Couple and Family Psychology:

Research and Practice

© 2022 American Psychological Association 2024, Vol. 13, No. 2, 81–99
ISSN: 2160-4096 https://doi.org/10.1037/cfp0000233

A Comprehensive Meta-Analysis on the Efficacy of Emotionally

Focused Couple Therapy

Paul M. Spengler1, Nicholas A. Lee2, Stephanie A. Wiebe3, and Andrea K. Wittenborn4

1
Department of Counseling Psychology, Social Psychology and Counseling, College of Health, Ball
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

State University
This document is copyrighted by the American Psychological Association or one of its allied publishers.

2
Open Door Health Services, Muncie, Indiana, United States
3
School of Counseling, Psychotherapy and Spirituality, Saint Paul University
4
Department of Human Development and Family Studies and Division of Psychiatry and Behavioral
Medicine, Michigan State University

Emotionally Focused Couple Therapy (EFT), as developed by Greenberg and Johnson

(1988) and Johnson (1996, 2004, 2020), is an evidence-based treatment that works with
a variety of presenting concerns (e.g., depression, posttraumatic stress disorder, sexual
dissatisfaction) and treatment characteristics, with good maintenance of treatment gains
up to 2 years following treatment. In this first comprehensive meta-analysis of EFT, we
evaluated the entirety of efficacy outcome research including randomized controlled
trial (RCT), quasi-experimental, and dissertation studies. Previous meta-analyses of
EFT analyzed solely RCTs (Beasley & Ager, 2019), Johnson’s model of EFT combined
with Greenberg and Goldman’s (2008) conceptually different model (Rathgeber et al.,
2018), or EFT embedded and not disentangled from other models of couple therapy
(Roddy et al., 2020). Across 20 studies and 332 couples, we found medium to large
random effects weighted ds for (a) pretest–posttest, d = .93, 95% CI [.75, 1.12], (b) EFT
versus viable alternative couple interventions, d = .44, 95% CI [.03, .85], and (c) pretest
to follow-up, d = .86, 95% CI [.56, 1.15]. Comprehensive sensitivity and moderator
analyses were conducted. This meta-analysis provides support for EFT as an evidence-
based couple therapy approach, where 70% of couples will be symptom free at the end
of treatment and greater therapist fidelity to the model is associated with stronger couple
gains. Limitations and implications are discussed in relation to research, training, and
practice of EFT.

This article was published Online First September 22, 2022. of review and editing. Nicholas A. Lee played lead role in
Paul M. Spengler https://orcid.org/0000-0001-9539- funding acquisition, supporting role in conceptualization,
4214 formal analysis, project administration, resources, supervi-
Stephanie A. Wiebe https://orcid.org/0000-0001- sion, visualization, writing of original draft and writing of
5865-4531 review and editing and equal role in investigation, methodol-
Andrea K. Wittenborn https://orcid.org/0000-0003- ogy and validation. Stephanie A. Wiebe played supporting
4608-0635 role in conceptualization, formal analysis, resources, writing
The authors have no conflict of interest to disclose. of original draft and writing of review and editing. Andrea K.
Portions of this study were presented in August 2016 at Wittenborn played supporting role in conceptualization,
the annual meeting of the American Psychological Associa- investigation, methodology, validation, visualization, writing
tion in Denver, CO. of original draft and writing of review and editing.
Paul M. Spengler played lead role in conceptualization, Correspondence concerning this article should be addressed
data curation, formal analysis, investigation, methodology, to Paul M. Spengler, Department of Counseling Psychology,
project administration, resources, software and writing of Social Psychology and Counseling, College of Health, Ball
original draft, supporting role in funding acquisition and State University, Muncie, IN 47306, United States. Email:
equal role in supervision, validation, visualization and writing [email protected]

81
 82 SPENGLER, LEE, WIEBE, AND WITTENBORN

Public Significance Statement

This empirical synthesis of existing emotionally focused couple therapy (EFT)
outcome research provides support for EFT is an effective treatment for couples
experiencing significant relationship distress. Of the 330 couples studied, 70%
were symptom free at the end of treatment and gains last up to 2 years post
treatment.

Keywords: emotionally focused therapy, emotionally focused couple therapy, couple

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

Supplemental materials: https://doi.org/10.1037/cfp0000233.supp

Couple therapy as a field has undergone a great EFT Research

deal of growth particularly in the past two dec-
ades. Models for treating couple distress have EFT has accumulated a large body of both
become increasingly systematized and research process and outcome studies (for review, see
has expanded in volume and rigor (Carr, 2018; Wiebe & Johnson, 2016). Previous meta-
Lebow et al., 2012). Synthesizing the growing analyses of EFT have contributed to establishing
research literature on couple therapy is important EFT as an evidence-based couple therapy, not-
because it mitigates biases and limitations inher- withstanding significant limitations in methodol-
ent in conclusions based on reading of individual ogy. Meta-analyses consist of analysis of effects
studies and efforts at narrative reviews (Shadish from individual studies that are then aggregated
& Baldwin, 2003). Findings from meta-analyses and analyzed in an overall analysis, thus the
reveal moderate-to-large overall effects for cou- term meta-analysis. Johnson et al. (1999) carried
ple therapy (Roddy et al., 2020; Shadish & out the first empirical synthesis of EFT research,
Baldwin, 2003) with clinically significant gains which included a narrative overview of nine EFT
in 65%–70% of treated couples (Halford et al., studies but meta-analyzed only four that were
2015; Lebow et al., 2012). considered by the authors to be of high-quality
randomized controlled trial studies (RCT;
Goldman & Greenberg, 1992; Gordon-Walker
Emotionally Focused Couple Therapy et al., 1996; James, 1991; Johnson & Greenberg
1985b). They reported a weighted mean effect of
Emotionally focused couple therapy (EFT; d = 1.31, a large effect by any convention (Cohen,
Johnson, 1996, 2004, 2020) is one of the modali- 1988), with 70%–73% of couples symptom free
ties with the strongest research support (Sprenkle, following treatment. As with many other couple
2012; Wiebe & Johnson, 2016) and is considered therapy meta-analyses, change in couple relation-
to be evidence-based (Sexton et al., 2011). EFT— ship satisfaction served as the outcome measure.
an attachment-based couple therapy built on More recently, Rathgeber et al. (2018) con-
humanistic and systemic principles—views the ducted a meta-analysis of EFT and behavioral
negative cycles of interaction typically presented couple therapy (BCT). They found a medium
by distressed couples as a reflection of an insecure effect for the combination of these two models,
attachment bond. The goal in EFT is to help g = .60,1 with a larger effect for EFT of g = 0.73,
partners create a safe and secure bond through compared with BCT, g = 0.53. At 6-month
the acknowledgement, exploration, and experi- follow-up, the effect was medium for EFT, g =
ential sharing of vulnerable attachment-based 0.66, and small for BCT, g = 0.35 (g = 0.44
emotions and needs. For more information out- overall). Rathgeber et al. (2018) assessed
lining the stages and steps of EFT, see the most moderators, which are basically interaction or
recent treatment manual (Johnson, 2020) and
research, training, and practice resources posted 1
For purposes of comparison of these meta-analyses,
by the International Centre for Excellence in Hedges’ g is comparable to Cohen’s d but is considered to
Emotionally Focused Couple Therapy (ICEEFT; provide correction for bias due to small sample sizes (Hedges
www.iceeft.com). & Olkin, 1985).
 META-ANALYSIS OF EMOTIONALLY FOCUSED COUPLE THERAPY 83

prediction effects, across both approaches. They gains, which is not surprising given more room
found that greater pretest couple distress and for measurable change and regression to the
couples who had been in longer relationships mean. Several potentially important conceptual
was associated with better outcomes. A notable and methodological moderators, addressed in the
limitation of this meta-analysis is that the current meta-analysis, were not assessed.
researchers combined two distinct forms of These meta-analyses, limitations notwith-
emotion(ally) focused couple therapy under the standing, are informative and suggest that couple
acronym EFT that have similarities but are, in therapy, in general, and EFT in particular produce
fact, different models known respectively as moderate-to-large improvements for distressed
emotionally focused couple therapy (Johnson, couples in relationship satisfaction. Despite these
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

2020) and emotion-focused couple therapy important contributions, there remains the need
This document is copyrighted by the American Psychological Association or one of its allied publishers.

(Greenberg & Goldman, 2008). This oversight for a comprehensive rigorous synthesis of EFT,
resulted in a conflation of the results attributable that is inclusive of RCTs, quasi-experimental
to the unique characteristics of each model. studies, and dissertations, without EFT embed-
This is problematic because of significant theo- ded or confused with other couple therapies.
retical differences between these models. John- While EFT has been extensively researched
son focuses on attachment-related emotions and and has had wide impact on the training and
needs whereas Greenberg and Goldman focus in practice of couple therapy, to date a fully com-
a more general manner on emotions and personal prehensive meta-analysis (CMA) of all existing
needs related to self and the relationship. EFT outcome studies has not been conducted.
Rathgeber et al. (2018) also isolated their analy- Our CMA focuses more precisely on a population
ses to only RCT studies and, thereby, excluded estimate by including 100% of extant EFT
about 50% of EFT research that is based on quasi- research, not making decisions a priori about
experimental pretest–posttest experimental the strength or quality of the research, and by
group only designs. analyzing for methodological differences (e.g.,
As with Rathgeber et al. (2018), Beasley and RCTs and quasi-experimental studies) that may
Ager (2019) synthesized only RCT EFT studies be associated with the overall effect rather than
and omitted the substantial number of quasi- eliminating studies a priori. In response to this
experimental studies. They reported a surpris- absence, we used meta-analytic techniques to
ingly large overall Hedge’s g of 2.09. This is assess the efficacy of EFT and to test an array
because two peer-reviewed studies included in of relevant moderators that are both (a) standard
the analyses arguably should have been identified to meta-analyses, and (b) arise from our observa-
as statistical outliers (reported effect sizes of g = tions of couple therapy and EFT literature.
4.46, Najafi et al., 2015 and g = 6.25, Solymani Assessing methodological moderators, such as
ahmadi et al., 20142) and were, in fact, eliminated selection bias, attrition bias, and study quality
from the present meta-analysis (see Greenhouse (Cooper et al., 2009), are important to under-
& Iyengar, 2009). Although the aim was to standing the strengths and limitations of a body of
include rigorous studies that met RCT standards, research. Identifying practice oriented and
some were not peer reviewed and there were no clinically meaningful moderators, such as thera-
moderation analyses for methodological rigor. pist experience with EFT, EFT treatment fidelity,
Most recently, Roddy et al. (2020) conducted a and the number or length of EFT sessions, has
meta-analysis of couple therapy studies that potential to aid with future model development,
included both RCT and quasi-experimental de- training, and practice of EFT. Unlike other meta-
signs. Only nine EFT studies, however, were analyses of EFT, we implemented a sensitivity
included, less than 50% of the extant body of analysis strategy (Greenhouse & Iyengar, 2009)
research, and these also were not disentangled in by examining factors such as the quality of the
the analyses from other couple therapies. For the data (e.g., inclusion or exclusion of outlier effects,
combined couple therapies, they found a large random vs. fixed effects analyses) and other
effect for changes in relationship satisfaction of commonly considered threats to validity in
g = 1.12, that is roughly maintained at follow-ups meta-analyses (publication bias, funnel plot,
of 6 months to 2 years (g = .91). As with
Rathgeber et al. (2018), greater couple pretest 2
Solymani ahmadi et al. (2014) is erroneously cited as
distress was associated with the greatest couple Ahmadi et al. (2014) by Beasley and Ager (2019).
 84 SPENGLER, LEE, WIEBE, AND WITTENBORN

file drawer analysis). A sensitivity analysis clari- study did not assess couple distress it was not
fies if the findings are robust to analysis and included (e.g., impact of EFT on depression,
design choices. Our hope is that the results of Alder et al., 2018). Additionally, the EFT model
this first CMA advances future research, practice, needed to be derived from either the initial
and training of EFT. Greenberg and Johnson (1988) or Johnson
(1996, 2004, 2020) treatment manuals. A study
was excluded if it contained missing information
Method
needed to calculate effect size estimates, or it was
not an efficacy outcome study (e.g., process
Literature Search Process research, effectiveness study, case application,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

theoretical or conceptual article). Concerted

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Prior meta-analyses and reviews on EFT (e.g.,

efforts were made to contact authors for missing
Beasley & Ager, 2019; Johnson et al., 1999;
data, or reanalysis of their data that would allow
Johnson & Wittenborn, 2012; Rathgeber et al.,
for calculation of a study effect size (e.g.,
2018; Roddy et al., 2020; Wiebe & Johnson,
Wittenborn et al., 2019). The search process
2016; Wood et al., 2005) and the ICEEFT website
yielded 299 possible studies; 264 were excluded
were examined for EFT outcome studies. For-
because they did not evaluate the efficacy of EFT,
ward and backward searching of identified stud-
two were excluded because they were statistical
ies was conducted for all identified studies.
outliers (Najafi et al., 2015; Solymani ahmadi et
Database searches were then performed using
al., 2014), and three were excluded because
PsycINFO, PsycBooks, PsycNET, and ProQuest
information needed to compute an effect size
dissertations and Theses Global. The first out-
was unobtainable. Eight studies shared the
come study on EFT was published in 1985 by the
same sample of couples; we analyzed only the
main authors of the model (Johnson & Greenberg,
original study in the pretest–posttest effects.
1985a). Therefore, our search parameters began
Some studies provided follow-up data in subse-
with articles published in 1985 and ended with
quent reports. In total, 19 studies were included in
studies dated through January of 2021. The fol-
the pretest–posttest analysis and one study was
lowing keywords were used: EFT, EFCT, emo-
included only in the EFT versus viable treatment
tionally focused therapy, and emotionally
analysis, Denton et al. (2012),3 making a total of
focused couple therapy. When searching Pro-
20 studies. Ten of the 19 pretest–posttest studies
Quest dissertations, additional keywords were
provided follow-up data; three follow-up data
used: outcome, randomized, efficacy, pretest–
sets were obtained from subsequent publications.
posttest, and quasi-experimental. This decision
was made after searching with the initial key-
words yielded over 6,500 results. If a dissertation Coding Procedures
was later published, we used the published article,
except for Dessaulles et al. (2003) where we used The search process and coding were con-
the dissertation (Dessaulles, 1991) to code nec- ducted by the authors of this study, three psy-
essary statistics. chologists and one doctoral marriage and family
therapist. All coders completed advanced train-
ing in EFT (4-day externship, year-long core-
Inclusion and Exclusion Criteria
skills training, and EFT clinical supervision).
To be included, a study had to be an efficacy Three of the coders are certified EFT therapists.
study where the outcome of EFT was compared to All four work in academia and regularly treat
a wait-list control condition, compared with one couples using EFT. The coding scheme con-
or more viable treatments, or assessed in a sisted of the following: primary outcomes for
pretest–posttest only design. The term “efficacy” treatment, comparison, and control conditions
is customarily contrasted with the use of the term (e.g., M, SD, N); and study, therapist, and
“effectiveness.” Efficacy refers to highly con- 3
trolled experimental conditions, such as RCTs, We decided to include Denton et al. (2012), but only in
the EFT versus viable treatment comparison because EFT
whereas effectiveness refers to outcome research was modified by combing it with antidepressant medication.
in the natural delivery of couple therapy or treat- This combined model was contrasted with a medication
ment-as-usual by practitioners. If an outcome management only treatment condition.
 META-ANALYSIS OF EMOTIONALLY FOCUSED COUPLE THERAPY 85

participant characteristics. Study characteristics decided a priori to use random effects weighted d
included assessment of design, exclusion rate, models for three analyses: (a) EFT pretest–
attrition, researcher allegiance, session length, posttest, (b) EFT versus a comparison treatment,
length of overall treatment, fidelity and adher- and (c) EFT pretest to follow-up. A random
ence checks, unit of data analysis, and study effects model is most appropriate when the intent
quality. Study quality was rated as acceptable, is to generalize to diverse research scenarios,
good, and high based on Shadish et al.’s (2002) compared with a fixed effects model which is
threats to validity. Cooper (1998) addressed more appropriate for closely replicated studies,
the advantages and disadvantages of using a such as in RCT medication trials (Borenstein
global rating for study quality; global ratings et al., 2009). We chose random effects modeling
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tend to have better rater agreement with compa- because extant EFT outcome research poses a
This document is copyrighted by the American Psychological Association or one of its allied publishers.

rable heuristic value. Therapist characteristics diverse set of questions, for example, in terms of
included EFT specific training and experience methodology, rigor, and study of different clini-
and general therapist experience level. Partici- cal populations. Pretest–posttest correlations
pant characteristics included mean length of were set at .50. In calculation of the three overall
relationship, mean age of couples, initial level effect sizes, each study effect was weighted by
of distress, the focus of treatment (e.g., general the inverse variance related to sample size
relationship distress vs. depression in the context (Borenstein et al., 2005). Additionally, we con-
of relationship distress), and participant race or ducted a priori planned mixed-effects moderator
ethnicity. In the case of missing information to and regression analyses when tests of heteroge-
compute an effect size attempts were made to neity of the effects was equal to or greater than
contact study authors by email and phone. Kappa 50% (Higgins et al., 2003). Only one moderator
agreement was calculated using Cohen’s κ (participant gender) could be assessed using
(Cohen, 1960). Agreement coefficients between subgroup analyses, otherwise for the overall
the raters ranged from κ = .67–1.0. According to and the moderator analyses one effect size was
conventional benchmarks this is high moderate analyzed from each study.
to full agreement (Landis & Koch, 1977). The
categories with the lowest interrater agreement
were (a) assessment of EFT specific training and Results
experience (prior experience, study specific
training, unknown), and (b) type of fidelity Overall Analyses
checks (bug/mirror supervision, supervision,
videotape review, supervision, and videotape EFT Pretest–Posttest
review). Disagreements by coders were resolved Across 19 studies (9 RCT and 10 pretest–
by consensus (Card, 2012). posttest), assessing the use of EFT in the
treatment of 308 couples, we found a weighted
Calculation of Effects random effect d = .93 reflecting a large treat-
ment effect.4 The 95% confidence interval, CI
The dependent measure for all studies as- [.75, 1.12], does not include zero meaning this
sessed couple satisfaction (e.g., Dyadic Adjust- is a statistically significant effect ( p < .001).
ment Scale [DAS]; Spanier, 1976; Revised The studies, outcome measures, effects, and
Dyadic Adjustment Scale [RDAS], Busby et study moderators along with a forest plot are
al., 1995). The RDAS is a revised shorter version presented in Figure 1. The forest plot is widely
of the DAS (14 items, 18 fewer than DAS). Both used in meta-analyses and provides a readily
the DAS and RDAS have good reliability and understandable and visual representation of
established construct and criterion validity; the meta-analytic findings. All effect sizes were pos-
DAS and RDAS total scores correlate .97 (Crane itive and ranged from .15 indicating a small EFT
et al., 2000). Means and standard deviations or treatment effect to 1.89 reflecting a large impact
other statistics (e.g., t tests, F tests), along with of EFT. The overall effect is heterogeneous in
participant sizes for EFT, comparison, and/or
wait-list control groups at pretest, posttest, and 4
Random effects d = 1.30, 95% CI [.95, 1.64], including
follow-up were entered in the software program the two studies removed from the analysis, because they were
CMA Version 2 (Borenstein et al., 2005). We assessed as outliers; discussed in the sensitivity analyses.
 86 SPENGLER, LEE, WIEBE, AND WITTENBORN

Figure 1
Random Effects (d) Pretest–Posttest EFT Only and Versus Control Group

Study name Subgroup within study Outcome Statistics for each study Std diff in means and 95% CI

Std diff Lower Upper Relative

in means limit limit p-Value weight

Dalgleish et al. (2015) Couple DAS 0.904 0.492 1.315 0.000 6.58
Dalton et al. (2013) Individuals DAS 0.468 -0.134 1.069 0.128 4.73
Dandeneau & Johnson (1994) Couple DAS 0.465 -0.346 1.276 0.261 3.30
Dessaulles (1991) Combined DAS 0.752 0.043 1.460 0.038 3.92
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Goldman & Greenberg (1992) Individuals DAS 1.259 0.686 1.833 0.000 4.97
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Gordon Walker et al. (1996) Couple DAS 1.125 0.379 1.870 0.003 3.68
James (1991) Couple DAS 0.870 0.322 1.418 0.002 5.19
Johnson & Greenberg (1985a) Couple DAS 1.886 1.025 2.746 0.000 3.04
Johnson & Greenberg (1985b) Couple DAS 0.935 0.491 1.379 0.000 6.22
Johnson & Talitman (1997) Couple DAS 1.277 0.947 1.608 0.000 7.50
Lee et al. (2017) Combined DAS 1.392 0.655 2.129 0.000 3.74
MacIntosh & Johnson (2008) Couple DAS 1.510 0.869 2.152 0.000 4.41
MacPhee et al. (2008) Individuals DAS 0.591 0.187 0.996 0.004 6.65
Makinen & Johnson (2006) Combined DAS 0.635 0.264 1.007 0.001 7.02
McLean et al. (2008) Combined RDAS 0.843 0.556 1.131 0.000 7.98
McLean et al. (2013) Combined RDAS 1.632 1.124 2.141 0.000 5.57
Sims (1999) Individuals DAS 0.145 -0.414 0.704 0.611 5.09
Weissman et al. (2017) Combined DAS 0.497 -0.063 1.056 0.082 5.09
Wittenborn et al. (2018) Combined DAS Satisfaction 0.987 0.452 1.522 0.000 5.31
Overall random effect d 0.933 0.753 1.112 0.000

-2.00 -1.00 0.00 1.00 2.00

Negative Effect Positive Effect

Note. EFT = emotionally focused couple therapy; DAS = Dyadic Adjustment Scale; RDAS = Revised Dyadic Adjustment
Scale. Positive d effect indicates positive treatment effect for EFT pretest–posttest or in comparison to control group. The
squares represent the d effects proportionately sized to inverse-weights used in calculating the overall effect. The lines reflect
95% confidence intervals (CI) for each study. The diamond reflects mean weighted overall EFT effect; width of the diamond
represents 95% CI.

variance reflected by a statistically significant treatment as usual couple therapy (Wittenborn et

homogeneity index, Q(18) = 41.69, p = .001. al., 2019). Two studies contrasted EFT alone
The I-squared index is I2 = 56.82 meaning 57% of (Dessaulles, 1991) or EFT combined with medi-
the variance between studies is due to heteroge- cation management (Denton et al., 2012) with
neity, and not chance, which is a moderate degree antidepressant medication. In both of these studies,
of heterogeneity (Higgins et al., 2003). This medication management was provided to hetero-
means that it is unlikely that all of the variance sexual women diagnosed with major depressive
between studies is due to sampling error; covari- disorder. The overall weighted random effect d =
ates (or moderators) may account for differences 0.44 reflects that EFT produces stronger treatment
between studies. effects on couple satisfaction than alternative
treatments. The 95% CI [.03, .85] does not include
EFT Versus Comparison Treatment zero reflecting a statistically significant difference
in favor of EFT (p = .03). All effects were positive
Four studies compared EFT with other couple ranging from small (d = .07) to large (d = 1.50).
treatments, specifically cognitive marital therapy A summary of study characteristics and a forest
(Dandeneau & Johnson, 1994), integrated sys- plot are presented in Figure 2. The overall effect is
temic therapy (Goldman & Greenberg, 1992), homogeneous, Q(5) = 8.13, p = .15. The I-squared
BCT (Johnson & Greenberg, 1985a), and index, I2 = 38.50, reflects that 39% of the variance
 META-ANALYSIS OF EMOTIONALLY FOCUSED COUPLE THERAPY 87

Figure 2
Random Effects (d) EFT Versus Comparison Treatment

Study name Subgroup within study Outcome Statistics for each study Std diff in means and 95% CI

Std diff Lower Upper Relative

in means limit limit Z-Value p-Value weight

Dandeneau & Johnson (1994) Couple DAS 0.070 -0.730 0.871 0.172 0.864 16.35

Denton et al. (2012) Women QMI 1.504 0.192 2.817 2.247 0.025 7.91

Dessaulles (1991) Combined DAS 0.034 -0.848 0.915 0.075 0.941 14.42
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Goldman & Greenberg (1992) Individuals DAS 0.055 -0.469 0.579 0.205 0.838 25.62
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Johnson & Greenberg (1985a) Couple DAS 0.849 0.102 1.596 2.227 0.026 17.80

Wittenborn et al. (2018) Combined DAS Satisfaction 0.786 0.042 1.530 2.071 0.038 17.89

Overall random effect d 0.441 0.034 0.848 2.125 0.034

-2.00 -1.00 0.00 1.00 2.00

Favors Comparison Favors EFT

Note. Positive d effect indicates positive treatment effect for EFT versus comparison treatment. All studies compared EFT
with a viable alternative treatment. Denton et al. (2012) compared EFT plus medication management with medication
management only for heterosexual women with depression. QMI = quality of marriage index; EFT = emotionally focused
couple therapy; DAS = Dyadic Adjustment Scale.

is due to chance, which is a low to moderate Sensitivity Analyses

degree of heterogeneity.
Greenhouse and Iyengar (2009) suggest con-
EFT Pretest to Follow-Up ducting sensitivity analyses to evaluate the
impact of methodological and analysis decisions
Of the 19 pretest–posttest EFT studies, ten on meta-analytic results. To this end, we as-
provided follow-up data either in those studies sessed the impact of outlier effects on the
or in subsequent publications. Time of follow- overall analyses using the outlier-labeling rule
up ranged from 2 to 24 months after couples suggested by Tukey (1977) and refined by
completed EFT. In each instance, we selected Hoaglin and Iglewicz (1987). With a lower
the longest time period available to contrast critical value (d = −1.18) and an upper critical
with pretest scores. A summary of study charac- value (d = 3.25), two outliers were identified in
teristics and a forest plot for these analyses are initial pretest–posttest comparisons (d = 5.72,
presented in Figure 3. The overall weighted ran- Najafi et al., 2015; d = 6.49, Solymani ahmadi et
dom effects d = .86 for pretest to follow-up is al., 2014). Including these outliers resulted in an
similar in magnitude to the pretest–posttest d = .93, inflated d = 1.30, 95% CI [.95, 1.64]. Removing
suggesting little decline over time in gains from these outliers resulted in a more centralized d =
EFT. This is a statistically significant effect (p < .93, 95% CI [.75, 1.11], that we consider to be
.001), reflected by a 95% CI [.56, 1.15] that does the best estimate of the effect of EFT on couple
not cross zero. All effects were positive ranging satisfaction. The lower (d = −2.07) and upper
from small (d = .05) to large (d = 1.92) suggesting (d = 3.13) critical values for EFT versus com-
that there are lasting positive effects of EFT. parison treatments identified no outliers. No
However, there is considerable heterogeneity outliers existed for follow-up analyses using
between studies, Q(9) = 33.54, p < .001, with calculated lower (d = −0.95) and upper (d =
73% of the variance not due to chance, I2 = 73.17. 2.83) critical values.
Therefore, any interpretation of EFT gains being Dandeneau and Johnson (1994) assessed EFT
enduring should be viewed tentatively as other versus cognitive marital Therapy, for relation-
factors may account for these results (e.g., attrition ship enhancement in couples with no marital
bias, time of follow-up). distress. Because this was the only study with
 88 SPENGLER, LEE, WIEBE, AND WITTENBORN

Figure 3
Random Effects (d) EFT Pretest to Follow-Up

Study name Subgroup within study Outcome Statistics for each study Std diff in means and 95% CI

Std diff Lower Upper Relative

in means limit limit p-Value weight

Dalgleish et al. (2015a) Couple DAS 0.441 0.112 0.769 0.009 12.29

Dandeneau & Johnson (1994) Couple DAS 1.559 0.963 2.154 0.000 9.08

Goldman & Greenberg (1992) Individuals DAS 0.449 0.011 0.887 0.045 10.97
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Gordon-Walker et al. (1996) Couple DAS 0.747 0.312 1.181 0.001 11.02
This document is copyrighted by the American Psychological Association or one of its allied publishers.

James (1991) Couple DAS 0.934 0.490 1.378 0.000 10.90

Johnson & Greenberg (1985a) Couple DAS 1.919 1.295 2.543 0.000 8.76

Lee et al. (2017) Combined DAS 1.256 0.552 1.960 0.000 7.90

Makinen & Johnson (2006) Combined DAS 0.725 -0.036 1.486 0.062 7.34

McLean et al. (2008) Combined RDAS 0.795 0.505 1.086 0.000 12.73

Sims (2000) Individuals DAS 0.052 -0.549 0.653 0.864 9.02

Overall random effect d 0.855 0.563 1.148 0.000

-2.00 -1.00 0.00 1.00 2.00

Negative Effect Positive Effect

Note. EFT = emotionally focused couple therapy; DAS = Dyadic Adjustment Scale; RDAS = Revised Dyadic Adjustment
Scale. Positive d effect indicates positive treatment effect for EFT pretest to follow-up. Follow-up for three studies were
obtained in subsequent publications: (a) Dalgleish et al. (2015; Wiebe, 2014), (b) Gordon-Walker et al. (1996; Cloutier et al.,
2002), and (c) Makinen & Johnson (2006; Halchuk et al., 2010). Original studies are entered in the forest plot.

nondistressed couples, we decided to report treatment (Higgins et al., 2003). With a magni-
effect sizes without Dandeneau and Johnson tude of more than 50%, moderator analyses were
(1994): (a) EFT pretest–posttest random effects conducted for both (a) pretest–posttest, and (b)
d = .95, 95% CI [.77, 1.13], (b) EFT versus a pretest to follow-up analyses. No moderators
comparison treatment random effects d = .53, were assessed for the EFT versus comparison
95% CI [.05, 1.03], and (c) EFT pretest to follow- treatment effect due to homogenous variance.
up random effects d = .85, 95% CI [.57, 1.14].
The only real difference by removing this Categorical Moderator Analyses
study was an increase in the effect for EFT versus
a comparison treatment (d = .53 vs. d = .44). We Mixed effect moderator analyses were used for
also examined differences between using categorical variables using a procedure analogous
random effects and fixed effects models. These to analysis of variance (ANOVA; Hedges &
comparisons show no real differences in the Olkin, 1985). Mixed effects models are thought
estimated effect sizes: (a) EFT pretest–posttest to be a more stringent moderator test than fixed
fixed effects d = .92, 95% CI [9.81, 1.03], (b) effects due to a lower likelihood of Type I errors
EFT versus a comparison treatment fixed (Overton, 1998). Each study was treated as the
effects d = .39, 95% CI [.08, .70], and (c) unit of analysis, except for a small number of
EFT pretest to follow-up fixed effects d = .78, studies that reported results separately for men
95% CI [.63, .92]. and women (Dessaulles, 1991; Lee et al., 2017;
Wittenborn et al., 2019), allowing for more
Moderator Analyses powerful gender subgroup analyses. Other EFT
studies reported results for additional subgroups
Moderator analyses are another important step (e.g., injured partner, offending partner; Makinen
in a sensitivity analysis, especially when 50% or & Johnson, 2006; veterans, partners; Weissman
more of the variance is unaccounted for by the et al., 2017) but were not sufficiently replicated
 META-ANALYSIS OF EMOTIONALLY FOCUSED COUPLE THERAPY 89

and we could not meaningfully categorize these between moderators are statistically significant
for subgroup analyses. when respective 95% CIs do not overlap (Borman
Table 1 presents categorical moderator analy- & Grigg, 2009). With the exception of unit of
ses for the pretest–posttest overall analyses. analysis, Q(2) = 6.32, p = .04, there were no other
When reviewing these findings, the between- statistically significant categorical moderators.
class effect (QB) is equivalent to a one-way Studies that averaged couple scores (d = 1.10)
ANOVA with levels included below. Moderators were associated with larger treatment effects than
that are statistically significant are reflected by a studies that analyzed couples’ individual scores
95% CI that does not cross zero, and differences (d = .68). Both effects are statistically significant,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Table 1
Categorical Models for Overall EFT Pretest–Posttest Effects

Between class effect Mean weighted 95% CI

Moderators and levels (QB) p k effect (d) Lower Upper

Conceptual moderators
Population 3.30 .35
Distressed couples 8 .95 .67 1.22
Enrichment couples 1 .47 −.54 1.48
Parents with child needs 4 1.20 .80 1.61
Couple psych symptoms 6 .46 .48 1.12
Researcher allegiance .35 .55
EFT Founder 13 .89 .67 1.12
No EFT founder 6 1.01 .69 1.34
Therapist experience EFT .20 .90
Prior EFT 8 .89 .61 1.67
Study-specific training 8 .98 .66 1.30
Unknown 3 .97 .50 1.45
Therapist experience .30 .86
Graduate degree 8 .88 .59 1.18
Graduate trainee 9 .99 .71 1.28
Mixed 2 .90 .37 1.44
Gender differences .47 .49
Men 3 .91 .40 1.41
Women 3 1.17 .62 1.71
Method moderators
Research design .07 .80
Experimental 9 .91 .62 1.20
Quasi-experimental 10 .96 .71 1.13
Study quality 3.78 .29
Acceptable 4 .82 .41 1.24
Good 7 .88 .60 1.15
Excellent 7 .94 .65 1.24
Inclusion criteria 2.54 .28
Exclusive 3 1.08 .67 1.50
Typical 14 .94 .73 1.16
Inclusive 2 .47 −.18 1.11
Treatment fidelity 9.25 .06
None 2 1.06 .61 1.51
Bug/mirror/videotape 1 1.39 .52 2.27
Supervision only 5 .62 .32 .93
Videotape only 1 1.63 .94 2.32
Supervision/videotape 10 .96 .73 1.19
Unit of analysis 6.32 .04
Averaged couple scores 10 1.10 .88 1.31
Couple’s individual scores 8 .68 .43 .93
One partner’ scores 1 1.13 .25 2.00
Note. Positive effects reflect EFT efficacy. Gender effects were calculated for k = 3 studies using subgroup analyses. 95%
CI = confidence interval; EFT = emotionally focused couple therapy.
 90 SPENGLER, LEE, WIEBE, AND WITTENBORN

as confidence intervals do not cross zero, and percentages ranging from 0 (Johnson &
reflect real comparative differences, as there is no Greenberg, 1985b) to 94 (Dalgleish et al.,
overlap between the confidence intervals. Treat- 2015), M = 46.4.6 Rate of prestudy exclusion
ment fidelity approached the convention of p < had no relation to the overall effect, Q(1) = 0.49,
.05 for statistical significance. The largest effects p = .53. Fifteen studies provided information
for treatment fidelity were associated with Lee on couples accepted for the study who dropped
et al. (2017; d = 1.39) who reported use of the out from the point of study inclusion to pretest
most rigorous fidelity steps (bug-in-the-ear, live measurement, with percentages ranging from 0
mirror, and videotape review), and McLean et al. (Dessaulles, 1991; James, 1991; Johnson &
(2013; d = 1.63) where Sue Johnson supervised Greenberg, 1985a; Lee et al., 2017) to 52
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

treatment fidelity for 25% of the videotaped (Goldman & Greenberg, 1992), M = 13.4. There
This document is copyrighted by the American Psychological Association or one of its allied publishers.

sessions. Caution should be exercised in drawing was no association between dropout rates of
conclusions about treatment fidelity and the couples prior to pretest and benefit from EFT,
efficacy of EFT due to the small number of studies Q(1) = 0.55, p = .46. All 19 studies provided
(k = 2) and overlapping 95% CIs. information on attrition of couples in treatment
The categorical moderator analyses for the from pre- to post-test, with a modal dropout
pretest to follow-up analyses are similar in find- percent of zero (k = 9) and an upper range of
ings (see Supplemental Table S1). Few categori- 53% (Weissman et al., 2017), M = 12.27. Drop-
cal moderators were statistically significant out rates for couples during treatment was also
predictors of the follow-up treatment effect, not associated with benefit from EFT, Q(1) =
with the exception of therapist specific experi- 1.41, p = .24.
ence with EFT, Q(1) = 4.80, p = .03. Studies that Participant Characteristics. Sixteen studies
included therapists who received study specific reported the mean age of treatment couples,
training in EFT (d = 1.17) were associated with ranging from 36.9 (Walker et al., 1996) to
larger follow-up effects, compared with those 51.83 years (McLean et al., 2013), M = 41.33.
that utilized therapists with only prior EFT train- Age of couples was unrelated to benefit from
ing (d = .57) but no study specific training and EFT, Q(1) = 0.85, p = .36. Seven studies pro-
feedback. The 95% CI between these two types of vided some or minimal information about
EFT specific experience overlap indicating cau- participants’ race. Percentage of participants
tion is warranted in reaching conclusions of true who identified as Caucasian ranged from 64.3
differences. (Lee et al., 2017) to 100 (Goldman & Greenberg,
1992) and had no association with gains from
Regression Moderator Analyses EFT, Q(1) = 0.01, p = .91. All studies except
Lee et al. (2017) reported mean years of couples
Ten continuous moderator variables of the over- being together, ranging from 6.9 (Johnson &
all pretest–posttest analyses were analyzed using Greenberg, 1985b) to 22.43 (McLean et al.,
mixed-effects (maximum likelihood) regression 2013), M = 13.03. Length of couple relationship
analyses (Borenstein et al., 2009).5 Of these mod- was unrelated to gains from EFT, Q(1) = 0.42, p =
erators, three were measures of selection and .52. With the exception of Wittenborn et al.
attrition bias (exclusion rate by researchers, attri- (2019), who used the DAS Satisfaction subscale,
tion of couples from eligible to pretest, attrition all other studies assessed participants using
of couples from pre- to post-test), four assessed total scores from the DAS or the RDAS. Scores
participant characteristics (age, race, years from the RDAS were converted to DAS total
together, couple pretest DAS/RDAS), and three score equivalencies following recommendations
assessed treatment characteristics (number of EFT by Crane et al. (2000). Initial level of participants’
sessions, length of EFT sessions, percent of EFT distress did not predict response to EFT, Q(1) =
sessions checked for fidelity). Two continuous 0.88, p = .35.
moderators were assessed for the pretest to
follow-up analyses: (a) months to follow-up,
and (b) attrition posttest to follow-up. 5
Graphs of continuous moderators regressed on standard-
Selection and Attrition Bias. Ten studies pro- ized differences in mean scores are available upon request.
vided rates of researcher exclusion of couples 6
All means used in these regression moderator analyses
while screening during study recruitment, with are reported as unweighted means.
 META-ANALYSIS OF EMOTIONALLY FOCUSED COUPLE THERAPY 91

Treatment Characteristics. The mean number for the data at the longest reported time period.
of EFT sessions in the 19 pretest–posttest studies Length of follow-up was associated with a decline
ranged from 6 (Dandeneau & Johnson, 1994) to in the treatment effect from pretest to time of
31 (Weissman et al., 2017), M = 13.89. Number follow-up, Q(1) 4.71, p < .05. Figure 4 provides a
of sessions was associated with less benefit from visual presentation of change from termination to
treatment, Q(1) = 5.48, p < .05. Length of time for follow-up. Attrition from termination to follow-
sessions was reported in 13 of the 19 pretest– up ranged from 0% (Dandeneau & Johnson,
posttest studies; six studies used 60 min, five 1994; James, 1991; Lee et al., 2017) to 77.78%
provided 75 min, and two reported 90 min (Makinen & Johnson, 2006), M = 25.53. Higher
each session. The length of time for sessions dropout rates at follow-up was also associated
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

was unrelated to the EFT treatment effect, with less benefit for the couples who participated
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Q(1) = 0.92, p = .39. Percent of sessions checked in follow-up, Q(9) = 7.46, p < .01.
for treatment fidelity with therapist feedback
could be assessed in 18 of the pretest–posttest Publication Bias
studies and ranged from 0.00 (Johnson &
Greenberg, 1985b; MacIntosh & Johnson, For the sensitivity analysis we further analyzed
2008; MacPhee et al., 1995; McLean et al., pre- and post-test studies for publication bias.
2008; Weissman et al., 2017; Wittenborn et al., Publication bias occurs when reviewers and edi-
2019) to 0.50 (Dandeneau & Johnson, 1994), M = tors are, for example, more inclined to publish
18.89. Higher percentage of sessions checked for statistically significant results compared with
treatment fidelity was associated with larger EFT statistically nonsignificant (Rothstein et al.,
treatment effects, Q(17) = 6.19, p < .05. 2005). A fail-safe analysis (Rosenthal, 1979)
Follow-Up Characteristics. Follow-up data resulted in a two-tailed Z-value of 15.56 ( p <
were available for 10 studies reported in either .00001) for the 19 pre- and post-test studies. To
the original study or subsequent publication. obtain a statistically nonsignificant Z-value,
Length of follow-up varied from 2 (Johnson & under the cutoff of 1.96, a total of 1,180 zero
Greenberg, 1985a) to 24 months (e.g., Cloutier et effect EFT studies would need to exist, which
al., 2002). In three studies, more than one time- seems implausible. A criticism of the fail-safe
stamp was reported for follow-up (e.g., 6, 12, 18, analysis is the assumption that missing studies
24 months; Wiebe, 2014). Due to nonindepen- would have zero effects, with no consideration of
dence of these data we elected to assess follow-up study weights (Becker, 2005). In response to this

Figure 4
Growth From Termination to Follow-Up
Effect Size (Cohen's d)

Termination Follow-up

Note. See the online article for the color version of this figure.
 92 SPENGLER, LEE, WIEBE, AND WITTENBORN

criticism, we visually inspected a funnel plot with research reviews (see Lambert, 2013), and for
EFT pre- and post-test d effects on the x-axis and a couple and family therapy, ranging from d =
measure of study sample size, or precision (one .59–.84 (Dunn & Schwebel, 1995; Shadish &
standard error) in this instance, on the y-axis (see Baldwin, 2003). Interestingly, we found no sta-
Supplemental Figure S1). Larger studies cluster tistical difference between effects in studies that
toward the top and smaller studies toward the originated within Johnson’s research lab and
bottom of the funnel. Visual inspection reflects no those from outside. In fact, the overall mean d
asymmetry suggesting there is no bias to publish effect for those studies originating outside of
EFT studies related to the magnitude of their Johnson’s lab was higher (d = 1.01 vs. .89),
effect. Egger et al. (1997) proposed, to further tentatively dispelling notions of researcher alle-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

test this assumption, using the inverse of the giance or bias. Moreover, we did not find any
This document is copyrighted by the American Psychological Association or one of its allied publishers.

standard error (precision) to predict the EFT evidence that the design (RCT vs. quasi-experi-
pre- and post-test effect which results in a two- mental) or study quality moderated this effect,
tailed t(17) = 0.51, p = .66. This finding means the highlighting the importance we placed on ana-
distribution of effect sizes is symmetrical giving lyzing 100% of EFT efficacy research, then ana-
further support to there being no evidence of lyzing design differences, versus excluding about
publication bias. 50% of studies, a priori, as in previous meta-
analyses.
Discussion We found a pretreatment to follow-up d effect of
.86 from 10 studies for those couples who partici-
In this study we sought to provide a compre- pated in follow-up assessments at lengths ranging
hensive review of the outcome research pertain- from 2 to 24 months. This is also a large effect per
ing to the efficacy of EFT (Johnson, 2020). the benchmarks outlined by Cohen (1988) and is
Previous EFT meta-analyses have either included on par with the gains made pre- to post-test in EFT,
RCTs only, combined different models of couple suggesting overall relative stability in couple
therapy, or mixed theoretically related, but con- gains. Conducting follow-up assessment in psy-
ceptually different models of couple therapy with chotherapy generally, and couple therapy specifi-
the acronym EFT. We solely focused on Sue cally, has been notoriously hard to do regardless
Johnson’s model for EFT given its distinct theo- of the model being studied. In perhaps the largest
retical and conceptual assumptions compared to, and longest follow-up study of couple therapy,
for example, the model of emotion-focused ther- Christensen et al. (2010) followed 134 couples for
apy put forth by Greenberg and Goldman (2008). 5 years after completing treatment with either
Additionally, our meta-analysis was not limited integrative behavioral couple therapy (IBCT) or
to RCTs, but included quasi-experimental stud- traditional behavioral couple therapy (TBCT). At
ies, all pretest–posttest designs, and one disserta- 5 years follow-up they found a pretreatment to
tion. The overall results show that EFT is a robust follow-up d effect of 1.03 for the IBCT group and
treatment for couple distress, that it compares .92 for the TBCT group. We urge caution in
well to a limited sampling of alternative treat- simply comparing d effects at follow-up between
ments, and that the gains achieved at the end of IBCT/TBCT and EFT without embedding this
treatment are sustained up to 2 years with modest comparison within context. The Christensen et
decline over time. al. (2010) project was a large-scale RCT where
Johnson et al. (1999) first reported a d effect of therapists received strict training and oversight in
1.3 after meta-analyzing four existing RCTs of deploying their respective treatment protocols.
EFT published at that time. We expected a down- Additionally, couples were paid increasing rates
ward correction in the overall d effect given the at each time point for their participation incentiv-
increase in EFT studies of varying scope and izing their likelihood to follow through. The
design. Across this body of outcome studies, corpus of EFT follow-up research presented in
we found a random effects weighted d of .93 this meta-analysis varies in design, the degree to
pretest–posttest for improvements in relationship which EFT fidelity was monitored, and appears
adjustment. Using Cohen’s (1988) conventions, to be influenced by moderators such as attrition
this is a large effect that exceeds d effects found in and therapist fidelity to the model.
general for individual psychotherapy, ranging In addition to examining the effect of EFT from
on average from d of .70–.80 s noted in broad pretest–posttest, and through follow-up, when
 META-ANALYSIS OF EMOTIONALLY FOCUSED COUPLE THERAPY 93

available we examined the differential effect of Moderators are essentially interaction effects
EFT compared to other interventions in six stud- assessed in meta-analytic studies and can be
ies. We found an overall d = .44 favoring EFT informative for future researchers, practitioners,
treatment. This effect was homogenous across and trainers. The strongest moderator finding is
studies; meaning, the variability in effects across the association between EFT treatment adher-
comparison studies was largely attributable to ence, or fidelity, and EFT efficacy. Studies that
differences in treatment and not chance. A criti- had higher percentages of sessions checked for
cism often cited against differences between treatment adherence, with feedback provided to
treatment effects is unequal comparisons, such clinicians, were associated with better couple
as researchers pitting a well-established treatment outcomes. Likewise, therapists who received
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

protocol against a nonbonafide approach. Of study specific EFT training produced substan-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

the six studies that compared EFT to some tially better follow-up outcomes (d = 1.17),
other form of treatment, two interventions were compared with therapists who only had prior
pharmacotherapy-based, two were a general form EFT experience but no on-study training (d =
of cognitive interventions, one was integrated- .57). The method used to monitor treatment
systemic, and the final was treatment as usual. As adherence, such as bug-in-the-ear with mirror
such, it is important to situate our finding of supervision and videotape feedback, or videotape
EFT being more effective to comparison treat- feedback only, only approached the convention
ments within a context of what kinds of interven- for statistical significance ( p = .06). These
tions were used. Not all these interventions, in findings collectively lend support to the notion
our opinion, rise to the level of what Wampold that following the EFT model, pursuing EFT
et al. (1997) called well-established, or bonafide. training and supervision, and striving for
Future research is needed to address this question model fidelity will lead to greater therapeutic
by comparing EFT to manualized and well- benefits for couples (Wiebe & Johnson, 2016).
established couple treatments conducted by clin- Sexton et al. (2011) called for evidence-based
icians equally trained in each model. Comparing treatments to demonstrate a relation between
effect sizes obtained from different meta-analyses model adherence and treatment efficacy, which
of couple therapies is problematic due to differ- is supported by these moderator analyses. Future
ences in the way the studies are conducted (e.g., research should systematically assess for which
Christensen et al., 2010). methods of EFT supervision and feedback are
most beneficial to couple outcomes. An extensive
body of research exists on EFT supervision and
What Accounts for EFT Treatment training that could be applied to this question
Differences Across Studies? (e.g., see Sandberg, 2011). Likewise, it could
be productive to conduct process-outcome
Unlike previous meta-analyses we sought to research on differences between low and high-
comprehensively examine moderators of the effi- fidelity EFT therapists correlated with follow-up
cacy of EFT and did so in four domains: (a) outcomes. It could be that high-fidelity therapists
treatment characteristics, (b) client characteris- are smoother, deeper, more incisive in their
tics, (c) therapist characteristics and (d) method- delivery of the treatment, or more educational
ological differences. Not all EFT meta-analyses resulting in greater generalization and mainte-
assessed for heterogeneity in variance between nance of gains made by couples.
studies (e.g., Beasley & Ager, 2019; Byrne et al., It was curious that more EFT sessions provided
2004). Failure to do so can lead to misleading to couples was associated with less benefit over-
findings and an overabundance of trust in the all. Several variables may be at play here, such as
meaning of a singular index (Spielmans & attrition and the need for more sessions for
Flückiger, 2018). Notably, we found significant more severely distressed couples. Consequently,
variance between studies in the pretest–posttest we inspected those studies with the highest num-
and follow-up analyses, which makes our confi- ber of sessions and found, in addition to relation-
dence in these summary d effects more tentative. ship distress, those couples were presenting with
Heterogeneity of variance could reflect differ- comorbid concerns such as sexual abuse and
ences that are clinically significant and/or signifi- trauma, posttraumatic stress, and major depres-
cant in terms of methodological variables. sive disorder. It is possible, therefore, that this
 94 SPENGLER, LEE, WIEBE, AND WITTENBORN

finding has less to do with the number of sessions EFT can be converted to a success rate for treated
being provided and more to do with the nature of couples (Norcross et al., 2017). For a d effect of
complex distress these couples were experienc- .93 this translates to 70% of couples in EFT are
ing. As any practitioner treating couples can successfully treated for their relationship distress.
attest to rarely do couples in routine clinical This finding is consistent with the frequently
settings present for issues pertaining to relation- cited finding where Johnson et al. (1999) con-
ship concerns alone (Whisman, 2007). Length of cluded that 70%–73% of couples are symptom
follow-up and higher attrition rates at follow-up free at the end of EFT.
were also associated with less treatment gains.
Further EFT research is needed to fully unpack
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

who is not responding best to EFT and not Limitations and Implications for
This document is copyrighted by the American Psychological Association or one of its allied publishers.

maintaining gains post treatment. The fact that Future Research

therapists with more study specific EFT supervi-
sion produced better follow-up outcomes seems While our study was more comprehensive,
promising. Perhaps therapists who adhere more and we were able to glean more meaningful
closely to the EFT model inoculate couples from moderator analyses, like other couple therapy
future couple distress by being better at experi- meta-analyses few statistically significant mod-
ential and reflective “teaching” of new relation- erators of the EFT effects were found. This
ship skills. could be an artifact of low statistical power
The unit of analysis used by researchers, either related to the modest sample size of 20 studies
analyses of averaged couple scores (d = 1.10) or with a mean of 14.3 couples per study (for
analyses of couples’ individual scores (d = .68), further discussion, see Hedges & Pigott,
was a significant methodological moderator. 2004). Further, because of restrictions in how
With few exceptions (e.g., Wittenborn et al., the data are reported nearly all the moderator
2019), existing EFT outcome studies did not analyses were conducted on a between study
use dyadic methods for data analysis (e.g., nesting basis versus the more powerful subgroups or
individuals within couples) to account for couple within-studies basis (Wittenborn et al., 2013).
interdependence. This issue of unit of analysis This artifact of the EFT research led us to
created a conundrum for us, as it resulted in comparisons with lower certainty because the
studies that analyzed at the individual level being moderators assess each study as a whole and
weighted more heavily than warranted. For exam- not the specific attribute. Wood and Eagly
ple, if there were 20 couples and one study (2009) argue that the moderator may then be
analyzed as N = 20 and the other analyzed as a proxy with other study attributes, thereby
N = 40 that means the second study received reducing statistical conclusion validity.
more weight in the overall meta-analysis. We We agree with Norcross and Cooper (2021)
recommend that future EFT researchers utilize who stated, “It is a false, and to be blunt,
methodologically and statistically sound dyadic misleading presupposition in RCTs that patients
research methods that account for couple interde- are homogeneous” (p. ii). Therefore, one area
pendence and thereby provide unbiased estimates that could be extended in EFT research is to
of treatment effects (for resource on dyadic examine relevant subgroups within studies. We
research, see Wittenborn et al., 2013). were able to do this only for gender and in only
three studies, thereby limiting our ability to
Practical Meaning of Overall EFT Effect assess differences between cisgender heterosex-
ual men’s and women’s responses to EFT. We
It is important to assess the practical meaning sought to code other meaningful subgroups
of research findings so they can be easily con- (e.g., offending partner vs. nonoffending part-
sumed by practitioners and shared with couples. ner) but too few studies existed. Several sub-
A d of .93 means that couples who participate groups might be systematically explored in
in EFT under experimental conditions, for exam- accordance with attachment theory and the
ple, with additional supervision, fidelity checks, EFT model, for example, (a) in studies of het-
and other characteristics of outcome research, are erosexual couples men and women given the
on average better off than 81% of couples who do idea that men are not as emotionally attuned and
not receive treatment. The overall effect size for tend to be withdrawers, (b) withdrawers versus
 META-ANALYSIS OF EMOTIONALLY FOCUSED COUPLE THERAPY 95

blamers, and (c) attachment style. Several mod- Clinical Implications and Applications
erators were not statistically significant (e.g.,
length of session), and we urge caution in inter- Despite these limitations practitioners can
preting these findings based on the low statistical conclude that EFT is an efficacious approach
power of the moderator analyses. to treating couple distress and, with caveats
Like other couple therapy research, the parti- noted, that treatment gains are relatively lasting.
cipants in this body of EFT research are Despite working with couples on a regular basis
predominantly White, middle-class, and moder- very few practitioners receive substantive train-
ately to mildly distressed, and are, or are pre- ing in evidence-based couple treatments beyond
sumed to be, cisgender, heterosexual couples a single course or two in graduate school (Lee &
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(for a comprehensive review of sexual and gen- Spengler, 2017). Based on this meta-analysis
This document is copyrighted by the American Psychological Association or one of its allied publishers.

der identity in couple therapy research, see and findings from process and other areas of
Spengler et al., 2020). The differences between EFT research (Wiebe & Johnson, 2016), practi-
couples who participate in efficacy studies and tioners can feel confident studying and providing
in naturalistic treatment settings have been EFT to couples. ICEEFT has a training library of
widely discussed (e.g., Wright et al., 2007). master videotapes with real couples in real treat-
To make meaningful assumptions about the ment; a network of consultants and supervisors;
generalizability of EFT to diverse couples training opportunities in large and small groups;
help-seeking in routine practice, EFT research- and a competency-based certification program
ers need to tackle this issue directly by actively aimed at increasing the fidelity with which prac-
recruiting and conducting studies with more titioners practice EFT. Chow et al. (2015) found
diverse participants. As discussed by Spengler that therapists who engaged in regular deliberate
et al. (2020), the implicit or explicit inclusion practice, such as is available in studying EFT,
or exclusion of sexual and gender minority achieve better client outcomes than those who
couples, and a general lack of consideration of did not. The results of this meta-analysis suggest
other sociocultural factors such as race and practitioners can confidently engage in such
ethnicity, in couple therapy outcome research study and practice of EFT to the benefit of their
risks making these couples invisible. Future EFT couples.
researchers are urged to expand tests of the
efficacy of the EFT model and to routinely report
References
on self-identified measures of couple character-
istics such as sexual, gender, racial, and ethnic References with an asterisk indicate studies in the
identity. In our opinion (Spengler et al., 2020), meta-analysis.
these are important areas for the future develop-
ment of research on EFT and, for that matter, all Alder, M. C., Dyer, W. J., Sandberg, J. G., Davis,
models of couple therapy. S. Y., & Holt-Lunstad, J. (2018). Emotionally-
Of course, questions of “transportability” or focused therapy and treatment as usual comparison
groups in decreasing depression: A clinical pilot
external validity exist with RCTs and quasi- study. The American Journal of Family Therapy,
experimental studies (see Lebow et al., 2012). 46(5), 541–555. https://doi.org/10.1080/01926187
As with other couple therapies more research is .2019.1572478
needed on the effectiveness of EFT in naturalistic Beasley, C. C., & Ager, R. (2019). Emotionally
settings. Mitchell and Spengler (2022) recently focused couples therapy: A systematic review of
conducted a naturalistic study of EFT in routine its effectiveness over the past 19 years. Journal of
clinical settings. Across seven therapists and 11 Evidence-Based Social Work, 16(2), 144–159.
couples being followed for 18 months, Mitchell https://doi.org/10.1080/23761407.2018.1563013
and Spengler found a statistically significant d Becker, B. J. (2005). The failsafe N or file-drawer
number. In H. R. Rothstein, A. J. Sutton, & M.
effect of .46 for relationship improvement. Such
Borenstein (Eds.), Publication bias in meta-analy-
shrinkage in effect size from efficacy to effective- sis: Prevention, assessment and adjustments (pp.
ness studies is common and should lead practi- 111–126). Wiley.
tioners to be cautious in assuming the findings Borenstein, M., Hedges, L. V., Higgins, J., & Roth-
from this EFT meta-analysis will generalize to stein, H. (2005). Comprehensive meta-analysis ver-
their clinical practice. sions 2. Biostat.
 96 SPENGLER, LEE, WIEBE, AND WITTENBORN

Borenstein, M., Hedges, L. V., Higgins, J., & Roth- Therapy, 28(1), 53–60. https://doi.org/10.1080/019
stein, H. (2009). Introduction to meta-analysis. 261800261815
Wiley. https://doi.org/10.1002/9780470743386 *Dalgleish, T. L., Johnson, S. M., Burgess Moser, M.,
Borman, G. D., & Grigg, J. A. (2009). Visual and Wiebe, S. A., & Tasca, G. A. (2015). Predicting key
narrative interpretation. In H. Cooper, L. V. Hedges, change events in emotionally focused couple ther-
& J. C. Valentine (Eds.), The handbook of research apy. Journal of Marital and Family Therapy, 41(3),
synthesis and meta-analysis (pp. 497–520). Sage 260–275. https://doi.org/10.1111/jmft.12101
Publications. *Dalton, E. J., Greenman, P. S., Classen, C. C., &
Busby, D. M., Crane, D. R., Larson, J. H., & Christensen, Johnson, S. M. (2013). Nurturing connections in the
C. (1995). A revision of the Dyadic Adjustment Scale aftermath of childhood trauma: A randomized con-
for use with distressed and nondistressed couples: trolled trial of emotionally focused couple therapy
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Construct hierarchy and multidimensional scales. Jour- for female survivors of childhood abuse. Couple &
This document is copyrighted by the American Psychological Association or one of its allied publishers.

nal of Marital and Family Therapy, 21(3), 289–308. Family Psychology, 2(3), 209–221. https://doi.org/
https://doi.org/10.1111/j.1752-0606.1995.tb00163.x 10.1037/a0032772
Byrne, M., Carr, A., & Clark, M. (2004). The efficacy *Dandeneau, M. L., & Johnson, S. M. (1994). Facili-
of behavioral couples therapy and emotionally tating intimacy: Interventions and effects. Journal of
focused therapy for couple distress. Contemporary Marital and Family Therapy, 20(1), 17–33. https://
Family Therapy, 26(4), 361–387. https://doi.org/10 doi.org/10.1111/j.1752-0606.1994.tb01008.x
.1007/s10591-004-0642-9 *Denton, W. H., Wittenborn, A. K., & Golden, R. N.
Card, N. (2012). Applied meta-analysis for social (2012). Augmenting antidepressant medication
science research. Guilford. treatment of depressed women with emotionally
Carr, A. (2018). Couple therapy, family therapy and focused therapy for couples: A randomized pilot
systemic interventions for adult-focused problems: study. Journal of Marital and Family Therapy,
The current evidence base. Journal of Family Ther- 38(Suppl. 1), 23–38. https://doi.org/10.1111/j.17
apy, 41(4), 492–536. https://doi.org/10.1111/1467- 52-0606.2012.00291.x
6427.12225 *Dessaulles, A. (1991). The treatment of clinical
Chow, D. L., Miller, S. D., Seidel, J. A., Kane, R. T., depression in the context of marital distress [Doc-
Thornton, J. A., & Andrews, W. P. (2015). The role toral Dissertation, University of Ottawa]. ProQuest
of deliberate practice in the development of highly Dissertations Publishing.
effective psychotherapists. Psychotherapy, 52(3), *Dessaulles, A., Johnson, S. M., & Denton, W. H.
337–345. https://doi.org/10.1037/pst0000015 (2003). Emotion-focused therapy for couples in the
Christensen, A., Atkins, D. C., Baucom, B., & Yi, J. treatment of depression: A pilot study. The Ameri-
(2010). Marital status and satisfaction five years can Journal of Family Therapy, 31(5), 345–353.
following a randomized clinical trial comparing https://doi.org/10.1080/01926180390232266
traditional versus integrative behavioral couple ther- Dunn, R. L., & Schwebel, A. I. (1995). Meta-analytic
apy. Journal of Consulting and Clinical Psychology, review of marital therapy outcome research. Jour-
78(2), 225–235. https://doi.org/10.1037/a0018132 nal of Family Psychology, 9(1), 58–68. https://
*Cloutier, P. F., Manion, I. G., Walker, J. G., & doi.org/10.1037/0893-3200.9.1.58
Johnson, S. M. (2002). Emotionally focused inter- Egger, M., Davey Smith, G., Schneider, M., & Minder,
ventions for couples with chronically ill children: A C. (1997). Bias in meta-analysis detected by a sim-
2-year follow-up. Journal of Marital and Family ple, graphical test. The BMJ, 315(7109), 629–634.
Therapy, 28(4), 391–398. https://doi.org/10.1111/j https://doi.org/10.1136/bmj.315.7109.629
.1752-0606.2002.tb00364.x *Goldman, A., & Greenberg, L. (1992). Comparison
Cohen, J. (1960). A coefficient of agreement for of integrated systemic and emotionally focused
nominal scales. Educational and Psychological approaches to couples therapy. Journal of Consult-
Measurement, 20(1), 37–46. https://doi.org/10 ing and Clinical Psychology, 60(6), 962–969.
.1177/001316446002000104 https://doi.org/10.1037/0022-006X.60.6.962
Cohen, J. (1988). Statistical power analysis for the *Gordon-Walker, J., Johnson, S. M., Manion, I., &
behavioral sciences (2nd ed.). Lawrence Erlbaum. Cloutier, P. (1996). Emotionally focused marital
Cooper, H. (1998). Synthesizing research: A guide for intervention for couples with chronically ill chil-
literature reviews (3rd ed.). Sage Publications. dren. Journal of Consulting and Clinical Psychol-
Cooper, H., Hedges, L. V., & Valentine, J. C. (2009). ogy, 64(5), 1029–1036. https://doi.org/10.1037/
The handbook of research synthesis and meta-anal- 0022-006X.64.5.1029
ysis (2nd ed.). Sage Publications. Greenberg, L. S., & Goldman, R. N. (2008). Emotion-
Crane, D. R., Middleton, K. C., & Bean, R. A. (2000). focused couples therapy. American Psychological
Establishing criterion scores for the Kansas Mar- Association. https://doi.org/10.1037/11750-000
ital Satisfaction Scale and the Revised Dyadic Greenberg, L. S., & Johnson, S. M. (1988). Emotion-
Adjustment Scale. The American Journal of Family ally focused therapy for couples. Guilford Press.
 META-ANALYSIS OF EMOTIONALLY FOCUSED COUPLE THERAPY 97

Greenhouse, J. B., & Iyengar, S. (2009). Sensitivity Science and Practice, 6(1), 67–79. https://doi.org/10
analysis and diagnostics. In H. Cooper, L. V. .1093/clipsy.6.1.67
Hedges, & J. C. Valentine (Eds.), The handbook *Johnson, S. M., & Talitman, E. (1997). Predictors of
of research synthesis and meta-analysis (pp. 417– success in emotionally focused marital therapy.
434). Sage Publications. Journal of Marital and Family Therapy, 23(2),
Halchuk, R. E., Makinen, J. A., & Johnson, S. M. 135–152. https://doi.org/10.1111/j.1752-0606.1997
(2010). Resolving attachment injuries in couples .tb00239.x
using emotionally focused therapy: A three-year Johnson, S. M., & Wittenborn, A. K. (2012). New
follow-up. Journal of Couple and Relationship research findings on emotionally focused therapy:
Therapy, 9(1), 31–47. https://doi.org/10.1080/153 Introduction to special section. Journal of Marital
32690903473069 and Family Therapy, 38(Suppl. 1), 18–22. https://
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Halford, W. K., Pepping, C. A., & Petch, J. (2015). doi.org/10.1111/j.1752-0606.2012.00292.x

This document is copyrighted by the American Psychological Association or one of its allied publishers.

The gap between couple therapy research efficacy Lambert, M. J. (2013). Outcome in psychotherapy:
and practice effectiveness. Journal of Marital and The past and important advances. Psychotherapy,
Family Therapy, 42(1), 32–44. https://doi.org/10 50(1), 42–51. https://doi.org/10.1037/a0030682
.1111/jmft.12120 Landis, J. R., & Koch, G. G. (1977). The measurement
Hedges, L. V., & Olkin, I. (1985). Statistical methods of observer agreement for categorical data. Bio-
for meta-analysis. Academic Press. metrics, 33(1), 159–174. https://doi.org/10.2307/
Hedges, L. V., & Pigott, T. D. (2004). The power of 2529310
statistical tests for moderators in meta-analysis. Lebow, J. L., Chambers, A. L., Christensen, A., &
Psychological Methods, 9(4), 426–445. https:// Johnson, S. M. (2012). Research on the treatment of
doi.org/10.1037/1082-989X.9.4.426 couple distress. Journal of Marital and Family
Higgins, J. P. T., Thompson, S. G., Deeks, J. J., & Therapy, 38(1), 145–168. https://doi.org/10.1111/j
Altman, D. G. (2003). Measuring inconsistency in .1752-0606.2011.00249.x
meta-analyses. The BMJ, 327(7414), 557–560. *Lee, N. A., Furrow, J. L., & Bradley, B. A. (2017).
https://doi.org/10.1136/bmj.327.7414.557 Emotionally focused couple therapy for parents
Hoaglin, D. C., & Iglewicz, B. (1987). Fine tuning raising a child with an autism spectrum disorder:
some resistant rules for outlier labeling. Journal of A pilot study. Journal of Marital and Family Ther-
the American Statistical Association, 82(400), apy, 43(4), 662–673. https://doi.org/10.1111/jm
1147–1149. https://doi.org/10.1080/01621459.19 ft.12225
87.10478551 Lee, N. A., & Spengler, P. (2017). Training counseling
*James, P. S. (1991). Effects of a communication psychologists in couple and family therapy. In J. L.
training component added to an emotionally Lebow, A. L. Chambers, & D. C. Breunlin (Eds.),
focused couples therapy. Journal of Marital and Encyclopedia of couple and family therapy.
Family Therapy, 17(3), 263–275. https://doi.org/10 Springer Publishing Company. https://doi.org/10
.1111/j.1752-0606.1991.tb00894.x .1007/978-3-319-15877-8_707-1
Johnson, S. M. (1996). The practice of emotionally *MacIntosh, H. B., & Johnson, S. (2008). Emotionally
focused couple therapy: Creating connection. Focused Therapy for couples and childhood sexual
Routledge. abuse survivors. Journal of Marital and Family
Johnson, S. M. (2004). The practice of emotionally Therapy, 34(3), 298–315. https://doi.org/10.1111/j
focused couple therapy: Creating connection (2nd .1752-0606.2008.00074.x
ed.). Routledge. *MacPhee, D. C., Johnson, S. M., & Van der Veer,
Johnson, S. M. (2020). The practice of emotionally M. M. C. (1995). Low sexual desire in women: The
focused couple therapy: Creating connection (3rd effects of marital therapy. Journal of Sex & Marital
ed.). Routledge. Therapy, 21(3), 159–182. https://doi.org/10.1080/
*Johnson, S. M., & Greenberg, L. S. (1985a). Differ- 00926239508404396
ential effects of experiential and problem-solving *Makinen, J. A., & Johnson, S. M. (2006). Resolving
interventions in resolving marital conflict. Journal attachment injuries in couples using emotionally
of Consulting and Clinical Psychology, 53(2), 175– focused therapy: Steps toward forgiveness and rec-
184. https://doi.org/10.1037/0022-006X.53.2.175 onciliation. Journal of Consulting and Clinical
*Johnson, S. M., & Greenberg, L. S. (1985b). Emo- Psychology, 74(6), 1055–1064. https://doi.org/10
tionally focused couples therapy: An outcome .1037/0022-006X.74.6.1055
study. Journal of Marital and Family Therapy, *McLean, L. M., Jones, J. M., Rydall, A. C., Walsh,
11(3), 313–317. https://doi.org/10.1111/j.1752- A., Esplen, M. J., Zimmermann, C., & Rodin, G. M.
0606.1985.tb00624.x (2008). A couples intervention for patients facing
Johnson, S. M., Hunsley, J., Greenberg, L. S., & advanced cancer and their spouse caregivers: Out-
Schindler, D. (1999). Emotionally focused couples comes of a pilot study. Psycho-Oncology, 17(11),
therapy: Status and challenges. Clinical Psychology: 1152–1156. https://doi.org/10.1002/pon.1319
 98 SPENGLER, LEE, WIEBE, AND WITTENBORN

*McLean, L. M., Walton, T., Rodin, G., Esplen, M. J., Guidelines for classifying evidence-based treat-
& Jones, J. M. (2013). A couple-based intervention ments in couple and family therapy. Family Pro-
for patients and caregivers facing end-stage cancer: cess, 50(3), 377–392. https://doi.org/10.1111/j
Outcomes of a randomized controlled trial. Psycho- .1545-5300.2011.01363.x
Oncology, 22(1), 28–38. https://doi.org/10.1002/ Shadish, W. R., & Baldwin, S. A. (2003). Meta-
pon.2046 analysis of MFT interventions. Journal of Marital
Mitchell, A. M., & Spengler, P. M. (2022). A pilot and Family Therapy, 29(4), 547–570. https://
study of the effectiveness of naturally delivered doi.org/10.1111/j.1752-0606.2003.tb01694.x
emotionally focused couple therapy [Manuscript Shadish, W. R., Cook, T. D., & Campbell, D. T.
submitted for publication]. (2002). Experimental and quasi-experimental de-
*Najafi, M., Soleimani, A. A., Ahmadi, K., Javidi, N., signs for generalized causal inference. Houghton-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

& Kamkar, E. H. (2015). The effectiveness of Mifflin and Company.

This document is copyrighted by the American Psychological Association or one of its allied publishers.

emotionally focused therapy on enhancing marital *Sims, A. E. B. (2000). Working models of attach-
adjustment and quality of life among infertile cou- ment: The impact of emotionally focused marital
ples with marital conflicts. International Journal of therapy [Doctoral Dissertation, University of
Fertility & Sterility, 9(2), 238–246. https://doi.org/ Ottawa]. ProQuest Dissertation Publishing.
10.22074/ijfs.2015.4245 *Solymani ahmadi, F., Zarei, E., & Fallahchai, S. R.
Norcross, J. C., & Cooper, M. (2021). Personalizing (2014). The effectiveness of emotionally-focused
psychotherapy: Assessing and accommodating couple therapy in resolution of marital conflicts
patient preferences. American Psychological between the couples who visited the consultation
Association. https://doi.org/10.1037/0000221-000 centers. Journal of Education and Management
Norcross, J. C., Hogan, T. P., Koocher, G. P., & Studies, 4(1), 118–123.
Maggio, L. A. (2017). Clinicians’ guide to Spanier, G. (1976). Measuring dyadic adjustment:
evidence-based practices: Behavioral health and New scales for assessing the quality of marriage
addictions (2nd ed.). Oxford University Press. and similar dyads. Journal of Marriage and Family,
https://doi.org/10.1093/med:psych/9780190621933 38(1), 15–28. https://doi.org/10.2307/350547
.001.0001 Spengler, E. S., DeVore, E. N., Spengler, P. M., &
Overton, R. C. (1998). A comparison of fixed-effects Lee, N. A. (2020). What does “couple” mean in
and mixed (random effects) models for meta- couple therapy outcome research? A systematic
analysis tests of moderator variable effects. Psycho- review of implicit and explicit, inclusion and
logical Methods, 3(3), 354–379. https://doi.org/10 exclusion of gender and sexual minority indivi-
.1037/1082-989X.3.3.354 duals and identities. Journal of Marital and Family
Rathgeber, M., Bürkner, P. C., Schiller, E. M., & Therapy, 46(2), 240–255. https://doi.org/10.1111/
Holling, H. (2018). The efficacy of emotionally jmft.12415
focused couples therapy and behavioral couples Spielmans, G. I., & Flückiger, C. (2018). Moderators
therapy: A meta-analysis. Journal of Marital Ther- in psychotherapy meta-analysis. Psychotherapy
apy, 45(3), 447–463. https://doi.org/10.1111/jm Research, 28(3), 333–346. https://doi.org/10.10
ft.12336 80/10503307.2017.1422214
Roddy, M. K., Walsh, L. M., Rothman, K., Hatch, Sprenkle, D. H. (2012). Intervention research in cou-
S. G., & Doss, B. D. (2020). Meta-analysis of ple and family therapy: A methodological and
couple therapy: Effects across outcomes, designs, substantive review and an introduction to the special
timeframes, and other moderators. Journal of Con- issue. Journal of Marital and Family Therapy,
sulting and Clinical Psychology, 88(7), 583–596. 38(1), 3–29. https://doi.org/10.1111/j.1752-0606
https://doi.org/10.1037/ccp0000514 .2011.00271.x
Rosenthal, R. (1979). The file drawer problem and Tukey, J. W. (1977). Exploratory data analysis.
tolerance for null results. Psychological Bulletin, Addison Wesley.
86(3), 638–641. https://doi.org/10.1037/0033-2909 *Walker, J. G., Johnson, S., Manion, I., & Cloutier, P.
.86.3.638 (1996). Emotionally focused marital intervention
Rothstein, H. R., Sutton, A. J., & Borenstein, M. for couples with chronically ill children. Journal
(Eds.). (2005). Publication bias in meta-analysis: of Consulting and Clinical Psychology, 64(5),
Prevention, assessment and adjustments. Wiley. 1029–1036. https://doi.org/10.1037/0022-006X.64
Sandberg, J. G. (2011). Introduction to the special .5.1029
section on learning Emotionally Focused Couples Wampold, B. E., Mondin, G. W., Moody, M., Stitch,
Therapy. Journal of Marital and Family Therapy, M., Benson, K., & Ahn, H. (1997). A meta-analysis
37(4), 377–379. https://doi.org/10.1111/j.1752- of outcome studies comparing bona fide psy-
0606.2011.00251.x chotherapies: Empirically, “all must have prizes.”
Sexton, T., Gordon, K. C., Gurman, A., Lebow, J., Psychological Bulletin, 122(3), 203–215. https://
Holtzworth-Munroe, A., & Johnson, S. (2011). doi.org/10.1037/0033-2909.122.3.203
 META-ANALYSIS OF EMOTIONALLY FOCUSED COUPLE THERAPY 99

*Weissman, N., Batten, S. V., Rheem, K. D., Wiebe, (2019). Randomized controlled trial of emotion-
S. A., Pasillas, R. M., Potts, W., Barone, M., Brown, ally focused couple therapy compared to treat-
C. H., & Dixon, L. B. (2017). The effectiveness of ment as usual for depression: Outcomes and
emotionally focused couples therapy with veterans mechanisms of change. Journal of Marital and
with PTSD: A pilot study. Journal of Couple & Family Therapy, 45(3), 395–409. https://doi.org/
Relationship Therapy, 17(1), 25–41. https://doi.org/ 10.1111/jmft.12350
10.1080/15332691.2017.1285261 Wood, N. D., Crane, D. R., Schaalje, B., & Law, D. D.
Whisman, M. A. (2007). Marital distress and DSM-IV (2005). What works for whom: A meta-analytic
psychiatric disorders in a population-based national review of marital and couples therapy in reference
survey. Journal of Abnormal Psychology, 116(3), to marital distress. The American Journal of Family
638–643. https://doi.org/10.1037/0021-843X.116 Therapy, 33(4), 273–287. https://doi.org/10.1080/
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

.3.638 01926180590962147
This document is copyrighted by the American Psychological Association or one of its allied publishers.

*Wiebe, S. A. (2014). A two-year longitudinal study Wood, W., & Eagly, A. H. (2009). Advantages of
of emotionally focused therapy for couples: Main- certainty and uncertainty. In H. Cooper, L. V.
tenance and predictors of relationship change Hedges, & J. C. Valentine (Eds.), The handbook
[Doctoral Dissertation, University of Ottawa]. of research synthesis and meta-analysis (pp. 455–
Wiebe, S. A., & Johnson, S. M. (2016). A review of 472). Sage Publications.
the research in emotionally focused therapy for Wright, J., Sabournin, S., Mondor, J., McDuff, P., &
couples. Family Process, 55(3), 390–407. https:// Mamodhoussen, S. (2007). The clinical representa-
doi.org/10.1111/famp.12229 tiveness of couple therapy outcome research. Fam-
Wittenborn, A. K., Dolbin-MacNab, M. L., & Keiley, ily Process, 46(3), 301–316. https://doi.org/10
M. K. (2013). Dyadic research in marriage and family .1111/j.1545-5300.2007.00213.x
therapy: Methodological considerations. Journal of
Marital and Family Therapy, 39(1), 5–16. https://
doi.org/10.1111/j.1752-0606.2012.00306.x Received July 7, 2021
*Wittenborn, A. K., Liu, T., Ridenour, T. A., Lach- Revision received June 23, 2022
mar, E. M., Mitchell, E. A., & Seedall, R. B. Accepted July 8, 2022 ▪