
Smith, B., Duffee, D., Steinke, C., Haung, Y., &
Journal of Counseling Psychology, 42(1), 79–84. Larkin, H. (2008). Outcomes in residential

McLeod, B., & Weisz, J. (2005). The therapy pro- treatment for youth: The role of early engage-
cess observational coding system-alliance scale: ment. Children and Youth Services Review, 30(12),
Measure characteristics and prediction of outcome 1425–1436.
in usual clinical practice. Journal of Consulting and Spielman, G., Gatlin, E., & McFall, J. (2010). The
Clinical Psychology, 73(3), 323–333. efficacy of evidence-based psychotherapy versus
Meeks, J. (1971). The fragile alliance: An orientation usual care for youths: Controlling confounds
to the outpatient psychotherapy of the adolescent. in a meta-analysis. Psychotherapy Research, 20,
Oxford England: Williams & Wilkins. 234–246.
Miller, S., Wampold, B., & Varhely, K. (2008). ∗
Tetzlaff, B. T., Kahn, J. H., Godley, S. H.,
Direct comparisons of treatment modalities for Godley, M. D., Diamond, G. S., & Funk, R. R.
youth disorders: A meta-analysis. Psychotherapy (2005). Working alliance, treatment satisfaction,
Research, 18(1), 5–14. and patterns of post-treatment use among ado-
Rogers, C. (1957). The necessary and sufficient lescent substance users. Psychology of Addictive
conditions for therapeutic personality change. Behavior, 19, 199–207.
Journal of Consulting Psychology, 21, 95–103. Weiss, B., & Weisz, J. (1995). Relative effective-
Russell, R., Shirk, S., & Jungbluth, N. (2008). ness of behavioral versus nonbehavioral child
First-session pathways to the working alliance psychotherapy. Journal of Consulting and Clinical
in cognitive-behavioral therapy for adolescent Psychology, 63(2), 317–320.
depression. Psychotherapy Research, 18(1), 15–27. Weisz, J., Jensen-Doss, A., & Hawley, K. (2006).
Sandler, J., Kennedy, H., & Tyson, R. (1980). The Evidence-based youth psychotherapy versus
techniques of child psychoanalysis: Discussions with usual clinical care: A meta-analysis of direct com-
Anna Freud. Cambridge, MA: Harvard University parisons. American Psychologist, 61, 671–689.
Press. Weisz, J., Weiss, B., Han, S., Granger, D., &
Shirk, S. (Ed.). (1988). Cognitive development and Morton, T. (1995). Effects of psychotherapy
child psychotherapy. New York: Plenum. with children and adolescents revisited: A meta-

Shirk, S., Gudmundsen, G., Kaplinski, H., & analysis of treatment outcome studies. Psycho-
McMakin, D. (2008). Alliance and outcome logical Bulletin, 117(3), 450–468.
in cognitive-behavioral therapy for adolescent Weisz, J. R., Weiss, B., Alicke, M. D., & Klotz, M. L.
depression. Journal of Clinical Child and (1987). Effectiveness of psychotherapy with
Adolescent Psychology, 37(3), 631–639. children and adolescents: A meta-analysis for
Shirk, S., & Karver, M. (2003). Prediction of clinicians. Journal of Consulting and Clinical
treatment outcome from relationship variables Psychology, 55, 542–549.
in child and adolescent therapy: A meta-analytic Wintersteen, M., Mensinger, J., & Diamond, G.
review. Journal of Consulting and Clinical (2005). Do gender and racial differences between
Psychology, 71, 462–471. patient and therapist affect therapeutic alli-
Shirk, S., & Karver, M. (2006). Process issues ance and treatment retention in adolescents?
in cognitive-behavioral therapy for youth. In Professional Psychology: Research and Practice,
P. C. Kendall (Ed.), Child and adolescent therapy: 36(4), 400–408.
Cognitive-behavioral procedures (pp. 465–491). Wright, L., Everett, F., & Roisman, L. (1986).
New York: Guilford Press. Experiential psychotherapy with children. Baltimore:
Shirk, S., & Russell, R. L. (1996). Change processes Johns Hopkins University Press.
in child psychotherapy: Revitalizing treatment and Zack, S., Castonguay, L., & Boswell, J. (2007).
research. New York: Guilford Press. Youth working alliance: A core clinical construct
Shirk, S., & Russell, R. L. (1998). Process issues in in need of empirical maturity. Harvard Review of
child psychotherapy. In A. Bellack & M. Hersen Psychiatry, 15, 278–285.
(Eds.), Comprehensive clinical psychology. London: ∗
Zaitsoff, S., Doyle, A., Hoste, R., & le Grange, D.
Pergamon. (2008). How do adolescents with bulimia
Shirk, S., & Saiz, C. (1992). Clinical, empirical, nervosa rate the acceptability and therapeu-
and developmental perspectives on the thera- tic relationship in family-based treatment?
peutic relationship in child psychotherapy. International Journal of Eating Disorders, 41(5),
Development and Psychopathology, 4, 713–728. 390–398.

s h i rk , k a rve r 91

4 Alliance in Couple and Family Therapy

Myrna L. Friedlander, Valentín Escudero, Laurie Heatherington, and Gary M. Diamond

Although the salience of the working of our original meta-analysis of the CFT
alliance in couple and family therapy (CFT) alliance outcome studies published through
was recognized over 20 years ago, it has 2008. We summarize the literature on
received far less theoretical and empirical moderators, mediators, and client contri-
attention than has the alliance in individual butions to CFT alliances, discuss the limi-
psychotherapy. In their seminal work on tations of the research, and conclude with
CFT alliances, Pinsof and Catherall (1986; recommended clinical practices on the basis
Catherall, 1984) took Bordin’s (1979) of the meta-analysis.
conceptualization of the alliance as a point
of departure and applied the goal, task, and Definitions and Measures
bond constructs to three interpersonal Creating and sustaining CFT alliances are
facets of the alliance in family treatment complicated by the fact that family members
(self-with-therapist, other-with-therapist, often seek psychotherapy as a last resort,
and group-with-therapist). The rationale when the conflicts among them seem irre-
was that not only do family members vary concilable. Moreover, it is common for
in the degree to which they form a personal family members to have different motives
bond and agree with the therapist about and motivational levels for treatment, and
treatment goals and tasks, but also each sometimes disagree about whether there
person observes, can report on, and is influ- even exists a problem that requires profes-
enced by how others in the family feel sional attention. Even when a problem is
about the therapy and by how the couple jointly acknowledged, for example, “We
or family unit as a whole is responding to fight all the time,” therapy may not be seen
what is taking place in treatment (Pinsof & as the solution, or individuals’ goals may
Catherall, 1986). Thus, from its first differ (“You need to stop drinking” versus
introduction into the literature, the CFT “We need to be a couple—it’s like we’re
alliance was described as unique, complex, living parallel lives”) (Friedlander, Escudero,
and multilayered. It is no longer in question & Heatherington, 2006; Lambert, Skinner,
that, as a treatment format, CFT demands & Friedlander, in press). Consequently,
a unique conceptualization of the alliance. family members’ willingness to engage in
In this chapter we define CFT alliances, various therapy tasks may also differ (“Why
summarize the major observational and should we argue about my drinking if you
self-report measures, and offer an extended don’t even want to stay married to me?”).
clinical example. We then report the results Indeed, it is not uncommon for one partner

or family member to feel like a therapy (Friedlander, 2000). Breaches of safety can
hostage (“Come to therapy with me or severely undermine a client’s trust in the
else …”) or expect the therapist to take therapist and the therapeutic process.
sides, particularly if the problem is defined Moreover, the degree of safety felt by family
in zero-sum terms (e.g., to divorce or not, members can change as new problems are
to have a baby or not, to relocate or not) revealed and explored and as different
(Friedlander, Escudero, & Heatherington, family members join or leave treatment
2006). (Beck, Friedlander, & Escudero, 2006).
CFT alliances develop simultaneously on What feels safe to the children when only
an individual level (self-with-therapist) and their father is there, for example, might
a group level (group-with-therapist). Just as feel quite unsafe when their stepmother is
in individual therapy, alliances in CFT present. Likewise, it may seem safer in
involve the creation of a strong emotional couples therapy to discuss conflicts over
bond as well as negotiation of goals and parenting than to explore expectations
tasks with the therapist. A unique charac- about intimacy or sexuality. In CFT, the
teristic of CFT, though, is that at any point conjoint nature of the treatment and ever-
in treatment there are multiple alliances changing composition of sessions makes
that interact systemically. For example, the creating a safe environment both compli-
degree to which a mother likes the psycho- cated and critical.
therapist and is engaged in the treatment A related construct is the group aspect of
may have a facilitating (or hindering) effect the alliance, which has alternately been
on her son’s willingness to trust the thera- conceptualized as allegiance (Symonds &
pist. The son’s involvement also depends on Horvath, 2004), within-family alliance
the mother–son bond and whether he agrees (Pinsof, 1994), and shared sense of purpose
with his mother about the nature of the (Friedlander, Escudero, & Heatherington,
problems, goals, or need for treatment. 2006). A complex part of the conjoint
Moreover, the family members’ degree of therapy process, this “we-ness” refers not
comfort with one another affects each only to a willingness to collaborate in treat-
person’s willingness to negotiate goals with ment but also to a strong emotional bond
the others and with the therapist. In other between and among family members. In
words, every individual simultaneously other words, the within-family alliance has
creates a personal alliance with the thera- more to do with family members’ thoughts,
pist, and each person’s alliance with the feelings, and behavior toward one another
therapist can negatively or positively affect than it does with any one person’s alliance
the others’. with the therapist considered in isolation.
An important aspect of CFT alliances is Moreover, the within-family alliance
the degree to which family members feel develops simultaneously and in interaction
safe and comfortable with each other in with all of the individual alliances. Indeed,
the therapeutic context. The revelation of research shows that family members often
secrets and in-session exploration of conflicts see their personal relationships with the
are not easily left behind in the consulting therapist differently from their allegiance
room at the end of the session. Family with each other (e.g., Beck et al., 2006;
members go home together, and therapy Friedlander, Lambert, Escudero, & Cragun,
can only progress if they feel that the 2008; Lambert et al., in press). For this
material discussed in-session is not used reason, a complete picture of the alliance
against them during the course of the week requires some accounting of how well the

f r i e d l a n d e r, e s c u d e ro , h e at h e r i n g to n , d i a m o n d 93
family works together in therapy as well (Pinsof, Zinbarg, & Knobloch-Fedders,
as how similarly individuals feel about the 2008) only has 12 items in 3 subscales: Self/
therapist. Group, Other, and Within. Self and Group
When alliances are “split” (Heatherington were combined because these subscales
& Friedlander, 1990; Pinsof & Catherall, were indistinguishable both statistically and
1986) or “unbalanced” (Robbins, Turner, experientially, and factor analyses did not
Alexander, & Perez, 2003), at least one support the independence of the original
family member has a stronger bond with the Goals, Tasks, and Bonds subscales.
therapist than do other family members. Only one measure of the alliance
There is ample evidence that, in both cou- includes the element of safety. In the System
ples and family therapy, split alliances occur for Observing Family Therapy Alliances
frequently and vary in severity (Heatherington (SOFTA; Friedlander, Escudero, &
& Friedlander, 1990; Mamodhoussen, Heatherington, 2006; Friedlander, Escudero,
Wright, Tremblay, & Poitras-Wright, 2005; Horvath et al., 2006) or Sistema de la
Muñiz de la Peña, Friedlander, & Escudero, Observación de la Alianza en Terapia Familiar
2009). In family therapy, although we might (SOATIF; Escudero & Friedlander, 2003),
expect parents to feel a greater connection Safety is one of four alliance dimensions.
with the therapist than do their adolescents, In brief, Safety within the Therapeutic
and indeed such a split alliance pattern has System reflects each client’s degree of comfort
been found in several studies, in at least one taking risks, being vulnerable, and exploring
study several of the adolescents felt closer to conflicts with a therapist and other family
the therapist than did their parents (Muñiz members, Engagement in the Therapeutic
de la Peña et al., 2009). Although severely Process reflects Bordin’s (1979) agreement
split alliances often lead to premature termi- with the therapist on tasks and goals,
nation, this is not invariably the case (Muñiz Emotional Connection with the Therapist is
de la Peña et al., 2009), and splits may occur similar to Bordin’s concept of client–therapist
in reverse as new topics or are explored or bond, and Shared Sense of Purpose within the
secrets are revealed. Family refers to productive family collabora-
To assess CFT alliances, the most widely tion (the within-family alliance).
used self-report measures are the Couple The pantheoretical SOFTA contains
Therapy Alliance Scale (CTAS; Pinsof & observational (SOFTA-o; Friedlander,
Catherall, 1986) and the Family Therapy Escudero, & Heatherington, 2006;
Alliance Scale (FTAS), which were revised Friedlander, Escudero, Horvath, et al., 2006;
to include the Within Alliance, “My part- Friedlander, Lambert, & Muñiz de la Peña,
ner and I….” (couple) and “Some of the 2008) and 16-item self-report (SOFTA-s;
other members of my family and I …” Friedlander, Escudero, & Heatherington,
(family). Like the couple version of the 2006; Friedlander, Lambert et al., 2008;
Working Alliance Inventory (WAI-Co; Lambert & Friedlander, 2008) measures
Symonds, 1999), the CTAS and FTAS from both client and therapist perspectives.
reflect Bordin’s (1979) concept of goals, Whereas the client version reflects the
tasks, and bonds. However, unlike the strength of the alliance, the therapist version
63-item WAI-Co, which also has a thera- assesses alliance-related interventions. Using
pist version and yields a total alliance score the SOFTA-o, trained raters observe a vid-
as well as 9 subscale scores (Goals, Tasks, eotaped or live session, tallying the frequency
and Bonds crossed with Self, Partner, and of specific positive behaviors, for example,
Group), the shortened CTAS-r and FTAS-r “Client introduces a problem for discussion”

94 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
(Engagement), and negative behaviors, for Because it was obvious that conjoint ses-
example, “Family members try to align with sions with the four family members would
the therapist against each other” (Shared not be productive at this stage of treatment,
Purpose), and using these tallies to make a the therapist proposed holding the next
global rating for each alliance dimension. To two sessions with the teens and the parents
date there has only been one published study separately. Indeed, this seemed to be the
with the therapist version (Friedlander, only arrangement that could provide even a
Lambert et al., 2008), a comparison of good modest amount of safety. In the children’s
and poor outcome cases with the same session, the boy willingly expressed concern
therapist. about his sister, but he cavalierly dismissed
CFT alliances have also been studied his own problems. For her part, the girl
using the Vanderbilt Therapeutic Alliance denied being anxious or having eating
Scale (Hartley & Strupp, 1983), an obser- problems and, instead, complained about
vational measure developed for individual how her brother constantly annoyed her.
psychotherapy and subsequently revised for As siblings, they collaborated minimally,
CFT (VTAS-R; Diamond, Liddle, Dakof, each one only willing to talk about the oth-
& Hogue, 1996). Recently, the 26 Patient er’s problems. When asked about relations
Contribution items in the VTAS-R were with their parents, both teens remained
pared down through factor and item analy- silent. Finally, the son asked if the therapist
ses to a 5-item scale (Shelef & Diamond, thought he could help their parents, but
2008) that includes three client behaviors the boy refused to clarify the basis for this
reflecting bonds and tasks and two client + request. Notably, the alliance seemed split:
therapist behaviors reflecting goals and The boy was visibly more involved and con-
tasks. The VTAS-R requires raters to nected with the (male) therapist than was
provide global judgments of each client’s his sister, whose sense of safety appeared to
overall behavior and therapist–client inter- be quite fragile.
actions across an entire session, for example, In their conjoint session, mother and
“To what extent did the patient indicate father demonstrated even greater unease
that she experiences the therapist as under- with each other. Not only did they not
standing and supporting her?” (Shelef & make eye contact or confer with one
Diamond, 2008, p. 439). another, but they also sat on either end of
the couch, their bodies positioned in oppos-
Case Example ing directions. Although both parents
A middle-aged couple brought their cooperatively described the children’s prob-
reluctant 15-year-old son and 13-year-old lems, they refused to discuss their own
daughter to psychotherapy. The girl, who relationship. Finally, the husband haltingly
exhibited anxiety and an eating disorder explained that “after something happened,”
(only at home), refused to speak in the he and his wife had agreed that the marriage
session, as did the boy, who had vandalized was finished. This “emotional divorce” was
a neighbor’s car and was failing in school. unknown to the children, however. Because
While the parents barely glanced at each neither parent was willing to leave the
other, both adamantly insisted that their home, they planned to continue living
children were in desperate need of help. together until both children grew up and
Thus ended the first session, which clearly moved out. After that, they would separate.
evidenced a lack of safety all around and an Neither client was willing to consider
exceedingly poor within-family alliance. couples work, as they were in agreement

f r i e d l a n d e r, e s c u d e ro , h e at h e r i n g to n , d i a m o n d 95
that their marriage was a “lost cause.” They to keep the family together.” For the first
were, however, willing to come for sessions time, the spouses looked at and spoke
if it would help their children. directly to one another. When the husband
Interestingly, within each subsystem made a joke that his wife smiled at, the ther-
there was a shared sense of purpose, at least apist commented that they both seemed to
about why they would continue coming to be experiencing “deep hurts” that they were
therapy: The children agreed to be seen so afraid to express. He said that he wanted it
that the therapist would help their parents, to “see if there was another way for [their
and the parents agreed to come in order to marital] relationship to improve, if only to
help their children. keep on helping the children.”
Given this curious arrangement and The mother, who seemed to trust the
everyone’s clear fear of taking emotional therapist a great deal, admitted thinking
risks, the therapist continued to see each that the children were “reacting” to the
subsystem weekly. He found it relatively emotional divorce. Moving closer to her,
easier to develop a personal bond with the the therapist softly commented that, “As a
son and the mother. By working hard to parent myself, I know it’s extremely hard to
enhance his connection with the daughter realize that something I’ve done has hurt
and father, the therapist gradually became a my children.” The mother responded with
trusted figure, and slowly everyone began tears, and at the end of the session admitted
to engage more freely in the therapeutic that the children “deserved to be told” about
work. Because the adolescents adamantly the status of the marriage. The husband
refused to acknowledge their own problems agreed, albeit reluctantly.
and were clearly protecting their parents— The parents chose to reveal the secret at
never criticizing them or even acknow- home rather than in a family session. In their
ledging their parents’ overt hostility—the next session (alone) with the therapist, the
therapist focused solely on improving the teens no longer felt the need to protect
sibling bond. In one homework assign- their parents. They responded positively to
ment, for example, the girl was asked to the therapist’s empathic response to their
choose a set of digital family photos that expressions of resentment and sadness. When
held happy memories for her, and her the daughter burst out, “But we’re not a real
brother was asked to arrange these pictures family!,” the therapist replied by proposing a
into a slideshow that he would set to new shared sense of purpose, in other words,
music. a common goal: “I disagree. Both of your
As brother and sister began fighting less parents care for you and want the best for
at home and cautiously started to enjoy you, and both of you feel the same for your
each other’s company, both parents began parents. I’m sure you can learn to work
to trust the therapist more. However, they together so that everyone has a happier life.”
rarely looked at one another in session, and Over the next month, the teens pushed
the chasm between them remained as deep their parents into committing to couples
as ever. therapy, with the goal of either working out
Alone with the parents in Week 5, the their differences or deciding to separate.
therapist made some strategic moves. Although the son’s grades in school did not
Focusing first on the within-system alliance, improve substantially, he had no further
he praised their mutual dedication to their delinquent offenses. The daughter remained
children, pointing out how they were both highly stressed but ate normally and began
willing to “sacrifice [their] personal happiness spending more time with friends.

96 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
This case illustrates how an alliance- Seven of the 24 were studies of couples
empowering approach can potentially therapy (2 of which were conducted in
repair seriously broken within-family groups), and the remaining 17 were family
attachments. By strategically focusing on studies in which at least a portion of the
different alliances and different aspects of treatment was conducted conjointly. The
each alliance, this therapist moved a stalled total number of clients who participated in
treatment forward. He began by separating these 24 studies is 1,461. Studies examined
the parents and children to enhance safety both treatment as usual as well as specifically
and negotiate different problem definitions defined approaches, including cognitive-
and goals within each subsystem. With the behavioral therapy, functional family
adolescents, the therapist relied heavily on therapy, family-based therapy, systemic
five interventions that have been shown to and ecosystemic-oriented therapy, emotion-
improve poor alliances with teens (Diamond focused therapy, integrative problem-
et al., 1999): He emphasized trust, honesty centered therapy, multidimensional family
and confidentiality; he explained the impor- therapy, brief strategic family therapy,
tance of collaborating in therapy; he defined parent management training, and psycho-
personally meaningful goals for each child; educative family therapy. Most of the treat-
and, most importantly, he presented him- ments were no more than 20 sessions in
self as an ally in the one thing the children length, and the majority (65%) described
agreed on—helping their parents. Then, by the therapy as manualized treatment,
encouraging engagement in therapy tasks although only a few studies provided infor-
through his personal bond with each family mation about treatment integrity.
member, the therapist eventually redefined The problems targeted in these treat-
the family’s problem and the treatment ments ranged from parent–adolescent
goals in a way that was both respectful and communication difficulties to substance
challenging. The success of this process abuse, child abuse or neglect, and schizo-
goal, to create a within-family shared sense phrenia. Some studies were highly specific
of purpose, seemed largely due to the thera- in defining the presenting problems or
pist’s attending to and emphasizing the disorders, but many other studies identi-
strong parent–child bonds. fied the clients’ problems in a general way.
The samples in the more naturalistic studies
Meta-Analysis tended to have a variety of presenting
Table 4.1 summarizes 24 studies in which issues.
CFT alliances, self-reported and observed, The instruments and methods used to
were used to predict treatment retention, evaluate outcome reflect the variability in
improvement midtreatment, and/or final problems treated. In terms of measures,
outcomes. To obtain as comprehensive a roughly 50% of the studies used an obser-
sample as possible, we cross-referenced vational methodology. Most evaluated the
articles known to us and searched electron- alliance early in the therapy, and only a few
ically (PsycINFO, PubMed, Social Sciences studies assessed the alliance at different
Citation Index) for additional alliance- stages of treatment (early, middle, and
related studies in CFT. Unpublished late). The observational instruments were
dissertations were excluded, as were analog primarily the SOFTA-o and the VTAS; the
studies. Only articles published in English WAI and CTAS/FTAS were most often
with validated measures of the alliance were used to measure self-reported alliance.
included. Five of the 24 studies did not measure

f r i e d l a n d e r, e s c u d e ro , h e at h e r i n g to n , d i a m o n d 97
Table 4.1 Summary of Alliance Outcome Studies in CFT
Study Treatment Alliance Outcome Overall
effect size
Therapy Format Rater Measure Time Measure Wt. Ave.
model Ave. r N
Bourgeois et al. (1990) CSP Couples C, T CTAS E Dyadic Adjustment 0.43 63
group Scale, Marital Happiness
Scale, Potential Problem
Brown & O’Leary (2000) CBT, PE Couples O WAI E Psychological 0.53 70
group Maltreatment of Women
Scale, Modified Conflict
Tactics Scale
Escudero et al. (2008) BFT Family O SOFTA E, L Perceived therapeutic 0.22 68
Flicker et al. (2008) FFT Family O VTAS-R E Completion vs. dropout 0.25 86
Friedlander et al. (2008) FS Family O SOFTA E Perceived improvement- 0.35 33
Greenberg et al. (2010) EFT Couple C CTAS E Enright Forgiveness 0.35 40
Hawley & Garland (2008) FS Family C, T WAI E, L Youth Symptom 0.33 36.7
Hogue et al. (2006) MDFT Family O VTAS-R E Child Behavior 0.05 44
Checklist (internalizing,
externalizing), Timeline
Follow-Back interview
for substance use
Johnson et al. (2006) EcoS Family, C FTAS L Outcome 0.46 32
home (tasks) Questionnaire,
based Inventory of Parent
and Peer Attachment
Johnson & Ketring (2006) EcoS Family, C FTAS L Outcome Questionnaire 0.10 430
home (bond) 45.2 -Symptom Distress
based subscale, Conflict Tactics
Scale - Physical Aggression
Johnson & Talitman EFT Couple C CTAS E Dyadic Adjustment 0.54 23
(1997) Scale, Miller Social
Intimacy Scale
Kazdin et al. (2005) PE Family T, C WAI, E, L Treatment Improvement 0.32 49
CTAS Scale, Marital Satisfaction
Knobloch et al. (2007) IPCT Couple C CTAS E, L Marital Satisfaction 0.39 37
Inventory Revised
Pereira et al. (2006) FBT Family O WAI E, L Eating Disorders 0.32 31.4
(Continued )

Table 4.1 Continued
Study Treatment Alliance Outcome Overall
effect size
Therapy Format Rater Measure Time Measure Wt. Ave.
model Ave. r N
Pinsof et al. (2008) IPCT Couple C CTAS-R E, L COMPASS Treatment 0.31 80
Assessment System,
Marital Satisfaction
Inventory Revised
Quinn et al. (1997) FS Family C FTAS E Goal achievement 0.53 19
and expectation of
Raytek et al. (1999) CBT Marital O VTAS-R E Attrition status 0.37 66
(completers, partial
completers, vs. early
Robbins et al. (2003) FFT Family O VTAS-R E Completion vs. dropout 0.29 34
Robbins et al. (2006) MDFT Family O VTAS-R E Completion vs. dropout 0.35 30
Robbins et al. (2008) BSFT Family O VTAS-R E Completion vs. dropout 0.36 31
Shelef et al. (2005) MDFT Family C, O WAI E Global Appraisal 0.24 59
of Individual Needs,
Substance Problem
Shelef & Diamond MDFT Family O VTAS-R E, L Completion vs. dropout, 0.41 45
(2008) days of cannabis use
Smerud & Rosenfarb PE Family O SOFTA L Brief Psychiatric Rating 0.54 28
(2008) Scale, Social Adjustment
Scale-II, days until first
rehospitalization, days
until first use of rescue
medication, Patient
Rejection Scale
Symonds & Horvath NS Couple C WAI-Co E Marital Satisfaction 0.37 22.4
(2004) Scale (female)
Note: N refers to the average sample size for the various correlations used in the within-study meta-analysis, i.e., not the N in the entire study.
CSP = Couple Survival Program; CBT = cognitive-behavioral therapy; PE = psycho-educative; BFT = brief family therapy; FFT = functional
family therapy; FS = family systems; EFT = emotion-focused therapy; MDFT = multidimensional family therapy; EcoS = ecosystemic therapy;
FBT = family-based therapy; IPCT= integrative problem-centered therapy; BMT = behavioral marital therapy; BSFT = brief strategic family
therapy; NS = not specified. C = client self-report; T = therapist self-report; O= external observer. CTAS = Couple Therapy Alliance Scale
(Pinsof & Catherall, 1986); CTAS-r = Couple Therapy Alliance Scale-Revised (Pinsof et al., 2008); FTAS = Family Therapy Alliance Scale
(Pinsof & Catherall, 1986); SOFTA-o: System for Observing Family Therapy Alliances - observer (Friedlander, Escudero, & Heatherington,
2006); VTAS-R = Vanderbilt Therapeutic Alliance Scale - Revised (Diamond et al., 1996); WAI = Working Alliance Inventory (Horvath &
Greenberg, 1986); WAI-Co = Working Alliance Inventory - couple (Symonds & Horvath, 2004). E = early; L = late.

client outcome but, rather, only explored and Weinberger (2009), which are based on
associations between alliance and treatment Hunter and Schmidt’s (1990) random-
retention. effects approach. For studies that reported
For the meta-analysis of correlation coef- statistics other than correlation coefficients
ficients, we used the recommendations and (e.g., a t test to compare the alliance in
computation program of Diener, Hilsenroth, families that completed therapy versus those

f r i e d l a n d e r, e s c u d e ro , h e at h e r i n g to n , d i a m o n d 99
that dropped out), we calculated the corre- differences in the designs, measures, treat-
sponding conversions to r (using Diener ment formats, and problems treated in the
et al.’s (2009) computation program). group of 24 studies made it essential to test
Because of the complex structure of for variables that may moderate the average
alliance in family therapy (multiple partici- effect size. We performed a chi-square test
pants generating multiple levels of analysis), to examine the homogeneity of various
most of the 24 studies reported more than subsamples.
a single alliance–outcome correlation. In The resulting χ2 = 43.52, p < 0.005,
these cases, we calculated a meta-analytic indicated that the null hypothesis of homo-
statistic within each study in order to geneity was rejected. In other words, there
maintain statistical assumption of indepen- was unaccounted-for variability among the
dence. These calculations were conducted effects produced by the various studies.
in the same way as the meta-analytic calcu- Consequently, we explored various moder-
lations for the entire group of 24 studies. ators that might explain the heterogeneity.
That is, the effect sizes listed in the far right
column of Table 4.1 were calculated from Moderators and Mediators
aggregated correlations within each study We computed a series of moderator analy-
(Diener et al., 2009; Hunter & Schmidt, ses with various subsamples. First, analysis
1990). We also calculated the weighted of the 17 family studies showed a similar
average effect size of the 24 studies, mini- average weighted effect size (r = 0.24;
mizing sampling error by weighting each z = 6.55, p < 0.005), and the null hypothesis
study by its sample size. of homogeneity was rejected (χ2 = 27.77,
The resulting weighted average effect size p < 0.05). In other words, the group of
for the 24 studies was r = 0.26. The standard- studies with a family therapy format had
ized normal test for determining whether an significant unexplained variability in the
aggregate r is statistically significant yielded relation between alliance and outcome.
z = 8.13, which is sufficiently large to reject (The subsample of seven couple studies was
the null hypothesis (p < 0.005). The upper too small to test for either effect size or
and lower limits of the weighted average homogeneity.)
effect size calculated for a 95% confidence Second, we analyzed the 13 studies that
interval were 0.33 and 0.20, respectively. used observational (rather than self-report)
These results indicate that the association measures of the alliance. This result showed
between alliance and outcome in couple a somewhat higher global effect size
and family therapy was statistically signifi- (weighted average r = 0.33; z = 8.85,
cant. According to conventional bench- p < 0.001), and the variability in these stud-
marks, an r of 0.26 (d = 0.53), which is a ies was not large enough to reject the null
medium effect size in the behavioral hypothesis of homogeneity (χ2 = 16.48, ns).
sciences, is quite similar to the r = 0.275 By contrast, the 11 studies that used self-
reported in Chapter 2 of this book for reported alliances had a slightly lower effect
the alliance–outcome relation in individual size (weighted average r = 0.22; z = 4.46,
therapy. p < 0.01). In this analysis, the homoge-
In addition to statistical significance, neity hypothesis was rejected (χ2 = 22.39,
the meta-analytic results demonstrated that p < 0.02), possibly due to the diversity of
alliance accounted for a substantial propor- perspectives (parents’, children’s, partners’
tion of variance in CFT retention and/or and therapists’) in self-report studies of
outcome. At the same time, evident alliance.

100 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Taken together, these results mean that permutations (mother–child, father–
in every subsample tested, the alliance in mother, etc.). For example, with adoles-
CFT accounted for a substantial proportion cents who have externalizing problems
of variance in treatment retention and/or (Shelef & Diamond, 2008) or anorexia
outcome. However, the most homogeneous (Pereira, Lock, & Oggins, 2006), research
studies were the observational ones, likely has found that the parents’ (but not the
due to the fact that trained external observ- youths’) alliances predicted treatment
ers use well-defined behavioral criteria to completion. Further, in other studies with
assess the alliance. By contrast, in the family externalizing adolescents (Robbins et al.,
studies and in the studies that relied on self- 2006, 2008), both the children’s and the
report to assess alliance, the alliance–outcome parents’ alliances with the therapist discrim-
relation was more variable. inated dropout from completer families.
These divergences in effect sizes across Unbalanced alliances tend to be negatively
the outcome studies, as well as complex related to retention, and this relation is also
findings within these studies, raise ques- moderated in complex ways by family role.
tions about the circumstances under which Unbalanced father-adolescent alliances
the alliance figures more or less strongly in (where the first person in the pair is the one
outcomes. Because very few direct tests of with the higher alliance) discriminated
moderators and mediators have been dropout from completer families in
conducted, a meta-analysis was not possible. functional family therapy (FFT; Robbins,
In this section, we summarize what is Turner, Alexander, & Perez, 2003), and
known and what has been suggested about increasingly unbalanced mother–adolescent
moderators and mediators of the alliance– alliances and unbalanced mother–father
outcome relation in CFT. alliances characterized families who
dropped out of brief strategic family therapy
Alliance and Treatment Retention (Robbins et al., 2008). Further, unbalanced
Good outcomes depend on treatment parent–adolescent alliances in FFT discrim-
attendance, and retention in family therapy inated dropout from completer Hispanic
is particularly challenging, as any one families but not Anglo families (Flicker,
person’s strong negative feelings can lead to Turner et al., 2008).
premature termination for the entire family. Therapist experience has not been system-
For this reason there has been significant atically manipulated in any study, although
work in CFT on strategies for engaging and experience did differ across studies, prompt-
retaining families, especially families with ing some thoughtful speculation (Flicker &
drug-using adolescents (cf. Liddle et al., Turner, 2008) about how it might account
1992; Szapoznick et al., 1988). for differing results. Therapist experience
Regarding retention, the only clear was positively associated with the alliance
moderator is family role—parent, spouse, (as measured by observer ratings of thera-
child. First, we note that the composite pist behavior and errors in technique) in
index of CFT alliance, that is, an average conjoint alcoholism treatment for couples
of all family members’ alliances, is not (Raytek, McCready, Epstein, & Hirsch,
predictive of retention (or outcome). 1999). A qualitative analysis revealed that
Rather, more nuanced indices of alliance experienced therapists were relatively more
matter: (a) the interplay of each individual active, more responsive to topics initiated
family member’s alliance with the therapist, by clients, more flexible in following man-
and (b) unbalanced alliances, in various ualized treatment guidelines, and better

f r i e d l a n d e r, e s c u d e ro , h e at h e r i n g to n , d i a m o n d 101
at managing the couples’ negativity. The parent measures did not. Moreover, adoles-
authors suggested that such responsiveness cent alliance predicted outcome only in
and flexibility strengthened the emotional cases in which the parent–therapist alliance
bond with the therapist and the clients’ was moderate to strong. In a study of
involvement in treatment, which in turn outpatient psychotherapy “as usual” that
facilitated retention in therapy. combined individual and conjoint parent–
teen sessions, youths’ alliance predicted
Couples Therapy Outcomes outcomes (youth symptom improvement,
In general, with respect to gender, the man’s family functioning) as reported by all family
alliance tends to be more strongly associ- members, whereas parents’ alliance predicted
ated with outcome in both group marital fewer outcomes and only their own (i.e.,
therapy (Bourgeois, Sabourin, Wright, not their childrens’) ratings of treatment
1990; Brown & O’Leary, 2000) and success (Hawley & Garland 2008).
couples therapy (Symonds & Horvath, Interestingly, there is no evidence that
2004). Less frequently, the woman’s therapist gender, race/ethnicity, or thera-
alliance is the stronger predictor of out- pist–family ethnic match are significant
come (Knobloch-Fedders, Pinsof, & Mann, factors in the strength of alliance. Nor have
2007). Explanations for the gender differ- they been found to moderate the CFT
ence focus on the documented greater alliance–outcome relation.
reluctance of men to engage in treatment, Type of treatment may moderate the
as well as their relative power in some alliance–outcome relation, given the differ-
couples (especially where there is abuse), ences in findings across studies that
and women’s relatively higher commitment employed different kinds of treatment. In a
and “ability to work toward positive study of behavioral family management
outcomes regardless of the relative strength treatment for schizophrenia (Smerud &
of their relationship with the therapist” Rosenfarb, 2008), only the relatives’
(Symonds & Horvath, 2004, p. 453). observed alliances predicted the patient’s
reoccurrence of symptoms, a finding that
Family Therapy Outcomes underscores the importance of family
In outcome studies of conjoint family environment in preventing relapse in major
therapy, family role emerged as the most mental illness. Interestingly, patients’
consistent (albeit complex) potential mod- alliance predicted less rejection by relatives
erator; its effects vary depending on the and less care burden, suggesting that
measures used and the treatment adminis- alliances in one subsystem may have posi-
tered. A study of family treatment for tive effects on others. Another study (Hogue
anorexia nervosa (Pereira et al., 2006), et al., 2006) that compared cognitive-
for example, found that adolescents’ (but behavior therapy (CBT) and multidimen-
not parents’) observed alliance with the sional family therapy (MDFT) for adolescent
therapist predicted early weight gain, whereas substance abuse found that in CBT, the
parents’ alliance later in therapy was associ- adolescents’ alliance was not associated with
ated with teens’ overall weight gain. outcome, whereas in MDFT both the youths’
Similarly, in a study of family treatment and the parents’ alliances were associated
for adolescent substance abuse (Shelef et al., with outcomes, albeit in different ways.
2005), observer measures (but not self- Finally, in a study of home-based family
report measures) of adolescents’ alliance therapy, the youths’ attachment, as measured
predicted posttreatment outcomes, whereas by trust in each parent, in tandem with the

102 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
“tasks” dimension of the alliance, predicted the alliance, just as some authors have
symptom reduction (Johnson, Ketring, claimed that all therapeutic interventions
Rohacs, & Brewer, 2006). The quality of are indistinguishable from alliance building
parents’ attachment relationships with their and maintenance (e.g., Hatcher & Barends,
children was not a significant moderator, 2006). To some extent, clients’ collabora-
however. tion in therapy depends on the therapist’s
We located only one study that specifi- theoretical approach. Couples in behavioral
cally tested a mediating model (Friedlander therapy, for example, spend less time access-
et al., 2008). In this study, the within- ing primary emotions than do couples in
family alliance (shared sense of purpose emotion-focused therapy. Yet, on a differ-
within the family) mediated the relation- ent process level, clients’ alliance-related
ship between the parents’ observed sense of behaviors cut across therapy approaches
safety in Session 1 and their ratings of and formats. That is, like successful clients
improvement-so-far in Session 3. In other in individual therapy, successful family
words, parents who felt comfortable in the members form a close, trusting bond with
first session were more likely to exhibit a their therapists and negotiate (and renego-
strong within-family alliance that, in turn, tiate) treatment goals and tasks. Regardless
predicted their perceptions of improvement of the kinds of in-session or out-of-session
after the third session. In another study tasks, clients who have a shared sense of
(Escudero et al., 2008), within-family purpose listen respectfully to one another,
alliance was the only observed alliance validate each other’s perspective (even when
dimension to increase over time; this they disagree), offer to compromise, and
dimension predicted therapists’ perceptions avoid excessive cross-blaming, hostility, and
of the alliance and ratings of improvement- sarcasm. Family members who feel safe
so-far after Session 6. Although within- and comfortable in therapy are emotion-
family alliance was not tested as a mediator, ally expressive, ask each other for feed-
the latter findings suggest that a stronger back, encourage one another to open up
shared sense of purpose may indeed be an and speak frankly, and share thoughts
important step along the way to treatment and feelings, even painful ones, that have
success. never been expressed at home (Friedlander,
Escudero, & Heatherington, 2006).
Client Contributions
By its very definition, the alliance construct Couples Therapy
implies interaction and collaboration. For While scant, the literature offers some
this reason, isolation of family members’ evidence about the personal characteristics
contributions to the alliance is somewhat and in-session behaviors of clients who
artificial. On the other hand, considering develop strong working relationships in cou-
client contributions is essential as, during ples therapy. Research suggests that whereas
the session, therapists must gauge clients’ psychiatric symptoms are not associated
receptivity and reactions to therapeutic with alliance formation (Knobloch-Fedders,
change attempts. Pinsof, & Mann, 2004; Mamodhoussen
The clinical CFT literature focuses almost et al., 2005), greater trust in the couple rela-
exclusively on therapist behavior, with far tionship (Johnson & Talitman, 1997) and
less emphasis on client participation in treat- less marital distress (Johnson & Talitman,
ment. One could well argue that all client 1997; Knobloch-Fedders et al., 2004) are
behavior contributes to (or detracts from) predictive of more favorable alliances. In one

f r i e d l a n d e r, e s c u d e ro , h e at h e r i n g to n , d i a m o n d 103
study (Knobloch-Fedders et al., 2004), a study of family-based therapy for anorexia
alliance development differed for men nervosa found that teens with relatively
and women. For men, recalling positive more weight and eating concerns found it
experiences in the family of origin was particularly difficult to establish an alliance
most critical for early alliance development, with the therapist (Pereira et al., 2006). On
whereas marital distress had a negative the other hand, the nature of adolescents’
impact on the alliance later on. For women, emotional problems played no role in
sexual dissatisfaction was negatively associ- alliance development in a study of MDFT
ated with the alliance throughout therapy, for drug-using adolescents (Shelef &
and women’s family-of-origin distress Diamond, 2008). That study showed no
contributed to a split alliance early in the variability in teens’ alliance-related behav-
process. ior based on pretreatment externalizing or
With respect to in-session behavior, the internalizing behaviors.
findings from one study (Thomas et al., Not surprisingly, alliances are stronger
2005) reflect the complexity of CFT when family members respond favorably to
alliances. Results showed that men were therapists’ alliance-building interventions.
less likely to agree with the therapist on the In a comparison of two families treated by
goals for treatment when their partners the same experienced therapist (Friedlander,
made negative statements about them, Lambert et al., 2008), clients in the poor-
whereas women tended to feel more nega- outcome case were less likely than those in
tive about therapy tasks when they were the good-outcome case to respond posi-
challenged by their partners. Both men and tively to the therapist’s alliance-related
women had a stronger bond with the thera- behaviors. Another small sample study,
pist when their partners self-disclosed, and although not directly assessing the alliance,
they felt more distant from the therapist has implications for client contributions
when their partners challenged or made to a shared sense of purpose. In this
negative comments about them. study, family members moved from disen-
gagement with each other to productive
Family Therapy in-session collaboration when, with the
In community-based family therapy, therapist’s help, they were willing and able
parental differentiation of self, assessed prior to explore the underlying basis for their
to the beginning of treatment, predicted disengagement and recognize some moti-
stronger perceived alliances after Session 3 vation for breaking through the impasse
(Lambert & Friedlander, 2008). Well- (Friedlander, Heatherington, Johnson, &
differentiated individuals are able to balance Skowron, 1994).
thinking and feeling, autonomy and togeth-
erness (Bowen, 1978). The most closely Limitations of the Research
associated aspects of these two constructs The body of research covered in the meta-
were emotional reactivity and safety. That analysis is small but solid. Diverse client
is, parents who reported being generally populations and therapy approaches have
less emotionally reactive tended to feel safer been sampled, and many of the treatments
and more comfortable in conjoint family studied have strong empirical support
therapy. and/or have been delivered by experienced
There is some evidence to suggest that therapists. Under these conditions, the
diagnosis or presenting problem maybe finding that alliances predict treatment
associated with the alliance. For example, retention and outcome over and above

104 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
specific therapy methods strengthens importance of attachment for a couple’s
the case for the unique contribution of level of intimacy and for a family’s level of
relationship variables in CFT. Nonetheless, cohesion, attachment may well mediate or
there are limitations in this body of moderate the relationship between alliance
work that require caution in interpreting and treatment outcomes.
the findings and applying them to Although alliances develop and change
practice. over the course of psychotherapy, as several
Considerable variation across studies in studies have shown, we have little knowl-
alliance instruments and in timing and edge about how multiple alliances develop
rating source of the measurement make it over time and interact with each other.
challenging to interpret results. Under- Furthermore, we have little information
standably but unfortunately, sample sizes about how therapists behave in order to
in many of the studies are small, rendering best nurture and sustain working alliances
it difficult to test more than one or two with multiple clients over time. Finally, we
moderators with confidence. Thus, much still have not fully answered the question of
of what we know about moderators is spec- whether early symptom improvement in
ulative, based on results across different CFT actually prompts alliance develop-
studies. ment or is the consequence of a strong
The bulk of the research to date focuses alliance.
on drug-abusing, externalizing adolescents.
The alliance–outcome associations found,
as well as the kinds of therapist behaviors
Therapeutic Practices
shown to strengthen the alliance, may • The first, perhaps most important
largely be specific to these kinds of families. practice implication is that the therapeutic
For this reason, it is unwise to generalize alliance is a critical factor in the process
from these results to families with younger and outcome of CFT. Our meta-analysis
children or to families whose children have underscores the necessity to be aware of
internalizing problems (depression, anxiety, what is going within the system itself as
eating disorders). The effects of unbalanced well as to monitor the personal bond and
alliances, for example, may be weaker or agreement on goals and tasks with each
nonexistent in families in which, by virtue individual family member.
of the child’s age, symptoms, or psycho- • Knowing the importance of
logical dynamics, the children are less the alliance in CFT, practitioners are
inclined to resist treatment. Moreover, encouraged to systematically evaluate
there are few studies of the alliance in family their alliances with clients. One option
treatments for adults with major mental is periodically to ask clients to complete
disorders, such as family psychoeducation brief self-report measures of the alliance.
for bipolar disorder and schizophrenia, Doing so would not only provide the
despite the demonstrated efficacy of these therapist with crucial information
treatments. Finally, little research to date regarding each client’s private experience
has examined individual psychodynamics but also would provide the impetus for
(e.g., attachment styles) as moderators. directly addressing the quality of the
Attachment, in particular, has been a relationship and the therapy process
fruitful area of inquiry in the alliance and, if necessary, for focusing specifically
research on individual therapy (see Chapter on improving safety or repairing a
19 by Levy et al., this volume). Given the seriously split alliance.

f r i e d l a n d e r, e s c u d e ro , h e at h e r i n g to n , d i a m o n d 105
• Our findings suggest that evaluating to give me more space”) that compete with
the alliance based on observation is a one another. In this manner, therapists can
skill that can be taught. Observationally facilitate a shared sense
measured alliances were as predictive of of purpose between family members
outcome and more homogeneous in regarding the goals of treatment and how
their effect size than alliances measured they can productively collaborate to
using self-report. These findings suggest achieve these goals.
that therapists may be taught, or • Indeed, shared sense of purpose, a
alternatively may train themselves, to whole system aspect of the alliance, seems
validly assess the strength of their alliance to be a particularly important dimension
with different family members by of the alliance. Creating a safe space,
reviewing videotaped sessions (as was which is critical early on in therapy, is
shown to be effective in the exploratory important for all therapy participants.
training study of Carpenter, Escudero, & A therapist who allies too strongly with an
Rivett, 2008). adolescent may unwittingly damage his or
• The meta-analytic findings also her alliance with the parents, particularly
underscore the need to develop alliances when the latter are expecting the child to
with all family members. Therapists change but are not expecting to be
frequently identify more easily with, personally challenged by the therapist.
or feel a greater affinity to, one family • In short, each person’s alliance
member than another. However, the matters and alliances are not
studies we reviewed indicate that each interchangeable. Thus, clinicians should
and every alliance exerts both a direct build and maintain strong alliances with
and an interactive effect on the course each party and be cognizant of the ways in
of treatment. For example, whereas a which, depending on the couple or family
woman may feel connected to and dynamics, the whole alliance is not equal
aligned with the therapist from the to the sum of its parts.
outset of treatment, it may be the degree
to which her disenfranchised partner
develops a connection with the therapist References
that determines outcome. Findings from Beck, M., Friedlander, M. L., & Escudero, V.
studies examining unbalanced alliances (2006). Three perspectives on clients’ experiences
underscore this point, in that couples of the therapeutic alliance: A discovery-oriented
and families that experienced disparate investigation. Journal of Marital and Family
alliances with the therapist tended to Therapy, 32, 355–68.
fare more poorly in treatment. Bordin, E. S. (1979). The generalizability of the
psychoanalytic concept of the working alliance.
• One approach to developing multiple Psychotherapy, 16, 252–60.
alliances is to establish overarching Bourgeois, L., Sabourin, S., & Wright, J. (1990).
systemic goals (e.g., “It sounds like what Predictive validity of therapeutic alliance in
the two of you want is a relationship in group marital therapy. Journal of Consulting and
which you feel both connected and that Clinical Psychology, 58, 608–613.
you can sometimes do your own thing”) Bowen, M. (1978). Family therapy in clinical
practice. NY: Jason Aronson.
that resonate for all family members rather Brown, P. D., & O’Leary, K. D. (2000). Therapeutic
than focusing on first-order, individual alliance: Predicting continuance and success in
goals (i.e., “I want him to stop watching group treatment for spouse abuse. Journal of
sports on TV every Sunday” or “I want her Consulting and Clinical Psychology, 68, 340–45.

106 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Carpenter, J., Escudero, V., & Rivett, M. (2008). Friedlander, M. L., Bernardi, S., & Lee, H. (2010).
Training family therapy students in conceptual Better versus worse family therapy sessions as
and observation skills relating to the therapeutic reflected in clients’ alliance-related behavior.
alliance: An evaluation. Journal of Family Therapy, Journal of Counseling Psychology, 57, 198–204.
30, 409–22. Friedlander, M. L., Escudero, V., & Heatherington, L.
Catherall, D. (1984). The therapeutic alliance in (2006). Therapeutic alliances with couples and
individual, couple, and family therapy. Unpub- families: An empirically-informed guide to practice.
lished doctoral dissertation, Northwestern Washington, DC: American Psychological
University, Chicago, IL. Association.
Diamond, G. M., Liddle, H. A., Dakof, G. A., & Friedlander, M. L., Escudero, V., Horvath, A.,
Hogue, A. (1996). Revised version of the Vanderbilt Heatherington, L., Cabero, A., & Martens, M.
Therapeutic Alliance Scale. Unpublished manu- (2006). System for Observing Family Therapy
script, Temple University, Philadelphia, PA. Alliances: A tool for research and practice. Journal
Diamond, G. M., Liddle, H. A., Hogue, A., & of Counseling Psychology, 53, 214–25.
Dakof, G. A. (1999). Alliance-building inter- Friedlander, M. L., Heatherington, L., Johnson, B.,
ventions with adolescents in family therapy: & Skowron, B. (1994). “Commitment to
A process study. Psychotherapy, 36, 355–68. engage:” A change event in family therapy.
Diener, M. J., Hilsenroth, M. J., & Weinberger, J. Journal of Counseling Psychology, 41, 1–11.
(2009). A primer on meta-analysis of correlation Friedlander, M. L., Lambert, J. E., Escudero, V.,
coefficients: The relationship between patient- & Cragun, C. (2008). How do therapists
reported therapeutic alliance and adult attach- enhance family alliances? Sequential analyses of
ment style as an illustration. Psychotherapy therapist–client behavior in two contrasting
Research, 19, 519–26. cases. Psychotherapy: Theory, Research, Practice,
Escudero, V., & Friedlander, M. L. (2003). Training, 45, 75–87.
El sistema de observación de la alianza terapeutica Friedlander, M. L., Lambert, J. E., & Muñiz de la
en intervención familiar (SOATIF): Desarrollo Peña, C. (2008). A step toward disentangling the
trans-cultural, fiabilidad, y aplicaciones del alliance/improvement cycle in family therapy.
instrumento. Mosaico (Journal of the Spanish Journal of Counseling Psychology, 55, 118–24.
Federation of Family Therapy Associations), Greenberg, L. S., Warwar, S., & Malcolm, W.
25, 32–36. (2010). Emotion-focused couples therapy and
Escudero, V., Friedlander, M. L., Varela, N., & facilitation of forgiveness. Journal of Marital and
Abascal, A. (2008). Observing the therapeutic Family Therapy, 36, 28–42.
alliance in family therapy: Associations with Hartley, D., & Strupp, H. H. (1983). The
participants’ perceptions and therapeutic therapeutic alliance: Its relationship to outcome
outcomes. Journal of Family Therapy, 30, in brief psychotherapy. In J. Masling (Ed.).,
194–214. Empirical studies of psychoanalytic theories (Vol. 1,
Feeley, M., DeRubeis, R. J., & Gelfand, L. A. pp. 1–37). Hillsdale, NJ: Erlbaum.
(1999). The temporal relation of adherence and Hatcher, R. L., & Barends, A. W. (2006). How a
alliance to symptom change in cognitive therapy return to theory could help alliance research.
for depression. Journal of Consulting and Clinical Psychotherapy: Theory, Research, Practice, &
Psychology, 67, 578–82. Training, 43, 292–99.
Flicker, S. M., Turner, C. W., Waldron, H. B., Hawley, K. M., & Garland, A. F. (2008). Working
Ozechowski, T. J., & Brody, J. L. (2008). Ethnic alliance in adolescent outpatient therapy: Youth,
background, therapeutic alliance, and treatment parent, & therapist reports and association with
retention in functional family therapy with ado- therapy outcomes. Child and Youth Care Forum,
lescents who abuse substances. Journal of Family 27, 59–74.
Psychology, 22, 167–70. Heatherington, L., & Friedlander, M. L. (1990).
Friedlander, M. L. (2000). Observational coding Couple and family therapy alliance scales:
of family therapy processes: State of the art. In Empirical considerations. Journal of Marital and
A. P. Beck & C. M. Lewis (Eds.), The process of Family Therapy, 16, 299–306.
group psychotherapy: Systems for analyzing change. Hogue, A., Dauber, S., Stambaugh, L. F., Cecero, J. J.,
(pp. 67–84). Washington, DC: American & Liddle, H. A. (2006). Early therapeutic
Psychological Association. alliance and treatment outcome in individual

f r i e d l a n d e r, e s c u d e ro , h e at h e r i n g to n , d i a m o n d 107
and family therapy for adolescent behavior in couples undergoing couple therapy. Journal of
problems. Journal of Consulting and Clinical Marital and Family Therapy, 31, 159–69.
Psychology, 74, 121–29. Muñiz de la Peña, C., Friedlander, M. L., &
Horvath, A. O., & Greenberg, L. S. (1986). The Escudero, V. (2009). Frequency, severity, and
development of the Working Alliance Inventory. evolution of split family alliances: How observ-
In L. S. Greenberg & W. M. Pinsof (Eds.), The able are they? Psychotherapy Research, 19, 133–42.
psychotherapeutic process: A research handbook. Pandyaa, K., & Herlhyb, J. (2009). An exploratory
(pp. 529–56). New York: Guilford Press. study into how a sample of a British South Asian
Hunter, J. E., & Schmidt, F. L. (1990). Methods of population perceive the therapeutic alliances in
meta-analysis: Correcting error and bias in research family therapy. Journal of Family Therapy, 31,
findings. Newbury Park, CA: Sage. 384–404.
Johnson, L. N., & Ketring, S. A. (2006). The Pereira, T., Lock, J., & Oggins, J. (2006). Role of
therapy alliance: A moderator in therapy therapeutic alliance in family therapy for adoles-
outcome for families dealing with child abuse cent anorexia nervosa. International Journal of
and neglect. Journal of Marital and Family Eating Disorders, 39, 677–84.
Therapy, 32, 345–54. Pinsof, W. B. (1994). An integrative systems
Johnson, L. N., Ketring, S. A., Rohacs, J., & perspective on the therapeutic alliance:
Brewer, A. L. (2006). Attachment and the Theoretical, clinical, and research implications.
therapeutic alliance in family therapy. American In A. O. Horvath & L. S. Greenberg (Eds.), The
Journal of Family Therapy, 34, 205–218. working alliance: Theory, research, and practice
Johnson, S. M., & Talitman, E. (1997). Predictors (pp. 173–95). New York: Wiley & Sons.
of success in emotionally focused marital Pinsof, W. B., & Catherall, D. (1986). The integra-
therapy. Journal of Marital and Family Therapy, tive psychotherapy alliance: Family, couple, and
23, 135–53. individual therapy scales. Journal of Marital and
Kazdin, A. E., Marciano, P. L., & Whitley, M. K. Family Therapy, 12, 137–51.
(2005). The therapeutic alliance in cognitive- Pinsof, W. B., Zinbarg, R., &
behavioral treatment of children referred for Knobloch-Fedders, L. M. (2008). Factorial and
oppositional, aggressive, and antisocial behavior. construct validity of the revised short form
Journal of Consulting and Clinical Psychology, Integrative Psychotherapy Alliance Scales for
73, 726–30. family, couple, and individual therapy. Family
Knobloch-Fedders, L. M., Pinsof, W. M., & Process, 47, 281–301.
Mann, B. J. (2007). Therapeutic alliance and Quinn, W. H., Dotson, D., & Jordan, K. (1997).
treatment progress in couple psychotherapy. Dimensions of therapeutic alliance and their
Journal of Marital & Family Therapy, 33, 245–57. associations with outcome in family therapy.
Lambert, J. E., & Friedlander, M. L. (2008). Psychotherapy Research, 7, 429–38.
Relationship of differentiation of self to adult Rait, D. S. (1998). Perspectives on the therapeutic
clients’ perceptions of the alliance in brief family alliance in brief couples and family therapy.
therapy. Psychotherapy Research, 43, 160–66. In J. D. Safran & J. C. Muran (Eds.), The thera-
Lambert, J., Skinner, A., & Friedlander, M. L. peutic alliance in brief psychotherapy (pp. 171–91).
(in press). Problematic within-family alliances Washington, DC: American Psychological
in conjoint family therapy: A close look at five Association.
cases. Journal of Marital and Family Therapy. Raytek, H. S., McCready, B. S., Epstein, E. E., &
Liddle, H. A., Dakof, G., Diamond, G., Holt, M., Hirsch, L. S. (1999). Therapeutic alliance and
Aroyo, J., & Watson, M. (1992). The adolescent the retention of couples in conjoint alcoholism
module in multidimensional family therapy. treatment. Addictive Behaviors, 24, 317–30.
Adolescent substance abuse: Etiology, treatment, Robbins, M. S., Liddle, H. A., Turner, C. W.,
and prevention. In G. W. Lawson & A. W. Dakof, G. A., Alexander, J. F., & Kogan, S. M.
Lawson (Eds.), Adolescent substance abuse: (2006). Adolescent and parent therapeutic alli-
Etiology, treatment, and prevention (pp. 165–86). ances as predictors of dropout in multidimen-
Gaithersburg, MD: Aspen Publishers. sional family therapy. Journal of Family Psychology,
Mamodhoussen, S., Wright, J., Tremblay, N., & 20, 108–116.
Poitras-Wright, H. (2005). Impact of marital Robbins, M. S., Mayorga, C. C., Mitrani, V. B., Turner,
and psychological distress on therapeutic alliance C. W., Alexander, J. F., & Szapocznik, J. (2008).

108 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Adolescent and parent alliances with therapists prospective change process study. Journal
in brief strategic family therapy with drug- of Consulting and Clinical Psychology, 76,
using Hispanic adolescents. Journal of Marital & 505–510.
Family Therapy, 34, 316–28. Symonds, B. D. (1999). The measurement of alliance
Robbins, M. S., Turner, C. W., Alexander, J. F., & in short term couples therapy. Unpublished
Perez, G.A. (2003). Alliance and dropout in doctoral dissertation, Simon Fraser University,
family therapy for adolescents with behavior Burnaby, British Columbia, Canada.
problems: Individual and systemic effects. Symonds, B. D., & Horvath, A. O. (2004).
Journal of Family Psychology, 17, 534–44. Optimizing the alliance in couple therapy. Family
Shelef, K., & Diamond, G. M. (2008). Short Process, 43, 443–55.
form of the Vanderbilt Therapeutic Alliance Szapocznik, J., Perez-Vidal, A., Brickman, A.,
Scale: Development, reliability, and validity. Foote, F. H., Santisteban, D. A., Hervis, O.,
Psychotherapy Research, 18, 433–43. et al. (1988). Engaging adolescent drug users
Shelef, K., Diamond, G. M., Diamond, G. S., & and their families into treatment: A strategic
Liddle, H. A. (2005). Adolescent and parent structural approach. Journal of Consulting and
alliance and treatment outcome in multidimen- Clinical Psychology, 56, 552–57.
sional family therapy. Journal of Consulting and Thomas, S. E. G., Werner-Wilson, R. J., &
Clinical Psychology, 73, 689–98. Murphy, M. J. (2005). Influence of therapist and
Smerud, P. E., & Rosenfarb, I. S. (2008). The client behaviors on therapy alliance. Contemporary
therapeutic alliance and family psychoeducation Family Therapy: An International Journal, 27,
in the treatment of schizophrenia: An exploratory 19–35.

f r i e d l a n d e r, e s c u d e ro , h e at h e r i n g to n , d i a m o n d 109

5 Cohesion in Group Therapy

Gary M. Burlingame, Debra Theobald McClendon, and Jennifer Alonso

Cohesion is the most popular of several by providing a tabular summary of

relationship constructs (e.g., alliance, group therapeutic practices that have been linked
climate, group atmosphere) in the clinical to increased cohesion. Our intent in this
and empirical literature on groups. Over chapter is to illuminate the coherence in
time it has become synonymous with the cohesion literature, present the meta-
the therapeutic relationship in group analytic conclusions, and offer group leaders
psychotherapy (Burlingame, Fuhriman, & the measures and practices to improve treat-
Johnson, 2002). From the perspective of a ment outcomes.
group member, relationships are comprised
of three structural components: member– Definitions and Measures
member, member–group, and member– Definitions of cohesion have traveled a
leader. From the perspective of the therapist, serpentine trail (Bednar & Kaul, 1994;
relationships include the same three Crouch, Bloch, & Wanlass, 1994; Kivlighan,
components as well as two additional ones: Coleman, & Anderson, 2000) ranging
leader–group and, in the case of a cother- from broad and diffuse (e.g., forces that
apy, leader–leader. The complexity of these cause members to remain in the group,
multilevel structural definitions coupled “sticking-togetherness”) to focused (e.g.,
with the dynamic interplay among them attractiveness, alliance) and structurally
has created an array of competing cohesion coherent (e.g., tripartite relationship; Yalom
instruments and an absence of a consensual & Leszcz, 2005). At different times, review-
definition. ers have pleaded for definitional clarity with
In this chapter, we review the multiple two noting that “there is little cohesion in
measures of group cohesion and then discuss the cohesion research” (Bednar & Kaul,
a new measure that elucidates group rela- 1978, p. 800). Indeed, instruments tapping
tionships by suggesting two latent factors group acceptance, emotional well-being,
that explain common variance among these self-disclosure, interpersonal liking, and
group therapy relationship instruments— tolerance for personal space have been used
quality and structure. We provide clinical as measures of cohesion (Burlingame et al.,
examples to illustrate the multiple facets of 2002). Behavioral definitions have included
cohesion in group work. We then present attendance, verbal content, early termina-
an original meta-analytic review of cohesion’s tion, physical seating distance, amount
relation with treatment outcome and of eye contact, and the length of time
discuss potential moderators. We conclude group members engaged in a group hug

(Hornsey, Dwyer, & Oei, 2007). The defi- vertical and horizontal cohesion (Dion,
nitional challenges of cohesion are reflected 2000). Vertical cohesion represents a
by one team’s observation that “just about member–leader relationship and refers to a
anything that has a positive valence [with group member’s perception of the group
outcome] has been interpreted at some leader’s competence, genuineness, and
point as an index of cohesion” (Hornsey, warmth. Horizontal cohesion describes a
Dwyer, Oei, & Dingle, 2009, p. 272). group member’s relationship with other
Empirical investigations examining the group members and with the group as a
multidimensional structure of cohesion whole. The second dimension contrasts task
have reported as few as two and as many as cohesion (task performance) or the work of
five dimensions (Braaten, 1991; Cattell & the group with affective or emotional
Wispe, 1948; Griffith, 1988; Selvin & cohesion (interpersonal/emotional support;
Hagstrom, 1963) with common factors Griffith, 1988). In task cohesion, members
including vertical and horizontal cohesion are drawn to the group to accomplish a
as well as task and social/affective cohesion. given task, while in affective cohesion mem-
After reviewing the literature, we believe bers feel connected because of the emotional
there is ample evidence to support two support the group experience affords.
fundamental definitional dimensions of The measures of cohesion that have been
cohesion. The first dimension relates to the most frequently studied are summarized in
structure of the therapeutic relationship in Table 5.1. All but the Harvard measure
groups and is most often referred to as (Budman et al., 1989) are self-report,

Table 5.1 Cohesion measures that appear two or more times in the literature
Cohesion Measure Description of Measure Scales Dimensions:
Direction &
Group Climate • Engaged measures the degree of self-disclosure, cohesion, and • Horizontal
Questionnaire (GCQ; work orientation in the group. • Affective
MacKenzie, 1981, • Avoiding examines the degree to which individuals rely on the
1983) other group members or leaders, avoiding responsibility for
their own change process.
• Conflict examines interpersonal conflict and distrust.
Group Atmosphere • Group Cohesion includes: Autonomy, Affiliation, Involvement, • Horizontal & Vertical
Scale (GAS; Silbergeld Insight, Spontaneity, Support, and Clarity. • Affective
et al., 1975) • Submission examines group conformity.
• Aggression, Order, Practicality, and Variety contribute to other
aspects of perceived environment. Authors did not define
these scales.
Feelings about group • Modified from Schutz (1957) Cohesiveness Questionnaire; • Horizontal & Vertical
(Lieberman, Yalom & 13-item Likert scale • Affective
Miles, 1973) • No subscales; items ask members to reflect on group
participation, liking of group, inclusion in the group and
feelings about leader
• Designed to measure attractiveness of a group for its members
and perceived belonging or acceptance by other members in
the group.

bu r l i n g a me , mcc l e n d o n , a lo n s o 111
Table 5.1 Continued
Cohesion Measure Description of Measure Scales Dimensions:
Direction &
Gross (1957) Cohesion • No subscales reported • Horizontal
Scale Revised • Questions examine: group fit, perceived inclusion, attraction to • Affective & Task
(Lieberman et al., group activities, likability of members, how well the group
1973) works together, and the like.
Group Cohesion Member–member: • Horizontal & Vertical
(Piper et al., 1983) • Affective & Task
• Positive qualities examines likability, trust, and ease of
• Personal compatibility examines attraction, similarity, and
desire for personal friendship.
• Significance as a group member examines personal
• Positive qualities examines likability, trust, attraction, and
ease of communication.
• Dissatisfaction with leader’s role examines discontent with style,
communication, and level of personal disclosure.
• Personal compatibility examines similarity and desire for
• Mutual stimulation and effect examines engagement, inclusion,
and influence.
• Commitment to group examines attending the group and desire
for the group to continue.
• Compatibility of the group examines fit and attractiveness to
the group.
Group Environment • Relationship examines cohesion, leader support, and the • Horizontal & Vertical
Scale (GES; Moos, amount of freedom of action and expression of feelings • Affective & task
1986; Moos & encouraged in the group.
Humphrey, 1974) • Personal Growth examines independent action and expression
among members, the degree of the group’s task orientation,
the group’s encouragement of discussion of personal problems,
and anger and disagreement.
• System Maintenance and Change measures the degree of
organization, structure and rules in the group, role of the leader
in making decisions and enforcing rules, and how much the
group promotes diversity and change in its own process.
Group Attitude • 20-items no subscales reported, measures attraction to • Horizontal
Scale (Evans & Jarvis, group. • Affective
1986) • Illustrative items: “I feel involved in what is happening
in my group”, “If I were told my group would not meet today
I would feel bad” & “I feel it would make a difference to the
group if I were not here”.
• Initially validated against the Moos and Humphrey 1974
version of GES.

Table 5.1 Continued
Cohesion Measure Description of Measure Scales Dimensions:
Direction &
Group cohesion • 23-items based on combining selected items from the Group • Horizontal
questionnaire Attitude Questionnaire and Stokes (1983) 3-factor • Affective
(GCQ23: van Andel questionnaire; no subscales reported.
et al., 2003; Trijsburg • Illustrative items: “The group is honest and straightforward”,
et al., 2004). “I feel involved in what is happening in my group” & “There
are people in the group I would enjoy spending time with
outside the group session”.
Harvard Group • Behavioral process scale rated by clinicians from videotapes • Horizontal
Cohesiveness Scale • Five subscales and a global score: (1) withdrawal and self- • Affective & Task
(Budman et al., 1987, absorption vs. interest and involvement, (2) mistrust vs. trust,
1989) (3) disruption vs. cooperation, (4) abusiveness vs. expressed
caring and (5) unfocused vs. focused
• Global scale called fragmentation vs. global cohesiveness

and several were developed by modifying four commonly used relationship measures
previous measures. We have classified each by having 662 members from 111 different
measure in Table 5.1 by its use of the groups complete a copy of each. This study
structural and affective/task definitions of found that a two-dimensional model
cohesion. All assess horizontal cohesion (quality & structure) explained a majority
between members and their group, while of the common variance across the four
fewer than half focus on a member’s rela- measures. Specifically, positive bond, positive
tionship with the leader (vertical). Similarly, work, and negative relationship factors
affective bond is universally assessed by all explained how members perceived the
measures while the task cohesion is assessed quality of the relationship in both non-
by a third of the measures. clinical and clinical groups. Positive bond
A different relation with outcome might described the affective relationship members
result from using different cohesion mea- felt with their leader (vertical cohesion)
sures. In fact, we have proposed a two-factor and in member-to-member relationships
definition of the therapeutic relationship in (horizontal cohesion). Positive work equally
groups to potentially clarify mixed results captured the tasks and goals of the group
in the literature: (1) belonging and accep- while negative relationship captured empathic
tance factors—cohesion and member– failure with the leader and conflict in
leader alliance; and (2) interpersonal work the group. Interestingly, members were
factors—group working alliance, individ- unable to distinguish member-to-group
ual working alliance, and group climate from member-to-member relationships
(Burlingame, Mackenzie, & Strauss, 2004). yielding a two-factor structural dimension:
We undertook a series of studies to evaluate member–leader/–member.
a toolkit containing several therapeutic The three relationship qualities (positive
relationship measures (Strauss, Burlingame, bond, positive work, & negative relationship)
& Bormann, 2008). The first study and two structural factors (member–leader
(Johnson, Burlingame, Davies, & Gleave, & member–member) stimulated subsequent
2005) estimated the empirical overlap of studies that attempted to replicate these

bu r l i n g a me , mcc l e n d o n , a lo n s o 113
findings across clinical settings and coun- nonclinical process groups replicating the
tries. One study (Bormann & Strauss, 2007) first study (Johnson et al., 2005). Another
collected data from 67 inpatient psychody- population was added (seriously mentally
namic groups drawn from 15 hospitals in ill inpatients; SMI) to determine if the
Germany and Switzerland. Both the rela- model could be used with seriously ill
tionship quality and structure dimensions members; groups were primarily psycho-
emerged, but unlike the first study, these educational (Krogel, 2009; Krogel &
authors found support for three structural Burlingame, 2009). Using a sample of
components (member–member, member– 485 group members, they found the same
leader, & member–group). The next study two-dimensional relationship quality and
(Lorentzen, Høglend, & Ruud, 2008) tested structure model based on a 30-item
the same four measures and reported a simi- solution.
lar two-dimensional model that varied by These studies improve our consensual
stage of treatment (Bakali, Baldwin, & definition of cohesion. Table 5.2a weds the
Lorentzen, 2009). More specifically, early past definition of affective and task cohe-
sessions produced a strong member– sion dimensions from Table 5.1 with the
leader positive bond while later sessions relationship quality and structure model.
(10–11 & 17–18 sessions) included positive As can be seen, affective cohesion is split by
bonds with both other members and the the emotional valence of the item loading
group. either on the positive bond or negative
Findings from these studies led to an relationship dimensions. Table 5.2b depicts
item reduction process to determine if a how horizontal and vertical cohesion load
subset of “practice friendly” items could on the relationship structure dimension.
be identified from the original measures Specifically, horizontal cohesion captures
that contained over 80 items. A four-person the quality of member-to-member relation-
team with 75 collective years of clinical ships and vertical cohesion captures the
experience used both empirical (statistical quality of member-to-leader relationships.
fit with two-dimensional model) and Interestingly, the member–group dimen-
clinical criteria (does it provide actionable sion has been the most elusive theoretical
clinical information?) to produce a 40-item construct to empirically detect with
instrument called the Group Question- mixed findings from preceding studies.
naire that measured the three quality The positive and negative valence items
and structural factors (Burlingame, 2010). that consistently load on the member–
After item consensus was achieved, data group factor across the five studies were
was collected from counseling centers and principally drawn from MacKenzie’s

Table 5.2a Modified Framework for Understanding Cohesion Using Relationship Quality
Relationship Structure
Relationship Quality Member–member Member–leader Member–group
Positive Bond Affective cohesion—positive feelings & belonging
Positive Work Task cohesion
Negative Affective cohesion—empathic failure & conflict

114 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Table 5.2b Modified Framework for Understanding Cohesion Using Relationship Structure
Relationship Structure
Relationship Quality Member–member Member–leader Member–group
Positive Bond
Horizontal Vertical Group-as-a-whole
Positive Work Cohesion Cohesion Cohesion
cf. McClendon & Burlingame (in press)

(1983) Group Climate Questionnaire going after the last session (smiles at
(see Table 5.1). leader). Member–group positive work
Pete to group: I’ve had this rotten
Clinical Examples headache all day … it would have
The multidimensional complexity of group been real easy to stay home from
cohesion makes it impossible to provide a almost anything … but not from our
single, concrete example. However, the group. Member–group positive bond
relationship quality and structure model Mary to group: Yeah, today as I thought
(Tables 5.2a, 5.2b) provide a practice- about coming to group I knew that
friendly framework to recognize group Steve was going to make me laugh,
behavior that might facilitate cohesion. We everybody else in the group is so good
selected a transcript from Session 14 of a to give me their advice and support,
15-session group therapy (Burlingame & and I enjoy everybody so much.
Barlow, 1996). The segment begins with a Member–member/group positive bond
leader acknowledging that the next session Leader to group: That’s great. I really
will be the last and then probes regarding think every single person needs this
the work achieved over the course of the kind of a positive association, maybe
group. Group members don’t respond to not in a formal setting like this, but
the work probe but instead focus on the somehow or another like this, we
positive bond. The quality and structure need it. We really do. Every human
categories from Table 5.2 are identified by being needs it. Leader–member/group
italics. positive bond
Leader to group: This is our next-to-last The relationship quality/structure model
session. In thinking about our group, also accounts for the multiplicity of relation-
I wondered if anyone would care to ships in the group. This allows a leader to
speak to how they met their goals consider multiple aspects of the therapeutic
over the past 14 weeks? Leader–group relationship as they plan interventions. The
positive work probe following dialogue includes all three relation-
Mary to leader/group: Well … I think ship structures (member–member, member–
we’ve all had fun, I know I have. leader, and member–group) and begins with
In fact, we talked after you left last a leader probe regarding a conflictual event
night. We’re gonna keep our group that happened at the end of the last group.

bu r l i n g a me , mcc l e n d o n , a lo n s o 115
Leader to Steve: Steve, you OK? You Steve to Susan: Well, I’m glad you’re
seemed upset at the end of our last here … because I’ve been worried
group meeting. Leader–member, about you [Steve goes on to inquire
negative relationship probe about Susan’s situation and tell her
Steve: I need to apologize to you all his thoughts about it. This goes on
because I was a little bit abrupt with for quite a while.] member–member
you last week and I … thought that positive bond
was kinda tacky, uh … even though I Susan to Steve: Thank you. The reason
said it was none of you damned I tore out of work so fast to get here
business. is because I knew I’d get the reception
[Group laughs] … But uh, what I I just got. [Susan starts to cry and
meant was I’m not handling it well group laughs lightly, leader pats
and, therefore, I can’t share anything Susan on the shoulder and Susan
with you. I have nothing to give pats Mary on the knee] member–
[laughs] because I … uh, I’m not group/member positive bond
handling it well. Member–leader Steve to Susan: I apologize for being
negative relationship abrupt with you last week. That was
Leader: You’ve done a lot of good work tactless. I’m sorry. Member–member
over the past few months, but right negative relationship
now you feel like you’ve got nothing Susan to Steve: It didn’t bother me, but
to give—that you’re no longer I accept your apology. It means a lot
handling it well. Leader–member to me that you’d check in with me on
positive work that. Member–member positive bond
Steve: I also feel badly that Susan is
In this dialogue we see Steve interacting
not here today, I miss her. Member–
with a notable level of interpersonal risk
member positive bond [later interaction
with the group leader. When Susan arrives
will reveal an underlying member–
we see multiple levels of positive bond,
member negative relationship valence]
which undoubtedly supports Steve’s ability
Steve: I’ve been thinking about her and
to handle a second negative relationship
her crisis a great deal, and I almost
concern from the last session with Susan.
called you [leader] up to get her
The next segment reflects the end of the
phone number. I know we’re not
group session with continued evidence of
supposed to interact outside …
member–group cohesion as another group
[Steve goes on to tell the group what
member who reinitiates discussion about
he has been thinking about Susan’s
continuing to meet after the group has for-
situation. As they are talking Susan
mally ended:
comes into the group and the whole
group cheers when she enters]. Mary to group: I would like to
Group-member positive bond see us work more on what we
toward Susan discussed last week, and that’s
Leader to Susan: We wanted you to continue it all until it finishes.
to be here so bad, some of us were I really am very interested in that …
thinking that you had a crisis and for your information … sort of a,
we were worried. Leader–member you know, forming after the
positive bond [Susan explains why group … Member–group
she is late]. positive work/bond

116 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Susan to Mary, leaning on her terms: group psychotherapy, group therapy,
shoulder: I don’t know if we can live support groups, group counseling, cohesion,
without each other [dramatically]. group cohesion, cohesiveness, and group cli-
[Group laughs.] Member–member mate. Each abstract was reviewed for fit
positive bond with the above inclusion criteria and, if
Mary to group: Uh, yeah . . . uh once deemed promising, the article was retrieved
a month or something like that or and again reviewed for fit. A total of 24
whatever … I’m easy … but just to articles were included using this method.
get together and see how we’re doing Next, the reference sections of obtained
and talk it over and support each articles were reviewed, and 42 unduplicated
other. Member–group positive bond studies were identified and reviewed result-
ing in 6 studies being included. Finally, six
of the most frequently used cohesion mea-
Meta-Analytic Review
sures (Group Environment Scale, Piper’s
Before undertaking this meta-analysis,
Cohesion Questionnaire Scale, Group
we reviewed the literature for similar or
Climate Questionnaire, Group Atmosphere
related meta-analyses. Three cohesion meta-
Scale, Shulz’s Cohesion Questionnaire,
analyses were located (Evans & Dion, 1991;
Gross Cohesion Scale; cf. Table 5.1) across
Gully, Devine, & Whitney, 1995; Mullen
the 30 identified studies were searched using
& Copper, 1994), but none focused on
Google Scholar, yielding 1,027 abstracts.
cohesion in group psychotherapy; all
Ten additional studies were added, yielding
examined cohesion’s relationship to task
a final data set of 40 studies.
performance in nontherapeutic settings.
Thus, we relied upon five published group
Coding and Analysis
therapy meta-analyses to develop inclusion
We selected and coded 19 variables, many
criteria herein (Burlingame, Fuhriman, &
of which had been found to moderate
Mosier, 2003; Hoag & Burlingame, 1997;
outcome in previous group therapy meta-
Kosters, Burlingame, Nachitgall, & Strauss,
analyses. Five variables that assessed study
2006; Lipsey & Wilson, 2001; McRoberts,
characteristics were coded: year of publica-
Burlingame, & Hoag, 1998). To be included
tion, type of cohesion, outcome measure,
in our meta-analysis, studies must have
and time when cohesion was assessed. Three
included: (a) a group that was comprised of
leader (experience, orientation, single leader
at least three members, (b) groups meeting
vs. co-led groups) and three member
for the purpose of counseling, psycho-
characteristics (gender, diagnosis, treatment
therapy, or personal growth, (c) at least one
setting) were examined. The largest numbers
quantitative measure of both cohesion and
of variables were associated with the group
outcome, (d) data that allowed the calcula-
itself. Specifically, we were interested in
tion of effect sizes as weighted correlations,
the degree of structure associated with the
and (e) English text.
group treatment given the recent emphasis
on manual-based treatments. We coded for
Search Strategy specific practices in the studies that were
Articles were obtained by searching used to increase cohesion. We also coded
PsycINFO, MedLine, and Google Scholar for groups that allowed greater interaction
for publications between January 1969 among members, believing this might lead
and May 2009. A total of 1,506 abstracts to higher levels of cohesion than those that
were retrieved using the following search are more problem focused.

bu r l i n g a me , mcc l e n d o n , a lo n s o 117
Group treatments varied from psycho- If heterogeneity is found, variability among
educational through psychotherapy/ the study’s effect size mean would be higher
counseling to personal growth groups, so than what would be expected from sam-
we also coded this variable. In an earlier pling error. Moderator results are ultimately
meta-analysis (McRoberts et al., 1998), interpreted with more confidence when
homogeneous groups were associated with heterogeneity exists.
greater improvements. Thus, we coded for A random-effects model was used to
identical or similar diagnoses and present- determine whether differences in the cohe-
ing problems (homogeneous) contrasted sion–outcome relationship existed across
with heterogeneous member composition the 19 variables. Random effects assume
to determine if the correlation between that studies are selected from a population
cohesion and outcome might be greater for of studies and that variability between stud-
homogeneous groups. ies is the result of sampling error. This ana-
The task group literature has sug- lytic model is recommended as a more
gested that group size may moderate the conservative test (Hedges & Vevea, 1998;
cohesion–outcome link, so we coded for Lipsey & Wilson, 2001).
small, medium, and large groups. Finally,
since dose of therapy has been shown to Results
moderate overall outcome, we coded for A summary of study characteristics is pro-
both session length and number of group vided in Table 5.3. As one might expect,
sessions. most of the groups (80%) had a therapy
Eight raters (one graduate student and focus. The majority (58%) of studies were
seven undergraduate students) were trained published after 2000, although over a
on a codebook to rate articles unrelated to fourth were published prior to 1990, cap-
the studies herein using an 85% criterion turing several classic papers (e.g., Braaten,
level of agreement with interrater reliability 1989; Budman et al., 1989; Roether &
being high (kappa = 0.73). After achieving Peters, 1972; Yalom, Houts, Zimerberg, &
this criterion, raters were paired and inde- Rand, 1967). We elected to include a few
pendently coded the same article contained personal growth group studies that met our
in our meta-analysis. Complete agreement criteria and used a task format (e.g., Flowers,
was required with discrepancies resolved by Booraem, & Hartman 1981; Hurley, 1989;
the graduate student. In a few instances, Kivlighan & Lilly, 1997) since all are
the first and second authors met with the frequently cited in cohesion–outcome
graduate student (third author) to clarify literature.
discrepancies. Past reviewers have concluded that
A number of studies used several outcome cohesion has shown a positive relation with
and cohesion measures, thus creating mul- patient improvement in nearly every
tiple cohesion–outcome correlations from published report (Tschuschke & Dies,
a single study. When this occurred, we aver- 1994). Our own previous, narrative review
aged the several values (weighted by n) so concluded that approximately 80% of
that only one correlation per study was published studies demonstrated a positive
included in subsequent analyses. Following association between group cohesion and
calculation of the aggregate correlation, treatment outcome (Burlingame et al.,
we examined the degree of heterogeneity 2002).
in the results across studies using the The results from our meta-analysis with
Q-statistic (Berkeljon & Baldwin, 2009). each study depicted in Figure 5.1 show

118 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Table 5.3 Study Characteristics
Variable % N

Number of Studies 40
Year of publication (median) 1,997.7
Overall number of clients 3,323
Average age of clients 36.4
Average number of sessions 23.5
Theoretical orientation of group
Cognitive/behavioral 33 13
Psychodynamic/Existential 25 10
Humanistic/Interpersonal/Supportive 20 8
Eclectic 8 3
Unknown 20 8
Primary Diagnosis
Informal 35 14
Anxiety Disorder 13 5
Mood Disorder 18 7
Substance Disorder 3 1
Eating Disorder 5 2
Personality Disorder 13 5
Medical Condition (not Somatic disorder) 5 2
Unknown 18 7
North America 50 20
Europe 23 9
Canada 18 7
Australia 5 2
Role of Group
Only group/group as primary treatment 10 4
Part of milieu of treatment (e.g. medication, individual therapy) 23 9
Unknown 68 27
Inpatient 15 6
Outpatient 68 27
Unknown 18 7

Table 5.3 Continued
Variable % N

University Counseling Center 3 1
Clinic or Private Practice 13 5
Hospital 45 18
Community Mental Health Center 05 2
Community Location 3 1
Classroom Setting 13 5
Unknown 20 8
Type of Outcome Measure
General psychological distress 38 15
Depression 30 12
Anxiety 15 6
Quality of Life/General well being 20 8
Interpersonal Problems / Relationships 23 9
Self Esteem 13 5
Other 45 18
Unknown1 8 3
Number of Cohesion Measure administrations
Once 10 4
Twice 20 8
Three times 20 8
Four times 0 0
Five or more times 48 19
Unknown 3 1
Note: 1. Values don’t add up to 40 because some studies used multiples

a less glowing picture than past reviews, overall conclusion from 40 studies published
including our own. Only 43% of the studies across a 4-decade span is a positive relation
posted a statistically significant correlation between cohesion and outcome.
between cohesion and patient improvement.
Nonetheless, the weighted aggregate correla- Moderators and Mediators
tion for the 40 studies was a statistically sig- Until recently, there have been few empirical
nificant r = 0.25 with a 95% confidence studies examining moderator or mediator
interval of .17 to .32 (SE = .04) which is con- variables for the cohesion–outcome link
sidered to be a moderate effect. Thus, the (Hornsey et al., 2007). Mediators have

120 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Study name Statistics for each study

Effect Lower Upper Z- p- Correlation and 95% CI

size limit limit value value
Andel et al (2003) 0.19 -0.14 0.48 1.15 0.25
Antonuccio et al (1987) 0.00 -0.19 0.19 0.00 1.00
Beutal et al (2006) 0.23 0.06 0.39 2.68 0.01
Braaten (1989) 0.21 0.02 0.38 2.20 0.03
Budman et al (1989) 0.63 0.48 0.74 6.91 0.00
Crowe & Grenyer (2008) 0.30 -0.06 0.60 1.63 0.10
Falloon 1981 0.16 -0.12 0.42 1.12 0.26
Flowers et al (1981) 0.56 0.09 0.83 2.28 0.02
Gillaspy et al (2002) 0.19 -0.10 0.45 1.30 0.19
Grabhorn et al (2002) 0.18 -0.11 0.44 1.22 0.22
Hilbert et al (2007) 0.24 0.08 0.39 2.84 0.00
Hoberman et al (1988) 0.38 0.09 0.61 2.50 0.01
Hurley (1997) 0.35 0.28 0.41 9.49 0.00
Hurley (1989) 0.70 0.64 0.75 16.7 0.00
Joyce et al (2007) 0.09 -0.10 0.28 0.92 0.36
Kipnes et al (2002) 0.00 -0.57 0.57 0.00 1.00
Kivlinghan & Lilly (1997) 0.36 0.00 0.64 1.96 0.05
Levenson & Macgowan (2004) 0.33 0.09 0.54 2.61 0.01
Lipman et al (2007) 0.15 -0.18 0.45 0.89 0.37
Lorentzen et al (2004) 0.30 -0.33 0.75 0.93 0.35
Mackenzie & Tschuschke (1993) 0.46 -0.05 0.78 1.79 0.07
Marmarosh et al (2005) 0.54 0.39 0.67 6.01 0.00
Marziali et al (1997) 0.19 -0.32 0.62 0.72 0.47
May et al (2008) 0.18 0.01 0.34 2.06 0.04
Norton et al (2008) 0.30 0.04 0.53 2.21 0.03
Oei & Brown (2006) -0.04 -0.19 0.12 -0.51 0.61
Ogrodniczuk & Piper (2003) 0.22 0.03 0.39 2.28 0.02
Ogrodniczuk et al (2005) 0.22 -0.01 0.43 1.90 0.06
Ogrodniczuk (2006) 0.42 0.12 0.65 2.69 0.01
Ratto & Hurley (1995) 0.23 -0.12 0.53 1.28 0.20
Rice (2001) 0.00 -0.26 0.26 0.00 1.00
Roether & Peters (1972) -0.18 -0.43 0.10 -1.26 0.21
Rugel & Barry (1990) 0.10 -0.28 0.46 0.50 0.62
Ryum et al (2009) 0.15 -0.24 0.50 0.74 0.46
Taft et al (2003) 0.18 -0.01 0.36 1.86 0.06
Taube-Schiff et al (2007) 0.43 0.11 0.67 2.56 0.01
Tschuschke & Dies (1994) 0.72 0.35 0.90 3.27 0.00
Woody & Adesky (2002) 0.17 -0.12 0.43 1.15 0.25
Wright & Duncan (1986) 0.13 -0.26 0.49 0.64 0.52
Yalom et al (1967) 0.11 -0.30 0.48 0.52 0.60
−1 −0.5 0 0.5 1 1.5

Favors negative relationship Favors positive relationship

Fig. 5.1 Weighted effect size for cohesion-outcome relationship

been proposed (e.g., member acceptance, from the Group Climate Questionnaire
support, self-disclosure, and feedback), but (MacKenzie, 1983), which posted the
there has been little progress due to the second highest weighted correlation (0.35).
varied definitions and confounds with Interestingly, one of the oldest cohesion
group cohesion (Hornsey et al., 2007). measures (Group Environment Scale)
Our analysis of the studies found only a posted the lowest r value.
handful of moderators among the 19 coded The cohesion–outcome relationship was
variables. explored to see if it varied by outcome
measure. As with the cohesion instrument
Study Characteristics analysis, most measures were used too
None of the study characteristics (publica- infrequently to test for reliable differences.
tion year, outcome or cohesion measure, or However, two measures (SCL-90, BDI)
time of assessment) explained variability that assess general psychiatric and depres-
among the studies. However, there were a sive symptoms, respectively, were each used
number of cohesion measures that were in a dozen studies posting reliable values
used two or more times, and their weighted that were at or above the meta-analytic
averages are depicted in Figure 5.2. The average. Higher weighted averages were
most frequently used was the engaged scale found on both interpersonal and self-esteem

bu r l i n g a me , mcc l e n d o n , a lo n s o 121
Wtd. Ave. Lower Upper Times
Cor. Cor. limit limit used
Group Atmosphere Scale (GAS-C; Silbergeld et al, 1975) 0.25 0.21 0.01 0.45 5
Group Cohesion (GC ; Piper et al 1983) 0.15 0.17 0.00 0.43 5
Group Environment Scale (GES; Moos, 1986) 0.05 0.07 -0.08 0.19 2
Stuttgarter Bogen (Czogalik & Koeltzow, 1987) 0.35 0.52 0.09 0.72 3
Harvard Group Cohesiveness Scale (GCS; Budman, et al, 1987) 0.58 0.38 0.00 0.70 2
Group Climate Questionnaire (GCQ; MacKenzie, 1981, 1983) 0.35 0.29 0.00 0.70 12
Gross Cohesion (Gross, 1957; Lieberman et. al, 1973) 0.23 0.20 0.00 0.58 6
Outcome Questionnaire (OQ; Lambert, et al, 1997)) 0.26 0.27 0.24 0.31 2
Inventory of Interpersonal Problems (IIP; Horowitz et al, 1988) 0.40 0.36 0.25 0.42 3
Therapy Project List (Braaten, 1989) 0.21 0.21 0.18 0.23 2
Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965) 0.54 0.51 0.45 0.58 2
Profile of Mood States (POMS; McNair, et al, 1981) 0.32 0.32 0.24 0.40 2
Beck Anxiety Inventory (BAI; Beck, et al, 1988) 0.13 0.11 -0.08 0.27 3 −0.2 0 0.2 0.4 0.6 0.8
Beck Depression Inventory(BDI; Beck, et al, 1996) 0.23 0.20 0.00 0.59 11
Symptom Checklist (SCL-90; Derogatis, 1977) 0.35 0.27 0.00 0.66 12

Fig. 5.2 Weighted Correlations and range by cohesion and outcome measure

measures (IIP, RSES), but these results are demonstrated across all three major diagnos-
heavily influenced by the two student tic classifications; Axis I (r = 0.17), Axis II
growth group studies (Hurley, 1989; (r = 0.41), and V-code (r = 0.26).
Kivlighan & Lilly, 1997). Thus, it remains
unclear how these outcome measures might Leader Variables
operate in clinical populations. A conclusion No evidence was found to support that
that seems warranted is that the cohesion– leader experience or single versus co-led
outcome relationship is well supported groups explained differences in the
when outcome is defined by general psychi- cohesion–outcome correlations. However,
atric and depressive symptoms. Moreover, there was a difference in the cohesion–
since the SCL-90 and BDI are also two of outcome relation when one considered the
the most frequently used instruments to differences across the theoretical orienta-
evaluate the effectiveness of group psycho- tion of the group leader; Q = 23.56, df = 9,
therapy, the generalizability to outcome p < 0.05. Leaders espousing an interper-
appears sound. sonal orientation posted the highest
cohesion–outcome relation (r = 0.58), with
Member Variables psychodynamic (r = 0.25) and cognitive-
Of the member variables, only one explained behavioral (r = 0.18) orientations posting
differences in the cohesion–outcome link. the lowest values. The remaining orienta-
The average age of participants was nega- tions posted either statistical trends
tively associated with effect size magnitude (humanistic, r = 0.21) or no reliable rela-
within studies (r = −.63; p < 0.0001). Studies tionship (behavioral, eclectic). This argues
with relatively younger group members for cohesion being considered as an evi-
tended to yield effect sizes of higher magni- dence-based relationship factor for groups
tude than studies with relatively older group using a cognitive-behavioral, psychody-
members; Q = 14.92, df = 1, p < 0.05. This namic, and interpersonal orientation.
finding was not explained by client symp-
tom severity; cohesion was reliably related to Group Variables
outcome in both inpatient and outpatient Four group variables proved useful in
settings (r = 0.29 and 0.24, respectively). explaining differences in the cohesion–
Furthermore, the positive association of outcome association. In the past, we
group cohesion and client outcome was (Burlingame et al., 2004; Fuhriman &

122 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Burlingame, 1994) have suggested that the higher cohesion–outcome correlations than
best test of the cohesion–outcome relation did groups lasting 12 or fewer sessions
would be to examine studies that empha- (r = 0.17). Interestingly, there was no statisti-
sized the importance of cohesion as a thera- cally significant difference between groups
peutic strategy. If group cohesion is lasting 13–19 sessions (r = 0.36) and for 20
undervalued or neglected by a group leader, or more sessions (r = 0.31).
its presence would likely be diminished and
perhaps attenuate its relationship with out- Other Potential Moderators
come. Two previous studies which included Two studies that fell outside our meta-
cognitive pre-training to enhance cohesion analytic review parameters—because they
showed higher cohesion for the pre-trained were recently published or the text is in
groups (Santarsiero, Baker & McGee, 1995; German—suggest a potential patient mod-
Palmer, Baker & McGee, 1997). The one erator variable. In the first study, members’
study herein that described procedures for interpersonal style moderated the relation-
enhancing cohesion in their methods section ship between cohesion and outcome
(Kivlighan & Lilly 1997) pre-trained par- with 73 depressed German inpatients
ticipants using videotapes and produced a (Schauenburg, Sammet, Rabung & Strack,
cohesion–outcome correlation (r = .36) that 2001). Specifically, patients with interper-
was slightly higher yet not statistically sonal problems described as “too friendly”
different from the weighted average. improved more when their cohesion
Another significant group variable decreased during therapy, whereas patients
(group focus) found a difference between with cold or hostile interpersonal problems
problem specific and interactive groups; improved most when their experience of
Q = 4.75, df = 1, p < 0.05. Problem specific cohesion increased during the group.
groups were comprised of members with A recent replication of that study involved
similar diagnoses, and group time appeared 327 mixed-diagnosis adults treated on a psy-
to be principally focused on this common chodynamically oriented inpatient psycho-
theme. Interactive groups had members therapy unit in Germany (Dinger &
who were more interactive, and group time Schauenburg, 2010). Higher levels of cohe-
appeared to be less structured. Interactive sion as well as an increase in cohesion over
groups posted a higher cohesion–outcome the life of the group were associated with
correlation than problem-specific groups greater symptom improvement, replicating
(r = 0.38 and r = 0.21, respectively). the findings of our meta-analysis. Once
The cohesion–outcome relationship again patients who described themselves as
proved to be statistically significant when too cold and who reported increased cohe-
examining the size of the group, Q = 4.54, sion posted the greatest improvement; the
df = 1, p < 0.05. Groups comprised of opposite was evident for those who described
5–9 members in each session posted the themselves as too friendly. The value of this
strongest cohesion–outcome relationship study is that it offers one theory-driven
(r = 0.35) whereas groups of any other size explanation of both the positive and neutral
(fewer than 5 members present or more than relationship findings in this meta-analysis.
9 members) were much weaker (r = 0.16). Could a member’s interpersonal style explain
Finally, there were differences in the past mixed cohesion–outcome findings?
cohesion–outcome correlations by number Unfortunately, the “jury is out” on this
of sessions; Q = 6.87, df = 2, p < 0.05. question since the primary measure assess-
Groups lasting more than 12 sessions posted ing cohesion in both studies falls short on

bu r l i n g a me , mcc l e n d o n , a lo n s o 123
psychometric support, thus attenuating our due to the scant number of studies that focus
confidence in its conclusions. on the leader (Burlingame et al., 2004).
Even if one were to uncover a handful of
Limitations of the Research studies testing the same member or leader
One of the clear challenges in understand- moderator, it is highly likely that different
ing and utilizing cohesion as an evidence- measures of cohesion would be used.
based principle has been the variability in We believe the biggest limitation to our
definition and measurement. The two- findings and the research in general is the
dimensional model (structure and quality) heterogeneity of study characteristics that
offers a promising, parsimonious, and is often hidden in meta-analyses. Even
empirically based definition of the latent though we used 40 studies, when one con-
structure inherent in measures of group siders the possible interactions among dif-
relationship. It suggests that leaders pay ferent diagnoses, settings, orientations, and
attention to the “who” (member, group, & type of groups, considerable caution must
leader), “what” (are we getting work done?), be invoked in interpreting our results.
and “how” (positive and negative emotional There are simply too few studies to ade-
valence) of group relationships. quately test for potential interactions
A second challenge in the literature has between the characteristics tested herein.
been the mixed findings regarding cohesion’s For example, the larger relation between
relation with treatment outcome. Some cohesion and outcome makes sense for
studies support its relationship with out- groups that last longer than 12 sessions.
come; others show no association. The meta- However, what we don’t know is how this
analysis clarifies this confusion by pointing plays out for different theoretical orienta-
out differences between measures, theoreti- tions and clinical settings.
cal orientations, group length, and group Finally, the question of causality cannot be
focus. For instance, cognitive-behavioral addressed in these correlational studies.
groups make up 80% of the recent group Perhaps the strongest evidence to support a
literature (Burlingame & Baldwin, in press), causal relationship was the finding from
and it’s clear that a cohesion–outcome link studies that intentionally used interventions
exists for these groups. Similarly, longer and to enhance cohesion that resulted in a stron-
more interactive groups produce larger ger cohesion–outcome relation. However, as
cohesion–outcome correlations, but even pointed out above, these studies were not
short groups (fewer than 12 sessions) still based on groups comprised of members
show a cohesion–outcome link. having a formal psychiatric diagnosis.
There are several specific limitations to
the findings herein. First, it is virtually
Therapeutic Practices
impossible to assess potential member or
We see the following therapeutic practices
leader moderators because of the absence of
supported by our meta-analysis:
research on this topic in the group therapy
literature. For instance, low psychological • Cohesion is reliably associated
mindedness and more severe symptoms have (r = 0.25) with group outcome when
been linked to early dropout (Burlingame outcome is defined as reduction in
et al., 2004), but to our knowledge, these symptom distress or improvement in
have never been formally tested as modera- interpersonal functioning. All group
tors of cohesion–outcome. Additionally, leaders should foster cohesion in its
there is no consensus on leader moderators multiple manifestations.

124 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
• Cohesion is certainly involved with Table 5.4. These behaviors track onto a
patient improvement in groups using a behavioral rating scale (Group Psychotherapy
cognitive-behavioral, psychodynamic, or Interventions Rating Scale; GPIRS) devel-
interpersonal orientation. oped from interventions suggested in our
• Group leaders emphasizing member first cohesion chapter (Burlingame et al.,
interaction, irrespective of theoretical 2002). In two studies (Sternberg & Trijsburg,
orientation, post higher cohesion– 2005; Snijders, Trijsburg, De Groot, &
outcome links than problem-focused Duivenvoorden, 2005), group structure,
groups. Thus, it is important to encourage verbal interaction, and emotional climate
member interaction. interventions were positively correlated with
• Cohesion explains outcome member-reported cohesion. A recent North
regardless of the length of the group, but American study (Chapman, Baker, Porter,
is strongest when a group lasts more than Thayer & Burlingame, 2010) translated the
12 sessions and is comprised of 5–9 Dutch GPIRS into English and replicated
members. Group cohesion obviously these findings. Small-to-moderate correla-
requires the correct balance of member tions were found with each GPIRS subscale,
interaction and time to build. suggesting that leader interventions intended
• Younger group members experience to affect emotional climate, manage verbal
the largest outcome gains when cohesion interaction, and maintain group structure were
is present within their groups. Fostering moderately correlated with member-reported
cohesion may be particularly useful for levels of cohesion. Leader interventions facil-
those working with young people (e.g., itating structure, emotional climate, and
counseling centers and adolescents). managing verbal interaction were positively
• Cohesion contributes to group related to cohesion and negatively related
outcome across different settings to interpersonal distrust and conflict, repli-
(inpatient and outpatient) and diagnostic cating the Dutch findings on a different
classifications. Thus, all leaders should measure of cohesion.
actively engage in interventions that foster Cohesion is integrally related to the
and maintain cohesion. success of group therapy, and the research
has identified specific behaviors that
In this regard, we would point to therapist enhance cohesion. For these reasons, we
behaviors that can enhance group cohesion. recommend the behaviors in Table 5.4 to
These specific interventions are depicted in group practitioners.

Table 5.4 Group Psychotherapy Intervention Rating Scale (GPIRS)

Group Structuring
Setting treatment expectation Set group agendas (such as discussion topics or group activities)
Described rationale underlying treatment
Establishing group procedures Discussed group rules (such as time, attendance, absences, tardiness,
confidentiality, participation)
Identified and discussed fears/concerns regarding self disclosure
Structured exercises that focus on emotional expression and exchange

bu r l i n g a me , mcc l e n d o n , a lo n s o 125
Table 5.4 Continued
Role preparation Discussed member roles and responsibility
Discussed leader roles and responsibility
Verbal Interaction
Verbal style Modeled giving personal information in the “here and now”
and interaction Modeled appropriate member-member behavior
Modeled appropriate self disclosure
Modeled appropriate feeling disclosure
Maintained moderate control
Facilitated appropriate member-member interaction
Self disclosure Encouraged self disclosure without “forcing it”
Encouraged self disclosure relevant to the current group agenda
Helped members understand that disclosed issues achieve more resolution than
undisclosed issues
Encouraged here-and-now vs. story-telling disclosure
Interrupted ill-timed or excessive member disclosure
Elicited member-member feeling disclosure (versus informational disclosures)
Leader shared relevant personal experience from outside of therapy (without
being judgmental or overly-intellectual)
Feedback Reframed injurious feedback (interrupting, if necessary)
Restated corrective feedback by member
Used consensus to reinforce feedback (toward therapist or group member)
Balanced positive and corrective leader-to-member feedback
Encouraged positive feedback
Gave structured feedback exercise
Helped balance positive and corrective member-to-member feedback
Therapist helped members apply in-group feedback to out-of-group situations
Creating and Maintaining a Therapeutic Emotional Climate
Leader contribution Maintained balance in expressions of emotional support and confrontation
Showed understanding of the members and their concerns
Refrained from conveying personal feelings of hostility and anger in response to
negative member behavior
Leader was not defensive when interventions failed
Leader was not defensive when confronted by a member
Maintained an active engagement with the group and its work
Used nonjudgmental language with members
Modeled expressions of open and genuine warmth
Encouraged active emotional engagement between group members
Fostered a climate of both support and challenge
Responded at an emotional level
Developed and/or facilitated relationships with and among group members
Helped members recognize why they feel a certain way (identifying underlying
concerns or motives)
Member contribution Prevented or stopped attacking and judgmental expressions between members
Assisted members in describing their emotions
Recognized and responded to the meaning of groups members’ comments
Prevented situations in which members felt discounted, misunderstood,
attacked, or disconnected
Involved members in describing and resolving conflict (instead of avoiding
Elicited verbal expressions of support among group members
Encouraged members to respond to other members’ emotional expression (such
as acceptance, belonging, empathy)

Braaten, L. (1991). Group cohesion: A new multi-
dimensional model. Group, 15, 39–55.
An asterisk (∗) indicates studies included in ∗
Braaten, L. J. (1989). Predicting positive goal
the meta-analysis. attainment and symptom reduction from early

Antonuccio, D. O., Davis, C., Lewinsohn, P. M., group climate dimensions. International Journal
& Breckenridge, J. S. (1987). Therapist variables of Group Psychotherapy, 39(3), 377–87.
related to cohesiveness in a group treatment for Broker, M., Rohricht, F., & Priebe, F. (1995). Initial
depression. Small Group Research, 18, 557–64. assessment of hospital treatment by patients with
Bakali, J., Baldwin, S., & Lorentzen, S. (2009). paranoid schizophrenia: A predictor of outcome.
Modeling group process constructs at three Psychiatry Research, 58(1), 77–81.
stages in group psychotherapy. Psychotherapy Budman, S. H., Demby, A., Feldstein, M.,
Research, 19, 332–43. Redondo, J., Scherz, B., Bennett, M. J., et al.
Beck, A. T., Epstein, N., Brown, G., & Steer, R. A. (1987). Preliminary findings on a new instru-
(1988). An inventory for measuring clinical ment to measure cohesion in group psychother-
anxiety: Psychometric properties. Journal of apy. International Journal of Group Psychotherapy,
Consulting and Clinical Psychology, 56, 893–97. 37, 75–94.

Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Budman, S. H., Soldz, S., Demby, A., Feldstein, M.,
Manual for the Beck Depression Inventory-II. San Springer, T., & Davis, M. S. (1989). Cohesion,
Antonio, TX: Psychological Corporation. alliance and outcome in group psychotherapy.
Bednar, R. L., & Kaul, T. J. (1978). Experiential Psychiatry, 52(3), 339–50.
group research: Current Perspectives. In Burlingame, G. (2010). Small group treatments:
S. L. Garfield & A. E. Bergin (Eds.), Handbook Introduction to special section. Psychotherapy
of psychotherapy and behavior change (2th ed., Research, 20(1), 1–7.
pp. 769–815). New York: Wiley. Burlingame, G., & Baldwin, S. (2010). History of
Bednar, R. L., & Kaul, T. J. (1994). Experiential group psychotherapy. In History of Psychotherapy
group research: Can the canon fire? In (2nd ed.), J. Norcross, G. VandenBos, &
A. E. Bergin & S. L. Garfield (Eds.), Handbook D. Freedheim. (Eds). (pp. 505–515), Washington
of psychotherapy and behavior change (4th ed., DC: American Psychological Association.
pp. 631–63). New York: Wiley. Burlingame, G. M., & Barlow, S. (1996). Outcome
Berkeljon, A., & Baldwin, S. A. (2009). An and process differences between professional and
introduction to meta-analysis for psychotherapy nonprofessional therapists in time-limited group
outcome research. Psychotherapy Research, 19(4), psychotherapy. International Journal of Group
511–518. Psychotherapy, 46, 455–78.

Beutal, M. E., Knickenberg, R. J., Krug, B., Burlingame, G. M., MacKenzie, K. R., & Strauß, B.
Mund, S., Schattenburg, L., & Zwerenz, R. (2004). Evidence-based small group treatments.
(2006). Psychodynamic focal group treatment In M. Lambert, A. E. Bergin, & S. L. Garfield
for psychosomatic inpatients—with an emphasis (Eds.), Handbook of psychotherapy and behavior
on work-related conflicts. International Journal change (5th ed.). New York: Wiley & Sons.
of Group Psychotherapy, 56(3), 285–305. Burlingame, G. M., & Fuhriman, A. (1994).
Bormann, B., & Strauß, B. (2007). Gruppenklima, Epilogue. In A. Fuhriman & G. M. Burlingame
Kohäsion, Allianz und Empathie als Kompo- (Eds.), Handbook of group psychotherapy: An
nenten der therapeutischen Beziehung in empirical and clinical synthesis (pp. 559–62).
Gruppenpsychotherapien–Überprüfung eines New York: Wiley & Sons.
Mehrebenen-Modells. [Group climate, cohesion, Burlingame, G., Fuhriman, A., & Johnson, J.
alliance, and empathy as components of the thera- (2002). Cohesion in group psychotherapy. In
peutic relationship within group psychotherapy–Test J. C. Norcross (Ed). Psychotherapy Relationships
of a multilevel model]. Gruppenpsychotherapie und That Work (pp. 71–87). New York: Oxford
Gruppendynamik, 43, 1–20. University Press.
Bormann, B., & Strauß, B. (2009, June). The Burlingame, G. M., Fuhriman, A. F., & Mosier, J.
German Group Questionnaire: A multi-site valida- (2003). The differentiated effectiveness of group
tion study. Paper presented at the annual meeting psychotherapy: A meta-analytic review. Group
of the Society for Psychotherapy Research, Dynamics: Theory, Research, and Practice,
Santiago, Chile. 7(1), 3–12.

bu r l i n g a me , mcc l e n d o n , a lo n s o 127
Cattell, R. B., & Wispe, L. G. (1948). The dimen- abuse treatment outcomes. Psychotherapy
sions of syntality in small groups. Journal of Research, 12(2), 213–29.

Social Psychology, 28, 57–78. Grabhorn, R., Kaufhold, J., & Overbeck, G.
Chapman, C., Baker, E., Porter, G., Thayer, S., & (2002). The role of differentiated group experi-
Burlingame, G. (2010). Rating group therapist ence in the course of inpatient psychotherapy. In
interventions: The validation of the Group Serge P. Shohov (Ed.), Advance in psychology
Psychotherapy Rating Scale. Group Dynamics: research, Volume 12, (pp. 141–54). Hauppauge,
Theory, Research, & Practice, 14(1), 15–31. NY: Nova Science Publishers.
Crouch, E. C., Bloch, S., & Wanlass, J. (1994). Griffith, J. (1988). Measurement of group cohesion
Therapeutic factors: Interpersonal and intraper- in U.S. Army units. Basic and Applied Social
sonal mechanisms. In A. Fuhriman & Psychology, 9, 149–71.
G. M. Burlingame (Eds.), Handbook of group Gross, E. F. (1957). An empirical study of the concepts
psychotherapy (pp. 269–315). New York: Wiley. of cohesiveness and compatibility. Honors thesis,

Crowe, T. P., & Grenyer, B. F. S. (2008). Is therapist Harvard University, Cambridge, MA.
alliance or whole group cohesion more influen- Gully, S. M., Devine, D. J., & Whitney, D. J.
tial in group psychotherapy outcomes? Clinical (1995). A meta-analysis of cohesion and
Psychology and Psychotherapy, 15, 239–46. performance: Effects of level of analysis and
Czogalik, D., & Koltzow, R. (1987). Zur Normierung task interdependence. Small Group Research,
des Stuttgarter Bogens. Gruppenpsychotherapie 26, 497–520.
und Gruppendynamik, 23, 36–45. Hedges, L. V., & Vevea, J. L. (1998). Fixed- and
Derogatis, L. R. (1977). SCL-90: Administration, random-effects models in meta-analysis.
scoring, and procedures manual-R (revised). Psychological Methods, 3, 486–504.

Baltimore: John Hopkins University School of Hilbert, A., Saelens, B. E., Stein, R. I., Mockus, D. S.,
Medicine, Clinical Psychometrics Research Unit. Welch, R. R., Matt, G. E., et al. (2007).
Dinger, U., & Schauenburg, H. (2010). Effects of Pretreatment and process predictors of outcome in
individual cohesion and patient interpersonal interpersonal and cognitive behavioral psychother-
style on outcome in psychodynamically oriented apy for binge eating disorder. Journal of Consulting
inpatient group psychotherapy. Psychotherapy and Clinical Psychology, 75(4), 645–51.
Research, 20(1), 22–29. Hoag, M. J., & Burlingame, G. M. (1997).
Dion, K. L. (2000). Group cohesion: From “field of Evaluating the effectiveness of child and adoles-
forces” to multidimensional construct. Group cent group treatment. Journal of Clinical Child
Dynamics, 4, 7–26. Psychology, 26, 234–46.

Evans, C. R., & Dion, K. L. (1991). Group cohesion Hoberman, H. M., Lewinsohn, P. M., & Tilson, M.
and performance: A meta-analysis. Small Group (1988). Group treatment of depression: Individual
Research, 22, 175–86. predictors of outcome. Journal of Consulting and
Evans, N. J., & Jarvis, P. A. (1986). The Group Clinical Psychology, 56(3), 393–98.
Attitude Scale: A measure of attraction to group. Hornsey, M., Dwyer, L., & Oei, T. (2007). Beyond
Small Group Behavior, 17, 203–216. cohesiveness: Reconceptualizing the link between

Falloon, I. R. H. (1981). Interpersonal variables group processes and outcomes in group psycho-
in behavioural group therapy. British Journal of therapy. Small Group Research, 38, 567–92.
Medical Psychology, 54, 133–41. Hornsey, M., Dwyer, L., Oei, T., & Dingle, G. A.

Flowers, J. V., Booraem, C. D., & Hartman, K. A. (2009). Group processes and outcomes in group
(1981). Client improvement on higher and lower therapy: Is it time to let go of cohesiveness?
intensity problems as a function of group International Journal of Group Psychotherapy,
cohesiveness. Psychotherapy: Theory, Research, and 59(2), 267–78.
Practice, 18, 246–251. Horowitz, L., Rosenberg, S., Baer, B., Ureno, F., &
Fuhriman, A., & Burlingame, G., (Eds.). (1994). Villasenor, V. (1988). Inventory of interpersonal
Handbook of group psychotherapy: An empirical problems: Psychometric properties and clinical
and clinical synthesis. New York: John Wiley applications. Journal of Consulting and Clinical
& Sons. Psychology, 56(6), 885–92.
∗ ∗
Gillaspy, J. A., Jr., Wright, A. R., Campbell, C., Hurley, J. R. (1989). Affiliativeness and outcome
Stokes, S., & Adinoff, B. (2002). Group alliance in interpersonal groups: Member and leader
and cohesion as predictors of drug and alcohol perspectives. Psychotherapy, 26(4), 520–23.

128 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
∗ Lieberman, M., Yalom, I., & Miles, M. (1973).
Hurley, J. R. (1997). Interpersonal theory and
measures of outcome and emotional climate Encounter groups: First facts. New York: Basic
in 111 personal development groups. Group Books.

Dynamics: Theory, Research, and Practice, 1(1), Lipman, E. L., Waymouth, M., Gammon, T.,
86–97. Carter, P., Secord, M., Leung, O., Mills, B., &
Johnson, J. E., Burlingame, G. M., Olsen, J. A., Hicks, F. (2007). Influence of group cohesion on
Davies, D. R., & Gleave, R. L. (2005). Group maternal well-being among participants in a
climate, cohesion, alliance, and empathy in support/education group program for single
group psychotherapy: Multilevel structural mothers. American Journal of Orthopsychiatry,
equation models. Journal of Counseling Psychology, 77(4), 543–49.
52, 310–21. Lipsey, M. W., & Wilson, D. B. (2001). Practical
∗ meta-analysis. Applied Social Research Methods
Joyce, A. S., Piper, W. E., & Ogrodniczuk, J. S.
(2007). Therapeutic alliance and cohesion vari- Series, Vol. 49). Thousand Oaks, CA: Sage.
ables as predictors of outcome in short-term Lorentzen, S., Høglend, P., & Ruud, T. (2008,
group therapy. International Journal of Group June). The efficacy of short- and long-term
Psychotherapy, 57(3), 269–96. psychodynamic group psychotherapy: A Norwegian
∗ multi-center study. Panel presentation at the
Kipnes, D. R., Piper, W. E., & Joyce, A. S. (2002).
Cohesion and outcome in short-term psychody- annual meeting of the Society for Psychotherapy
namic groups for complicated grief. International Research, Barcelona, Spain.

Journal of Group Psychotherapy, 52(4), 483–509. Lorentzen, S., Sexton, H. C., & Hoglend, P.
Kivlighan, D. M., Coleman, M. N., & (2004). Therapeutic alliance, cohesion and
Anderson, D. C. (2000). Process, outcome, and outcome in a long-term analytic group. A prelim-
methodology in group counseling research. In inary study. Nordic Journal of Psychiatry,
S. D. Brown & R. W. Lent (Eds.), Handbook 58, 33–40.
of counseling psychology (3rd ed., pp. 767–96). MacKenzie, K. R. (1981). Measurement of
New York: Wiley. group climate. International Journal of Group
∗ Psychotherapy, 31, 287–95.
Kivlighan, D. M., & Lilly, R. L. (1997).
Developmental changes in group climate as MacKenzie, K. R. (1983). The clinical application
they relate to therapeutic gain. Group Dynamics: of a group climate measure. In R. R. Dies &
Theory, Research, and Practice, 1, 208–21. K. R. MacKenzie (Eds), Advances in group
Kosters, M., Burlingame, G., Nachtigall, C., & psychotherapy: Integrating research and practice
Strauss, B. (2006). A meta-analytic review of the (pp. 159–70). New York: International
effectiveness of inpatient group psychotherapy. Universities Press.

Group Dynamics: Theory, Research, & Practice. Mackenzie, R. K., & Tschuschke, V. (1993).
10(2), 143–63. Relatedness, group work, and outcome in long-
Krogel, J. (2009). The Group Questionnaire: A new term inpatient psychotherapy groups. Journal of
measure of the group relationship. Provo, UT: Psychotherapy Practice and Research, 2, 147–56.

Brigham Young University. Marmarosh, C., Holtz, A., & Schottenbauer, M.
Krogel, J., & Burlingame, G. (2009). The group (2005). Group cohesiveness, group-derived
questionnaire: A new measure of the therapeutic collective self-esteem, group-derived hope, and
group relationship. Paper presented at the annual the well-being of group therapy members. Group
meeting of the Society for Psychotherapy Research. Dynamics: Theory, Research, and Practice, 9(1),
Santiago, Chile. 32–44.

Lambert, M. J., Hansen, N. B., Umpress, V., Marziali, E., Munroe-Blum, H., & McCleary, L.
Lunnen, K., Okiiski, J., & Burlingame, G. M. (1997). The contribution of group cohesion and
(1997). Administration and scoring manual for group alliance to the outcome of group psycho-
the OQ-45.2. Stevenson, MD: American therapy. International Journal of Group
Professional Credentialing Services. Psychotherapy, 47(4), 475–97.
∗ ∗
Levenson, J. S., & Macgowan, M. J. (2004). May, A. M., Duivenvoorden, H. J., Korstjens, I.,
Engagement, denial, and treatment progress Weert, E. V., Hoekstra-Weebers, J. E.,
among sex offenders in group therapy. Sexual Borne, B. V. D., et al. (2008). The effect of group
Abuse: A Journal of Research and Treatment, cohesion on rehabilitation outcome in cancer
16(1), 49–63. survivors. Psycho-Oncology, 17, 917–25.

bu r l i n g a me , mcc l e n d o n , a lo n s o 129
McClendon-Theobald, D., & Burlingame, G. (in Piper, W. E., Marrache, M., Lacroix, R.,
press). Group climate: Construct in search of Richardsen, A. M., & Jones, B. D. (1983).
clarity. In R. Conyne (Ed.), Oxford handbook of Cohesion as a basic bond in groups. Human
group counseling. New York: Oxford University Relations, 36, 93–108.

Press. Ratto, R., & Hurley, J. R. (1995). Outcomes of
McNair, D. M., Lorr, M., & Droppleman, L. F. inpatient group psychotherapy associated with
(1981). Profile of mood states manual. San Diego, dispositional and situational affiliativeness.
CA: Educational and Industrial Testing Group, 19(3), 163–72.

Services. Rice, A. H. (2001). Evaluating brief structured
McRoberts, C., Burlingame, G. M., & Hoag, M. J. group treatment for depression. Research on
(1998). Comparative efficacy of individual and Social Work Practice, 11(1), 53–78.

group psychotherapy: A meta-analytic perspec- Roether, H. A., & Peters, J. J. (1972). Cohesiveness
tive. Group Dynamics, 2(2), 101–117. and hostility in group psychotherapy. American
Moos, R. H. (1986). Group Environment Scale Journal of Psychiatry, 128(8), 1014–1017.
manual (2nd ed.). Palo Alto, CA: Consulting Rosenberg, M. (1965). Society and the adolescent self-
Psychologists Press. image. Princeton, NJ: Princeton University Press.

Moos, R., & Humphrey, B. (1974). Group Rugel, R. P., & Barry, B. (1990). Overcoming
environment scale, Palo Alto, CA: Consulting denial through the group: A test of acceptance
Psychologists Press. theory. Small Group Research, 21(1), 45–58.

Mullen, B., & Copper, C. (1994). The relationship Ryum, T., Hagen, R., Nordahl, H. M., Vogel, P. A.,
between group cohesiveness and performance: & Stiles, T. C. (2009). Perceived group climate as
An integration. Psychological Bulletin, 115, a predictor of long-term outcome in randomized
210–27. controlled trial of cognitive-behavioral group

Norton, P. J., Hayes, S. A., & Springer, J. R. therapy for patients with comorbid psychiatric
(2008). Transdiagnostic cognitive-behavioral disorders. Behavioural and Cognitive Psychotherapy,
group therapy for anxiety: Outcome and process. 37, 497–510.
International Journal of Cognitive Therapy, Santarsiero, L. J., Baker, R. C., & McGee, T. F.
1(3), 266–79. (1995). The effects of cognitive pretraining on

Oei, T. P. S., & Browne, A. (2006). Components of cohesion and self-disclosure in small groups: An
group processes: Have they contributed to the analog study. Journal of Clinical Psychology,
outcome of mood and anxiety disorder patients 51(3), 403-409.
in a group cognitive-behaviour therapy program? Schauenburg, H., Sammet, I., Rabung, S., &
American Journal of Psychotherapy, 60(1), 53–70. Strack, M. (2001). Zur differentiellen Bedeutung

Ogrodniczuk, J., & Piper, W. (2003). The effect of des Gruppenerlebens in der stationaren
group climate on outcome in two forms of short- Psychotherapie depressiver Patienten [On the
term group therapy. Group Dynamics: Theory, differential importance of group experience in
Research, and Practice, 7, 64–76. inpatient psychotherapy of depressive patients].

Ogrodniczuk, J. S., Piper, W. E., & Joyce, A. S. Gruppenpsychotherapie und Gruppendynamik,
(2005). The negative effect of alexithymia on the 37, 349–64.
outcome of group therapy for complicated grief: Selvin, H. C., & Hagstrom, W. O. (1963). The
What role might the therapist play? Comprehensive empirical classification of formal groups.
Psychiatry, 46, 206–213. American Sociological Review, 28, 399–411.

Ogrodniczuk, J. S., Piper, W. E., & Joyce, A. S. Silbergeld, S., Koenig, G. R., Manderscheid, R. W.,
(2006). Treatment compliance among patients Meeker, B. F., & Hornung, C. A. (1975).
with personality disorders receiving group Assessment of environment-therapy systems: The
psychotherapy: What are the roles of inter- Group Atmosphere Scale. Journal of Consulting
personal distress and cohesion? Psychiatry: and Clinical Psychology, 43, 460–69.
Interpersonal and Biological Processes, 69(3), Snijders, J. A., Trijsburg, R. W., De Groot, M. H.,
249–61. & Duivenvoorden, H. J. (2005, March). Group
Palmer, K. D., Baker, R. C., & McGee, T. F. (1997). cohesion, working alliance, and therapeutic
The effects of pretraining on group psychother- interventions as factors predicting outcome in
apy for incest-related issues. International Group group psychotherapy for personality disorders:
Psychotherapy, 47(1), 71–89. First results of a follow-along study (2003–2006).

130 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Presented at AGPA Research Symposium, Questionnaire (GCQ). The Netherlands:
New York). University of Amsterdam/Rotterdam.

Sternburg, S., & Trijburg, W. (2005). The relation- Tschuschke, V., & Dies, R. R. (1994) Intensive
ship between therapeutic interventions and thera- analysis of therapeutic factors and outcome in
peutic outcome. Unpublished manuscript. long-term inpatient groups. International Journal
Stokes, J. (1983). Components of Group Cohesion: of Group Psychotherapy, 44(2), 185–208.

Intermember attraction, instrumental value van Andel, P., Erdman, R. A. M., Karsdorp, P. A.,
and risk taking. Small Group Research, 14 (2), Appels, A., & Trijsburg, R. W. (2003). Group
163-173. cohesion and working alliance: Prediction of
Strauss, B., Burlingame, G., & Bormann, B (2008). treatment outcome in cardiac patients receiving
Using the CORE-R batter in group psycho- cognitive behavioral group psychotherapy.
therapy. Journal of Clinical Psychology, 64(11), Psychotherapy and Psychosomatics, 72, 141–49.

1225–37. Woody, S. R., & Adessky, R. S. (2002). Therapeutic

Taft, C. T., Murphy, C. M., King, D. W., alliance, group cohesion, and homework compli-
Musser, P. H., & DeDeyn, J. M. (2003). Process ance during cognitive-behavioral group treatment
and treatment adherence factors in group of social phobia. Behavior Therapy, 33, 5–27.

cognitive-behavioral therapy for partner violent Wright, T. L., & Duncan, D. (1986). Attraction
men. Journal of Consulting and Clinical Psychology, to group, group cohesiveness, and individual
71(4), 812–20. outcome: A study of training groups. Small

Taube-Schiff, M., Suvak, M. K., Antony, M. M., Group Research, 17(4), 487–92.

Bieling, P. J., & McCabe, R. E. (2007). Group Yalom, I., Houts, P. S., Zimerberg, S. M., &
cohesion in cognitive-behavioral group therapy Rand, K. H. (1967). Prediction of improvement
for social phobia. Behaviour Research and Therapy, in group therapy: An exploratory study. Archives
45, 687–98. of General Psychiatry, 17, 159–68.
Trijsburg, R. W., Bogaerds, H., Letiche, M., Yalom, I. D., & Leszcz, M. (2005). The theory and
Bidzjel, L., & Duivenvoorden, H. J. (2004). practice of group psychotherapy (5th ed.). New York:
De ontwikkeling van de Group Cohesion Basic Books.

bu r l i n g a me , mcc l e n d o n , a lo n s o 131

6 Empathy

Robert Elliott, Arthur C. Bohart, Jeanne C. Watson, and Leslie S. Greenberg

Psychotherapist empathy has had a long (Decety & Ickes, 2009), which we will
and sometimes stormy history in psycho- address briefly in the next section.
therapy. Proposed and codified by Rogers
and his followers in the 1940s and 1950s, Definitions and Measures
it was put forward as the foundation of Defining Empathy
helping skills training popularized in the The first problem with researching empathy
1960s and early 1970s. Claims concerning in psychotherapy is that there is no con-
its universal effectiveness were treated with sensual definition (Batson, 2009; Bohart
skepticism and came under intense scru- & Greenberg, 1997; Duan & Hill, 1996).
tiny by psychotherapy researchers in the Recent neuroscience research on empathy
late 1970s and early 1980s. After that, begins to clarify some of the conceptual
research on empathy went into relative confusion, as a result of the concerted
eclipse, resulting in a dearth of research efforts of researchers using a variety of
between 1975 and 1995 (Duan & Hill, methods ranging from performance tasks,
1996; Watson, 2001). self-report, and neuropsychological assess-
Since the mid-1990s, however, empathy ment to fMRI and transcranial stimulation.
has once again become a topic of scientific Research examining the brain correlates of
interest in developmental and social psy- different component subprocesses of empa-
chology (e.g., Bohart & Greenberg, 1997; thy (Decety & Ickes, 2009) has extended
Ickes, 1997), particularly because empathy the initial discovery of “mirror neurons” in
came to be seen as a major part of “emotional the motor cortex of macaque monkeys
intelligence” (Goleman, 1995). We believe (e.g., Gallese, Fadiga, Fogassi, & Rizzolatti,
the time is ripe for the reexamination and 1996) to a broader range of affective and
rehabilitation of therapist empathy as a key perspective-taking components of empa-
change process in psychotherapy (Bohart thy in humans (Decety & Lamm, 2009).
& Greenberg, 1997). Indeed, the meta- The result of this research has been to
analytic results we will present clearly sup- deepen and clarify our understanding of
port such a conclusion. The most important therapist empathic processes (Watson &
development in the past 10 years, however, Greenberg, 2009), resulting in a growing
is the emergence of active scientific research consensus (e.g., Eisenberg & Eggum, 2009)
on the biological basis of empathy, as part that it consists of three major subprocesses,
of the new field of social neuroscience each with specific sets of neuroanatomical

correlates. First, there is an emotional simu- automatic, bodily-based emotional simula-
lation process that mirrors the emotional tion processes.
elements of the other’s bodily experience Nevertheless, it is easy to see both
with brain activation centering in the processes in Rogers’ (1980) definition of
limbic system (amygdala, insula, anterior empathy:
cingulate cortex) and elsewhere (Decety &
“the therapist’s sensitive ability and
Lamm, 2009; Goubert, Craig, & Buysse,
willingness to understand the client’s
2009). Second, a conceptual, perspective-
thoughts, feelings and struggles from
taking process operates, particularly local-
the client’s point of view. [It is] this ability
ized in medial and ventromedial areas of
to see completely through the client’s
prefrontal cortex as well as the temporal
eyes, to adopt his frame of reference . . .”
cortex (Shamay-Tsoory, 2009). Third, there
(p. 85) . . .“It means entering the private
is an emotion-regulation process that people
perceptual world of the other . . . being
use to reappraise or soothe their personal
sensitive, moment by moment, to the
distress at the other person’s pain or dis-
changing felt meanings which flow in
comfort, allowing them to mobilize com-
this other person . . . It means sensing
passion and helping behavior for the other
meanings of which he or she is scarcely
(probably based in orbitofrontal cortex, as
aware. . . .” (p. 142).
well as in the prefrontal and right inferior
parietal cortex). Defined this way, empathy is a higher-
Interestingly, the two therapeutic order category, under which different sub-
approaches that have most focused on types, aspects, expressions, and modes can
empathy—client-centered therapy and be nested. There are different ways one
psychoanalytic—have emphasized its cog- can put oneself into the shoes of the other:
nitive or perspective-taking (Selman, 1980) emotionally, cognitively, on a moment-
aspects, as well as its feeling aspects. That is, to-moment basis, or by trying to grasp an
they have focused on empathy as connected overall sense of what it is like to be that
knowing (Belenky et al., 1986), under- person. Within these subtypes different
standing the client’s frame of reference or aspects of the client’s experience can become
way of experiencing the world. By some the focus of empathy (Bohart & Greenberg,
accounts, 70% or more of Carl Rogers’ 1997). Similarly, there are many different
responses were to meaning rather than to ways of expressing empathy, including
feeling, despite the fact that his mode of empathic reflections, empathic questions,
responding is typically called “reflection of experience-near interpretations, empathic
feeling” (Brodley & Brody, 1990; Hayes & conjectures, as well as the responsive use of
Goldfried, 1996; Tausch, 1988). However, other therapeutic procedures. Accordingly,
understanding clients’ frames of reference empathy is best understood as a complex
does include understanding their affective construct consisting of a variety of different
experiences. In addition, empathy and acts used in different ways.
sympathy have typically been sharply dif- We distinguish between three main
ferentiated, with therapists such as Rogers modes of therapeutic empathy: empathic
disdaining sympathy but prizing empathy rapport, communicative attunement, and
(Shlien, 1997). In affective neuroscience person empathy. First, for some thera-
terms, this means that therapists in this pists empathy is primarily the establish-
tradition have often emphasized conscious ment of empathic rapport and support. The
perspective-taking processes over the more therapist exhibits a compassionate attitude

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 133

toward the client and tries to demonstrate measures have been developed. Within
that he or she understands the client’s expe- psychotherapy, the measures of therapist
rience, often in order to set the context for empathy fall into four categories: empathy
effective treatment. A second mode of rated by nonparticipant raters; client-rated
empathy consists of an active, ongoing effort empathy; therapists rating their own empa-
to stay attuned on a moment-to-moment thy; and empathic accuracy (congruence
basis with the client’s communications and between therapist and client perceptions of
unfolding experience. Client-centered and the client).
experiential therapists are most likely to Observer-Rated Empathy. Some of the
emphasize this form of empathy. The thera- earliest observer measures of empathy were
pist’s attunement may be expressed in many those of Truax and Carkhuff (1967) and
different ways, but most likely in empathic Carkhuff and Berenson (1967). These scales
responses. The third mode, person empathy asked raters to decide if the content of
(Elliott, Watson, Goldman, & Greenberg, the therapist’s response detracts from the
2003) or experience-near understanding of client’s response, is interchangeable with it,
the client’s world, consists of a sustained or adds to or carries it forward. Typically,
effort to understand the kinds of experi- trained raters listened to 2–5 minute sam-
ences the client has had, both historically ples from session tapes. Samples are usually
and presently, that form the background of drawn from the beginning, middle, and/or
the client’s current experiencing. The ques- the end of therapy. Scales such as these do
tion is: How have the client’s experiences not adequately reflect the client-centered
led him or her to see/feel/think and act as conception of empathy as an attitude
he or she does? This is the type of empathic because they focus narrowly on a particular
understanding emphasized by psychody- kind of response, often empathic reflec-
namic therapists. However, empathic rap- tions. Furthermore, the equation of a par-
port, communicative attunement, and person ticular response with empathy has also
empathy are not mutually exclusive, and the made these scales less appropriate for mea-
differences are a matter of emphasis. suring empathy in approaches other than
Many other definitions for empathy have client centered (Lambert, De Julio, & Stein,
been advanced: as a trait or response skill 1978).
(Egan, 1982; Truax & Carkhuff, 1967), as More recent observer empathy measures
an identification process of “becoming” the are based on broader understandings of
experience of the client (Mahrer, 1997), forms of empathic responding. Elliott and
and as a hermeneutic interpretive process colleagues’ (1982) measure breaks empathy
(Watson, 2001). Perhaps the most practical down into component elements and has
conception, and one that we will draw on shown good psychometric properties, but
in our meta-analysis, is Barrett-Lennard’s it has not been widely used. Watson and
(1981) operational definition of empathy Prosser (2002) developed a promising new
in terms of three different perspectives: that observer-rated measure of empathy that
of the therapist (empathic resonance), the assesses therapists’ verbal and nonverbal
observer (expressed empathy), and the behavior and shows convergent validity
client (received empathy). with client ratings on the Barrett-Lennard
Relationship Inventory.
Measuring Empathy In addition, the therapist’s general empa-
Reflecting the complex, multidimensional thy can also be rated by others who know or
nature of empathy, a confusing welter of have supervised the therapist. For instance,

134 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
therapists’ empathic capacities can be rated comparing these ratings to how clients
by their supervisors (Gelso, Latts, Gomez, actually rated themselves. For instance, one
& Fassinger, 2002). For purposes of our study compared how therapists rated clients
meta-analysis, we lumped together all on Kelly’s REP grid with how clients rated
observer perspective measures of empathy. themselves (Landfield, 1971). The measure
Client Ratings. The most widely used of empathy is the degree of congruence
client-rated measure of empathy is the between therapist and client ratings. This
empathy scale of the Barrett-Lennard can be referred to as predictive empathy,
Relationship Inventory (BLRI). Other because the therapist is trying to predict
client rating measures have been developed how clients will rate themselves. This is
(e.g., Hamilton, 2000; Lorr, 1965; Persons closer to a measure of the therapist’s ability
& Burns, 1985; Truax & Carkhuff, 1967). to form a global understanding of what it
Rogers (1957) hypothesized that clients’ is like to be the client (person empathy)
perceptions of therapists’ facilitative con- than it is to a process measure of ongoing
ditions (positive regard, empathy, and communicative attunement.
congruence) predict therapeutic outcome. Recent work on empathic accuracy,
Accordingly, the BLRI, which measures however, does provide a predictive mea-
clients’ perceptions, is an operational defi- sure of communicative attunement (Ickes,
nition of Rogers’ hypothesis. In several 1997, 2003). This line of research typically
earlier reviews, including our meta-analysis employs a tape-assisted recall procedure in
in the previous edition of this book, client- which therapists or observers’ moment-to-
perceived empathy predicted outcome better moment empathy is measured by compar-
than observer- or therapist-rated empathy ing their perceptions of client experiences
(Barrett-Lennard, 1981; Gurman, 1977; to clients’ reports of those experiences.
Bohart, Elliott, Greenberg, & Watson, Unfortunately, no process–outcome studies
2002; Orlinsky, Grawe, & Parks, 1994; using this promising but time-consuming
Orlinsky & Howard, 1978, 1987). method have yet been carried out.
Therapist Ratings. Therapist empathy Correlations among Different Empathy
self-rating scales are not so common, but Measures. Intercorrelations of different
the BLRI does have one. Earlier reviews empathy measures have generally been
(Barrett-Lennard, 1981; Gurman, 1977) weak. Low correlations have been reported
found that therapist-rated empathy nei- between cognitive and affective measures
ther predicted outcome nor correlated (Gladstein et al., 1987) and between pre-
with client-rated or observer-rated empa- dictive measures and the BLRI (Kurtz &
thy. However, we previously found that Grummon, 1972). Other research has
therapist-rated empathy did predict out- found that tape-rated measures correlate
come, but at a lower level than client or only moderately with client-perceived
observer ratings (Bohart et al., 2002). empathy (Gurman, 1977). These weak
Empathic Accuracy. Several studies use correlations are not surprising when one
measures of therapist–client perceptual con- considers what the different instruments
gruence, commonly referred to as “empathic are supposed to be measuring. Trying to
accuracy” (Ickes, 1997, 2003). These typi- predict how a client will fill out a symptom
cally consist of therapists rating clients check list is very different from responding
as they think the clients would rate them- sensitively and tentatively in a way that
selves on various measures, such as person- demonstrates subtle understanding of what
ality scales or lists of symptoms, and then the client is trying to communicate, while

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 135

checking and adjusting one’s emerging Clinical Example
understanding with that of the client. Mark presented to psychotherapy com-
Similarly, client ratings of therapist under- plaining of pervasive anxiety. He was a
standing may be based on many other 30-year-old unmarried man who had been
things than the therapists’ particular skill struggling since his early 20s to break into
in empathic reflection. Accordingly, we the movie business. When he entered ther-
should not expect different measures of this apy, he was working as a waiter. He came
complex construct to correlate (Gladstein from a traditional family, living in the
et al., 1987). southern United States. His brothers and
Confounding between Empathy and Other sisters all had successful careers and were
Relationship Variables. A related concern is married, with children. His parents were
the distinctiveness of empathy from other constantly pestering him about his not
relationship constructs. One early review being married and not having a stable
of more than 20 studies primarily using career. His anxiety attacks had begun a few
the BLRI found that, on average, empathy weeks after a visit home for the Christmas
correlated 0.62 with congruence and 0.53 holidays. When Mark came to his first
with positive regard, and 0.28 with uncon- appointment, he was clearly agitated. He
ditionality (Gurman, 1977). Factor analysis had previously called and had sounded
of scale scores found that one global factor desperate over the phone. The therapist
typically emerged, with empathy loading initially was concerned that Mark might be
on it along with congruence and posi- in a state of crisis.
tive regard (Gurman, 1977). Others have The therapist’s orientation was integra-
reported that the empathy scale loaded tive experiential/humanistic, based in the
0.93 on a global BLRI factor, with Positive principles of person-centered therapy. The
Regard loading 0.87 and Congruence load- therapist tried to understand the client’s
ing 0.92 (Blatt et al., 1996). Such results point of view actively and empathically
suggest that clients’ perceptions of empathy and to share that understanding, using her
are not clearly differentiated from their per- attunement to the client’s experience to
ceptions of other relationship factors. identify effective interventions, and to stay
On the other hand, reviews of several responsively attuned so that therapeutic
factor analytic studies where, instead of procedures could be adjusted to maximize
using scale scores, specific items were used learning. The following are two examples
have found empathy emerging as a separate of the therapist’s utilization of empathic
factor (Gurman, 1977). In addition, empa- responding during the first session:
thy tends to correlate more highly with
the bond component of the therapeutic C1: I’m really in a panic (anxious,
alliance than with the task and goal compo- looking plaintively at the therapist).
nents (Horvath & Greenberg, 1986). I feel anxious all the time. Sometimes
Thus, there is evidence both for and it seems so bad I really worry that
against the hypothesis that the Rogerian I’m on the verge of a psychotic break.
triad of empathy, unconditional positive I’m actually afraid of completely
regard, and congruence are separate and falling apart. Nothing like this has
distinct variables. We view empathy as a ever happened to me before. I always
relationship component that is both con- felt in charge of myself before, but
ceptually distinct and part of a higher-order now I can’t seem to get any control
relationship construct. over myself at all.

136 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
T1: So a real sense of vulnerability— During the first few sessions, the client
kind of like you don’t even know had repeatedly expressed the suspicion that
yourself anymore. something about his early relationships
C2: Yes! That’s it. I don’t know myself with his parents played an important role
anymore. I feel totally lost. The in his current problems. Initially, the thera-
anxiety feels like a big cloud that just pist had not taken this too seriously, since
takes me over, and I can’t even find progress was apparently being made through
myself in it anymore. I don’t even the collaborative use of other procedures.
know what I want, what I trust . . . Because the therapist was not psychoge-
I’m lost. netic and past oriented, she had not tuned
T2: Totally lost, like, “Where did Mark into this. The therapist’s lack of person
go? I can’t find myself anymore.” empathy (i.e., grasping of how figural this
C3: No, I can’t (sadly, and was for the client within the client’s frame
thoughtfully). of reference) for the larger meaning of the
client’s interest in this topic had effectively
The dialogue continued like this, and shut off this avenue of exploration.
soon the therapist’s empathic recognition Eventually, the therapist listened,
provided the client with a sense of being responded in an invitational way to the
understood. This fostered a sense of safety, client, and the client began to explore his
and gradually the client moved from agita- childhood. This illustrates how empathy
tion into reflective sadness. The client then not only gives permission, but also provides
began to reflect on his experience in a more active support for exploration. It also illus-
productive, exploratory manner. He talked trates how sensitive empathic understand-
about the basic conflict in his life: over ing of the client’s way of seeing the problem
whether to continue to pursue an acting is sometimes crucial for therapeutic prog-
career or to find a “real job” and life partner, ress (Hubble, Duncan, & Miller, 1999).
given that he was now 30 and had shown no This led to a breakthrough moment. In
signs of making a breakthrough in acting. reviewing his childhood, Mark became
Later, the client role-played a dialogue emotionally aware of how neglected he
between two sides of himself. One side, his had felt as a child by his high-achieving
critic or “should” side, said that he should parents, who were not mean and cruel, but
get a stable job and get married and criti- who were not themselves highly empathic.
cized him for not being married. The other As a child, the client had always been
side was the “want” side—or in this case, unusually interested in fantasy activities,
the “don’t want” side—which said “I don’t and was a rather “inner” person, in contrast
want to live an ordinary life; I want to live to his siblings, who were more conventional
a creative life.” This side came out in the and high-achievers at school. The parents
form of defensive rebellion. Empathic had not known what to make of their
sensitivity was used to help the therapist unique child and were unable to respond in
tune into the client’s point of view and an empathic and supportive way to his
to focus the client’s exploratory activities emerging uniqueness.
during the role-play. What emerged from The result was that he had had to adopt
this role-play was that there was a longing a defensive “I have a right to be different”
for a “normal” lifestyle underlying Mark’s attitude. He was rarely able to genuinely
defensive rebellion, in conflict with a desire consider whether he wanted to be conven-
to do something creative. tional or not. Underlying this was a longing

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 137

for conventionality. Accessing this in the “counseling,” or “counselling”. Additionally,
context of his family life helped him accept we consulted the tables of contents of rele-
that he was different and to mourn the fact vant journals such as: Psychotherapy, Person-
that he was not conventional (and, in effect, Centered Journal, Psychotherapy Research,
mourn that he might never be what his Journal of Counseling Psychology, and Person-
family wanted him to be). Over the course Centered and Experiential Psychotherapies.
of this work, Mark’s anxiety decreased.
Eventually he made a decision to continue Inclusion Criteria
to pursue an acting career, for a while at Our inclusion criteria were as follows:
least; and his crisis abated. (a) a specific measure of empathy was used,
(b) empathy was related to some measure
Meta-Analytic Review of therapy outcome, (c) the client sample
In this section we report the results of an involved genuine clinical problems, (d) the
original meta-analysis conducted on available average number of sessions was three or
research relating empathy to psychother- more, (e) the study was available in English,
apy outcome. We addressed the following (f ) the study included at least five clients,
questions: (a) What is the association (g) the study was available in published
between therapist empathy and client out- form, and (h) the study contained suffi-
come? (b) Do different forms of psycho- cient information to calculate a weighted
therapy yield different levels of association effect size.
between empathy and outcome? (c) Does
the type of empathy measure predict the Characteristics of the Studies
level of association between empathy To examine variables that might moderate
and outcome? (d) What other study and the empathy–outcome association, we eval-
sample characteristics predict an association uated the studies on a wide range of sample
between empathy and outcome (i.e., sample and methodological features. For measures
size, treatment setting, therapy format and of outcome, we included a study as long
length, level of client severity, therapist as there was some assessment of the effects
experience, type of outcome measure, unit of therapy, even if only at the session level
of process)? (immediate outcome). For example, we
included abstinence from drinking (Miller
Search Strategy et al., 1980), level of depression (Burns &
Articles were culled from previous reviews Nolen-Hoeksema, 1992), MMPI scores
(Beutler, Crago, & Arizmendi, 1986; (Kiesler et al., 1967), client satisfaction
Gurman, 1977; Lambert, DeJulio & Stein, (Lorr, 1965), supervisors’ ratings of client
1978; Mitchell, Bozarth, & Krauft, 1977; improvement (Gelso et al., 2002), client
Orlinsky & Howard, 1986; Orlinsky, and therapist posttherapy ratings of amount
Grawe, & Parks, 1994; Parloff et al., 1978; of change (Hamilton, 2000), and postses-
Truax and Mitchell, 1971; N. Watson, sion ratings of progress (Orlinsky & Howard,
1984). We also searched PsycINFO and 1967). There is some conceptual overlap
PsycLIT forward from 1992 (2 years between feeling understood and client
before the publication of the last major satisfaction, but this one outcome measure
review of empathy research in Orlinsky represented only 6% of effects; we subse-
et al., 1994), using the search terms, “empa- quently examined type of outcome mea-
thy” or “empathic” and “psychotherapy,” sure as a moderator variable. The resulting

138 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
sample consisted of 224 separate tests of Coding Procedure
the empathy–outcome association, aggre- The following variables were coded: therapy
gated into 59 different samples of clients format (individual or group); theoretical
(from 57 studies) and encompassing a total orientation; experience level of therapists;
of 3,599 clients. Table 6.1 summarizes rel- treatment setting (inpatient, outpatient);
evant study characteristics. number of sessions (typically the mean); type
of problems (mixed neurotic, depression,
Estimation of Effect Size anxiety, severe problems such as psychosis);
For effect sizes, we used Pearson correla- source of outcome measure (therapist rating,
tions if available. Our strategy was to extract client rating, objective, and other mea-
all possible effects. Therefore, we used the sures); when outcome was measured (e.g.,
following conventions (extensions of those postsession, posttherapy, followup); type
used in Smith, Glass, & Miller, 1980) to of outcome measured (symptom change,
estimate r : First, if we had a significance improvement, global); source of empathy
level, we converted it to r. If the result was measure (objective ratings, therapist, client,
nonsignificant, but we had enough infor- therapist/client congruence, trait measure);
mation to calculate a t and then convert, and unit of measure (2–5 minute samples,
we did so. If we had no other information session, therapy to date).
than that the effect was nonsignificant, we We conducted two sets of analyses: by
set r at 0. If the authors indicated a “nonsig- effects and by studies. First, we analyzed
nificant trend” but did not report a correla- the 224 separate effects in order to examine
tion (for instance, a key study, Kiesler et al., the impact of perspective of empathy mea-
1967, indicated several trends on MMPI surement and type of outcome. Second,
scales), we estimated the trend by assigning study-level analyses used averaged indi-
an ES of half the size of a significant r. vidual effects within client samples using

Table 6.1 Study Characteristics

Parametric characteristics: M SD Range
Sample size: 61 59.6 6–320
Length of therapy (sessions) 24 42.4 3–228
Effects per study 3.8 5.7 1–42
Categorical characteristics: Modal categories %
Time period (range: 1961–2008) Before 1980 49
Theoretical orientation Mixed, eclectic, or unknown 40
Modality Individual 74
Client presenting problem Mixed neurotic (mixed anxiety/depression) 40
Therapist experience level Recent PhD or M.D. 36
Outcome assessment time point Posttreatment 60
Empathy perspective Client (mostly Barrett-Lennard) 39
(Observer, mostly Truax-Carkhuff: 34%)
Empathy measurement unit Therapy to date 60

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 139

Fisher r-to-Z conversions to correct for dis- outcome. This effect size is on the same order
tributional biases before further analysis, of magnitude as, or slightly larger than, pre-
thus avoiding problems of nonindepen- vious analyses of the relationship between
dence and eliminating bias due to variable the alliance in individual therapy and treat-
numbers of effects reported in different ment outcome (i.e., Horvath, Flückiger, &
studies (Lipsey & Wilson, 2001; e.g., one Symonds, this volume, Chapter 2: 0.275;
study, Kurtz & Grummon, 1972, contri- Martin, Garske, & Davis, 2000: 0.22).
buted 42 effects). For analyses across stud- Overall, empathy typically accounts for
ies, we weighted studies by inverse error more outcome variance than do specific
and analyzed for heterogeneity of effects treatment methods (compare Wampold’s,
using Cochrane’s Q (following the Hunter- 2001, estimate of 1% to 8% for interven-
Schmidt method, using the program in tion effects).
Diener, Hilsenroth, & Weinberger, 2009), However, the 0.30 figure conceals sta-
and also I 2, an estimate of the proportion of tistically significant variability in effects,
variation due to true variability as opposed as indicated by a study-level Cochrane’s Q
to random error (Higgins, Thompson, Deeks, of 205.8 (p < 0.001); in addition, I 2 was
& Altman, 2003). Finally, where necessary 67%, considered to be a large value. This
in the correlational analyses of moderator means that a further examination of possi-
variables to correct for nonindependence, ble moderators of the empathy–outcome
we used effects weighted by the inverse of association is not only justified but is in fact
number of analyses per study. necessary (Lipsey & Wilson, 2001).

Results Moderators and Mediators

The single best summary value, as shown in We divide this section on moderators
Table 6.2, is the study-level, weighted r and mediators into two parts: meta-analytic
of 0.30, a medium effect size. Average analyses of moderator variables and thera-
effects were 0.22 for analyses of the 224 pist-mediating factors.
nonindependent separate effects, probably
an underestimate due to smaller effects Meta-Analytic Moderator Analyses
found in one study (Kurtz & Grummon, The significant Q and large I 2 statistics
1972). These values were very similar to our point to the existence of important moder-
previous review (Bohart et al., 2002) and ator variables or sources of heterogeneity
mean that in general empathy accounts but do not specify what those are. We began
for about 9% of the variance in therapy our search by testing the hypothesis that

Table 6.2 Empathy–Outcome Correlations: Overall Summary Statistics

Effect level (N = 224) Study level (N = 59)

Weighted mean r 0.22∗ 0.33 0.30∗ 0.13

Cochrane’s Q 646.22∗ 174.65∗
I2 65.49 66.79

p < 0.001.
Note: Fisher’s r-to-z transformation used to calculate means and SDs. Weighted rs use inverse variance (i.e., n−3) as weights and are tested
against mean r = 0 following the Hunter-Schmidt method using Deiner’s (2010) program.

140 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
different empathy–outcome correlations than observer-rated measures (0.25; n = 27)
might be obtained for different theoretical and therapist measures (0.20; n = 11); each
orientations. For example, one might expect of these mean effects was significantly
the association to be larger in those thera- greater than zero (p < 0.001). In contrast,
pies for which empathy is held to be a key empathy accuracy measures were unrelated
change process, such as person-centered to outcome (0.08; n = 5, ns). Although the
therapies. However, our analyses, summa- overall Q value for between-group hetero-
rized in Table 6.3, turned up little evi- geneity was not significant, comparison of
dence of such a trend, but significant, large confidence intervals indicated that client-
amounts of nonchance heterogeneity within perceived empathy significantly predicted
the CBT and Other/Unspecified therapy outcome better than accuracy measures
samples. This finding contrasts with our (p < 0.05). A word of caution: All perspec-
previous meta-analysis (Bohart et al., 2002), tives except empathic accuracy are character-
where we found tantalizing evidence that ized by large (>50%), statistically significant
empathy might be more important to amounts of nonchance heterogeneity.
outcome in cognitive-behavioral therapies Clarification of the source of this heteroge-
than in others. However, our present analy- neity awaits further research; however, for
sis failed to confirm that conjecture but now it seems fair to say that clients’ feelings
points to important sources of variability of being understood and observer ratings
that need to be explored (and to a lesser extent, therapist impres-
In Table 6.4 we chart relations between sions) appear to carry significant weight as
specific types of empathy measures and far as outcome goes, but that empathic
outcome, using effect-level analyses aggre- accuracy measures do not, in spite of their
gated within studies (n = 82). As we intuitive appeal.
expected, and has been noted by previous Finally, in Table 6.5, we examine several
reviewers (e.g., Barrett-Lennard, 1981; other variables that might account for
Parloff, Waskow & Wolfe, 1978), the per- some of the heterogeneity of the effect sizes:
spective of the empathy rater made a differ- year of publication, sample size, outpa-
ence for empathy–outcome correlations. tient versus inpatient treatment, treatment
Specifically, client measures predicted out- format (individual vs. group), length of
come the best (mean corrected r = 0.32; therapy, client severity, therapist experi-
n = 38), slightly but not significantly better ence level, globalness of outcome measures

Table 6.3 Mean Effects across Theoretical Orientation

Theoretical orientation n Mean weighted r Within group Q I2
Experiential/ humanistic 8 0.26∗∗ 7.68 8.91
Cognitive-behavioral 10 0.31∗∗ 24.55∗ 63.34
Psychodynamic 4 0.19∗∗ 2.01 0
Other/unspecified 37 0.31∗∗ 138.01∗∗ 74.64
Between groups Q 2.39 (df = 3, 55, ns)

p < 0.01; ∗∗p < 0.001.
Note: Mean correlations calculated using Fisher’s z-scores. Significance tests for mean correlations are against the null hypothesis of mean r = 0.
Q tests for heterogeneity are evaluated as a chi-square test, using Diener et al.’s (2009) program and the Hunter-Schmidt method. Within-
groups Q is analogous to a one-way ANOVA with study samples as levels; between-groups Q calculated as difference between total sample Q
and within-group Q, following Lipsey and Wilson (2001).

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 141

Table 6.4 Mean Effects across Empathy Measurement Perspectives
Measurement perspective n Mean weighted r Within group Q I2
Observer 27 0.25∗∗ 93.14∗∗ 72.09
Client 38 ∗∗ ∗∗ 69.00
0.32 119.35
Therapist 11 0.20∗∗ 21.05∗ 52.50
Empathic accuracy 5 0.08 5.91 32.35
Total 82 0.27 258.08∗∗ 68.61
Between-groups Q 2.39 (df = 3, 78, ns)
∗ ∗∗
p < 0.05; p < 0.001.
Note: See note for Table 6.3.

(individualized to satisfaction ratings), and possibility that empathy is slightly more

size of empathy unit (5 min segment to predictive of positive outcome in group
whole therapy). Using ordinary (that is, therapy, with more severely distressed cli-
unweighted correlations), none of these ents, in more recent studies, and with more
were statistically significant. On the other global outcome measures (i.e., satisfaction
hand, analyses using weighting for inverse ratings, which begin to overlap conceptu-
error (i.e., sample size minus 3) were sig- ally with empathy).
nificant for all variables except outcome On the other hand, it may be that the
globality and size of empathy unit; how- empathy relationship is slightly less predic-
ever, these suffer from nonindependence tive of positive outcome in inpatient set-
within studies and will require a substan- tings, and with more experienced therapists
tially larger set of studies or more sophisti- (study level mean r = −0.19; effect level =
cated, multilevel meta-analytic methods to −0.29); the latter is the largest of this set of
verify. Briefly, these analyses point to the correlations and is consistent with our 2002

Table 6.5 Correlations between Empathy–Outcome Effect Size and Selected Moderator Variables
Predictor Unweighted Weighted
r n r n
Year of publication 0.14 59 0.12∗ 3422
No. of clients in study 0.06 59 0.15∗ 3422
Setting (1 = outpatient; 2 = inpatient) −0.13 58 −0.08∗ 3305
Format (1 = individual; 2 = group) 0.12 54 0.15∗ 2807
Length of therapy (in sessions) 0.04 41 −0.08∗ 2074
Client severity (3-point scale) 0.10 41 ∗ 2320
Therapist experience level (6-point scale) −0.19 51 −0.29∗ 2820
Outcome globality (6-point scale: individualized to satisfaction ratings) 0.17 59 0.00 3360
Size of empathy unit (4-point scale) −0.06 59 -0.02 3443

p < 0.001.
Note: Weighted analyses used inverse error (i.e., degrees of freedom) but are not corrected for nonindependence of participants within studies;
analyses of outcome globality and size of empathy unit analyses were also inverse weighted by number of effects per study to correct for
nonindependence of effects within studies.

142 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
meta-analysis. As we previously speculated, (Duan & Hill, 1996; Watson, 2001). Other
there are at least two possible reasons for research has shown that responses that are
this: To begin with, inexperienced thera- just ahead of the client seem to be more
pists may vary more in empathy, while effective than responses that are either at
smaller correlations for experienced thera- the same level as the client, or at a more
pists may reflect a restriction of range or global level (Sachse, 1990a, 1990b; Tallman
ceiling effect. Alternatively, experienced et al., 1994; Truax & Carkhuff, 1967). And
therapists may have developed additional a qualitative study of clients’ experience
skills such as effective problem solving, so of empathy, interrupting, failing to main-
that clients are more likely to forgive tain eye contact, and dismissing the client’s
empathic misattunements. position while imposing the therapist’s own
position were all perceived as unempathic
Therapist-Mediating Factors (Myers, 2000). Conversely, being nonjudg-
As noted earlier, affective neuroscience mental, attentive, open to discussing any
researchers have proposed that empathy topic, and paying attention to details were
involves three interlinked skills or processes: perceived as empathic.
affective simulation, perspective taking, and
regulation of one’s own emotions (Decety Client Contributions
& Jackson, 2004). Supporting this, research Clinical and research experience suggest that
has found a relationship between various the amount of therapist empathy varies as a
measures of cognitive complexity, such as function of the client. Early studies (Kiesler
those of perspective taking or abstract abil- et al.,1967), for example, found that levels
ity, and empathy in both developmental of empathy were higher with clients who
psychology and in psychotherapy (Eisenberg had less pathology, who were brighter, but
& Fabes, 1990; Henschel & Bohart, 1981; yet who were lower in self-esteem. Therefore,
Watson, 2001). With respect to affective the client him or herself almost certainly
simulation and emotion regulation, thera- influences therapist empathy. As Barrett-
pists who were open to conflictual, coun- Lennard (1981) pointed out, the client’s
tertransferential feelings were perceived revealing of their experiencing is an essential
as more empathic by clients (Peabody & link in the cycle of empathy. Clients who are
Gelso, 1982). more open to and able to communicate
The degree of similarity between therapist their inner experiencing will be easier to
and client (Duan & Hill, 1996; Gladstein empathize with. Empathy truly appears to
& Associates, 1987; Watson, 2001) also be a mutual process of shared communica-
influences the level of empathy. Similarity tive attunement (Orlinsky et al., 1994).
and familiarity between the target of empa- On the other hand, not all clients respond
thy and the empathizer have been found favorably to explicit empathic expressions. In
to be important modulators of empathy their review, Beutler, Crago, and Arizmendi
in neuroscientific studies of mirror neu- (1986, p. 279) cite evidence that suggests
rons (Watson & Greenberg, 2009). Another that “patients who are highly sensitive,
important factor is therapist nonlinguistic suspicious, poorly motivated, and reac-
and paralinguistic behavior. This encom- tive against authority perform relatively
passes therapists’ posture, vocal quality, abil- poorly with therapists who are particu-
ity to encourage exploration using emotion larly empathic, involved, and accepting.”
words, and the relative infrequency of talking Another study (Mohr & Woodhouse, 2000)
too much, giving advice, and interrupting found that some clients prefer businesslike

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 143

rather than warm, empathic therapists. It is with the research on empathy. In addition
worth noting, however, that when thera- to the well-known difficulty of inferring
pists are truly empathic they attune to their causality from correlational data, these
clients’ needs and accordingly adjust how entail: (a) the questionable validity of some
and how much they express empathy. outcome measures (e.g., client satisfaction);
More broadly, Duan and Hill (1996) (b) lack of appropriate, sensitive outcome
speculated that different types of empathy measures; (c) restricted range of predic-
may be hindering or helpful to clients at tor and criterion variables; (d) confounds
different times. Hill and her colleagues among variations in time of assessment,
(Hill et al., 1992; Thompson & Hill, 1991) experience of raters, and sampling meth-
found that when clients had negative ods; (e) reliance on obsolete diagnostic
in-session reactions to their therapists, the categories; and (f ) incomplete reporting
therapist’s awareness or understanding of of methods and results. In fact, these and
the reaction “led to interventions that were other problems are not restricted to empa-
perceived as less helpful than when the thy research but are common to all process–
awareness was absent” (p. 269). In such outcome research (Elliott, 2010).
relational ruptures, it is probably useful for The restricted range of predictor and
therapist empathy to be accompanied and criterion variables is particularly a problem.
deepened by genuine warmth, openness, In the Mitchell, Truax, Bozarth, and Krauft
and concern for the clients’ feelings, rather (1973) study, for instance, most of the
than defending oneself and blaming the therapists scored below the minimum con-
client (also see Safran, Muran, & Eubanks- sidered to be effective, and outcome was
Carter, this volume, Chapter 11). only modest to moderate in the study. It is
Keeping in mind the notion of empathy not surprising that no significant correla-
as not only getting inside the skin of the tions were found. Furthermore, in a few
client, but getting inside the skin of the cases, results were reported as either sig-
relationship (O’Hara, 1984), it may be that nificant in the positive direction or nonsig-
in some cases the therapist is more empathic nificant, possibly disguising weak negative
by not expressing empathy. Martin (2000, effects. This is particularly a problem for
pp. 184–185) notes: “Think of the insensi- calculating effect sizes based on limited
tive irony of a therapist who says, ‘I sense information, thus introducing error into
the sadness you want to hide. It seems like the process.
you don’t want to be alone right now The key question of whether empathy is
but you also don’t want somebody talking causally related to therapeutic outcome—as
to you about your sadness . . .’ ” This opposed to being merely a correlate of it—
response might technically seem empathic, cannot be answered definitively from our
but in fact at a higher level, it is unem- meta-analysis. This is the central limitation
pathic, controlling, and intrusive, because of the process–outcome research reviewed
it violates the client’s need for interpersonal here. However, data from several studies
distance. Variations among clients in desire shed light on the question. First, Burns and
for and receptivity to different expressions Nolen-Hoeksema (1992) and Cramer and
of empathy need further research. Takens (1992) have used causal modeling
(structural equation modeling, path analysis)
Limitations of the Research to explore the relationship between empa-
Many reviewers (e.g., Watson, 2001; thy and outcome. Second, in another study
Patterson, 1984) have discussed problems (Miller et al., 1980), ratings of therapist

144 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
empathy were made by supervisors before caring, transparent, empathic stance in all
and independent of knowing about out- professional contacts, it is therefore both
come data. Yet empathy showed a strong impractical and unethical to randomize
(r = 0.82) relationship to outcome in a clients to demonstrably empathic versus
cognitive-behavioral program for drinking. unempathic therapists. In such cases, meta-
Third, Anderson (1999) measured thera- analyses can provide a valid alternative
pists’ facilitative interpersonal skills, includ- to randomized clinical trials (Berman &
ing accurate empathy, before therapy, by Parker, 2002), providing that the identifi-
having them respond to videotapes of cation and analysis of observational studies
clients who presented in difficult interper- has been done carefully and systematically.
sonal ways. Anderson found statistically
significant relationships between this prior Therapeutic Practices
measure of therapist interpersonal skills The most consistent and robust evidence is
and client outcome in subsequent psycho- that clients’ perceptions of feeling under-
therapy, a finding recently replicated with a stood by their therapists relate favorably to
larger, practice-based sample of therapists outcome. As we have shown, empathy is
and clients (Anderson, Ogles, Patterson, a medium-sized predictor of outcome in
Lambert, & Vermeersch, 2009). psychotherapy. It also appears to be a gen-
On the other hand, Burns and Nolen- eral predictor across theoretical orientations,
Hoeksema (1992) note that structural treatment formats, and client severity levels.
equation modeling cannot definitely show This repeated finding, in both dozens of
causality but only explore and elaborate individual studies and now in multiple
particular causal models. Miller et al. (1980) meta-analyses, leads to a series of clinical
had supervisors rate supervisees’ levels of recommendations.
empathy, but it is possible that these ratings
were influenced by supervisees’ reports of • It is important for psychotherapists
how well therapy was going with the cli- to make efforts to understand their clients,
ents. In Anderson et al.’s (2009) study in and to demonstrate this understanding
which empathy was measured indepen- through responses that address the needs
dently of therapy, empathy is confounded of the client as the client perceives them
with other facilitative interpersonal skills. on an ongoing basis. The empathic
Even though empathy is the predominant therapist’s primary task is to understand
process in client-centered and related ther- experiences rather than words.
apy, it is not the only process. • Empathic therapists do not parrot
The evidence we have presented is clearly clients’ words back or reflect only the
compatible with a causal model implicat- content of those words; instead, they
ing therapist empathy as a mediating pro- understand overall goals as well as
cess leading to client change. It is true that moment-to-moment experiences,
correlational studies can only probe into or both explicit and implicit. Empathy
lend support for or against causal models entails capturing the nuances and
of therapeutic change. As is the case for implications of what people say, and
much of the behavioral sciences, establishing reflecting this back to them for their
conclusive evidence for particular hypoth- consideration.
esized causal processes is notoriously difficult • Empathic responses follow the
and may ultimately prove elusive. Insofar “moving point” of the focus of the
as codes of professional ethics stipulate a client’s concerns as therapy progresses.

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 145

• Research has identified a range of your boat from being sucked in
useful types of empathic responses, several or capsizing.
of which we illustrate here with a running
example. Empathic understanding responses Empathic explorations are attempts by
convey understanding of client experience. therapists to get at that which is implicit in
For example: clients’ narratives and focus on information
that has been in the background but not
Client: I have been trying to push yet articulated:
things away, but every time I sit
down to do something it is like C: I keep responding to him, like it’s
I forget what I am doing. against what I want to do.
Therapist: Somehow you are not in a T: Somehow you can’t let go. It is just
space to work, it’s hard for you to so hard to walk away.
concentrate. • Empathic therapists assist clients to
Empathic affirmations are attempts by the symbolize their experience in words and
therapist to validate the client’s perspective: track their emotional responses, so that
clients can deepen their experience and
C: And my cat is still lost, so we have reflexively examine their feelings, values,
been staying up at night in case he and goals. To this end they need to attend
returns, so last night was another to that which is not said, or that which is
night without sleep . . . and work has at the periphery of awareness as well as
been so busy and I have been so tired that which is said and is in focal awareness
and P needs my attention. I have (Watson, 2001).
been going around in circles and, oh, • Empathy entails individualizing
everything is just a big mess, you responses to particular patients. For
know? example, certain fragile clients may find
T: Yeah, really hard, being pulled in a the usual expressions of empathy too
million different directions and there intrusive, while hostile clients may find
hasn’t been time for you, no wonder empathy too directive; still other clients
it feels like things are a mess. may find an empathic focus on feelings
Empathic evocations try to bring the cli- too foreign (Kennedy-Moore &Watson,
ents’ experience alive using rich, evocative, 1999). Therapists therefore need to know
concrete, connotative language and often when—and when not—to respond
have a probing, tentative quality: empathically. When clients do not want
therapists to be explicitly empathic, truly
C: I don’t know what I’m going to do. empathic therapists will use their
I have two hundred dollars this perspective-taking skills to provide an
month, everything’s behind, there optimal therapeutic distance (Leitner,
isn’t enough work, and I have been 1995) in order to respect their clients’
doing other things, and then my Dad boundaries.
was here. Things are just swirling • There is no evidence that accurately
around me. I don’t know how to predicting clients’ own views of their
keep my stuff together enough for problems or self-perceptions is effective.
me even to survive. Therapists should neither assume that
T: It’s like being caught in a they are mind readers nor that their
whirlpool as if it is hard to keep experience of understanding the client

146 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
will be matched by the client feeling neuroscience of empathy (pp. 3–15). Cambridge,
understood. Empathy should always be MA: MIT Press.
Belenky, M. F., Clinchy, B. M., Goldberger, N. R.,
offered with humility and held lightly,
& Tarule, J. M. (1986). Women’s ways of know-
ready to be corrected. ing: The development of self, voice, and mind.
• Finally, because research has shown New York: Basic Books.
empathy to be inseparable from the other ∗
Bergin, A. E., & Jasper, L. G. (1969). Correlates of
relational conditions, therapists should empathy in psychotherapy: A replication. Journal
seek to offer empathy in the context of of Abnormal Psychology, 74, 477–81.
Berman, N.G., & Parker, R. (2002). Meta-analysis:
positive regard and genuineness. Empathy
Neither quick nor easy. BMC Medical Research
will not be effective unless it is grounded Methodology, 2(10). Accessed online on June
in authentic caring for the client. We 19, 2010 at
encourage psychotherapists to value 1471-2288/2/10
empathy as both an “ingredient” of a Beutler, L. E., Crago, M., & Arizmendi, T. G.
healthy therapeutic relationship as well (1986). Research on therapist variables in psycho-
therapy. In S. L. Garfield & A. E. Bergin (Eds.),
as a specific, effective response that
Handbook of psychotherapy and behavior change
strengthens the self and deeper (3rd ed., pp. 257–310). New York: Wiley.
exploration. ∗
Beutler, L. E., Johnson, D. T., Neville, C. W., &
Workman, S. N. (1972). “Accurate empathy”
and the AB dichotomy. Journal of Consulting and
References Clinical Psychology, 38, 372–75.
An asterisk (∗) indicates studies included in Blatt, S. J., Zuroff, D. C., Quinlan, D. M., &
the meta-analysis. Pilkonis, P. A. (1996). Interpersonal factors in
Anderson, T. (1999). Specifying non-‘specifics’ in brief treatment of depression: Further analyses of
therapists: The effect of facilitative interpersonal the National Institute of Mental Health
skills in outcome and alliance formation. Paper Treatment of Depression Collaborative Research
presented at the 30th annual meeting of the Program. Journal of Consulting and Clinical
International Society for Psychotherapy Research, Psychology, 64, 162–71.
Braga, Portugal. Bohart, A. C., Elliott, R., Greenberg, L. S., &
Anderson, T., Ogles, B. M., Patterson, C. L., Watson, J. C. (2002). Empathy. In J. Norcross
Lambert, M. J., & Vermeersch, D. V. (2009). (Ed.), Psychotherapy relationships that work (pp.
Therapist effects: Facilitative interpersonal skills 89–108). New York: Oxford University Press.
as a predictor of therapist success. Journal of Bohart, A. C., & Greenberg, L. S. (1997). Empathy:
Clinical Psychology, 65, 755–68. Where are we and where do we go from here?
∗ In A. C. Bohart & L. S. Greenberg (Eds.),
Barrett-Lennard, G. (1962). Dimensions of thera-
pist response as causal factors in therapeutic Empathy reconsidered: New directions in psy-
change. Psychological Monographs, 76, 1–33, chotherapy (pp. 419–50). Washington, DC:
Whole Number 562. American Psychological Association.
Barrett-Lennard, G. T. (1981). The empathy cycle: Brodley, B. T., & Brody, A. F. (1990, August).
Refinement of a nuclear concept. Journal of Understanding client-centered therapy through
Counseling Psychology, 28, 91–100. interviews conducted by Carl Rogers. Paper pre-
∗ sented at the annual convention of the American
Barrington, B. L. (1967). The differential effective-
ness of therapy as measured by the Thematic Psychological Association, Boston, MA.

Apperception Test. In C. R. Rogers, et al. (Eds.), Buckley, P., Karasu, T. B., & Charles, E. (1981).
The therapeutic relationship and its impact (pp. Psychotherapists view their personal therapy.
337–52). Madison: University of Wisconsin Psychotherapy: Theory, Research, and Practice, 18,
Press. [combined with Kiesler, Mathieu, Klein & 299–305.

Schoeninger, 1967, and Van der Veen, 1967] Bugge, I., Hendel, D. D., & Moen, R. (1985).
Batson, C. D. (2009). These things called empa- Client evaluations of therapeutic processes and
thy: Eight related but distinct phenomena. In outcomes in a university mental health center.
J. Decety & W. Ickes (Eds.) (2009). The social Journal of American College Health, 33, 141–46.

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 147

∗ Egan, G. (1982). The skilled helper (2nd ed.).
Bullmann, F., Horlacher, K. D., & Kieser, B.
(2004). Clarifying as a mediator-variable in per- Monterey, CA: Brooks/Cole.
son-centered psychotherapy and therapists´ gender Eisenberg, N., & Eggum, N. D. (2009). Empathic
and therapy-school differences in empathy and posi- responding: Sympathy and personal distress. In
tive regard. Unpublished study, Psychological J. Decety & W. Ickes (Eds.) (2009). The Social
Institute, University of Heidelberg, Germany. Neuroscience of Empathy (pp. 71–83). Cambridge,
∗ MA: MIT Press.
Burns, D. D., & Nolen-Hoeksma, S. (1992).
Therapeutic empathy and recovery from depres- Eisenberg, N., & Fabes, R. A. (1990). Empathy:
sion in cognitive-behavioral therapy: A structural Conceptualization, assessment, and relations to
equation model. Journal of Consulting and prosocial behavior. Motivation and Emotion, 14,
Clinical Psychology, 60, 441–49. 131–49.
Carkhuff, R. R., & Berenson, B. (1967). Beyond Elliott, R. (2010). Psychotherapy change process
counseling and therapy. New York: Holt, Rinehart, research: Realizing the promise. Psychotherapy
& Winston. Research, 20, 123–35.
∗ Elliott, R., Filipovich, H., Harrigan, L., Gaynor, J.,
Cartwright, R. D., & Lerner, B. (1965). Empathy,
need to change, and improvement in psychother- Reimschuessel, C., & Zapadka, J. K. (1982).
apy. Journal of Consulting Psychology, 27, 138–44. Measuring response empathy: The development
∗ of a multi-component rating scale. Journal of
Clark, J. V., & Culbert, S. A. (1965). Mutually
therapeutic perception and self-awareness in a Counseling Psychology, 29, 379–87.
T-group. Journal of Applied Behavioral Science, 1, Elliott, R., Watson, J., Goldman, R., Greenberg, &
180–94. L. S. (2003). Learning emotion-focused therapy:
∗ The process-experiential approach to change.
Cooley, E. J., & Lajoy, R. (1980). Therapeutic
relationship and improvement as perceived by Washington, DC: APA.

clients and therapists. Journal of Clinical Filak, J., & Abeles, N. (1984). Posttherapy congru-
Psychology, 36, 562–70. ence on client symptoms and therapy outcome.
∗ Professional Psychology: Research & Practice, 15,
Cramer, D., & Takens, R. (1992). Therapeutic
relationship and progress in the first six sessions 846–55.

of individual psychotherapy: A panel analysis. Fretz, B. R. (1966). Postural movements in a coun-
Counselling Psychology Quarterly, 5, 25–36. seling dyad. Journal of Counseling Psychology, 13,
Decety, J., & Ickes, W. (Eds.) (2009). The social neu- 335–43.

roscience of empathy. Cambridge, MA: MIT Press. Fuertes, J. N., Mislowack, A., Brown, S.,
Decety, J., & Jackson, P. L. (2004). The functional Gur-Arie, S., Wilkinson, S., & Gelso, C. J.
architecture of human empathy. Behavioral and (2007). Correlates of the real relationship in
Cognitive Neuroscience Reviews, 3, 71–100. psychotherapy: A study of dyads. Psychotherapy
Decety, J., & Lamm, C. (2009). Empathy versus Research, 17, 423–30.

personal distress: Recent evidence from social Fuertes, J. N., Stracuzzi, T. I., Bennett, J.,
neuroscience. In J. Decety & W. Ickes (Eds.) Scheinholtz, J., Mislowack, A., Hersh, M., et al.
(2009). The Social Neuroscience of Empathy (pp. (2006). Therapist multicultural competency:
199–213). Cambridge, MA: MIT Press. A study of therapy dyads. Psychotherapy: Theory,
Diener, M. J., Hilsenroth, M. J. & Weinberger, J. Research, Practice, Training, 43, 480–490.

(2009). A primer on meta-analysis of correlation Gabbard, C. E., Howard, G. S., & Dunfee, E. J.
coefficients: The relationship between patient- (1986). Reliability, sensitivity to measuring
reported therapeutic alliance and adult attach- change, and construct validity of a measure of
ment style as an illustration. Psychotherapy counselor adaptability. Journal of Counseling
Research, 19, 519–26. Psychology, 33, 377–86.
∗ Gallese, V., Fadiga, L., Fogassi, L., & Rizzolatti, G.,
Dormaar, J. M., Dijkman, C. I., & de Vries, M. W.
(1989). Consensus in patient-therapist inter- (1996). Action recognition in the premotor
actions: A measure of the therapeutic relation- cortex. Brain, 119, 593–609.

ship related to outcome. Psychotherapy and Garfield, S. L., & Bergin, A. E. (1971). Therapeutic
Psychosomatics, 51, 69–76. conditions and outcome. Journal of Abnormal
Duan, C., & Hill, C. E. (1996). A critical review Psychology, 77, 108–114.

of empathy research. Journal of Counseling Gelso, C. J., Latts, M. G., Gomez, M. J., &
Psychology, 43, 261–74. Fassinger, R. E. (2002). Countertransference

148 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
management and therapy outcome: An initial D. C. Brink, & P. M. Raskin (Eds.), The psycho-
evaluation. Journal of Clinical Psychology, 58, therapy of Carl Rogers (pp. 357–74). New York:
861–67. Guilford.

Gillispie, R., Williams, E., & Gillispie, C. (2005). Henschel, D. N., & Bohart, A. C. (1981, August).
Hospitalized African American mental health The relationship between the effectiveness of a course
consumers: Some antecedents to service satis- in paraprofessional training and level of cognitive
faction and intent to comply with after- functioning. Paper presented at the American
care. American Journal of Orthopsychiatry, 75, Psychological Association, Los Angeles, CA.
254–61. Higgins, J. P. T., Thompson, S. G., Deeks, J. J., &
Gladstein, G. A., & Associates (1987). Empathy and Altman, D. G. (2003). Measuring inconsistency
counseling: Explorations in theory and research. in meta-analyses, British Journal of Medicine,
New York: Springer-Verlag. 327, 557–60.

Goldman, R., Greenberg, L., & Angus, L. ( June, Hill, C. E., Thompson, B. J., & Corbett, M. (1992).
2000). The York II Psychotherapy Study on The impact of therapist ability to perceive
Experiential Therapy of Depression. Paper pre- displayed and hidden client reactions on imme-
sented at Society for Psychotherapy Research, diate outcome in first sessions of brief therapy.
Chicago, IL. Psychotherapy Research, 2, 143–55.

Goleman, D. (1995). Emotional intelligence. Hoffart, A., Versland, S., & Sexton, H. (2002).
New York: Bantam. Self-understanding, empathy, guided discovery,
Goubert, L., Craig, K. D., & Buysee, A. (2009). and schema belief in schema-focused cogni-
Perceiving others in pain: Experimental and clin- tive therapy of personality problems: A process-
ical evidence of the role of empathy. In J. Decety outcome study. Cognitive Therapy and Research,
& W. Ickes (Eds.) (2009). The Social Neuroscience 26, 199–219.
of Empathy (pp. 153–65). Cambridge, MA: MIT Horvath, A. O., & Greenberg, L. S. (1986). The
Press. development of the Working Alliance Inventory.

Greenberg, L. S., & Webster, M. (1982). Resolving In L. S. Greenberg & W. M. Pinsof (Eds.), The
decisional conflict by means of two-chair dia- psychotherapeutic process: A research handbook (pp.
logue: Relating process to outcome. Journal of 529–56). New York: Guilford.

Counseling Psychology, 29, 468–77. Horvath, A. O., & Greenberg, L. S. (1989).

Gross, W. F., & DeRidder, L. M. (1966). Significant Development and validation of the working alli-
movement in comparatively short-term counsel- ance inventory. Journal of Counseling Psychology,
ing. Journal of Counseling Psychology, 13, 98–99. 36, 223–33.
Gurman, A.S. (1977). The patient’s perception of Hubble, M. A., Duncan, B. L., & Miller, S. D.
the therapeutic relationship. In A. S. Gurman (1999). Directing attention to what works. In
& A. M. Razin (Eds.), Effective psychotherapy: M. A. Hubble, M. A. Duncan, & S. D. Miller
A handbook of research (pp. 503–43). New York: (Eds.), The heart and soul of change: What
Pergamon. works in therapy (pp. 407–48). Washington, DC:

Hall, J. A., & Davis, M. H. (2000). Dispositional American Psychological Association.
empathy in scientist and practitioner psycholo- Ickes, W. (Ed.). (1997). Empathic accuracy. New York:
gists: Group differences and relationship to self- Guilford.
reported professional effectiveness. Psychotherapy, Ickes, W. (2003). Everyday mind reading: Under-
37, 45–56. standing what other people think and feel. Amherst,

Hamilton, J. C. (2000). Construct validity of NY: Prometheus Books.
the core conditions and factor structure of the Kennedy-Moore, E., & Watson, J. C. (1999).
Client Evaluation of Counselor Scale. The Person- Expressing emotion: Myths, realities, and therapeu-
Centered Journal, 7, 40–51. tic strategies. New York: Guilford.
∗ ∗
Hansen, J. C., Moore, G. D., & Carkhuff, R. R. Kiesler, D. J., Klein, M. H., Mathieu, P. L., &
(1968). The differential relationships of objective Schoeninger, D. (1967). Constructive personal-
and client perceptions of counseling. Journal of ity change for therapy and control patients. In
Clinical Psychology, 24, 244–46. C. R. Rogers, et al. (Eds.), The therapeutic rela-
Hayes, A. M., & Goldfried, M. R. (1996). Rogers’ tionship and its impact (pp. 251–94), Madison:
work with Mark: An empirical analysis and cog- University of Wisconsin Press. (Combined with
nitive-behavioral perspective. In B. A. Farber, Barrington, 1967, and Van Der Veen, 1967)

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 149

∗ treatment: Their reliable identification and influ-
Kurtz, R. R., & Grummon, D. L. (1972). Different
approaches to the measurement of therapist ence on behaviour change. Clinical Psychology
empathy and their relationship to therapy and Psychotherapy, 9, 395–405.
outcomes. Journal of Consulting and Clinical Martin, D. G. (2000). Counseling and therapy
Psychology, 39, 106–115. skills (2nd ed.). Prospect Heights, IL: Waveland
∗ Press.
Lafferty, P., Beutler, L. E., & Crago, M. (1989).
Differences between more and less effective Martin, D. J., Garske, J. P., & Davis, M. D. (2000).
psychotherapists: A study of select therapist Relation of the therapeutic alliance with out-
variables. Journal of Consulting and Clinical come and other variables: A meta-analytic review.
Psychology, 57, 76–80. Journal of Consulting and Clinical Psychology, 68,
Lambert, M. J., DeJulio, S. J., & Stein, D. M. 438–50.

(1978). Therapist interpersonal skills: Process, Martin, P. J., & Sterne, A. L. (1976). Post-hospital
outcome, methodological considerations, and adjustment as related to therapist’s in-therapy
recommendations for future research. Psycho- behavior. Psychotherapy: Theory, Research, and
logical Bulletin, 85, 467–89. Practice, 13, 267–273.

Landfield, A. W. (1971). Personal construct systems Melnick, B., & Pierce, R. M. (1971). Client evalu-
in psychotherapy. Chicago: Rand McNally. ation of therapist strength and positive-negative
∗ evaluation as related to client dynamics, objec-
Langhoff, C., Baer, T., Zubraegel, D., &
Linden, M. (2008). Therapist-patient alliance, tive ratings of competence and outcome. Journal
patient-therapist alliance, mutual therapeutic of Clinical Psychology, 27, 408–410.

alliance, therapist-patient concordance, and out- Miller, W., Taylor, C., & West, J. (1980). Focused
come of CBT in GAD. Journal of Cognitive versus broad spectrum behavior therapy for prob-
Psychotherapy, 22, 68–79. lem drinkers. Journal of Consulting and Clinical
Leitner, L. M. (1995). Optimal therapeutic dis- Psychology, 48, 590–601.
tance: A therapist’s experience of personal con- Mitchell, K. M., Bozarth, J. D., & Krauft, C. C.
struct psychotherapy. In R. A. Neimeyer & (1977). A reappraisal of the therapeutic effec-
M. J. Mahoney (Eds.), Constructivism in psy- tiveness of accurate empathy, nonpossessive
chotherapy (pp. 70). Washington, DC: American warmth, and genuineness. In A. S. Gurman &
Psychological Association. A. N. Razin (Eds.), Effective psychotherapy.
∗ New York: Pergamon.
Lerner, B. (1972). Therapy in the ghetto. Baltimore:

Johns Hopkins University Press. Mitchell, K. M., Truax, C. B., Bozarth, J. D., &
∗ Krauft, C. C. (1973, March). Antecedents to
Lesser, W. M. (1961). The relationship between
counseling progress and empathic understand- psychotherapeutic outcome. NIMH Grant Report
ing. Journal of Counseling Psychology, 8, 330–36. (12306). Arkansas Rehabilitation Research and
Lipsey, M. W., & Wilson, D. B. (2001). Practical Training Center, Arkansas Rehabilitation Services,
meta-analysis. Thousand Oaks, CA: Sage. Hot Springs, Arkansas.
∗ Mohr, J. J., & Woodhouse, S. S. (2000, June).
Lorr, M. (1965). Client perceptions of therapists:
A study of therapeutic relation. Journal of Clients’ visions of helpful and harmful psychother-
Consulting Psychology, 29, 146–49. apy: An approach to measuring individual differ-
Mahrer, A. R. (1997). Empathy as therapist-client ences in therapy priorities. Paper presented at the
alignment. In A. C. Bohart & L. S. Greenberg 31st Annual Meeting of the Society for
(Eds.), Empathy reconsidered: New directions in Psychotherapy Research, Chicago, IL.

psychotherapy (pp. 187–216). Washington, DC: Muller, J., & Abeles, N. (1971). Relationship of
American Psychological Association. liking, empathy and therapists’ experience to
∗ outcome in psychotherapy. Journal of Counseling
Marshall, W. L., Serran, G. A., Fernandez, Y. M.,
Mulloy, R., Mann, R. E., & Thornton, D. Psychology, 18, 39–43.
(2003). Therapist characteristics in the treatment Myers, S. (2000). Empathic listening: Reports on
of sexual offenders: Tentative data on their rela- the experience of being heard. Journal of
tionship with indices of behaviour change. Humanistic Psychology, 40, 148–73.
Journal of Sexual Aggression, 9, 25–30. O’Hara, M. M. (1984). Person-centered gestalt:
∗ Towards a holistic synthesis. In R. F. Levant &
Marshall, W. L., Serran, G. A., Moulden, H.,
Mulloy, R., Fernandez, Y. M., Mann, R. E., et al. J. M. Shlien (Eds.), Client-centered therapy and
(2002). Therapist features in sexual offender the person-centered approach: New directions in

150 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
theory, research and practice (pp. 203–21). ∗
Rabavilas, A. D., Boulougouris, J. C., & Perissaki, C.
New York: Praeger. (1979). Therapist qualities related to outcome
Orlinsky, D. E., Grawe, K., & Parks, B. K. (1994). with exposure in vivo in neurotic patients.
Process and outcome in psychotherapy—noch Journal of Behaviour Therapy and Experimental
einmal. In A. E. Bergin & S. L. Garfield (Eds.), Psychiatry, 410, 293–94.
Handbook of psychotherapy and behavior change ∗
Roback, H. B., & Strassberg, D. S. (1975).
(4th ed., pp. 270–378). New York: Wiley. Relationship between perceived therapist-offered

Orlinsky, D. E., & Howard, K. I. (1967). The good conditions and therapeutic movement in group
therapy hour: Experiential correlates of patients’ psychotherapy with hospitalized mental patients.
and therapists’ evaluations of therapy sessions. Small Group Behavior, 6, 345–52.
Archives of General Psychiatry, 12, 621–32. Rogers, C. R. (1957). The necessary and sufficient
Orlinsky, D. E., & Howard, K. I. (1978). The rela- conditions of therapeutic personality change.
tion of process to outcome in psychotherapy. In Journal of Consulting Psychology, 21, 95–103.
Garfield, S. L., & Bergin, A. E. (Eds.), Handbook Rogers, C. R. (1980). A way of being. Boston:
of psychotherapy and behavior change (2nd ed., pp. Houghton Mifflin.
283–330). New York: Wiley. Sachse, R. (1990a). Concrete interventions are
Orlinsky, D. E., & Howard, K. I. (1986). Process crucial: The influence of the therapist’s process-
and outcome in psychotherapy, In S. L. Garfield ing proposals on the client’s intrapersonal explo-
& A. E. Bergin (Eds.), Handbook of psychother- ration in client-centered therapy. In G. Lietaer,
apy and behavior change (3rd ed., pp. 311–84). J. Rombauts, & R. Van Balen (Eds.), Client-
New York: Wiley. centered and experiential psychotherapy in the nine-

Pantalon, M. V., Chawarski, M. C., Falcioni, J., ties (pp. 295–308). Leuven, Belgium: Leuven
Pakes, J., & Schottenfeld, R. S. (2004). Linking University Press.
process and outcome in the community rein- Sachse, R. (1990b). The influence of therapist pro-
forcement approach for treating cocaine depen- cessing proposals on the explication process of
dence: A preliminary report. American Journal of the client. Person-Centered Review, 5, 321–44.
Drug and Alcohol Abuse, 30, 353–67. ∗
Saltzman, C., Leutgert, M. J., Roth, C. H.,
Parloff, M. B., Waskow, I. E., & Wolfe, B. E. (1978). Creaser, J., & Howard, L. (1976). Formation of
Research on therapist variables in relation to a therapeutic relationship: Experiences during
process and outcome. In Garfield, S. L., & the initial phase of psychotherapy as predictors
Bergin, A. E. (Eds.), Handbook of psychotherapy of treatment duration and outcome. Journal of
and behavior change (2nd ed., pp. 233–82). New Consulting and Clinical Psychology, 44, 546–55.
York: Wiley. ∗
Sapolsky, A. (1965). Relationship between patient-
Patterson, C. H. (1984). Empathy, warmth, and doctor compatibility, mutual perceptions, and
genuineness: A review of reviews. Psychotherapy, outcome of treatment. Journal of Abnormal
21, 431–38. Psychology, 70, 70–76.
∗ ∗
Payne, A., Liebling-Kalifani, H., & Joseph, S. Saunders, S. M. (2000). Examining the relation-
(2007). Client-centred group therapy for survi- ship between the therapeutic bond and the
vors of interpersonal trauma: A pilot investiga- phases of treatment outcome. Psychotherapy, 37,
tion. Counselling & Psychotherapy Research, 7, 206–218.
100–105. Selman, R. I. (1980). The growth of interpersonal
Peabody, S. A., & Gelso, C. J. (1982). understanding. Orlando, FL: Academic Press.
Countertransference and empathy: The complex Shamay-Tsoory, S. (2009). Empathic processing:
relationship between two divergent concepts Its cognitive and affective dimensions and neuro-
in counseling. Journal of Counseling Psychology, anatomical basis. In J. Decety & W. Ickes (Eds.),
29, 240–45. The social neuroscience of empathy (pp. 215–32).

Peake, T. H. (1979). Therapist-patient agreement Cambridge, MA: MIT Press.
and outcome in group therapy. Journal of Clinical Shlien, J. (1997). Empathy in psychotherapy:
Psychology, 35(3), 637–46. A vital mechanism? Yes. Therapist’s conceit? All
Persons, J. B., & Burns, D. D. (1985). Mechanisms too often. In A. C. Bohart & L. S. Greenberg
of action of cognitive therapy: Relative contribu- (Eds.), Empathy reconsidered: New directions in
tion of technical and interpersonal intervention. psychotherapy (pp. 63–80). Washington, DC:
Cognitive Therapy and Research, 9, 539–51. American Psychological Association.

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 151

Smith, M. L., Glass, G. V., & Miller, T. I. (1980). ∗
Truax, C. B., & Wittmer, J. (1971). The effects
The benefits of psychotherapy. Baltimore: The of therapist focus on patient anxiety source and
Johns Hopkins University Press. the interaction with therapist level of accu-

Staples, F. R., Sloane, R. D., Whipple, K., rate empathy. Journal of Clinical Psychology, 27,
Cristol, A. H., & Yorkston, N. (1976). Process 297–99.
and outcome in psychotherapy and behavior ∗
Truax, C. B., Wittmer, J., & Wargo, D. G. (1971).
therapy. Journal of Consulting and Clinical Effects of the therapeutic conditions of accurate
Psychology, 44, 340–50. empathy, nonpossessive warmth, and genuine-

Strupp, H. H., Fox, R. E., & Lessler, K. (1969). ness on hospitalized mental patients during
Patients view their psychotherapy. Baltimore: Johns group therapy. Journal of Clinical Psychology,
Hopkins Press. 27, 137–42.
Tallman, K., Robinson, E., Kay, D., Harvey, S., & ∗
Van der Veen, F. (1967). Basic elements in the
Bohart, A. (1994, August). Experiential and non- process of psychotherapy: A research study.
experiential Rogerian therapy: An analogue study. Journal of Consulting Psychology, 31, 295–303.
Paper presented at the American Psychological (Combined with Barrington, 1967, and Kiesler,
Association Convention, Los Angeles. Mathieu, Klein & Schoeninger, 1967)
Tausch, R. (1988). The relationship between emo- Wampold, B. E. (2001). The great psychotherapy
tions and cognitions: Implications for therapist debate: Models, methods, and findings. Mahwah,
empathy. Person-Centered Review, 3, 277–91. NJ: Erlbaum.
Thompson, B., & Hill, C. (1991). Therapist percep- Watson, J. C. (2001). Re-visioning empathy. In
tion of client reactions. Journal of Counseling and D. Cain & J. Seeman (Eds.), Humanistic psy-
Development, 69, 261–65. chotherapies: Handbook of research and practice

Truax, C. B. (1966). Therapist empathy, warmth, (pp. 445–71). Washington, DC: APA.
and genuineness and patient personality change ∗
Watson, J. C., & Geller, S. M. (2005). The relation
in group psychotherapy: A comparison between among the relationship conditions, working alli-
interaction unit measures, time sample measures, ance, and outcome in both process-experiential
and patient perception measures. Journal of and cognitive-behavioral psychotherapy. Psycho-
Clinical Psychology, 22, 225–29. therapy Research, 15, 25–33.
Truax, C. B., & Carkhuff, R. R. (1967). Toward Watson, J. C., & Greenberg, L. S. (2009). Empathic
effective counseling and psychotherapy: Training resonance: A neuroscience perspective. In
and practice. Chicago: Aldine. J. Decety & W. Ickes (Eds.) (2009). The social

Truax, C. B., Carkhuff, R. R., & Kodman, F., Jr. neuroscience of empathy (pp. 125–38). Cambridge,
(1965). Relationships between therapist-offered MA: MIT Press.
conditions and patient change in group psycho- Watson, J. C., & Prosser, M. (2002). Development
therapy. Journal of Clinical Psychology, 21, 327–29. of an observer rated measure of therapist
Truax, C. B., & Mitchell, K. M. (1971). Research empathy. In J. C. Watson, R. Goldman, &
on certain therapist interpersonal skills in rela- M Warner (Eds.), Client-centered and experien-
tion to process and outcome. In A. E. Bergin & tial psychotherapy in the 21st century: Advances in
S. L. Garfield (Eds.), Handbook of psychother- theory, research and practice (pp. 303–314). Ross
apy and behavior change (1st ed., pp. 299–344). on Wye, UK: PCCS Books.
New York: Wiley. Watson, N. (1984). The empirical status of Rogers’

Truax, C. B., Wargo, D. G., Frank, J. D. Imber, S. hypotheses of the necessary and sufficient condi-
D., Battle, C. C., Hoehn-Saric, R., et al. (1966). tions for effective psychotherapy. In R. F. Levant
Therapist empathy, genuineness and warmth and & J. M. Shlien (Eds.), Client-centered therapy
patient therapeutic outcome. Journal of Consulting and the person-centered approach (pp. 17–40).
Psychology, 30, 395–401. New York: Praeger.

152 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p

7 Goal Consensus and Collaboration

Georgiana Shick Tryon and Greta Winograd

This chapter focuses on one element of the Definitions and Measures

therapeutic contract between patient and Goal Consensus
psychotherapist—goal consensus—that sets At the beginning of treatment, psychother-
the parameters of treatment and one apists effect a contract with their patients
therapeutic operation—collaboration—that that outlines the conditions of their work
implements the contract that should contri- together. This “. . . therapeutic contract
bute to a satisfactory treatment outcome. In is their ‘understanding’ about their goals
support of this statement, our chapter in the and conditions for engaging each other as
first edition of this volume (Tryon & patient and therapist” (Orlinsky, Grawe,
Winograd, 2002) presented evidence that & Parks, 1994, p. 279). Agreement about
goal consensus and collaboration were posi- treatment goals and the processes by which
tively associated with measures of adult patient and therapist will achieve these
patient psychotherapy outcome. The current goals is the essence of goal consensus. Goal
chapter updates and improves upon this consensus is part of the pantheoretical
work by examining via meta-analyses results working alliance that includes patient–
of more recent studies, published from 2000 therapist agreement on the therapy goals
through 2009, that relate goal consensus and and the tasks to reach those goals as well as
collaboration to therapy outcome. formation of a bond between the members
Our chapter begins with definitions of of the therapeutic dyad (Bordin, 1979;
terms followed by a clinical example of goal Chapter II of this book).
consensus and collaboration. We then As in our chapter in the first edition of
describe and present results of two meta- this book, we define goal consensus as:
analyses (one on goal consensus and one on
collaboration). We also present results of a (a) patient therapist agreement on
small meta-analysis using data from studies goals; (b) the extent to which a therapist
that relate goal consensus and collaboration explains the nature and expectations of
to each other. Following this, we discuss therapy, and the patient’s understanding of
the patients’ contributions and the perspec- this information; (c) the extent to which
tives that they bring to these elements of goals are discussed, and the patient’s belief
the therapeutic relationship. We also dis- that goals are clearly specified; (d) patient
cuss the limitations of the research reviewed. commitment to goals; and (e) patient–
The chapter concludes with suggestions for therapist congruence on the origin of
therapeutic practice. the patient’s problem, and congruence

on who or what is responsible for problem earlier chapter (Tryon & Winograd, 2002)
solution (Tryon & Winograd, 2001, covered the goal consensus literature
pp. 385–386). prior to 2000. The instruments comprised
scales of measures that assess the working
The third column of Table 7.1 shows alliance, such as the Working Strategy
that the studies included in the goal Consensus scale of the California Psycho-
consensus meta-analysis for this chapter therapy Alliance Scale (CALPAS; Marmar,
measured goal consensus using several Gaston, Gallagher, & Thompson, 1989);
different instruments. These studies were the Goals and Tasks scale for patients, and
published between 2000 and 2009. Our the Shared Goals and Goal and Task

Table 7.1 Descriptions of Studies of Collaboration Outcome and Goal Consensus Outcome (2000–2009)
with Effect Sizes
Effect Size (95% CI)
Study N Goal consensus Collaboration Outcome measure(s) GC Cb
measure(s) measure(s)
Ablon et al. (2006) 17 Psychotherapy Symptom Checklist 90 .18
Process Q-Set - R, Anxiety Sensitivity (−.33
Index, Panic Disorder – 0.62)
Severity Scale
Abramowitz et al. 28 Therapy rationale Treatment Yale-Brown Obsessive .65 .68
(2002) acceptance rating compliance, Compulsive Scale (0.37 (0.42
homework – 0.83) – 0.84)
Ackerman et al. 128 Combined Combined Session Evaluation .66 .55
(2000) Alliance Short Alliance Short Questionnaire (0.56 (0.42
Form–Evaluation Form–Evaluation – 0.76) – 0.67)
scales scales
Addis & Jacobson 150 Treatment Homework Beck Depression .35 .29
(2000) acceptance completion Inventory, Hamilton (0.20 (0.14
rating rating Rating Scale for – 0.49) – 0.43)
Bogalo & 30 Homework IBS-Symptom Severity .16
Moss-Morris (2006) Assessment Scale, Subject’s Global (−.22
Tool Assessment of Relief – 0.50)
Brocato & Wagner 124 Working Alliance Stages of Change .05
(2008) Inventory Readiness & Eagerness (−.13
for Treatment Scale, – 0.23)
treatment retention
Burns & Spangler 521 Homework Beck Depression .34
(2000) completion Inventory, Hopkins (0.27
Symptom Checklist – 0.42)
Busseri & Tyler 46 Patient & therapist Patient improvement .14
(2004) goal agreement ratings, Post Therapy (−.16–.42)

154 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Table 7.1 Continued

Effect Size (95% CI)

Study N Goal consensus Collaboration Outcome measure(s) GC Cb
measure(s) measure(s)
Caspar et al. (2005) 21 Plan Analysis Beck Depression .33
Inventory, Global (−.12
Assessment of – 0.66)
Functioning, Hamilton
Rating Scale for
Depression, Symptom
Checklist 90 - R
Clemence et al. (2005) 113 Combined Alliance Combined Help Received Scale, .37 .44
Short Form Alliance Short Patient’s Estimate of (0.20 (0.29
Form Improvement to Date – 0.53) – 0.58)
Cowan et al. (2008) 576 Homework Beck Depression .08
completion Inventory, Hamilton (0.00
Rating Scale for – 0.16)
Depression, Perceived
Social Support Scale,
Social Support
Dunn et al. (2006) 29 Rating of Positive and Negative −.13
homework Signs of Schizophrenia (−.48
Fitzpatrick et al. 48 Working Alliance Session Impact Scale .28
(2005) Inventory (0.00
– 0.53)
Gabbay et al. (2003) 128 Problem agreement Dropout .02
– 0.20)
Gonzalez et al. (2006) 123 Homework % positive urine, .12
completion treatment retention (−.06
rating – 0.29)
Graf et al. (2008) 44 Post writing Satisfaction rating, .45
questionnaire insight rating (0.18
– 0.67)
Hegel et al. (2002) 179 Rating of patient Homework Hamilton Rating Scale .25 .21
understanding of completion for Depression (0.11 (0.07
therapy – 0.38) – 0.35)
Lingiardi et al. (2005) 47 California California Dropout .42 .45
Psychotherapy Psychotherapy (0.15 (0.19
Alliance Scale Alliance Scale – 0.64) - 0.66)
Long (2001) 24 Working Alliance Causal Dimension Scale .21
Inventory, Goal II, Global Assessment (−.21
Statement Inventory of Functioning, Target – 0.57)

Table 7.1 Continued
Effect Size (95% CI)
Study N Goal consensus Collaboration Outcome measure(s) GC Cb
measure(s) measure(s)
Principe et al. (2006) 91 Working Alliance Dropout .12
Inventory (−.09
– 0.32)
Schönberger et al. 45 c Working Alliance Client’s d2 Test of Concentration, .18 .52
(2006, 2007) Inventory Compliance European Brain Injury (−.05 (0.27
Scale Questionnaire, – 0.40) – 0.71)
Neurosensory Center
Examination for Aphasia,
Ravens Advanced
Progressive Matrices,
success ratings, Trail
Making Test, WAIS-R,
Wisconsin Card Sorting
Test, word fluency
Stein et al. (2004) 53 Adherence to Hamilton Rating Scale .31
intervention for Depression (0.04
checklist – 0.54)
Wettersten et al. 64 Working Alliance Brief Symptom .18
(2005) Inventory Inventory, Counseling (−.07
Center Follow-up – 0.41)
Whittal et al. (2004) 59 Homework Yale-Brown Obsessive .34
Compliance Compulsive Scale (0.09
– 0.56)
Woods et al. (2002) 82 Homework Behavioral Avoidance .05
hours, Test, target symptoms (−.17
homework – 0.27)
Yovel & Safren (2007) 15 Homework AD/HD Rating Scale, .39
adherence, Clinical Global (−.15
symptom Impression Scale, Global – 0.76)
change Assessment of
Functioning, Hamilton
Anxiety Scale, Hamilton
Rating Scale for
Zane et al. (2005) 60 Goals measure, Global Assessment .24
Perceptual Rating of Functioning, (−.01
Scale Session Evaluation – 0.46)
Note: Of the 15 goal consensus effect sizes reported in column 6, 11 were based on zero-order correlations, 3 were derived from partial beta
coefficients (Abramowitz et al., 2002; Brocato & Wagner, 2008; Zane et al., 2005), and 1 originated from a combination of zero order
correlations and partial beta coefficients (Wettersten et al., 2005). Of the 19 collaboration effect sizes reported in column 7, 15 were based
on zero-order correlations, and 4 (Ablon et al., 2006; Abramowitz et al., 2002; Cowan et al., 2008; Woods et al., 2002) were derived from
partial beta coefficients with two or more independent variables.
GC = Goal Consensus. b C = Collaboration. c Both Schönberger et al. (2006) and Schönberger et al. (2007) used the same study sample.
The 2006 paper assessed collaboration (N = 45). Both papers assessed goal consensus (N = 72).

Disagreement scales for therapists of the involvement of patient and therapist in a
Combined Alliance Short Form (CASF; helping relationship as well as (b) patient
Hatcher, 1999; Hatcher & Barends, 1996); cooperation and (c) role involvement.
and the Goal and Task scales of the Working Another indicator of, but not a measure of,
Alliance Inventory (WAI; Horvath & collaboration is (d) patient completion of
Greenberg, 1989). Other studies assessed assigned homework.
goal consensus using rating scales specifi- The fourth column of Table 7.1 presents
cally designed to assess actual patient– collaboration measures used by the studies
therapist goal agreement, such as the Goal in our meta-analysis. These studies were
Statement Inventory (GSI; McNair & Lorr, published between 2000 and 2009. Our
1964) and the Causal Dimensions Scale II earlier chapter (Tryon & Winograd, 2002)
(CDS-II; McAuley, Duncan, & Russell, covered the collaboration literature prior
1992). Still others (e.g., Gabbay et al., 2003; to 2000. Collaboration measures include
Hegel, Barrett, Cornell, & Oxman, 2002) scales from working alliance measures
used goal consensus measures unique to such as the Therapist Understanding and
their studies. Involvement scale of the CALPAS (Marmar
Column 5 in Table 7.1 shows that the et al., 1989); the Confident Collaboration
studies also used several measures of therapy scale of the patient CASF (Hatcher &
outcome associated with goal consensus. Barends, 1996); and the Patient Working
Most outcome measures assessed patient Engagement, Patient Confidence and
improvement (e.g., as in Busseri & Tyler, Commitment, and Therapist Confident
2004), while others examined treatment Collaboration scales of the therapist CASF
dropout (e.g., as in Gabbay et al., 2003), (Hatcher, 1999). Other studies used rat-
and others considered session impact (e.g., ings of patient treatment compliance (e.g.,
as in Fitzpatrick, Iwakabe, & Stalikas, as in Abramowitz, Franklin, Zoellner, &
2005). In our previous review of the goal DiBernardo, 2002), treatment acceptance
consensus literature (Tryon & Winograd, (e.g., as in Addis & Jacobson, 2000), and
2001), we found that 68% (n = 17) of the patient adherence to the treatment (e.g.,
studies reviewed (n = 25) “revealed a Stein et al., 2004) that were unique to the
positive relationship between goal consen- study. Studies that assessed homework
sus and outcome on at least one measure compliance also generally used measures
completed by patient, therapist, or observer” unique to their studies (e.g., as in Yovel &
(p. 386). Safren, 2007).
The studies in the collaboration outcome
Collaboration meta-analysis used several types of outcome
To implement the therapeutic contract, measures (see fifth column of Table 7.1).
patient and therapist must function as a These ranged from patient symptom
team. Collaboration represents the active improvement (e.g., as in Ablon, Levy, &
process of working together to fulfill Katzenstein, 2006), session evaluation (e.g.,
treatment goals. As with goal consensus, col- as in Ackerman, Hilsenroth, Baity, &
laboration is a pantheoretical concept that Blagys, 2000), treatment retention (e.g., as
applies to all types of therapies. Collaboration in Brocato & Wagner, 2008), and patient
“. . . is largely defined by the instruments satisfaction (e.g., as in Graf, Gaudiano, &
devised to assess the concept” (Bachelor, Geller, 2008). In our previous review of the
Laverdière, Gamache, & Bordeleau, 2007, collaboration outcome literature (Tryon &
p. 175). These instruments assess (a) mutual Winograd, 2001), “we combined results

t ryo n , w i n o g r a d 157
from 24 studies and found that 89% of basis while she continues to work toward
the time, collaborative involvement and recovery.
outcome were significantly positively related The excerpt below is from Hope and her
on at least one measure completed by therapist’s second session together. Elements
patient, therapist, or observer” (p. 387). of goal consensus and collaboration in their
interaction are indicated in brackets.
Relationship of Goal Consensus
and Collaboration Therapist: Last time we talked about the
By definition, consensus implies an agree- challenges you’ve been facing as you
ment based on the opinions of the par- return to school and reconnect with
ties involved.1 Achievement of consensus friends and family members who
requires that those involved work coopera- know about your recent episode and
tively, which is the definition of collabora- hospitalization. I asked you to jot
tion.2 Since therapist and patient work down some of the thoughts that have
together to establish agreement on the goals been running through your mind
and tasks of psychotherapy, one might expect when you are around people at
that measurements of goal consensus and school, and some of the ways people
collaboration would be related. Since 2000, act or the things they say that you
however, only seven of the published studies find upsetting.
that we found reported this relationship Hope: Yes, I did both of those
(Abramowitz et al., 2002; Ackerman et al., things [collaboration: homework
2000; Addis & Jacobson, 2000; Clemence, completion]. I realized that even
Hilsenroth, Ackerman, Strassle, & Handler, with classmates I hardly know,
2005; Hegel et al., 2002; Lingiardi, I find myself worrying constantly
Filippucci, & Baiocco, 2005; Schönberger, about whether they know what
Humle, & Teasdale, 2006, 2007). happened, or if they can just tell
by the way I act that I was in the
hospital, that I have a mental illness.
Clinical Example
And when I’m thinking this way,
Hope is a 21-year-old college senior and
I start to act nervous and insecure,
honors student who took a leave of absence
and just find it hard to concentrate.
from school last semester after being hospi-
But when I’m around friends and
talized for several weeks in a mental health
family members who know about my
crisis center. A psychotic episode—her
crisis, I feel even worse. They act so
first—led to Hope’s hospitalization and to
differently from the way they acted
a diagnosis of bipolar disorder. Hope has
before my diagnosis. This is what I
recently returned to college on a part-time
wrote down: “They either act like I’m
made of glass and about to break, or
they keep their distance.” And I don’t
1 consensus. (2010). In Merriam-Webster know how to get them to just treat
Online Dictionary. Retrieved June 6, 2010, from me like a regular person again. Therapist: It sounds to me like some
of the people you are close to have
2 collaboration. (2010). In Merriam-Webster
Online Dictionary. Retrieved June 6, 2010, from disappointed you. You’d like to reconnect with them but aren’t sure
collaboration how to do this. It also sounds to me

158 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
as though even when other students Presuming Hope continued to experi-
don’t necessarily act differently, you ence her psychotherapy as a collabora-
are worried that they are seeing you tive enterprise, she would likely strongly
differently, or would see you endorse collaboration items from the revised
differently if they knew about your Helping Alliance Questionnaire (Haq-II;
recent life events. This makes it hard Luborsky et al., 1996) such as, “the thera-
to relax around them and focus on pist and I have meaningful exchanges” and
your schoolwork. “I feel I am working together with the thera-
Hope: Yes, I am quite uncomfortable pist in a joint effort” (p. 271). Hope would
around other people now, almost also be likely to endorse consensus items
all of the time [goal consensus: from the Working Alliance Inventory such
congruence on patient problem]. as, “we agree about what is important for
Therapist: I also recall that you me to work on” and “my therapist and I are
mentioned being quite isolated. working towards mutually agreed upon
Hope: Yes, that’s true. Since I don’t goals” (Horvath & Greenberg, 1989a,
know what to say or how to act, p. 226).
I’ve started to avoid people.
I basically just go directly to my Meta-Analytic Review
classes and then straight home. Inclusion Criteria and Study Selection
But I don’t think spending so much We included goal consensus and collabo-
time alone is good for my mood. It ration studies in their respective meta-
doesn’t even feel like me. I used to be analysis if they used: (a) at least one measure
a really social person [goal consensus: of goal consensus and/or one measure of
further congruence on patient collaboration defined according to criteria
problem]. listed above; (b) at least one psychotherapy
Therapist: I was thinking that over the outcome measure; (c) a group design;
next few sessions, we could work (d) individually conducted psychotherapy;
together to come up with ideas about (e) adult clients (aged 18 and older). Finally,
how to talk about your each study (f ) reported a correlation, its
hospitalization and recovery with equivalent (standardized β weight), or other
your friends and family statistic (t, F, or d ) that could be converted
[collaboration: mutual involvement of to a correlation between goal consensus
patient and therapist in a helping and/or collaboration scores and outcome
relationship]. I was also thinking we scores; and (g) was published in English in a
might try out some relaxation and refereed journal from 2000 through 2009.
thought replacement strategies for We and our students (a psychology doc-
when you get anxious around toral student, a psychology masters student,
classmates in school. and an undergraduate psychology major) con-
Hope: I like the sound of that. ducted advanced Google Scholar searches for
And I’d also like you to help articles from 2000 through 2009 using the
me experiment with gradually following terms: patient–therapist collabora-
coming out of my shell as I work on tion (1,290 references), patient–therapist goal
getting healthy again [goal consensus: consensus (631 references), homework comp-
discussion and specification of goals; liance and psychotherapy outcome (3,760
covllaboration: patient role references), patient–therapist goal consensus
involvement]. and psychotherapy outcome (459 references),

t ryo n , w i n o g r a d 159
patient–therapist agreement and psychother- in the course of therapy, and rater perspec-
apy outcome (796 references), and patient– tive. We also independently identified goal
therapist collaboration and psychotherapy consensus, collaboration, and outcome data
outcome (555 references). and calculated effect sizes for the relation-
We cross-tabulated the references, ships among these data. Our ratings and
inspected the article abstracts, and identi- calculations were in agreement for 498 out
fied 53 articles for in-depth examination by of 512 items (97%). When we disagreed,
the two authors of this chapter. After review- we discussed the item until we reached
ing each of the 53 studies independently, agreement. Table 7.1 presents the coded
the authors were in perfect agreement in information for the measures and effect
identifying 28 studies that met inclusion sizes.
criteria and 25 studies that did not. The We classified patient disturbance as mild
reasons for not meeting inclusion criteria (volunteers from college classes, no formal
were: other than individual therapy (n = 8), diagnosis, or seen at university-based train-
measure of collaboration or goal consensus ing clinics), moderate (formal diagnoses of
not used in outcome analysis (n = 8), no nonpsychotic mood disorder, seen in out-
measure of collaboration or goal consensus patient settings other than university train-
(n = 3), no outcome measure (n = 3), no ing clinics), and severe (seen in inpatient
treatment (n = 2), and results could not be settings, diagnosed with a psychosis, iden-
converted to effect size (n = 1). tified by study authors as severely disturbed).
Table 7.1 describes the 28 studies that met Eight studies had patients with severe
inclusion criteria. Thirteen of the 28 studies disturbances, 13 studies had patients with
(46%) provided effect sizes for collaboration moderate disturbances, and 7 studies had
and outcome only, 9 studies (32%) provided patients with mild disturbances.
effect sizes for goal consensus and outcome Psychotherapists were either experienced
only, and 7 studies (25%) provided effect (16 studies), trainees (2 studies), or a
sizes for both goal consensus and collabo- combination of experienced therapists and
ration with outcome. Thus, the goal consen- trainees (7 studies). Three studies did not
sus meta-analysis included results from specify therapist experience. Four studies
15 studies, and the collaboration meta- did not specify theoretical orientation. In
analysis included results from 19 studies. studies using a single theoretical orienta-
We were also in perfect agreement in catego- tion, therapy was behavioral (2 studies),
rizing the studies into these two groups. cognitive behavioral (9 studies), psychody-
namic (2 studies), or eclectic, interper-
Coding of Study Characteristics sonal, individualized, or solution focused
In keeping with editorial requirements for (1 study each). The remaining 7 studies
this chapter and similar to a prior meta- used therapies reflecting more than one ori-
analysis by the first author and colleagues entation (e.g., cognitive behavioral). We
(Tryon, Blackwell, & Hammel, 2007), we divided treatment length into 1–10 sessions
independently recorded and coded the (7 studies), 11–20 sessions (12 studies),
following information from each study: and 21 or more sessions (7 studies). Two
number of participants, severity of patient studies did not specify treatment length.
disturbance, therapist experience, treatment Table 7.1 shows the instruments used
theoretical orientation, treatment length, in studies for goal consensus (Column 3),
measures of goal consensus and/or collabo- collaboration (Column 4), and outcome
ration and outcome, time of measurement (Column 5). The majority (13 out of

160 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
15, 87%) of goal consensus studies used consensus and four for collaboration)
only one measure of goal consensus were derived from partial beta coefficients
(M = 1.20, SD = 0.56), and the majority of with two or more independent variables
collaboration studies (16 out of 19, 84%) (see table note). Partial beta coefficients
used only one collaboration measure underestimate the zero-order correlation
(M = 1.16, SD = 0.37). The studies tended coefficient.
to have more than one outcome measure When measures within a study were
(M = 2.14, SD = 1.53). Measures were completed or recorded by different numbers
often completed by more than one rater. In of participants or observers, we weighted
total, patients completed the most mea- each correlation by the number of patients
sures (n = 49), followed by therapists for whom the measures were completed and
(n = 37), and observers (n = 32). divided by the total number of patients
For each study, we coded the time in represented for each correlation.
therapy that measures were completed by
dividing the number of sessions in the study The Meta-Analyses
into thirds, and naming the thirds as early, We used a meta-analysis package (Schmidt
middle, and end of therapy (in all cases, & Le, 2005) that corrects for study artifacts
measures in the latter third were at the very (Hunter & Schmidt, 2004) such as the
end of treatment). Most measures were unreliability of measures used in each study.
completed at the end (54 measures) of ther- For the current meta-analyses, we corrected
apy, followed by early (17 measures) and for unreliability using coefficient alphas for
in the middle (3 measures) of therapy. the studies’ measures. Because each study
Twelve studies used at least one measure did not provide alphas for all measures used
(18 measures) that was completed conti- in the meta-analyses (i.e., collaboration,
nuously (in each third, often after each goal consensus, and outcome measures),
session) during therapy, and one study we used the artifact distribution option of
contained three measures completed at the program that allowed us to enter the
follow-up only. Finally, two studies did alphas that the studies provided as well as
not specify time of completion of measures alphas from other studies that used the
(7 measures). measures (contact corresponding author
for a reference list of additional studies
Estimation of Effect Sizes providing reliabilities). Effect sizes were
To obtain the effect sizes listed in Table 7.1, weighted by sample size and reliability so
for each study, we recorded correlations that effects that were more precise (i.e.,
or standardized β weights between goal derived from studies with larger sample
consensus measures and outcome measures sizes and greater reliability) were given
and/or collaboration measures and outcome more weight.
measures. We averaged correlations or stan-
dardized β weights in studies that had more Results: Goal Consensus
than one measure of goal consensus and/or Fifteen studies with a total sample size
collaboration or outcome to obtain one of 1,302 provided goal consensus–psycho-
effect size for the relationship of each thera- therapy outcome effect sizes for the meta-
peutic element (goal consensus, collabora- analysis (see Table 7.1). Weighting for
tion) to outcome for each study. Most effect sample size and unreliability of measures,
sizes were based on zero-order correlations; the meta-analysis yielded a mean correlation
however, a few effect sizes (three for goal of 0.34 (SD = 0.19) with a 95% confidence

t ryo n , w i n o g r a d 161
interval of 0.23 to 0.45. The variability standard deviation (d = 0.68; Lyons, 2003)
between studies after removal of the effects improvement associated with a 1 standard
of unreliability of the measures and varia- deviation boost in collaboration. Thus,
tion in sample size was 0.02. Because this patient experience and well-being appear
procedure left virtually no variability in to be considerably enhanced with a better
effects due to differences between studies, quality collaborative relationship between
there was no variability to be explained by patient and therapist.
moderators. Therefore, as recommended
(Hunter & Schmidt, 2004), we did not Results: Relation of Goal Consensus
conduct a moderator analysis. The results and Collaboration
signify a medium (Cohen, 1992), unmod- Of the 7 studies that had measures of both
erated effect between goal consensus and goal consensus and collaboration, only 4
psychotherapy outcome. Because an r of reported effect sizes, or information from
0.34 is equivalent to a d of 0.72 (Lyons, which to calculate effect sizes, for the
2003), a 1 standard deviation improvement association between these two variables.
in goal consensus predicts nearly a 3/4 stan- Studies by Addis and Jacobson (2000;
dard deviation improvement in outcome. r = 0.17, n = 150), Abramowitz et al.
This is a substantial relationship, especially (2002; r = 0.17, n = 28), Clemence et al.
considering outcomes as meaningful as (2005; r = 0.09, n = 125), and Lingiardi
retention in treatment, symptom reduc- et al. (2005, r = 0.19, n = 37) provided
tion, and adaptive functioning. data for the goal consensus–collaboration
relationship. Collectively, these four studies
Results: Collaboration provided a sample of 340 patients. Weigh-
Nineteen studies involving a total sample ing for sample size and unreliability of
of 2,260 patients provided collaboration– measures, the meta-analysis on these four
psychotherapy outcome effect sizes for the studies yielded a mean correlation of 0.19
meta-analysis (Table 7.1). Weighing for (SD = 0), which represents an effect size
sample size and unreliability of measures, between small and medium (Cohen, 1992).
the meta-analysis yielded a mean correla- Variability between studies after removal of
tion of 0.33 (SD = 0.17) with a 95% the effects of unreliability of the measures
confidence interval of 0.25 to 0.42. The and variation in sample size was 0.
variability between studies after removal of
the effects of unreliability of the measures File Drawer Analyses
and variation in sample size was 0.02. It is possible that studies not included
Because this procedure left virtually no in the meta-analyses (e.g., unpublished
variability in effects due to differences papers, dissertations, book chapters)
between studies, there was no variability to could have null results that, if included,
be explained by moderators. Therefore, as would have reduced the effects we found.
recommended (Hunter & Schmidt, 2004), Thus, we conducted file drawer analyses
we did not conduct a moderator analysis. to determine how many studies with
The results denote a medium (Cohen, null results would reduce the effect sizes
1992), unmoderated effect between colla- substantially. For the goal consensus meta-
boration and psychotherapy outcome. Across analysis, we would need to have found 87
outcomes including service use, satisfaction studies with null results to reduce the effect
with services received, and patient improve- size to 0.05. For collaboration, the number
ment, this effect corresponds to a 2/3 of studies needed is 106. Thus, it is unlikely

162 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
that we would find such large numbers of input and feedback. The verbal interchanges
well-designed, unpublished studies. involved in goal consensus reflect a negoti-
For the goal consensus–collaboration ation in which patients and therapists
meta-analysis, however, we would need to together refine the goals and tasks of
find only 11 studies with null effects to therapy.
reduce the effect size to 0.05. Conse- In addition to completing homework,
quently, the results of the goal consensus– the patient’s contribution to the ongoing
collaboration analysis should be interpreted collaborative work of therapy takes the
cautiously. form of offering information, insights, self-
reflections, elaborations and explorations
Potential Moderators and Mediators of important themes, and “work(ing)
Although in meta-analyses there may be actively with the therapist’s comments”
moderators or mediators of statistical (Colli & Lingiardi, 2009, p. 723). Patients
relationships between variables, after may not recognize the importance of these
removal of the effects of unreliability of the behaviors and the role they play in goal
measures and variation in sample size, consensus and collaboration (Hatcher &
results of the current meta-analyses indi- Barends, 2006). Indeed, they tend to
cated unmoderated relationships between emphasize the importance of what the
the therapy elements of collaboration and therapist does, and even when prompted,
goal consensus and therapy outcome. We, downplay their contribution to the work of
therefore, did not analyze for moderators. therapy (Bedi, Davis, & Williams, 2005).
Perhaps some patients have so little confi-
Patient Contribution dence in their efficacy in relation to their
Patients, particularly those who are new to problems (after all, they have been unable
therapy, may have an inaccurate percep- to solve their troubles on their own) that
tion of the role they are expected to play they do not acknowledge the importance
in the treatment process. In their experi- of their part in collaborating with thera-
ences with other health professionals, such pists toward a successful outcome.
as physicians, patients tend to play a rela-
tively passive, submissive role, presenting Limitations of the Research
their symptoms and receiving treatment. This chapter only included studies with
The goals of such treatment typically do adults that were published in English and
not involve much discussion, and there in refereed journals, and it did not include
may be little collaboration regarding treat- studies with child or adolescent patients.
ment beyond patient compliance in follow- Because goal consensus and collaboration
ing professional directives. are often considered to be part of the
Psychotherapy, in contrast, requires working alliance, articles included in this
active involvement by patients from initial chapter may also have been included in
goal setting to termination. The therapist the analysis presented in Chapter 2 of this
cannot effect treatment alone. Patients volume concerning the working alliance in
bring their concerns to therapists, and adult psychotherapy. In contrast to analyses
together they conceptualize treatment goals reported in other chapters in this volume,
and ways to achieve them. Although thera- which covered a more extensive time period,
pists frequently reconceptualize patients’ the meta-analyses in this chapter used
problems, refine goals, and suggest ways to studies that were published in the past
achieve those goals, they do so with patients’ 10 years. For a review of studies published

t ryo n , w i n o g r a d 163
prior to 2000, see our earlier chapter in the consensus and collaboration and measures
first edition of this volume (Tryon & that allow for their accurate assessment.
Winograd, 2002). Finally, while the results of the meta-
Although we included a goal consensus– analyses in this chapter indicate positive
collaboration meta-analysis using data from relationships between goal consensus and
articles in listed Table 7.1, we did not search outcome and between collaboration and
for additional articles that included these outcome, they do not provide proof that
two elements because their relationship was either goal consensus or collaboration
not the focus of the chapter. So, we advise causes positive outcomes.
readers to interpret the correlation between
goal consensus and collaboration reported
Therapeutic Practices
here cautiously.
The results of the primary meta-analyses
While the studies in the current meta-
indicate strong links between patient–
analyses represent improvements over those
therapist goal consensus and positive
included in our prior chapter, they were
therapy outcomes, as well as between their
not without limitations. Although many
collaboration and outcome. The results
studies reported results as effect sizes, several
point to a number of practices that psycho-
studies did not. Several studies also failed
therapists can profitably effect.
to report nonsignificant results. Statistical
correction of this problem through file • Begin work on client problems
drawer analyses or assignment of arbitrary only after you and the patient agree on
effect sizes is a poor substitute for having treatment goals and the ways you will go
the actual effect sizes. Also, some studies about reaching them.
reported results based on fewer participants • Rarely push your own agenda.
than indicated in their procedure sections. Listen to what your patients tell you and
Readers should bear these limitations, and formulate interventions with their input
those in the previous paragraph, in mind and understanding.
when interpreting results. The acceptance • Encourage patients’ contributions
of journal article reporting standards throughout psychotherapy by asking
(JARS; APA Publications and Communi- for their feedback, insights, reflections,
cations Working Board Group, 2008) by and elaborations. Regularly seek infor-
editors should address these difficulties mation from patients about their current
and allow for more precise meta-analytic functioning, motivation to change,
syntheses of research data. and social support and provide them
A glance at Table 7.1 shows the diver- with feedback about their progress
sity of goal consensus and collaboration (Harmon et al., 2007; Whipple et al.,
measures. These reflect the various defini- 2003; also see Chapter 10 by Lambert
tions of these elements in the literature, and & Shimokawa).
in the case of collaboration, the element • Educate patients about the
itself is defined by the measures used to importance of their collaborative
assess it (Bachelor et al., 2007). Instru- contribution to the success of therapy.
mentation frequently evolves from theoreti- Psychotherapists can do so by sharing with
cal advances, and we endorse continuing patients the results of research, such as
conceptualization of these elements. The those reviewed in this chapter, that link
goal should be to provide researchers and their collaborative contribution to
practitioners with clear definitions of goal successful outcomes.

164 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
• Encourage homework completion. research in psychology. American Psychologist,
To enhance homework completion, 63, 839–51.
Bachelor, A., Laverdière, O., Gamache, D., &
encourage patient collaboration in
Bordeleau, V. (2007). Clients’ collaboration in
formulating homework assignments; therapy: Self-perceptions and relationships with
assign homework that relates to client psychological functioning, interpersonal
treatment goals; begin with small, easily relations, and motivation. Psychotherapy: Theory,
accomplished assignments and build to Research, Practice, Training, 44, 175–192.
larger ones; define homework tasks clearly; Bedi, R. P., Davis, M. D., & Williams, M. (2005).
Critical incidents in the formation of the
give homework assignments in writing;
therapeutic alliance from the client’s perspective.
provide written reminders to complete Psychotherapy: Theory, Research, Practice, Training,
tasks; encourage and incorporate client 42, 311–323.
feedback on homework (Detweiler & ∗
Bogalo, L., & Moss-Morris, R. (2006). The
Whisman, 1999; Nelson, Castonguay, effectiveness of homework tasks in an irritable
& Barwick, 2007). bowel syndrome self-management programme.
New Zealand Journal of Psychology, 35, 120–125.
• Be “on the same page” with patients.
Bordin, E. S. (1979). The generalizability of the
Check frequently with patients to make psychoanalytic concept of the working alliance.
sure that you understand each other and Psychotherapy: Theory, Research, Practice, Training,
are working toward the same ends. 16, 252–260.

• Modify your treatment methods and Brocato, J., & Wagner, E. F. (2008). Predictors of
relational stance, if ethically and clinically retention in an alternative-to-prison substance
abuse treatment program. Criminal Justice and
appropriate, in response to patient
Behavior, 35, 99–119.
feedback. ∗
Burns, D. D., & Spangler, D. L. (2000). Does
psychotherapy homework lead to improvements
in depression in cognitive-behavioral therapy or
does improvement lead to increased homework
References marked with an asterisk (∗) indi- compliance? Journal of Consulting and Clinical
cate studies included in the meta-analyses. Psychology, 68, 46–56.
∗ ∗
Ablon, J. S., Levy, R. A., & Katzenstein, T. (2006). Busseri, M. A., & Tyler, J. D. (2004). Client-
Beyond grand names of psychotherapy: Identi- therapist agreement on target problems, working
fying empirically supported change processes. alliance, and counseling outcome. Psychotherapy
Psychotherapy, Theory, Research, Practice, Training, Research, 14, 77–88.
43, 216–231. ∗
Caspar, F., Grossmann, C., Unmüssig, C., &

Abramowitz, J. S., Franklin, M. E., Zoellner, L. A., Schramm, E. (2005). Complementary thera-
& DiBernardo, C. L. (2002). Treatment compli- peutic relationship: Therapist behavior, inter-
ance and outcome in obsessive-compulsive personal patterns, and therapeutic effects.
disorder. Behavior Modification, 26, 447–463. Psychotherapy Research, 15, 91–102.
∗ ∗
Ackerman, S. J., Hilsenroth, M. J., Baity, M. R., & Clemence, A. J., Hilsenroth, M. J., Ackerman, S. J.,
Blagys, M. D. (2000). Interaction of therapeutic Strassle, C. G., & Handler, L. (2005). Facets of
process and alliance during psychological assess- the therapeutic alliance and perceived progress in
ment. Journal of Personality Assessment, 75, psychotherapy: Relationship between patient
82–109. and therapist perspectives. Clinical Psychology

Addis, M. E., & Jacobson, N. S. (2000). A closer and Psychotherapy, 12, 443–454.
look at the treatment rationale and homework Cohen, J. (1992). A power primer. Psychological
compliance in cognitive-behavioral therapy for Bulletin, 112, 155–159.
depression. Cognitive Therapy and Research, 24, Colli, A., & Lingiardi, V. (2009). The Collaborative
313–326. Interactions Scale: A new transcript-based
APA Publications and Communications Board method for assessment of therapeutic alliance
Working Group on Journal Article Reporting ruptures and resolutions in psychotherapy.
Standards. (2008). Reporting standards for Psychotherapy Research, 19, 718–734.

t ryo n , w i n o g r a d 165
∗ ∗
Cowan, M. J., Freedland, K. E., Burg, M. M., Hegel, M. T., Barrett, J. E., Cornell, J. E., &
Saab, P. G., Youngblood, M. E., Cornell, C. E., Oxman, T. E. (2002). Predictors of response to
et al. (2008). Predictors of treatment response problem-solving treatment of depression in
for depression and inadequate social support– primary care. Behavior Therapy, 33, 511–527.
The ENRICHD randomized clinical trial. Horvath, A. O., & Greenberg, L. S. (1989a).
Psychotherapy and Psychosomatics, 77, 27–37. Development and validation of the Working
Detweiler, J. B., & Whisman, M. A. (1999). The Alliance Inventory. Journal of Counseling
role of homework assignments in cognitive ther- Psychology, 36, 223–233.
apy for depression: Potential methods for Horvath, A. O., & Greenberg, L. S. (1989b). The
enhancing adherence. Clinical Psychology: Science development of the Working Alliance Inventory.
and Practice, 6, 267–282. In L. S. Greenberg & W. M. Pinsof (Eds.), The

Dunn, H., Morrison, A. P., & Bentall, R. P. (2006). psychotherapeutic process: A research handbook
The relationship between patient suitability, (pp. 529–556). New York: Guilford Press.
therapeutic alliance, homework compliance, Hunter, J. E., & Schmidt, F. L. (2004). Methods of
and outcome in cognitive therapy for psycho- meta-analysis: Correcting error and bias in research
sis. Clinical Psychology and Psychotherapy, 13, findings (2nd ed.). Thousand Oaks, CA: Sage.

145–152. Lingiardi, V., Filippucci, L., & Baiocco, R. (2005).

Fitzpatrick, M. R., Iwakabe, S., & Stalikas, A. Therapeutic alliance evaluation in personality
(2005). Perspective divergence in the working disorders psychotherapy. Psychotherapy Research,
alliance. Psychotherapy Research, 15, 69–79. 15, 45–53.
∗ ∗
Gabbay, M., Shields, C., Bower, B., Sibbald, M., Long, J. R. (2001). Goal agreement and early
King, M., & Ward, E. (2003). Patient- therapeutic change. Psychotherapy, 38, 219–232.
practitioner agreement: Does it matter? Luborsky, L., Barber, J. P., Siqueland, L., Johnson,
Psychological Medicine, 33, 241–251. S., Najavits, L. M., Frank, A., et al. (1996). The

Gonzalez, V. M., Schmitz, J. M., & DeLaune, K. revised Helping Alliance Questionnaire (Haq-II):
A. (2006). The role of homework in cognitive- Psychometric properties. Journal of Psychotherapy
behavioral therapy for cocaine dependence. Practice and Research, 5, 260–271.
Journal of Consulting and Clinical Psychology, 74, Lyons, L. C. (2003, January 30). Meta-analysis:
633–637. Methods of accumulating results across research

Graf, M. C., Gaudiano, B. A., & Geller, P. A. domains. Retrieved from http://www.lyonsmorris.
(2008). Written emotional disclosure: A con- com/MetaA/studyfx1.htm
trolled study of expressive writing homework in Marmar, C. R., Gaston, L., Gallagher, D., &
outpatient psychotherapy. Psychotherapy Research, Thompson, L. W. (1989). Alliance and outcome
18, 389–399. in late-life depression. Journal of Nervous and
Harmon, S. C., Lambert, M. J., Smart, D. M., Mental Disease, 177, 464–472.
Hawkins, E., Nielsen, S. L., Slade, K., et al. McAuley, E., Duncan, T. E., & Russell, D. W.
(2007). Enhancing outcome for potential (1992). Measuring causal attributions: The
treatment failures: Therapist-client feedback and revised Causal Dimension Scale (CDS-II).
clinical support tools. Psychotherapy Research, 17, Personality and Social Psychology Bulletin, 18,
379–392. 566–573.
Hatcher, R. L. (1999). Therapists’ views of treat- McNair, D., & Lorr, M. (1964). Three kinds of
ment alliance and collaboration in therapy. psychotherapy goals. Journal of Clinical
Psychotherapy Research, 9, 405–423. Psychology, 20, 390–593.
Hatcher, R. L., & Barends, A. W. (1996). Patients’ Nelson, D. L., Castonguay, L. G., & Barwick, F.
view of the alliance in psychotherapy: Exploratory (2007). Directions for the integration of home-
factor analysis of three alliance measures. Journal work in practice. In N. Kazantzis & L. L’ Abate
of Consulting and Clinical Psychology, 64, (Eds.), Handbook of homework assignments in
1326–1336. psychotherapy: Research, practice, and prevention
Hatcher, R. L., & Barends, A. W. (2006). How (pp. 425–444). New York: Springer.
a return to theory could help alliance research. Orlinsky, D. E., Grawe, K., & Parks, B. K. (1994).
Psychotherapy: Theory, Research, Practice, Process and outcome in psychotherapy–Noch
Training, 43, 292–299. einmal. In A. E. Bergin & S. L. Garfield (Eds.),

166 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Hanbook of psychotherapy and behavior change Psychotherapy relationships that work: Therapist
(4th ed., pp. 270–376). New York, NY: Wiley. contributions and responsiveness to patients

Principe, J. M., Marci, C. D., Glick, D. M., & (pp. 109–125). New York: Oxford University
Ablon, J. S. (2006). The relationship among Press.

patient contemplation, early alliance, and con- Wettersten, K. B., Lichtenberg, J. W., &
tinuation in psychotherapy. Psychotherapy: Theory, Mallinckrodt, B. (2005). Associations between
Research, Practice, Training, 43, 238–243. working alliance and outcome in solution-

Schönberger, M., Humle, F., & Teasdale, T. (2006). focused brief therapy and brief interpersonal
Subjective outcome of brain injury rehabilitation therapy. Psychotherapy Research, 15, 35–43.
in relation to the therapeutic working alliance, Whipple, J. L., Lambert, M. J., Vermeersch, D. A.,
client compliance and awareness. Brain Injury, Smart, D. W., Nielsen, S. L., & Hawkins, E. J.
20, 1271–1282. (2003). Improving the effects of psychotherapy:

Schönberger, M., Humle, F., & Teasdale, T. (2007). The use of early identification of treatment
The relationship between clients’ cognitive func- failure and problem-solving strategies in routine
tioning and the therapeutic working alliance in practice. Journal of Counseling Psychology,
post-acute brain injury rehabilitation. Brain 50, 59–68.

Injury, 21, 825–836. Whittal, M. L., Thordarson, D. S., & McLean, P. D.
Schmidt, F. L., & Le, H. (2005). Software for the (2004). Treatment of obsessive-compulsive
Hunter-Schmidt meta-analysis methods (Version disorder: Cognitive behavior therapy vs. exposure
1.1) [Computer software]. Iowa City: University and response prevention. Behaviour Research and
of Iowa, Department of Management & Therapy, 43,1559–1576.

Organization. Woods, C. M., Chambless, D. L., & Steketee, G.

Stein, M. D., Solomon, D. A., Herman, D. S., (2002). Homework compliance and behavior
Anthony, J. L., Ramsey, S. E., Anderson, B. J., therapy outcome for panic with agoraphobia and
et al. (2004). Pharmacotherapy plus psychother- obsessive compulsive disorder. Cognitive
apy for treatment of depression in active injec- Behaviour Therapy, 31, 88–95.

tion drug users. Archives of General Psychiatry, Yovel, I., & Safren, S. A. (2007). Measuring
61, 152–159. homework utility in psychotherapy: Cognitive-
Tryon, G. S., Blackwell, S. C., & Hammel, E. F. behavioral therapy for adult attention-deficit
(2007). A meta-analytic examination of client- hyperactivity disorder as an example. Cognitive
therapist perspectives of the working alliance. Therapy and Research, 31, 385–399.

Psychotherapy Research, 17, 629–642. Zane, N., Sue, S., Chang, J., Huang, L.,
Tryon, G. S., & Winograd, G. (2001). Goal Huang, J., Lowe, S., et al. (2005). Beyond ethnic
consensus and collaboration. Psychotherapy: match: Effects of client-therapist cognitive match
Theory, Research, Practice, Training, 38, 385–389. in problem perception, coping orientation, and
Tryon, G. S., & Winograd, G. (2002). Goal consensus therapy goals on treatment outcomes. Journal of
and collaboration. In J. C. Norcross (Ed.), Community Psychology, 33, 569–585.

t ryo n , w i n o g r a d 167

8 Positive Regard and Affirmation

Barry A. Farber and Erin M. Doolin

Author Note. We gratefully acknowledge time—and still an enormously influen-

the invaluable research assistance provided tial one, though currently cast in some-
by Alex Behn, Sarah Bellovin-Weiss, and what different (e.g., more evidence-based)
Valery Hazanov. terms—was that technical expertise on the
part of the therapist, especially in terms of
The book [Client-Centered Therapy] . . . choice and timing of interventions, is the
expresses, I trust, our conviction that essential discriminating element between
though scientists can never make effective and noneffective therapy. Under
therapists, it can help therapy; that though the sway of Rogers’ burgeoning influence
the scientific finding is cold and abstract, in the late 1950s and throughout the 1960s,
it may assist us in releasing forces that are the notion that the relationship per se was
warm, personal, and complex; and that the critical factor in determining therapeu-
though science is slow and fumbling, it tic success, took hold.
represents the best road we know to the Over the years, a great many studies have
truth, even in so delicately intricate an attempted to investigate Rogers’ claims
area as that of human relationships. regarding the necessary and sufficient con-
—Rogers, 1951, p. xi ditions of therapy. There is, then, a sub-
stantial body of research to draw upon in
Over 50 years ago, in what is now looking at the association between the ther-
considered a classic paper, Carl Rogers apist’s positive regard for his or her patients
(1957) posited that psychotherapists’ pro- and therapeutic outcome. However, as
vision of positive regard (nonpossessive detailed below, drawing firm conclusions
warmth), congruence (genuineness), and from these efforts has been difficult. The
empathy were the necessary and sufficient problems that typically plague the investi-
conditions for therapeutic change. Rogers gation of complex psychological issues have
had been developing these views for many been played out in this area as well: incon-
years, some of which were expressed as early sistent findings, small sample sizes, lack of
as 1942 in his seminal work, Counseling standardized measures, and lack of opera-
and Psychotherapy. Still, the publication tional definitions of the concepts them-
of the 1957 article seems to have catalyzed selves. In addition, as the Rogerian influence
a shift in the way that many thought about on clinical practice has diminished in the last
the putative mechanisms of psychothe- three decades––or, more accurately, has been
rapeutic change. The prevailing view at the incorporated into the psychotherapeutic

mainstream with little awareness or This observation suggests that Freud was
explicit acknowledgment (Farber, 2007)–– unaware of, or at least underappreciated,
empirical studies based on Rogerian con- what may well have been the most potent
cepts have also waned. Similarly, the focus elements of his approach—that along with
of research and theory has shifted away whatever positive effects accrue as a result
from the individual contributions of each of accurate interpretations, psychoanalytic
of the participants in therapy toward a con- success has arguably always been based
sideration of the alliance or therapeutic substantially on the undervalued ability
relationship—what each member of the of the analyst to be empathic and, even
dyad contributes to the ongoing, interac- more to the point of this chapter, to be
tive process of the work. supportive and positively regarding of his
Whereas consideration of the therapeu- or her patients.
tic relationship as mutative began with In this chapter, we review Rogers’
Rogers, therapists of varying persuasions, ideas about the concept of positive regard
even those from theoretical camps that and discuss how the use of multiple terms
had traditionally emphasized more techni- (including positive regard, affirmation, respect,
cal factors, have begun to acknowledge the warmth, support, and prizing) has led to
importance of the relationship. Behaviorists conceptual confusion as well as empirical
and cognitive-behaviorists now suggest that difficulties in determining the link between
a good relationship may facilitate the pro- this phenomenon and therapeutic out-
vision of their technical interventions come. The emphasis of this chapter is on
(e.g., Beck, 1995), and many psychoana- meta-analytically reviewing the findings of
lysts have shifted their clinical perspective those empirical studies that have investi-
to emphasize “relational” factors (Mitchell gated the relation between therapist sup-
& Aron, 1999; Wachtel, 2009). But even port and treatment outcome in individual
before these relatively recent developments, psychotherapy. Most studies of positive
there is evidence to suggest that Freud’s regard are framed within a Rogerian (per-
psychoanalytic cases were only successful son-centered) paradigm; however, as noted
when he was supportive and positively above, nearly all schools of therapy now
regarding. As Breger (2009) has noted: either explicitly or implicitly promote
the value of this basic attitude toward
When Freud followed these [psychoanalytic] patients. Thus, the results of these studies
rules his patients did not make progress. have implications for the conduct not only
His well-known published cases are of person-centered therapists, but for virtu-
failures . . . in contrast are patients like ally all psychotherapists.
Kardiner and others—cases he never
wrote or publicly spoke about—all of Definitions and Measures
whom found their analyses very helpful.
With these patients, what was curative To the extent that the therapist finds
was not neutrality, abstinence, or himself experiencing a warm acceptance
interpretations of resistance, but a more of each aspect of the client’s experience
open and supportive relationship, as being a part of that client, he is
interpretations that fit their unique experiencing unconditional positive
experiences, empathy, praise, and the regard . . . it means there are no conditions
feelings that they were liked by their of acceptance . . . it means a ‘prizing’ of
analyst. (p. 105). the person . . . it means a caring for the

fa r b e r, d o o l i n 169
client as a separate person (Rogers, 1957, of therapist affirmation. We will use the
p. 101). phrase positive regard to refer to the general
constellation of attitudes encompassed by
From the beginning of his efforts to explicate this and similar phrases.
the essential elements of client-centered Further confusing the conceptual issues
therapy, Rogers focused on positive regard at play here, Rogers’ focus on accepting and
and warmth: “Do we tend to treat individ- affirming the client has, from the outset,
uals as persons of worth, or do we subtly been conflated with an emphasis on empa-
devaluate them by our attitudes and behav- thy and genuineness. The therapist’s attempt
ior? Is our philosophy one in which respect to “provide deep understanding and accep-
for the individual is uppermost?” (1951, tance of the attitudes consciously held at
p. 20). Implicit in this statement is his the moment by the client” could only be
disapproval of what he perceived as the accomplished by the therapist’s “struggle to
arrogance of, and strict hierarchical dis- achieve the client’s internal frame of refer-
tinctions between, psychotherapists and ence, to gain the center of his own percep-
patients held by the psychoanalytic com- tual field and see with him as perceiver”
munity at that time. Rogers did not believe (1951, pp. 30–31). Rogers seems to be sug-
that anyone, including a therapist, could gesting here that positive regard (including
be more expert or knowledgeable about the component of acceptance) can best be
a client than the client him or herself. He achieved through empathic identification
did not believe that a therapist’s neutrality, with one’s client. In a similar vein, Rogers
dispassionate stance, or even intellectual suggested that the therapist’s genuineness
understanding could facilitate a client’s or congruence was a prerequisite for his or
growth—no matter how astute the inter- her experience of positive regard and empa-
pretations emanating from such a therapy thy (Rogers & Truax, 1967).
might be. Instead, he believed that treat- Further problems with the concept of
ing clients in a consistently warm, highly positive regard have been identified (e.g.,
regarding manner would inevitably allow Lietaer, 1984). One is that there may be
them to grow psychologically, to fulfill their an inherent tension between this attitude
potential. and that of genuineness; that is, therapists’
To this day, agreeing on a single phrase own conflicts inevitably affect what they
to refer to this positive attitude remains can and cannot truly accept or praise in
problematic. It is most often termed posi- others. A second, related problem is that it
tive regard but early studies and theoretical is unlikely that any therapist can provide
writings preferred the phrase nonpossessive constant doses of unconditional positive
warmth. In his famous filmed work with regard in that we all reinforce selectively. As
Gloria (Shostrom, 1965), Rogers struggled Rogers himself anticipated:
to find a single phrase to illuminate this
concept: it is, he said, “real spontaneous The phrase ‘unconditional positive regard’
praising; you can call that quality accep- may be an unfortunate one, since it
tance, you can call it caring, you can call it a sounds like an absolute, an all-or-nothing
non-possessive love. Any of those terms tend dispositional concept . . . From a clinical
to describe it.” Some reviews of “accep- and experiential point of view I believe
tance, nonpossessive warmth, or positive the most accurate statement is that
regard” (Orlinsky, Grawe, & Parks, 1994, the effective therapist experiences
p. 326) grouped them under the category unconditional positive regard for the

170 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
client during many moments of his a person,” “I feel appreciated by her,” and
contact with him, yet from time to time “she is friendly and warm toward me.”
he experiences only a conditional positive Representative negative items include
regard—and perhaps at times a negative “I feel that she disapproves of me,” “She
regard, though this is not likely in is impatient with me,” and “At times she
effective therapy. It is in this sense that feels contempt for me.”
unconditional positive regard exists as Unconditionality of Regard is explained
a matter of degree in any relationship. by Barrett-Lennard (1986, p. 443) in terms
(p. 101). of its stability, “in the sense that it is
not experienced as varying with other or
How can one, then, assess a therapist’s otherwise dependently linked to particular
level of positive regard without implicitly attributes of the person being regarded.”
measuring empathy or genuineness as well? Examples of positively worded items: “How
In fact, reading transcripts of Rogers’ work much he likes or dislikes me is not altered
(e.g., Farber, Brink, & Raskin, 1996) makes by anything that I tell him about myself ”;
clear how difficult it is to tease out pure “I can (or could) be openly critical or appre-
examples of positive regard. Rogers is ciative of him without really making him
consistently “with” his clients, testing his feel any differently about me.” Examples
understanding, clarifying, and intent on of negatively worded items: “Depending
entering and grasping as much as possible on my behavior, he has a better opinion of
the client’s experiential world. For these me sometimes than he has at other times”;
reasons, most research focusing on the “Sometimes I am more worthwhile in his
effects of therapist positive regard have eyes than I am at other times.”
used measures, typically either the Barrett- Truax developed two separate instru-
Lennard Relationship Inventory (BLRI; ments for the measurement of Rogers’ facili-
1964, 1978) or the Truax Relationship tative conditions. One was a set of scales
Questionnaire (Truax & Carkhuff, 1967), to be used by raters in their assessment of
that include items reflecting multiple, over- these conditions as manifest in either live
lapping, relational elements. observations or through tape recordings of
The BLRI consists of 64 items across sessions. There are five stages on the scale
four domains (Level of Regard, Empathic that measures Nonpossessive Warmth. At
Understanding, Unconditionality of Regard, Stage 1, the therapist is “actively offering
Congruence). Eight items are worded posi- advice or giving clear negative regard”
tively, eight negatively in each domain; (Truax & Carkhuff, 1967, p. 60); at Stage 5,
each item is answered on a +3 (yes, strongly the therapist “communicates warmth with-
felt agreement) to −3 (no, strongly felt out restriction. There is a deep respect for
disagreement) response format. This ins- the patient’s worth as a person and his
trument can be used by the client, thera- rights as a free individual” (p. 66).
pist, or both. Both Level of Regard and The second instrument developed by
Unconditionality have been used in research Truax, The Relationship Questionnaire,
studies to investigate the influence of posi- was to be used by clients. This measure
tive regard. Level of Regard, according to consists of 141 items marked “true” or
Barrett-Lennard (1986, p. 440–441), “is “false” by the client. Of these items, 73 are
concerned in various ways with warmth, keyed to the concept of nonpossessive
liking/caring, and ‘being drawn toward’.” warmth; it is noteworthy, however, that
Positive items include “she respects me as many of these items are also keyed to the

fa r b e r, d o o l i n 171
other two facilitative conditions (genu- High scores on this dimension reflect the
ineness and empathy). That is, a “true” therapist’s ability to teach or encourage a
response on one item may count toward a patient in a kind or positive manner.
higher score on more than one subscale. Whereas the “bond” component of
Representative items on the Nonpossessive various alliance measures (e.g., Horvath &
Warmth scale: “He seems to like me no Greenberg, 1989; Tracey & Kokotovic,
matter what I say to him” (this item is also 1989) contains aspects of positive regard
on the “genuineness” scale); “He almost phenomena that have been elucidated
always seems very concerned about me”; above, its items primarily assume an inter-
“He appreciates me”; “I feel that he really action between patient and therapist, one
thinks I am worthwhile”; “even if I were to that reflects the contributions and charac-
criticize him, he would still like me”; and teristics of each. Thus, results from studies
“whatever I talk about is OK with him.” using alliance measures were not included
In addition to these scales, therapist pos- in our meta-analysis.
itive regard has been assessed via instru-
ments designed primarily to measure the Clinical Examples
strength of the alliance. In particular, the The case examples below have been pur-
Vanderbilt Psychotherapy Process Scale posely drawn from disparate theoretical
(VPPS) has been used in this manner. perspectives. Although the concept of posi-
The VPPS is “a general-purpose instru- tive regard originated with Rogers, the pro-
ment designed to assess both positive and vision of this facilitative condition can and
negative aspects of the patient’s and the does occur in the work of practitioners of
therapist’s behavior and attitudes that are quite distinct therapeutic traditions.
expected to facilitate or impede progress
in therapy” (Suh, Strupp, & O’Malley, Case Example 1
1986, p. 287). Each of 80 items is rated by
clinical observers on a 5-point, Likert-type Client: I can outsmart people. I won’t
scale, either from the actual therapy sessions be taken advantage of. I call the
or from video- or audiotapes of therapy. shots.
Factor analyses of these items have yielded Therapist: It seems important for you to
eight subscales, one of which, Therapist be dominant in every relationship.
Warmth and Friendliness, closely approxi- Client: Yes. I don’t show emotion and
mates the concept of positive regard. The I don’t put up with it in anyone else.
specific therapist attributes rated in this I don’t want someone to get all
subscale include “involvement” (the thera- hysterical and crying with me. I don’t
pist’s engagement in the patient’s experi- like it.
ence), “acceptance” (the therapist’s ability Therapist: How did you learn that
to help the patient feel accepted), “warmth being emotional is a sign of
and friendliness,” and “supportiveness” (the weakness?
therapist’s ability to bolster the patient’s Client: I don’t know.
self-esteem, confidence, and hope). Therapist Therapist: What if you meet your
positive regard has also sometimes been intellectual match, if you can’t
measured through the use of Structural “outsmart” them?
Analysis of Social Behavior (SASB; Client: (silence)
Benjamin, 1984), specifically through the Therapist: Okay, what if someone got
dimension of Helping and Protecting. to you through your feelings?

172 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Client: Last week you did. It bothered therapist might do) but in doing so, explic-
me all day. itly conveys the fact that he cares for this
Therapist: That you were weak? patient far more than she imagines to be
Client: Yeah. the case.
Therapist: I didn’t see you as submissive
or weak. In fact, since showing
emotion is so difficult for you I saw it Case Example 3
as quite the opposite. Client: It really hurts when I think
In this example, the therapist, primarily about the fact that it is over.
psychodynamic in orientation, initially Therapist: Yes, of course it hurts. It
tries to get the patient to open up about hurts because you loved him and it
his past and discuss his “faulty strategy” did not work out. It shows, I think,
of dominating relationships. It appears as your capacity to love and to care. But
if they are about to discuss transference it also hurts to have that ability.
issues. However, the therapist shifts at the Client: I don’t think I’ll ever feel
end, perhaps intuitively sensing that what that way.
would be most effective for this patient (at Therapist: Right now it may be
least at this moment) is a statement of true important for you to protect yourself
positive regard. Thus, the therapist is affirm- with that feeling. Perhaps we can
ing, suggesting that she views the client not look at what you have learned about
as weak or submissive but rather the oppo- yourself and your needs and the kind
site, as perhaps brave for doing something of man who would be right for you.
that was difficult for him. Client: What do you mean?
Therapist: I mean that you have a great
ability to love. But what can you
Case Example 2 learn about what you need in a man
that [Tom] lacked?
“You’re reading me entirely wrong. I don’t Client: I guess I learned not to get
have any of those feelings. I’ve been pleased involved with a married man.
with our work. You’ve shown a lot of Therapist: What do you think led you
courage, you work hard, you’ve never to think you’d be able to handle
missed a session, you’ve never been late, being involved with a married man?
you’ve taken chances by sharing so many Client: Well, after my marriage ended,
intimate things with me. In every way I guess I didn’t want to get too
here, you do your job. But I do notice that attached. So I thought that being
whenever you venture a guess about how involved with someone who is
I feel about you, it often does not jibe with married would keep me from being
my inner experience, and the error is hurt.
always in the same direction: You read Therapist: Perhaps you’ve learned that
me as caring for you much less than I do” you have such a strong ability to love
(Yalom, 2002, p. 24). that you can’t compartmentalize your
feelings that way (Leahy, 2001, p. 82).
In this example, Yalom, an existential
therapist, not only offers assumedly accu- In this example, Leahy, a cognitive thera-
rate feedback to his patient on her interper- pist, is not only consistently empathic (“of
sonal tendencies (much like a psychoanalytic course it hurts”) and not only attempting

fa r b e r, d o o l i n 173
to teach his patient something about her- Client: I actually surprised myself.
self and her needs and choices, but he It didn’t even feel so risky. I just
also makes sure that he contextualizes his “went with it.” He smiled,
interventions in a supportive, caring way, I smiled back . . .
emphasizing his patient’s “strong ability to Therapist: Good. And that was so
love.” courageous of you . . ..
Client: Yeah, that’s me, I guess. My
Case Example 4 mother used to say that my refusal
to not give in irritated the hell out
Client: (smiles) I think I’m having male of my father. I’m sure I disappointed
menopause. him as a son . . . But sometimes,
Therapist: (smiles) OK, but I think I’m glad he’s gone. I feel guilty
you’ll need to explain that condition thinking and saying it but if I had
to me. to choose which parent would go
Client: (laughs) I met this great guy first, I’m not sorry it was him and
coming out of the supermarket. glad it wasn’t my mother. I really
And we chatted right there on the still need her.
street and we exchanged phone Therapist: I know. And you know,
numbers. He probably won’t call. I want to tell you how much
He’s a lot younger than I am so I appreciate your honesty in
I don’t think he’s really interested, allowing yourself to think about
but, hey, I actually had a daylight these things that are sometimes
conversation with an attractive man hard to think about.
and he knows my name. Now that’s Client: Thanks. (Pause.) I hope he
something, huh? calls me.
Therapist: Yes, it is. And something Therapist: I hope so too.
different for you. Here, a relationally oriented psychody-
Client: Yeah, I’m feeling less creature- namic clinician banters somewhat (“you’ll
like. More human these days. have to explain that condition to me”) as
Like coming out a means to be connected and supportive.
from under a rock. Oh, I finished my Moreover, she values her patient’s efforts
painting . . . the one with the lost to change (“that was so courageous of
boy. I thought about what you said you”) and gives him credit for the work
about the boy feeling lost . . . When he’s doing in the here-and-now of the clini-
I was finishing the painting, I felt cal setting (“how much I appreciate your
like . . . it’s almost like you and honesty”).
I came up with that together.
Therapist: I feel that, too. I think that Case Example 5
we each contribute to our work here
together. The accomplishment of Client: Yeah, I don’t feel like it’s [filling
finishing the painting is all yours, out diary cards every day] for me.
though. And talking to a man you I don’t want to wake up every day,
find attractive, giving him your name and go “Oh I felt like suicide last
and phone number . . . Sounds well, Tuesday! Oh my god, I was sad last
something commonplace, but not week!” I don’t want to keep
for you. Not in a long time. remembering!

174 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Therapist: Oh, okay, so you want one Case Example 6
of those therapies where you don’t
remember things? Client: I feel like there are people who
Client: No, I just don’t want to keep do care and accept me. I do, but.
bringing it up all the time. “Oh, Therapist: But the person that
I was raped on so-and-so date, can’t accept and value you, is
let me remember what I felt at actually you.
the time.” Client: Yes, mostly.
Therapist: Yeah, it’s so painful to bring Therapist: It seems the person who
up this stuff. Why would anyone is hardest on you is you.
want that? Client: Yes. No one else would be as
Client: Yeah, exactly! cruel to me as I am.
Therapist: Now here’s the dilemma. Therapist: And make such harsh
We could not talk about your judgments, you’re pretty tough
problems, and if this would take on yourself.
away your pain and misery, I’d be Client: Yes, I wouldn’t judge my friends
all for it. On the other hand, if we the way I judge myself.
help you figure out how to tolerate Therapist: No, you’re not a very good
your bad feelings, then you won’t friend to yourself.
have to rely on your pain medicine Client: No, I wouldn’t treat anyone
or resort to thinking of killing the way I treat me.
yourself when these feelings Therapist: Maybe because you can
come up. see what is lovable in them, but
Client: But the feelings are horrible! not in yourself. To you, you’re
What am I supposed to do, just unlovable.
wave a magic wand to make Client: Maybe there are small pieces
them go away? You make it sound of me that are lovable.
so easy. Therapist: (pause) So there are parts
Therapist: It’s not easy at all. This is of you that you see as OK, as worthy
incredibly tough and painful for you, of being loved.
and I also believe you have what it Client: Yes, I guess. The child in me,
takes to do it. (Adapted from the child that struggled and survived.
McMain, Korman, & Dimeff, 2001, She, I, can still be playful and fun
p. 196). and warm.
In this dialogue, a dialectical behavior Therapist: Those are very wonderful
therapist offers a supportive qualities.
statement (“I believe you have Client: She’s strong, a survivor.
what it takes to do it”) that has Therapist: She’s a part of you that you
much in common with the can hold on to.
comments of the relationally Client: Yes.
oriented therapist in the previous Therapist: Do you think she’d judge
example. Here, the therapist’s you so harshly?
empathic response (“it’s not easy Client: No, she loves me.
at all”) is followed up by a more Therapist: To this special child part of
explicit statement of positive you, none of you is unforgivable.
regard. Client: No, she loves all of me.

fa r b e r, d o o l i n 175
In this final example, the client-centered Wolfe, 1978) later pointed out—namely,
therapist is clearly conveying to the patient that there are multiple ways of understand-
that she is worthy of respect and love. The ing such complex data. For example, of 108
therapist’s positive regard for the patient correlations noted in Truax and Mitchell’s
may allow her to begin to view herself report, only 34 were reported as significantly
as the therapist does. These last few exam- positive. While none of these correlations
ples are prime illustrations of the multi- were significantly negative, relatively few
ple aspects of positive regard, including were significantly positive.
affirmation, trust, understanding, warmth, In a follow-up review, Mitchell, Bozarth,
interest, and respect. and Krauft (1977) evaluated 11 studies
that investigated the relationship between
Previous Reviews positive regard (here again termed nonpos-
Before describing the results of our meta- sessive warmth) and treatment outcome.
analysis, we summarize several previous According to these authors, at most four
reviews of the association between positive of these studies offered support for the
regard and outcome. The first such effort proposition that higher levels of therapist-
was by Truax and Carkhuff (1967) in their provided warmth lead to better outcome.
book, Toward Effective Counseling and The following year, Orlinsky and Howard
Psychotherapy. Many of the studies they (1978) reviewed 23 studies, concluding
cited failed to report the separate associa- that approximately two thirds of these
tions of each of Rogers’ facilitative condi- indicated a significant positive association
tions to outcome, focusing instead on the between therapist warmth and therapeutic
aggregate results of all three conditions outcome, with the remaining one third
taken together. They did, however, review showing mostly null results. However, they
10 studies from which conclusions could also added several caveats, notably that
be drawn on the effects of positive regard the uneven quality and methodological
alone on therapeutic outcome, finding that flaws in the research made any firm con-
8 of these supported the hypothesis that clusions suspect. Their conclusion: “If they
nonpossessive warmth (the preferred term [warmth and empathy] do not by them-
at that time) is significantly associated with selves guarantee a good outcome, their
therapeutic improvement. presence probably adds significantly to the
Next, Bergin and Garfield’s first (1971) mix of beneficial therapeutic ingredients,
edition of Handbook of Psychotherapy and and almost surely does no harm” (p. 293).
Behavior Change included a chapter by As part of a comprehensive review of
Truax and Mitchell that summarized the process and outcome in psychotherapy,
results of 12 studies (involving 925 clients) Orlinsky and Howard (1986) conducted
that included nonpossessive warmth. The separate reviews of studies evaluating the
authors contended that the evidence was effects of therapist support and therapist
quite positive in regard to the relation- affirmation. They identified 11 studies that
ship between warmth and therapeutic out- included a support/encouragement variable;
come, noting that there was a statistically within this group of studies they focused
significant relationship between this vari- on 25 separate findings. Their conclusion:
able and a total of 34 specific outcome “Although 6 of the 25 are significantly
measures. Nevertheless, it is important to positive findings and none are negative,
reiterate what others (Mitchell, Bozarth, more than three-quarters show a null asso-
Truax, & Krauft, 1973; Parloff, Waskow, & ciation between specific therapist efforts to

176 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
give support and patient outcome” (p. 326). and outcome. Second, the results of the
In addition, the authors identified 94 find- 16 studies analyzed in that chapter were
ings on the association between thera- essentially evenly split between positive and
pist affirmation (essentially warmth, caring, nonsignificant effects. That is, 49% (27/55)
and acceptance) and outcome, with more of all reported associations were signifi-
than half (53%) demonstrating a signifi- cantly positive and 51% (28/55) were non-
cant relationship between these sets of significant. However, the authors noted
variables. Underscoring their emphasis on that the majority of nonsignificant find-
considering the perspective of raters, they ings occurred when an objective rater
noted that “the proportion of positive find- (rather than the therapist or patient) evalu-
ings is highest across all outcome catego- ated therapeutic outcome. Third, confirm-
ries when therapist warmth and acceptance ing the pattern noted by previous reviewers,
are observed from the patient’s process Farber and Lane found that when the
perspective” (p. 348). That is, in 30 cases patient rated both the therapist’s positive
where the patient’s ratings of therapist posi- regard and treatment outcome, a positive
tive regard were used, 20 outcome scores association between these variables was
were positively correlated with these ratings especially likely. Lastly, the effect sizes for
(aggregated over the outcome perspectives the significant results tended to be modest,
of patient, therapist, rater, and objective with the larger effect sizes occurring when
score), and no outcome scores (regardless positive regard was assessed in terms of its
of the source) were significantly negatively association to length of stay in therapy
correlated with patient ratings of therapist rather than outcome per se (for example,
positive regard. Najavits & Strupp, 1994).
In 1994 Orlinsky and colleagues studied
this general phenomenon under the rubric Meta-Analytic Review
of therapist affirmation, explained by the Literature Search and Study Selection
authors as a variable that includes aspects To find studies that documented a relation-
of acceptance, nonpossessive warmth, or ship between positive regard and outcome
positive regard. They found that 56% of in psychotherapy, we used the PsycINFO
the 154 results reviewed were positive, and database. Main root terms searched in the
that, again, the findings based on patients’ title or the abstract were positive regard,
process perspective (the patient’s rating warmth, nonpossessive warmth, therapist
of the therapist’s positive regard) yielded affirmation, unconditional positive regard,
even a higher rate of positive therapeutic affirmation, acceptance, and unconditional
outcomes, 65%. “Overall,” Orlinsky et al. regard. All these terms were crossed with
(p. 326) concluded, “nearly 90 findings psychotherapy, searching for the following
indicate that therapist affirmation is a sig- root terms in the title or the abstract: psy-
nificant factor, but considerable variation chotherapy, therapy, counseling, and client-
in ES [effect size] suggests that the contri- centered. Additional studies were located by
bution of this factor to outcome differs running a search with the root term “Barrett-
according to specific conditions.” Lennard” since this is the most widely used
Lastly, in the previous review of positive instrument to assess positive regard.
regard for this volume, the authors (Farber Our specific inclusion criteria were as
& Lane, 2002) highlighted several patterns. follows: (a) the study identified positive
First, no post-1990 study reported a nega- regard as either unconditional regard, posi-
tive relationship between positive regard tive regard, warmth, nonpossessive warmth,

fa r b e r, d o o l i n 177
affirmation, or acceptance; (b) positive 1991) was excluded from the revised analy-
regard (in any of these forms) was con- sis because it could no longer be located.
sidered as a predictor of outcome in the After scanning the literature with these
study; (c) the study reported quantitative criteria in mind, 44 studies were selected
outcome data and relevant statistics (e.g., for review of which 18 were found to be
correlations between positive regard ratings entirely consistent with these criteria and
and treatment outcome or mean outcome thus were included in the meta-analysis.
comparisons between groups with differen-
tial positive regard ratings) that could be Coding Potential Moderators
used to calculate effect sizes; (d) patients were The moderator variables were broken into
adults or adolescents; and (e) treatment was three categories: study characteristics, char-
individual psychotherapy. In addition, stud- acteristics of sample/treatment, and thera-
ies that reported the contribution of posi- pist factors. All studies were coded by the
tive regard to other process or relational junior author; a coding manual is available
variables and thus indirectly to treatment upon request.
outcome were excluded from our analysis. Study characteristics included: (a) publi-
For example, studies that looked at positive cation status (e.g., published article,
regard as a component of other predictor book chapter, or unpublished dissertation);
variables (e.g., empathy, therapeutic alli- (b) sampling (whether the sample was
ance) or were part of an aggregated factor random or a convenience sample); (c) rater
associated with outcome were not included perspective for both the independent and
in the analysis. In fact, many early studies dependent variables; and (d) total number of
looked at positive regard in the context participants. Characteristics of the sample/
of the entire constellation of facilitative treatment that were coded were: mean age,
conditions posed by Rogers—congruence, percentage of women, percentage of racial/
empathy, positive regard—without explic- ethnic minorities, frequency of treatment,
itly reporting the impact of positive regard average number of sessions involved in
as an individual variable. treatment, measure of relationship element,
In addition to searching for relevant point in time that the relationship element
studies in this manner, we consulted was assessed, and the theoretical orienta-
the 2002 chapter to determine which of tion that informed said treatment.
those 16 studies met our current criteria. The specific therapist factors coded for
Six potential articles (Gaston et al., 1990; this analysis involved mean age, percentage
Hynan, 1990; Klein, 2002; Meyer, 1990; of women, number of therapists used, and
Rothman, 2007; Schauble & Pierce, 1974; composition of therapists (e.g., trainees;
Schut, Castonguay, Flanagan, & Yamasaki, four years postgraduation).
2005) were excluded because there was
not enough information presented in the Effect Size Coding
results to compute the appropriate effect Because the purpose of this meta-analysis
size, and any contact with the original was to examine the relation between thera-
authors yielded no assistance. One article pist positive regard and treatment outcome,
(Russell, Bryant, & Estrada, 1996) included a simple correlation, r, was obtained to
in our 2002 review was excluded because measure the effect for each study. The effect
it did not explicitly examine the relation- sizes for several studies had to be recom-
ship between positive regard and therapeu- puted using the data the authors provided
tic outcomes. Another article (Schindler, and then converted to r (per Cooper,

178 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Hedges, & Valentine, 2009). After each of publication, total sample size, and 95%
study was coded for the moderator vari- confidence limits are provided in Table 8.1.
ables, effect sizes were again computed for The aggregate effect size was 0.27, indicat-
each of the 18 studies. Additionally, if there ing that positive regard has a moderate
was more than one effect size per study, association with psychotherapy outcomes;
within-study aggregation was performed. only two of the 18 studies had negative
A new statistical package available online effect sizes. Additionally, the 95% confi-
aided in the statistical analysis for this dence interval (CI) did not include zero
project (see Del Re, 2010; Del Re & Hoyt, (CI = 0.16, 0.38), which demonstrates that
2010). the effect of positive regard on outcome is
significantly different from zero.
Results To assess whether there were differences
A total of 18 effect sizes were yielded (after between these 18 studies above and beyond
aggregation) and were included in the sampling error, a homogeneity test was
analysis. A complete list of the studies conducted. Using the homogeneity statis-
included in the analysis, their authors, date tic, Q (Hedge, 1982), the assumption that

Table 8.1 Effect Sizes between Positive Regard and Outcome

Study n Study effect 95% Confidence limits for r
size (r) [lower, upper]
Bachelor (1991) 47 .49 [.24, 0.68]
Chisholm (1998) 173 −.04 [−.19, 0.11]
Coady (1991) 9 .71 [.09, 0.93]
Conte, Ratto, Clutz, & Karasu (1995) 138 .29 [.13, 0.44]
Cramer & Takens (1992) 37 .37 [.05, 0.62]
Eckert, Abeles, & Graham (1988) 77 .35 [.14, 0.53]
Garfield & Bergin (1971) 38 −.15 [−.45, 0.18]
Green & Herget (1991) 11 .83 [.46, 0.96]
Hayes & Strauss (1998) 32 .31 [−.04, 0.59]
Henry, Schacht, & Strupp (1990) 14 0 [−.53, 0.53]
Keijsers, Hoogduin, & Schaap (1994) 40 .12 [−.20, 0.42]
Litter (2004) 8 .53 [.29, 0.71]
Najavits & Strupp (1994) 12 .75 [.32, 0.93]

Quintana & Meara (1990) 48 .02 [−.26, 0.31]

Rabavilas, Boulougouris, & Perissaki (1979) 36 .09 [−.25, 0.41]

Sells, Davidson, Jewell, Falzer, & Rowe (2006) 83 .33 [.12, 0.51]
Williams & Chambless (1990) 33 .20 [−.15, 0.51]
Zuroff & Blatt (2006) 191 .20 [.06, 0.33]
Overall n or r 1067 .27 [.16, 0.38]
n of studies/samples 18

fa r b e r, d o o l i n 179
the studies selected were sampled from the willingness to deviate from the conventions
same population (i.e., were homogenous) of psychodynamic treatment. However,
was rejected, Q (17) = 50.52, p = 0.000. given that all of these moderators were
This indicates that there is a large amount significant, they all contribute in some way
of heterogeneity in the studies due to dif- to the large amount of heterogeneity in the
ferences among the studies. This degree overall effect of positive regard on thera-
of heterogeneity implies that the over- peutic outcomes.
all effect varies as a function of study
characteristics. Continuous Moderator Variables
We also conducted several univariate con-
Moderators tinuous moderator analyses; none were sig-
Categorical Moderator Analyses nificant at the .05 level However, because
In order to account for the systematic vari- of its increasing importance in contempo-
ance present in this sample of studies, we rary psychological research, we note that
conducted several univariate categorical the percentage of racial-ethnic minorities
moderator analyses. As Table 8.2 indicates, as a patient characteristic (R2 = .42, F (1, 7)
the following moderators were significant = 4.37, p = .08) approached significance.
(i.e., demonstrated significant heterogeneity The numbers indicated that as the percent-
in their aggregate effect sizes): publication age of racial/ethnic minorities increases in
status, rater perspective, origin of sample, the patient sample, the overall effect size
measure used to assess positive regard, time also increases. If this finding proves to be
in treatment when positive regard was mea- robust, it has implications for the field of
sured, and type of treatment. These vari- therapist-client matching, as well as multi-
ables moderate the overall effect of positive cultural competence (see Smith, Rodriguez,
regard on therapeutic outcome. In other & Bernal, Chapter 16, this volume).
words, each of these moderators accounts
for some portion of the unaccounted-for Patient Contribution
variance discussed in the previous section Although no patient characteristics emerged
on the overall effect. Most notably, as indi- as significant moderators in our analyses
cated in Table 8.2, the overall effect of posi- of the data, we hypothesize that some
tive regard on outcome tends to be higher patient factors, not assessed in the studies
when studies are published in journal arti- we examined, are likely to affect the thera-
cles, or when the type of treatment pro- pist’s provision of positive regard and the
vided is psychoanalytic/psychodynamic. extent to which this increases the likelihood
In regard to this last finding, our hypothesis of therapeutic success. First, most thera-
is that patients engaged in traditional pists’ behavior is a function, among other
(rather than more contemporary, relational) things, of the characteristics of the patients
psychodynamic treatment were particularly they work with. Simply put, some patients
affected by their therapists’ occasional and are more easily liked and therefore elicit
perhaps unexpected displays of support more affirmation than others. Patients who
and positive regard. In a manner analo- themselves are warm, empathic, and dis-
gous to the power of relatively infrequent closing are more easily liked and affirmed.
therapist disclosures wherein “less is more” Just as disclosure begets disclosure (Jourard,
(Knox & Hill, 2003), these patients’ treat- 1971), it is quite likely that warmth begets
ment outcomes may have been influenced warmth. Conversely, demanding, resistant,
significantly by their therapists’ ability and or angry patients can be difficult to like or

180 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
Table 8.2 Significant Moderators
Moderator No. of studies (k) Effect size (r) 95% Confidence limits for r
[lower, upper]
Publication status
Journal article 16 .26 [.20, 0.32]
Unpublished dissertation 2 .09 [−.05, 0.22]
Rater perspective
Patient 7 .29 [.20, 0.37]
Non-participant rater 6 .05 [−.06, 0.17]
More than one perspective 4 .46 [.29, 0.60]
Not specified 1 .20 [.17, 0.28]
Origin of Sample
University setting 4 .07 [−.04, 0.18]
Part of larger study 7 .27 [.17, 0.37]
Hospital setting 1 .29 [.13, 0.44]
General outpatient setting 4 .26 [.11, 0.40]
K–12 setting 1 .53 [.28, 0.70]
Not specified 1 .09 [−.25, 0.41]
Measure Used
VPPS: TWFS 1 .49 [.24, 0.68]
SASB: H&P 3 .23 [.02, 0.42]
TSS 1 .29 [.13, 0.44]
VPPS 1 .53 [.28, 0.70]
Measure created for study 3 .30 [.08, 0.50]
RSTCP 1 .00 [−.53, 0.53]
TRS 1 .20 [.15, 0.51]
BLRI 2 .24 [.12, 0.35]
More than one measure 2 .01 [−.14, 0.15]
Relationship 1 .12 [−.20, 0.42]
Other 2 .20 [.01, 0.37]
Time when positive regard was measured
Pre- to post- 3 .16 [−.02, 0.32]
Post-treatment 7 .37 [.27, 0.46]
After second and sixth sessions 1 .83 [.46, 0.96]
One month and three years 1 .31 [−.05, 0.59]
More than one of the above-listed time 3 .01 [−.11, 0.14]
Intake, four, eight, twelve, and sixteen 1 .20 [.06, 0.33]
Six and twelve months 1 .33 [.12, 0.51]
Follow-up 1.4 years later 1 .09 [−.25, 0.41]
Theoretical orientation/Type of treatment
Psychoanalytic/psychodynamic 4 .52 [.35, 0.65]
Combination/eclectic 7 .12 [.03, 0.20]
Not specified 3 .31 [.20, 0.42]
Peer-based 1 .33 [.12, 0.51]
In-vivo/exposure 3 .14 [−.06, 0.32]
VPPS: TWFS = Vanderbilt Psychotherapy Process Scale: Therapist Warmth and Friendliness; SASB: H&P = Structured Analysis of Social
Behavior: Helping & Protecting; TSS = Therapist Satisfaction Scale; VPPS = Vanderbilt Psychotherapy Process Scale; RSTCP = The Rating
Scale of Therapy Change Processes; TRS = Therapist Rating Scale; BLRI = Barrett-Lennard Relationship Inventory.

181 fa r b e r, d o o l i n 181
affirm (see Winnicott, 1949). Thus, we data indicate a greater overall treatment
suspect that those with Axis II pathologies, effect of positive regard when more than
especially individuals with borderline or one perspective (e.g., patient, therapist, and
narcissistic disorders, are less likely to con- nonparticipant rater) was assessed. These
sistently evoke positive regard from their multiple perspectives may serve as reliabil-
therapists. Many difficult patients are test- ity checks on the accuracy, including the
ing their sense of the world (e.g., their potential underevaluation, of the patient’s
lovableness), simultaneously desperate to perspective on the provision of therapist
have their worst fears unconfirmed but positive regard.
overdetermining through their behavior Our data indicated that the effects of
that they will be reconfirmed (Weiss & positive regard increased as a function of
Sampson, 1986). the racial-ethnic composition in a study,
A related client characteristic that may although this association was statistically
influence a therapist’s tendency to be posi- weak. Thus, we tentatively hypothesize that
tively regarding is the nature of the client’s therapists’ provision of positive regard may
needs at a particular point in therapy. For be a salient factor in treatment outcome
example, patients suffering acutely from when non-minority therapists work with
any of the many variants of depression, minority clients. In such cases, the possibil-
or dealing with the aftermath of a recently ity of mistrust and of related difficulties—
experienced trauma, may explicitly ask stemming in large part from our nation’s
for or more subtly indicate their need for troubled racial history as well as traditional
intensive doses of positive regard and affir- neglect of minority clients by the mental
mation. These requests may range from health community--may be attenuated by
“please tell me I’m OK,” to “I really need clear indications of the therapist’s positive
your support now,” to “no one cares about regard, in turn facilitating the likelihood of
me at all.” A third possible factor here a positive outcome (Sue & Sue, 1999).
is motivational status—that is, a patient’s
current stage of change (see Norcross, Krebs, Limitations of the Research
& Prochaska, Chapter 14, this volume). Our database was restricted to 18 studies, a
Patients who are more highly motivated to relatively small basis for conclusions about
do the work, who appear to be courageous a variable that has been part of psychother-
or risk taking, are more likely to evoke their apeutic lore for more than 50 years. In part,
therapist’s positive regard. this reflects the stringent criteria we used in
A consideration of how various patient deciding which studies were to be entered
characteristics may contribute to the expres- in the meta-analysis; in part, it represents
sion of positive regard in the therapeutic the fact that positive regard has been stud-
relationship also illuminates the relevance ied primarily within the realm of client-
of the perspective from which positive centered therapy, an orientation that no
regard is rated. Although Rogers (1957) longer attracts the attention of many prom-
believed that it is only the client’s perspec- inent researchers. In this respect, there have
tive that matters––that it is the client’s been very few studies of positive regard
experience of positive regard (or genuine- within the past 20 years. We believe that
ness or empathy) that “counts” and that the concept of positive regard hasn’t so
the therapist’s belief as to whether he or she much gone away in recent years as it has
has been positively regarding is essentially been folded into newer concepts in the field,
moot in regard to positive outcome––our particularly measures of the therapeutic

182 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
alliance (see Chapters 2, 3, and 4 in this types of problems at which point in
volume). therapy, is the provision of therapist regard
The restricted range of theoretical orien- most important?
tations in which positive regard has been
studied leads to another limitation: the Therapeutic Practice
possibility that the action of this variable The psychotherapist’s ability to provide
is restricted to a specific form of therapy positive regard is significantly associated
(person-centered) or interacts with a spe- with therapeutic success. However, our meta-
cific aspect of this therapy. Thus, Orlinsky analysis indicates a moderate relationship,
and Howard (1978) raised the possibility suggestive of the fact that, like many other
that empathy and warmth “interact differ- relational factors, it is a significant but not
entially with other aspects of therapist style” exhaustive part of the process–outcome
(p. 293). That is, they suggested that these equation. Extrapolating from the data, we
qualities might be significantly associated offer the following recommendations for
with outcome only when therapist direc- clinical practice:
tiveness is low––as is usually true of person-
centered or psychodynamic therapists––and • Therapists’ provision of positive
may not be the case when therapists prac- regard is strongly indicated in practice.
tice from a more heterogeneous or directive At a minimum, it “sets the stage” for
perspective that might reflect a CBT orien- other mutative interventions and that,
tation. In fact, our data allow us to know at least in some cases, it may be sufficient
only partially the answer to this question. by itself to effect positive change.
Whereas we found a significant moderat- • There is virtually no research-driven
ing effect for psychodynamic treatment–– reason to withhold positive regard. We are
patients in this form of therapy tended reminded of the oft-heard sentiment in
to improve more than others as a func- contemporary psychoanalytic circles that
tion of receiving positive regard from their one of Kohut’s major contributions was to
therapists––our database included no stud- provide a theoretical justification for being
ies of patients in any CBT-related treatment. kind to one’s patients.
It is nevertheless noteworthy that meta- • Positive regard serves many valuable
analytic results on the association between functions across the major forms of
therapist empathy and treatment outcome psychotherapy. From a psychodynamic
reveal no differences in effect sizes for perspective, positive regard serves to
different forms of therapy (Elliott, Bohart, strengthen the client’s ego (sense of self
Watson, & Greenberg, Chapter 6, this or agency) and belief in his or her capacity
volume). Nor were there any differential to be engaged in an effective relationship;
effect sizes as a function of the type of treat- from a behavioral perspective, the
ment in the relation between treatment therapist’s positive regard functions as a
outcome and the therapeutic alliance positive reinforcer for clients’ engagement
in individual therapy for adults (Horvath, in the therapeutic process, including
Del Re, & Flückiger, Symonds, Chapter 2, difficult self-disclosures; and from a more
this volume) or youth (Shirk & Karver, purely humanistic perspective, the
Chapter 3, this volume). therapist’s stance of caring and positive
More generally, the extant research has regard facilitates the client’s natural
not addressed the question of specificity: tendency to grow and fulfill his or her
For which patients, presenting with which potential.

fa r b e r, d o o l i n 183
• Positive regard may be especially alliance as seen by client and therapist.
indicated in situations wherein a non- Psychotherapy, 28, 534–49.
Barrett-Lennard, G. T. (1964). The Relationship
minority therapist is working with a
Inventory. Form OS-M-64 and OS-F-64 Form
minority client. MO-M-64 and MO-F-64. Armidale, New South
• Therapists cannot be content with Wales, Australia: University of New England.
feeling good about their patients but Barrett-Lennard, G. T. (1978). The Relationship
instead should ensure that their positive Inventory: Later development and applications.
feelings are communicated to them. This JSAS: Catalog of selected documents in psychology,
8, 68 (Ms. No. 1732, p. 55).
does not have to translate to a stream of
Barrett-Lennard, G. (1986). The relationship
compliments or to a gushing of positive inventory now: Issues and advances in theory,
sentiment that, in fact, may overwhelm or method and use. In L. S. Greenberg &
even terrify some clients; rather, it speaks W. M. Pinsof (Eds.), The Psychotherapeutic process:
to the need for therapists to communicate A research handbook (pp. 439–76). New York:
a caring, respectful, positive attitude that Guilford.
Beck, J. S. (1995). Cognitive therapy: Basics and
serves to affirm a client’s basic sense of
beyond. New York: Guilford.
worth. To many, if not most, clients, the Benjamin, L. (1984). Principles of prediction using
conviction that “my therapist really cares Social Analysis of Structural Behavior (SASB).
about me” likely serves a critical function, In R. A. Zucker, J. Aronoff, and A. J. Rabin
especially in times of stress. (Eds.), Personality and the prediction of behavior
• Therapists need to monitor their (pp. 121–73). New York: Academic Press.
Bergin, A. E., & Garfield, S. L. (Eds.) (1971).
positive regard and adjust it as a
Handbook of psychotherapy and behavior change:
function of the needs of particular An empirical analysis. New York: Wiley.
patients and specific clinical situations. Beutler, L. E., Machado, P. P. P., & Neufeldt, S. A.
The research demonstrates that therapists (1994). Therapist variables. In S. L. Garfield &
vary in the extent to which they are A. E. Bergin (Eds.), Handbook of psychother-
able to convey positive regard to their apy and behavior change (4th ed., pp. 229–69).
New York: Wiley.
patients, and clients vary in the extent
Chisholm, S. M. (1998). A comparison of the
to which they need, elicit, and/or benefit therapeutic alliances of premature terminators
from a therapist’s positive regard. In versus therapy completers. Unpublished doctoral
this regard, too, we suspect that the dissertation, Kent State University, Kent, OH.
inevitable ruptures in the therapeutic Coady, N. F. (1991). The association between client
alliance (Safran, Muran, & Eubanks- and therapist interpersonal processes and out-
comes in psychodynamic psychotherapy. Research
Carter, Chapter 11, this volume) that
on Social Work Practice, 1, 122–38.
occur over the course of therapy are Conte, H. R., Ratto, R., Clutz, K., & Karasu, T. B.
the result not only of a therapist’s (1995). Determinants of outpatients’ satisfaction
technical errors but also of the therapist’s with therapists. Journal of Psychotherapy Practice
occasional inability to demonstrate and Research, 4, 43–51.
minimally facilitative levels of positive Cooper, H., Hedges, L. V., & Valentine, J. C. (Eds.).
(2009). The handbook of research synthesis and
regard and support.
meta-analysis (2nd ed.). New York: Russell Sage
Cramer, D., & Takens, R. J. (1992). Therapeutic
relationship and progress in the first six sessions
An asterisk (∗) indicates studies included of individual psychotherapy: A panel analysis.
in the meta-analysis. Counseling Psychology Quarterly, 5, 25–36.
Bachelor, A. (1991). Comparison and relationship Del Re, A. C. (2010). RcmdrPlugin.MAc: Meta-
to outcome of diverse dimensions of the helping Analysis with correlations (MAc) Rcmdr Plug-in.

184 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
R package version 1.0.7. http://CRAN.R-project. Jourard, S. M. (1971). The transparent self. New York:
org/package=RcmdrPlugin.MAc Van Nostrand.
Del Re, A. C., & Hoyt, W. T. (2010). MAc: Meta- Keijsers, G. P., Hoogduin, C. A., & Schaap, C. P.
Analysis with correlations. R package version (1994). Predictors of treatment outcome in the
1.0.6. behavioral treatment of obsessive-compulsive
Eckert, P. A., Abeles, N., & Graham, R. N. (1988). disorder. The British Journal of Psychiatry, 165,
Symptom severity, psychotherapeutic process, 781–86.
and outcome. Professional Psychology: Research Knox, S., & Hill, C. E. (2003). Therapist self-
and Practice, 19, 560–64. disclosure: Research-based suggestions for prac-
Farber, B. A. (2007). On the enduring and substan- titioners. Journal of Clinical Psychology/In Session,
tial influence of Carl Rogers’ not-quite essential 59, 529–40.
nor necessary conditions. Psychotherapy: Theory, Leahy, R. L. (2001). Overcoming resistance in cogni-
Research, Practice, Training, 44, 289–294. tive therapy. New York: Guilford.
Farber, B. A., Brink, D. C., & Raskin, P. M. (1996). Lietaer, G. (1984). Unconditional positive regard:
The psychotherapy of Carl Rogers: Cases and com- A controversial basic attitude in client-centered
mentary. New York: Guilford. therapy. In R. F. Levant & J. M. Shlien (Eds.),
Garfield, S. L., & Bergin, A. E. (1971). Therapeutic Client-centered therapy and the person-centered
conditions and outcome. Journal of Abnormal approach: New directions in theory, research, and
Psychology, 77, 108–114. practice (pp. 41–58). New York: Praeger.
Gaston, L., Marmar, C. R., Gallagher, D., & Litter, M. (2004). Relationship-based psycho-
Thompson, L. W. (1991). Alliance prediction of therapy with court-involved youth: The therapy
outcome beyond in-treatment symptomatic relationship’s effect on outcome. Dissertation
change as psychotherapy processes. Psychotherapy Abstracts International, 65 (12), 4474. (AAT no.
Research, 1, 104–113. 3156920).
Green, R. J., & Herget, M. (1991). Outcomes of McMain, S., Korman, L. M., & Dimeff, L. (2001).
systemic/strategic team consultation: The impor- Dialectical behavior therapy and the treatment
tance of therapist warmth and active structuring. of emotion dysregulation. Journal of Clinical
Family Process, 30, 321–36. Psychology, 57(2), 183–96.
Gurman, A. (1977). The patient’s perception of Meyer, A. E. (1990). Nonspecific and common
the therapeutic relationship. In A. S. Gurman factors in treatment outcome: Another myth? Paper
& A. M. Razin (Eds.), Effective psychotherapy presented at the Annual Meeting, Society for
(pp. 503–43). New York: Pergamon. Psychotherapy Research, Wintergreen, VA.
Hayes, A. M., & Strauss, J. L. (1998). Dynamic sys- Mitchell, S. A., & Aron, L. (1999). Relational psy-
tems theory as a paradigm for the study of change choanalysis. New York: Analytic Press.
in psychotherapy: An application to cognitive Mitchell, K., Bozarth, J., & Krauft, C. (1977).
therapy for depression. Journal of Consulting and A reappraisal of the therapeutic effectiveness of
Clinical Psychology, 66, 939–47. accurate empathy, non-possessive warmth and
Hedges, L. V. (1982). Estimating effect sizes from a genuineness. In A. S. Gurman & A. M. Razin
series of independent experiments. Psychological (Eds.), Effective psychotherapy (pp. 482–502).
Bulletin, 92, 490–99. New York: Pergamon.
Henry, W. P., Schacht, T. E., & Strupp, H. H. Mitchell, K., Bozarth, J., Truax, C., & Krauft, C.
(1990). Patient and therapist introject, interper- (1973). Antecedents to psychotherapeutic outcome.
sonal process, and differential psychotherapy Arkansas Rehabilitation Research and Training
outcome. Journal of Consulting and Clinical Center, University of Arkansas (NIMH Final
Psychology, 58, 768–74. Report, MH 12306).
Horvath, A. O., & Greenberg, A. (1989). Devel- Najavits, L. M., & Strupp, H. H. (1994). Differences
opment and validation of the Working Alliance in the effectiveness of psychodynamic therapists: A
Inventory. Journal of Counseling Psychology, 36, process-outcome study. Psychotherapy, 31, 114–23.
223–33. Orlinsky, D. E., Grawe, K., & Parks, B. K. (1994).
Hynan, D. J. (1990). Client reasons and experi- Process and outcome in psychotherapy—noch
ences in treatment that influence termination of einmal. In A. E. Bergin & S. L. Garfield (Eds.),
psychotherapy. Journal of Clinical Psychology, 46, Handbook of psychotherapy and behavior change
891–95. (4th ed., pp. 27–376). New York: Wiley.

fa r b e r, d o o l i n 185
Orlinsky, D. E., & Howard, K. (1978). The rela- prozessforschung in der verhaltenstherapie [The
tion of process to outcome in psychotherapy. empirical analysis of the therapeutic relation:
In S. L. Garfield & A. E. Bergin (Eds.), Handbook Contribution to process research in behavior
of psychotherapy and behavior change (2nd ed., therapy]. Berlin, Heidelberg, Germany: Springer-
pp. 283–329). New York: Wiley. Verlag.
Orlinsky, D. E., & Howard, K. (1986). Process and Sells, D., Davidson, L., Jewell, C., Falzer, P., &
outcome in psychotherapy. In S. L. Garfield, & Rowe, M. (2006). The treatment relationship in
A. E. Bergin (Eds.), Handbook of Psychological peer-based and regular case management for
Behavior and Change (pp. 311–81) New York: clients with severe mental illness. Psychiatric
Wiley. services, 57, 1179–84.
Parloff, M. B., Waskow, I. E., & Wolfe, B. E. (1978). Sue, D. W., & Sue, D. (1999). Counseling the cultur-
Research on therapist variables in relation to ally different: Theory and practice (3rd edition).
process and outcome. In S. L. Garfield & New York: Wiley.
A. E. Bergin (Eds.), Handbook of psychotherapy Suh, C. S., Strupp, H. H., & O’Malley, S. S.
and behavior change (2nd ed., pp. 233–82). (1986). The Vanderbilt Process Measures:
New York: Wiley. The Psychotherapy Process Scale (VPPS) and
Quintana, S. M., & Meara, N. M. (1990). the Negative Indicators Scale (VNIS). In
Internalization of therapeutic relationships in L. S. Greenberg & W. M. Pinsof (Eds.), The
short-term psychotherapy. Journal of Counseling psychotherapeutic process: A research handbook
Psychology, 2, 123–30. (pp. 285–323). New York: Guilford.
Rabavilas, A. D., Boulougouris, J. C., & Tracey, T. J., & Kokotovic, A. M. (1989). Factor
Perissaki, C. (1979). Therapist qualities related structure of the working alliance inventory.
to outcome with exposure