Академический Документы
Профессиональный Документы
Культура Документы
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Valentn Escudero
Universidad de La Coruna
Laurie Heatherington
Gary M. Diamond
Williams College
Couple and family therapy (CFT) is challenging because multiple interacting working alliances
develop simultaneously and are heavily influenced by preexisting family dynamics. An original
meta-analysis of 24 published CFT alliance-retention/outcome studies (k 17 family and 7 couple
studies; N 1,416 clients) showed a weighted aggregate r .26, z 8.13 ( p .005); 95% CI
.33, .20. This small-to-medium effect size is almost identical to that reported for individual adult
psychotherapy (Horvath Del Re, Fluckiger, & Symonds, this issue, pp. 9 16). Analysis of the 17
family studies (n 1,081 clients) showed a similar average weighted effect size (r .24; z 6.55,
p .005; 95% CI .30, .16), whereas the analysis of the 7 couple therapy studies (n 335 clients)
indicated r .37; z 6.16, p .005; 95% CI .48, .25. Tests of the null hypothesis of
homogeneity suggested unexplained variability in the alliance-outcome association in both treatment
formats. In this article we also summarize the most widely used alliance measures used in CFT
research, provide an extended clinical example, and describe patient contributions to the developing
alliance. Although few moderator or mediator studies have been conducted, the literature points to
three important alliance-related phenomena in CFT: the frequency of split or unbalanced
alliances, the importance of ensuring safety, and the need to foster a strong within-family sense of
purpose about the purpose, goals, and value of conjoint treatment. We conclude with a series of
therapeutic practices predicated on the research evidence.
Keywords: couple and family therapy, meta-analysis, alliance, therapy relationships
Myrna L. Friedlander, Department of Educational and Counseling Psychology, University at Albany/State University of New York; Valentn Escudero,
Department of Psychology, Universidad de La Coruna; Laurie Heatherington,
Department of Psychology, Williams College; and Gary M. Diamond, Department of Psychology, Ben Gurion University of the Negev, Israel.
This article is adapted, by special permission of Oxford University
Press, from a chapter of the same title by the same authors in J.C. Norcross
(Ed.). (2011). Psychotherapy relationships that work (2nd ed.). New York:
Oxford University Press. The book project was co-sponsored by the APA
Division of Psychotherapy.
We are grateful to Cristina Muniz de la Pena for her valuable contribution to the meta-analysis, and to Gabriel Garza Sada, Hanna Seifert, and
Joshua Wilson for their assistance in the literature search.
Correspondence concerning this article should be addressed to Myrna L.
Friedlander, Ph.D., Department of Educational and Counseling Psychology, Education 220, University at Albany/SUNY, 1400 Washington Avenue, Albany, NY 12222. E-mail: mfriedlander@uamail.albany.edu
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
26
Clinical Example
A middle-aged couple brought their reluctant 15-year-old son
and 13-year-old daughter to psychotherapy. The girl, who exhibited anxiety and an eating disorder (only at home), refused to speak
in the session, as did the boy, who had vandalized a neighbors car
and was failing in school. While the parents barely glanced at each
other, both adamantly insisted that their children were in desperate
need of help. Thus ended the first session, which clearly evidenced
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
27
Meta-Analytic Review
Table 1 summarizes 24 studies in which CFT alliances, selfreported and observed, were used to predict treatment retention,
improvement midtreatment, and/or final outcomes (see Friedlander, Escudero, Heatherington, & Diamond, in press, for all
references). The studies were located by electronic searching
couple
family
family
EFT
FS
MDFT
EcoS
EcoS
EFT
PE
IPCT
FBT
IPCT
FS
CBT
FFT
MDFT
BSFT
MDFT
MDFT
PE
NS
O
O
O
O
O
C,O
C
O
C
T, C
C
O
C
C
C, T
O
O
O
O
C, T
Rater
(Bond)
(tasks)
WAI-Co
VTAS-R
SOFTA
VTAS-R
VTAS-R
VTAS-R
WAI
FTAS
VTAS-R
CTAS
WAI,
CTAS
CTAS
WAI
CTAS-R
FTAS
FTAS
CTAS
WAI
VTAS-R
SOFTA
VTAS-R
SOFTA
WAI
CTAS
Measure
Alliance
E,L
L
E
E
E
E
E
E
E
E,L
E,L
E,L
E,L
E
E,L
E
E,L
E
E
Time
Outcome Measure
.37
.41
.54
.29
.35
.36
.24
.53
.37
.54
.32
.39
.32
.31
.10
.46
.35
.33
.05
.22
.25
.35
.53
.43
Wt. Ave. r
22.4
45
28
34
30
31
59
19
66
23
49
37
31.4
80
430
32
40
36.7
44
68
86
33
70
63
Ave. N
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 Emotional-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. All studies in this table are referenced in Friedlander et al. (in press).
couple
family
family
family
family
family
family
family
marital
couple
family
couple
family
couple
family
family
family
couples group
BFT
FFT
FS
CBT, PE
couples group
Format
CSP
Therapy model
Study
Treatment
Table 1
Summary of Alliance-Outcome Studies in CFT
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
28
FRIEDLANDER, ESCUDERO, HEATHERINGTON, AND DIAMOND
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
(PsycInfo, PubMed, Social Sciences Citation Index) and crossreferencing of key articles. Unpublished dissertations were excluded, as were analogue studies, non-English articles, and studies
that did not use validated alliance measures.
Seven of the 24 studies were couples therapy (two of which
were conducted in groups), and the remaining 17 were family
studies in which at least a portion of the treatment was conducted
conjointly. The total number of clients in the 24 studies is 1,461.
Studies examined treatment-as usual as well as specifically defined
approaches, for example, cognitive behavioral therapy, functional
family therapy, emotion-focused therapy, psycho-educational family therapy. Most treatments were fewer than 20 sessions, and the
majority (65%) described the therapy as manualized treatment,
although only a few provided information about treatment integrity.
The target problems ranged from parent-adolescent communication difficulties to substance abuse, child abuse or neglect, and
schizophrenia; many of the studies identified the clients problems
generally. The instruments and methods used to evaluate outcome
reflect the variability in problems treated. Roughly 50% of the studies
used an observational methodology. Most evaluated the alliance early
in the therapy, and only a few studies assessed the alliance at different
stages of treatment (early, middle and late). The observational instruments were primarily the SOFTA-o and the VTAS; the WAI and
CTAS/FTAS were most often used to measure self-reported alliance.
Five of the 24 studies did not measure client outcome but rather only
explored associations between alliance and treatment retention.
For the meta-analysis of correlation coefficients, we used the
recommendations and computation program of Diener, Hilsenroth,
and Weinberger (2009), which are based on Hunter and Schmidts
(1990) random-effects approach. For studies that reported statistics
other than correlation coefficients (e.g., a t test to compare the
alliance in families that completed therapy vs. those that dropped
out), we calculated the corresponding conversions to r.
Because of the complex structure of alliance in CFT (multiple
participants generating multiple levels of analysis), most of the 24
studies reported more than a single alliance-outcome correlation.
In these cases, we calculated a meta-analytic statistic within each
study in order to maintain the statistical assumption of independence. Thus, the effect sizes listed in the far right column of Table
1 were calculated from aggregated correlations within each study.
The weighted average effect size for the 24 studies was r .26,
z 8.13 ( p.005), with a 95% confidence interval of .33 and .20,
indicating that the association between alliance and outcome was
statistically significant and accounted for a substantial proportion
of variance in CFT retention and/or outcome. According to conventional benchmarks, an r of .26 (d .53) is a small-to-medium
effect size in the behavioral sciences; this value is quite similar to
the r .275 reported by Horvath, Del Re, Fluckiger, and Symonds
(this issue, pp. 9 16) on the alliance in individual therapy.
29
effects produced by the various studies. Consequently, we explored one moderator (treatment format) that might explain the
heterogeneity by computing separate analyses with two subsamples (family and couple therapy).
Analysis of the 17 family studies (n 1,081 clients) showed a
average weighted effect size similar to what was found for the
entire sample (r .24; z 6.55, p .005; 95% CI .30, .16),
and the null hypothesis of homogeneity was also rejected (2
27.77, p .05). In other words, the group of studies with a family
therapy format had significant unexplained variability in the relation between alliance and outcome. The analysis of the seven
couple therapy studies (n 335 clients) showed a slight difference, whereas the average weighted effect size was similar (r
.37; z 6.16, p .005; 95% CI .48, .25). Although the
homogeneity in this subsample was somewhat higher (2 11.08,
p .08) in this subsample, two of the seven studies used a group
couple format. For this reason, we did not conclude that the
alliance-outcome association in couple therapy was more homogeneous than its counterpart in family therapy.
Divergences in effect sizes across the outcome studies, as well
as complex findings within these studies, raise questions about the
circumstances under which the alliance figures more or less
strongly in outcomes. Because few direct tests of moderators and
mediators were conducted in the original studies, a meta-analysis
was not possible. In this section, we summarize what is known and
has been suggested about moderators and mediators of the
alliance-outcome relation in CFT.
30
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Patient Contribution
The clinical CFT literature focuses almost exclusively on therapist behavior, with far less emphasis on client participation in
treatment. One could well argue that all client behavior contributes
to (or detracts from) the alliance. To some extent, clients collaboration in therapy depends on the therapists theoretical approach.
Couples in behavioral therapy, for example, spend less time accessing primary emotions than do couples in emotion-focused
therapy. Yet, on a different process level, alliance-related behavior
cuts across therapy approaches and formats. That is, like successful clients in individual therapy, successful family members form
a close, trusting bond with their therapists and negotiate (and
renegotiate) treatment goals and tasks. Regardless of the kinds of
in-session or out-of-session tasks, clients who have a shared sense
of purpose listen respectfully to one another, validate each others
perspective (even when they disagree), offer to compromise, and
avoid excessive cross-blaming, hostility, and sarcasm. Family
members who feel safe and comfortable in therapy are emotionally
expressive, ask each other for feedback, encourage one another to
open up and speak frankly, and share thoughts and feelings, even
painful ones, that have never been expressed at home (Friedlander
et al., 2006).
Couples Therapy
While scant, the literature offers some evidence about the personal characteristics and in-session behaviors of clients who develop strong working relationships in couples therapy. Research
suggests that whereas psychiatric symptoms are not associated
with alliance formation (Knobloch-Fedders, Pinsof, & Mann,
2004; Mamodhoussen et al., 2005), greater trust in the couple
relationship (Johnson & Talitman, 1997) and less marital distress
(Johnson & Talitman, 1997; Knobloch-Fedders et al., 2004) are
predictive of more favorable alliances. In the Knobloch-Fedders et
al. study, alliance development differed for men and women. For
men, recalling positive experiences in the family of origin was
most critical for early alliance development, whereas marital distress had a negative impact on the alliance later on. For women,
sexual dissatisfaction was negatively associated with the alliance
throughout therapy, and womens family of-origin distress contributed to a split alliance early in the process.
Regarding in-session behavior, Thomas, Werner-Wilson, and
Murphy.s (2005) results reflect the complexity of CFT alliances.
Men were less likely to agree with the therapist on the goals for
treatment when their partners made negative statements about
them, whereas women tended to feel more negative about therapy
tasks when they were challenged by their partners. Both men and
women had a stronger bond with the therapist when their partners
self disclosed, and felt more distant from the therapist when their
partners challenged or made negative comments about them.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Family Therapy
In community-based family therapy, parental differentiation of
self predicted stronger perceived alliances after Session 3 (Lambert & Friedlander, 2008); parents who reported being generally
less emotionally reactive tended to feel safer and more comfortable
in conjoint family therapy. Diagnosis or presenting problem may
also matter, for example, in family based therapy for anorexia
nervosa, teens with relatively more weight and eating concerns
found it particularly difficult to establish an alliance with the
therapist (Pereira et al., 2006). On the other hand, the nature of
adolescents emotional problems played no role in alliance development in a study of MDFT for drug-using adolescents (Shelef &
Diamond, 2008), where there was no variability in teens alliancerelated behavior based on pretreatment externalizing or internalizing behaviors.
Not surprisingly, alliances are stronger when family members
respond favorably to therapists alliance-building interventions. In
a comparison of two families treated by the same experienced
therapist (Friedlander, Lambert et al., 2008a), clients in the poor
outcome case were less likely than those in the good outcome case
to respond positively to the therapists alliance-related behaviors.
Another small study, although not directly assessing the alliance,
has implications for building a shared sense of purpose (Friedlander, Heatherington, Johnson, & Skowron, 1994). Family members were observed to move from disengagement with each other
to productive in-session collaboration when, with the therapists
help, they explored the underlying basis for their disengagement
and recognized motivations for breaking through it.
31
Therapeutic Practices
} The therapeutic alliance is a critical factor in the process and
outcome of CFT. Therapists need to be aware of what is going within
the family system while monitoring the personal bond and agreement
on goals and tasks with each individual family member. Periodically
asking clients to complete a brief measure of the alliance would
provide an opportunity for directly addressing and repairing problematic alliances.
} Shared sense of purpose within the family, a particularly important dimension of the alliance, involves establishing overarching systemic goals (e.g., It sounds like what the two of you want is a
relationship in which you feel both connected and that you can
sometimes do your own thing) rather than competing, first-order,
individual goals (i.e., I want him to stop watching sports on TV every
Saturday or I want her to give me more space). Creating a safe
space is critical, particularly early on in therapy, but doing so requires
caution. A therapist who allies too strongly with a resistant adolescent
may unwittingly damage the alliance with the parents, particularly
when the latter are expecting the teen to change but are not expecting
to be personally challenged by the therapist.
} Evaluating the alliance based on observation is a skill that can
be taught. Therapists may train themselves to validly assess their
alliances with different family members by reviewing videotaped
sessions (Carpenter, Escudero, & Rivett, 2008).
} In short, each persons alliance matters, and alliances are not
interchangeable. Each and every alliance exerts both direct and
interactive effects on the course of treatment. Thus, clinicians
should build and maintain strong alliances with each party and be
aware of the ways in which, depending on the familys dynamics,
the whole alliance is more than the sum of its parts.
References
Beck, M., Friedlander, M. L., & Escudero, V. (2006). Three perspectives on
clients experiences of the therapeutic alliance: A discovery-oriented investigation. Journal of Marital and Family Therapy, 32, 355368.
Bordin, E. S. (1979). The generalizability of the psychoanalytic concept of
the working alliance. Psychotherapy, 16, 252260.
Bourgeois, L., Sabourin, S., & Wright, J. (1990). Predictive validity of
therapeutic alliance in group marital therapy. Journal of Consulting and
Clinical Psychology, 58, 608 613.
Brown, P. D., & OLeary, K. D. (2000). Therapeutic alliance: Predicting
continuance and success in group treatment for spouse abuse. Journal of
Consulting and Clinical Psychology, 68, 340 345.
Carpenter, J., Escudero, V., & Rivett, M. (2008). Training family therapy
students in conceptual and observation skills relating to the therapeutic
alliance: An evaluation. Journal of Family Therapy, 30, 409 422.
Diamond, G. M., Liddle, H. A., Dakof, G. A., & Hogue, A. (1996). Revised
version of the Vanderbilt Therapeutic Alliance Scale. Unpublished manuscript, Temple University.
Diamond, G. M., Liddle, H. A., Hogue, A., & Dakof, G. A. (1999).
Alliance-building interventions with adolescents in family therapy: A
process study. Psychotherapy, 36, 355368.
Diener, M. J., Hilsenroth, M. J., & Weinberger, J. (2009). A primer on
meta-analysis of correlation coefficients: The relationship between
patient-reported therapeutic alliance and adult attachment style as an
illustration. Psychotherapy Research, 19, 519 526.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
32
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
33