Content uploaded by Joshua R. Novak
Author content
All content in this area was uploaded by Joshua R. Novak on Oct 16, 2015
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=uaft20
Download by: [129.130.22.177] Date: 16 October 2015, At: 09:29
The American Journal of Family Therapy
ISSN: 0192-6187 (Print) 1521-0383 (Online) Journal homepage: http://www.tandfonline.com/loi/uaft20
Measuring Fidelity in Emotionally Focused Couples
Therapy (EFT): A Pilot Test of the EFT Therapist
Fidelity Scale
Jonathan G. Sandberg, Andrew P. Brown, Lori C. Schade, Joshua R. Novak,
Wayne H. Denton & Julianne Holt-Lunstad
To cite this article: Jonathan G. Sandberg, Andrew P. Brown, Lori C. Schade, Joshua R. Novak,
Wayne H. Denton & Julianne Holt-Lunstad (2015) Measuring Fidelity in Emotionally Focused
Couples Therapy (EFT): A Pilot Test of the EFT Therapist Fidelity Scale, The American Journal of
Family Therapy, 43:3, 251-268, DOI: 10.1080/01926187.2015.1034637
To link to this article: http://dx.doi.org/10.1080/01926187.2015.1034637
Published online: 15 May 2015.
Submit your article to this journal
Article views: 83
View related articles
View Crossmark data
The American Journal of Family Therapy, 43:251–268, 2015
Copyright © Taylor & Francis Group, LLC
ISSN: 0192-6187 print / 1521-0383 online
DOI: 10.1080/01926187.2015.1034637
Measuring Fidelity in Emotionally Focused
Couples Therapy (EFT): A Pilot Test of the EFT
Therapist Fidelity Scale
JONATHAN G. SANDBERG, ANDREW P. BROWN, LORI C. SCHADE,
and JOSHUA R. NOVAK
Brigham Young University, Provo, Utah, USA
WAYNE H. DENTON
Florida State University, Tallahassee, Florida, USA
JULIANNE HOLT-LUNSTAD
Brigham Young University, Provo, Utah, USA
Data was drawn from 15 sessions of couple’s therapy where clin-
icians attempted to apply the EFT model. Results provide initial
support for the Emotionally Focused Therapy —Therapist Fidelity
Scale (EFT-TFS) as a reliable and useful measure of fidelity. In
addition, results suggest the measure can discriminate effectively
(95% correct classification) between low and high fidelity EFT. The
findings suggest that certain in-session therapist behaviors (main-
taining session focus on attachment and emotion) may be more
predictive of high fidelity EFT. Implications regarding the EFT-TFS
as a training and practice tool, as well as direction for future re-
search, are discussed.
LITERATURE REVIEW
Marital/Couples Therapy: Effectiveness and Fidelity of Treatment
Over the last 30 years, researchers have empirically established the effec-
tiveness of marital/couples therapy in both the treatment of couple distress
(Lebow, Chambers, Christensen, & Johnson, 2012) and for general relation-
ship enhancement (Markman & Rhoades, 2012). Two specific models (Be-
havioral Couples Therapy and Emotionally Focused Couples Therapy–EFT)
have shown the greatest empirical support (Christensen, Atkins, Baucom,
Address correspondence to Jonathan G. Sandberg, Brigham Young University, 266 TLRB,
Provo, UT 84602. E-mail: jgs@byu.edu
251
Downloaded by [] at 09:29 16 October 2015
252 J. G. Sandberg et al.
& Yi, 2010; Furrow, Johnson, & Bradley, 2011). Proponents of Behavioral
Couples Therapy were the first to attempt to measure therapist fidelity
in research on couples based therapies (Christensen et al., 2004). Within
the literature, there have been two main ways of defining fidelity: adher-
ence and competence. Adherence refers to the degree to which therapists
are delivering model specific techniques and interventions. Competence
refers to the skill with which these techniques or methods are implemented
(Barber et al., 2006; Webb, DeRubeis, & Barber, 2010). Therapist adherence
to, as well as competence within a specific model, is crucial to process and
outcome research where clinician/researchers attempt to identify and repli-
cate the in-session behaviors and processes that lead to effective outcomes
(Shaw et al., 1999). Despite its importance, there is a surprising lack of fidelity
research for couples therapy. This lack of research hampers the translation of
laboratory therapy to the front lines (community settings) of clinical practice
(Vivian et al., 2012). Without fidelity studies, it is impossible to know how
and if evidence based practices work with licensed, full time clinicians.
Emotionally Focused Couples Therapy: Efficacy and Therapist Fidelity
The second major model of couples’ therapy to clearly establish its effective-
ness is Emotionally Focused Couples Therapy (Johnson, 2004). Three differ-
ent meta-analyses have described EFT’s efficacy in reducing couple distress
and increasing relationship satisfaction (Wood, Crane, Schaalje, & Law, 2005;
see also www.eft.ca for a summary EFT research with additional presenting
problems and populations). Considering the formalized treatment manual
for EFT was published within the last 15 years (Johnson, 2004), it is not sur-
prising that only a few clinician/researchers have attempted to describe and
measure fidelity in EFT practice. In previous attempts, authors used close
clinical supervision (Denton, Burleson, Clark, Rodriguez, & Hobbs, 2000) or
adherence checklists (Makinen & Johnson, 2006) to ensure faithfulness to
the EFT model. In an effort to provide a clear and replicable measure of
fidelity to the EFT model, Denton, Johnson, and Burleson (2009) created the
Emotion-Focused Therapy—Therapist Fidelity Scale (EFT-TFS).
In the process of developing the EFT-TFS, the authors used EFT lit-
erature, author expertise, and survey data from over 100 EFT clinicians to
identify 13 skills that are central to EFT practice and created anchor points
and a scoring system for these skills based on the model provided by the
Cognitive Therapy Scale (Vallis, Shaw, & Dobson, 1986). The 13 skills (listed
below with brief explanation for lesser known terms) are related in theory
and practice to the well-established steps and stages of EFT practice (see
Johnson, 2004):
1. Alliance Making, Alliance Maintenance, Creating Safety
2. Validation of Each Partner
Downloaded by [] at 09:29 16 October 2015
Measuring Fidelity in EFT 253
3. Continually Reframing the Problem in Terms of the Cycle—pattern of
negative interaction between partners
4. Management of Couple’s Interaction—help to keep couple interaction
on track when derailed by conflict, joking, changing subject, non-
participation, etc.
5. Processing Emotion—access, explore, and put words to emotion
6. Working with Primary Emotion—help couples focus on core emotion
related to negative cycle, i.e. helps client move beyond anger to deep
emotion of fear
7. Placing Emerging Emotions into the Cycle—help clients see how cer-
tain emotional experiences are related and fit into specific steps in the
negative cycle
8. Therapeutic Use of Enactments—therapist facilitates interaction where
one partner speaks directly to the other partner
9. Managing Defensive Responses—therapist helps clients move past defen-
sive secondary emotions into vulnerable expression of relational needs
10. Maintaining Session Focus on Emotion, the Cycle, and Attachment Issues
11. Framing cycle, Problems, and Emotions in Terms of Attachment Needs
and Fears—therapist helps clients see how attachments needs and fears
drive emotion, cycle, and relational problems
12. Following Steps and Stages of EFT
13. Consolidation of Change and Development of New Narratives—therapist
helps clients move out of old cycle into new patterns where attachment
needs are expressed and met
The authors also developed a scoring sheet for raters and a manual to
describe in detail the development and use of the fidelity scale (which can be
obtained by contacting the first author of the 2009 article, Wayne Denton).
Yet to date, no published studies exist that use and describe the EFT-TFS as
a measure of the practice of EFT.
Therefore, the purpose of this study is to test and describe the psycho-
metric properties of the EFT-TFS and to identify therapist behaviors/skills
that are predictive of high fidelity EFT. We have used the development and
subsequent description of psychometric properties of the Cognitive Therapy
Scale as our guide in the writing of this article (Vallis et al., 1986).
Research Questions
The primary purpose of the current study is to address and attempt to answer
the following research questions:
•RQ1. Can the EFT-TFS be used to reliably rate therapist fidelity to the EFT
model?
Downloaded by [] at 09:29 16 October 2015
254 J. G. Sandberg et al.
•RQ2. What, if any, is the underlying factor structure of the EFT-TFS?
•RQ3. Can the EFT-TFS accurately distinguish between high and low fidelity
EFT?
•RQ4. Which of the EFT skills (of the 13 specific behaviors listed in the
EFT-TFS) are predictors of high fidelity to EFT?
METHOD
Overview and Design
The data for this study are drawn from a research project where therapists
provided 12 sessions of EFT for couples in a community mental health center
that is affiliated with a large University in the Western United States and
serves primarily uninsured and low income clients. Sixteen different couples
participated in the study, with each couple completing at least 12 sessions
of therapy. For the purposes of this study, three sessions per couple were
rated across different points in therapy (beginning, middle, and end). After
excluding session recordings that were used for rater training and those with
missing data, a total of 40 sessions were used in the analysis. Each session
was rated by two different raters.
Procedure
COUPLES THERAPY
After receiving Institutional Review Board (IRB) approval, couples were re-
cruited through flyers posted in the clinic building, on campus, and in de-
partments of other mental health related fields, mental health clinics, and
libraries in the community. Once couples met study inclusion requirements
and completed consent forms, they were enrolled in therapy. The couples
were given the monetary gift ($250) at the end of the complete round of
therapy sessions.
THERAPIST TRAINING
Once enrolled as therapists in the study, the clinicians received weekly
supervision from an EFT certified therapist (also EFT and AAMFT
supervisor-in-training; third author). This supervision focused on EFT based
conceptualization of cases and execution of EFT specific skills (steps and
stages of EFT). Every other week, all therapists met for supervision with
the lead researcher who is a certified EFT therapist and AAMFT approved
supervisor (first author). This supervision focused on research program ad-
herence, more difficult EFT cases, and the theoretical basis of EFT and
attachment.
Downloaded by [] at 09:29 16 October 2015
Measuring Fidelity in EFT 255
RATER TRAINING
Prior to rating, each rater read through the fidelity scale article (Denton,
Johnson, & Burleson, 2009) as well as the EFT-TFS manual and scoring sheet
(Denton, 2007). In the first stage of scoring sessions, the experienced raters
met as a group to watch and rate sessions as a triad, using the rating sheet to
score therapist behaviors. During the scoring of these two sessions, the first
three authors discussed and debated as a group each therapist behavior prior
to classification, yet each rater was left to score therapist behaviors according
to her/his own interpretation. After this introductory experience, the lead
author met with the developer of EFT-TFS manual and scoring sheet (fifth
author) to seek clarification regarding the distinction between categories as
well as to address questions that had arisen in the initial rating sessions. This
initial meeting with the lead author of the fidelity scale provided an external
credibility check early on in the rating process; it also helped to center the
training of other raters in the EFT-TFS manual and principles.
Next, the three lead authors rated a session individually and checked for
reliability among scores. Once it was clear that the three more experienced
raters could conceptualize and rate behaviors reliably, and in accordance
with the manual, the experienced raters were then partnered in dyads with
the novice raters. A similar training process (rating sessions together as dyad,
discussion/debate of each therapist behavior, check for reliability) was then
followed with the novice raters until all 6 raters could score sessions indi-
vidually in a manner consistent with both the scoring sheet and with her/his
assigned partner. From this point, sessions were assigned to dyads of one
experienced and one novice rater based on each rater’s work availability.
This two-rater system helped increase reliability, as described by Vallis and
colleagues (1986). One experienced rater was also a therapist in the study,
but in no case did he rate his own sessions. This process of assignment
continued until all sessions were rated.
Each rating of the same session was compared to its counterpart, when-
ever raters diverged on a specific therapist behavior by more than one point
on the Likert scale, the raters were required to meet, review the scoring, and
discuss the discrepancy to resolve the scoring difference. However, through-
out the study, this “mediation” process was used only 1.2% (only 6 out of
520 subscales where raters could potentially diverge by more than 1 point)
of the total possible times.
Sample Description
COUPLE DEMOGRAPHICS
The mean age for husbands and wives in the study was 35.50 and 31.69
respectively. The mean length of marriage was 9.68 years with a range
Downloaded by [] at 09:29 16 October 2015
256 J. G. Sandberg et al.
of less than one year to 29 years. On average, couples reported having
two children. Fifteen of the husbands and 15 of the wives self-identified as
Caucasian, one husband and one wife (not married to each other) described
themselves as Hispanic. Two couples reported an annual household income
of less than or equal to $10,000, five couples reported making between
$10,000–24,999, three couples reported between $25,000–39,000, one couple
reported making between $55,000–69,999, four couples reported between
$70,000–84,999, and one couple reported between $85,000–99,000.
THERAPIST DEMOGRAPHICS
The therapists in the study were two male and two female interns (all four MS
student therapists began the study with between 250–400 hours of clinical
experience). One student continued on during his 1st year as a doctoral
student in a COAMFTE-accredited marriage and family therapy graduate
program. As part of the study, they provided 12 sessions of therapy on
a weekly basis to the couples who had passed screening and agreed to
participate in the study.
RATER DEMOGRAPHICS
Six different therapists or therapists-in-training participated in the rating of
tapes. Three of the raters (the first, second and third authors) were considered
‘experienced’ because they had received formalized training in Emotionally
Focused Therapy from EFT certified trainers. The three other raters were
all masters students in a COAMFTE accredited masters’ program and had
participated in bi-weekly EFT group supervision meetings. Three of the raters
were female and three were male. Four raters were Caucasian and two were
African American.
Analysis Strategy
In order to answer the research questions, the analysis unfolded in 4 separate
stages.
RQ1. CAN THE EFT-TFS BE USED TO RELIABLY RATE THERAPIST FIDELITY TO THE
EFT MODEL?
In order to answer the first research question, this study will use both the
Intraclass Correlation Coefficient (ICC; Shrout & Fleiss, 1979) to assess relia-
bility between raters and Chronbach’s alpha to assess reliability within raters.
Our analyses will be evaluated using appropriate ranges and cut-off scores
as follows: ICC scores closer to +1.00 indicate better between-rater reliabil-
Downloaded by [] at 09:29 16 October 2015
Measuring Fidelity in EFT 257
ity (Lahey, Downy, & Saal, 1983) and Chronbach Alpha scores above 0.50
indicate acceptable within-rater reliability. For the between rater analysis, all
80 ratings of the 40 sessions will be used (two raters for each session). For
the within rater analysis, we will use all of the raters’ ratings for each of the
13 items for each session they rated.
RQ2. WHAT,IF PRESENT,IS THE UNDERLYING FACTOR STRUCTURE OF THE EFT-TFS?
In order to discover and interpret the underlying factor structure of the EFT-
TFS, a principal axis factoring (PAF) will be performed. This method was
chosen because of the non-normal distribution of data, small sample size, in
order to account for shared variance, and to avoid the inflation of estimates
of variance accounted for (Costello & Osbourne, 2005). Before performing
a factor analysis, a mean score will be taken of the two raters’ scores and
this score will be used in the analyses. An oblique rotation method was
selected as suggested by Costello and Osborne (2005) as a “more accurate,
and perhaps more reproducible, solution” (p. 3) than orthogonal rotation for
social science data. A scree plot test (Cattell, 1966) will be used to identify the
breaking point of factors. Additionally, appropriateness of a factor analysis
in regard to sample size will be evaluated.
RQ3. CAN THE EFT-TFS ACCURATELY DISTINGUISH BETWEEN HIGH AND LOW
FIDELITY EFT?
Currently, the EFT-TFS lists a score of 40 or greater as competent implemen-
tation of EFT (a score of 39 would represent an average of “3” on the EFT-TFS
items; Denton et al., 2009). Although the EFT-TFS defines it this way, this
article will attempt to statistically verify the classification of high/low fidelity
using discriminant analysis. Predictive discriminant analysis is a particular
type of discriminant analysis that attempts to classify subjects into one of
several groups (Betz, 1987; Stevens, 1996). Since the actual group mem-
bership of the sessions is known (High and Low Fidelity), the accuracy of
the discriminant analysis in classification can be determined. A discriminant
function will be calculated by regressing the predictor variables on a depen-
dent variable that represents group membership. The output of this analysis
will then reveal the proportion of cases correctly classified on the basis of
the discriminant function.
RQ4. WHICH OF THE EFT SKILLS (OF THE 13 SPECIFIC BEHAVIORS LISTED IN THE
EFT-TFS) ARE HIGHER PREDICTORS OF FIDELITY TO EFT?
A stepwise discriminant analysis will be conducted in order to understand
which items predict high fidelity EFT. Stepwise Discriminant Analysis is a
Downloaded by [] at 09:29 16 October 2015
258 J. G. Sandberg et al.
method that will provide an examination of the extent to which multiple
variables were related to group membership (MacNair & Corazzini, 1994; in
this case as either High Fidelity EFT or Low Fidelity EFT). An Fscore will be
calculated to provide information about the importance of variables. Higher
Fscores will identify items that increase the separation of groups. Using
this method, the stepwise discriminant analysis will identify which of the 13
items are more predictive of high fidelity to EFT.
RESULTS
Means, standard deviations, and bivariate correlations for all rating scores
can be seen in Table 1.
RQ1. CAN THE EFT-TFS BE USED TORELIABLY RATE THERAPISTC FIDELITY
TOTHE EFT MODEL?
The Intraclass Correlation Coefficient (ICC; Shrout & Fleiss, 1979) two-way
mixed average measures (3, k) method was used to assess reliability between
and within the six different raters for all sessions. For the purpose of this
analysis, each sessions’ raters were randomly assigned to be either in Group
A or B so that both ratings for the session were not in the same group. The
ICC score was .913 (range: .835–.954), F=11.45, (df =39), p<.001 (Rating
Group A: M=37.64, SD =8.91; Rating Group B: M=37.65, SD =9.01).
These results show that the raters were in agreement on the sum total score
of each individual EFT session. The reliability for the 13 individual items had
a range of .66 to .94, indicating high agreement on each individual item that
was rated (i.e., alliance, validation, managing defensive responses, etc.).
Within-rater reliability analysis (a single rater’s scores across sessions)
was performed for each rater, using their scores of all 13 items across all
sessions they rated. Each of the raters achieved acceptable reliability (Lahey,
Downy, & Saal, 1983). Rater 1 rated 16 sessions (α=0.91, p <.001), rater
2 rated 11 sessions (α=0.55, p =.021), rater 3 rated 14 sessions (α=
0.91, p <.001), rater 4 rated 4 sessions (α=0.90, p <.001), rater 5 rated
21 sessions (α=0.89, p <.001), and rater 6 rated 13 sessions (α=0.78,
p<.001).
RQ2. WHAT,IF PRESENT,IS THE UNDERLYING FACTOR STRUCTURE OF THE EFT-TFS?
The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy (with higher
values closer to 1.0 indicating appropriateness for factor analysis) and the
Bartlett’s test of Sphericity (pvalues less than .05 indicating appropriateness
for factor analysis; Cerny & Kaiser, 1977) indicated that the data was suitable
Downloaded by [] at 09:29 16 October 2015
TABLE 1 Correlations, Means, and Standard Deviations for All Items (Mean of Both Raters’ Score)
Variables 1234567891011121314
1. Alliance —
2. Validation .68∗∗ —
3. Reframe to Cycle .39∗.37∗∗ —
4. Management of
Couple’s Interaction
.15 .43∗∗ .31 —
5. Processing Emotion .16 .08 .53∗∗ .16 —
6. Working with
Primary Emotion
.10 .10 .48∗∗ .43∗∗ .74∗∗ —
7. Emotion in Cycle .14 .19 .47∗∗ .48∗∗ .66∗∗ .85∗∗ —
8. Enactments .04 .15 .39∗.32∗.50∗∗ .60∗∗ .60∗∗ —
9. Managing Defensive
Responses
.13 .20 .46∗∗ .66∗∗ .32∗.50∗∗ .49∗∗ .37∗—
10. Maintaining Session
Focus on Emotion,
Cycle, and Attachment
Needs
.29 .24 .32∗.61∗∗ .18 .46∗∗ .51∗∗ .48∗∗ .57∗∗ —
11. Framing Cy-
cle/Emotions/Problem
to Attachment Needs
.12 .28 .44∗∗ .59∗∗ .27 .58∗∗ .58∗∗ .49∗∗ .47∗∗ .58∗∗ —
12. Following Steps and
Stages of EFT
.23 .16 .45∗∗ .42∗∗ .28 .41∗∗ .51∗∗ .22 .57∗∗ .55∗∗ .46∗∗ —
13. Consolidation of
Change and
Development of New
Narratives
.23 .22 .20 .10 .22 .25 .12 .34 .09 .26 .46∗∗ .04 —
14. Total Score .38∗.47∗∗ .70∗∗ .71∗∗ .60∗∗ .79∗∗ .77∗∗ .68∗∗ .71∗∗ .72∗∗ .77∗∗ .59∗∗ .41∗∗ —
¯
X4.74 4.42 3.21 2.85 3.16 2.26 2.19 1.90 2.23 2.91 2.21 2.94 1.78 36.39
Standard Deviation .67 .88 1.23 1.20 1.13 0.89 .88 1.12 0.91 0.94 1.06 .89 .82 8.21
∗p<.05. ∗∗p<.01. ∗∗∗p<.001.
259
Downloaded by [] at 09:29 16 October 2015
260 J. G. Sandberg et al.
for factor analysis: KMO =.785, Bartlett’s: X2=320.37, df =78, p<.001.
Additionally, the diagonals of the anti-image correlation matrix were all over
.5, supporting the inclusion of each item in the factor analysis (Field, 2005).
The internal structure of the EFT-TFS was evaluated using principal axis
factoring (PAF). Initial PAF without fixing a number of factors to extract, 3
factors resulted with eigenvalues higher than 1. A scree plot test (Cattell,
1966) showed the initial breaking point after 1 factor. Subsequent PAFs with
manual factor extraction from 1 to 3 were performed. Clarity was maintained
in a single factor structure according to the criteria discussed in Costello and
Osborne (2005): item loadings above .30, no or few cross loadings, and no
factors with fewer than three items (p. 3). The one factor structure of the
EFT-TFS explained 38.15% of the variance in EFT skills (shown in Table 2).
SAMPLE SIZE
With a sample size of 40, the authors were concerned regarding the appro-
priateness of the analysis. Drawing upon the work of Costello and Osborne
(2005), the following criteria was met: Communalities of items were between
.41 and .85 indicating an acceptable range for the social sciences; at least 3
items loaded on a factor at .5 or above; no cross-loadings were above .5;
no Heywood cases (factor loadings >1.0); and actual convergence led to a
solution. Thus, the sample size of 40 was appropriate.
RQ3. CAN THE EFT-TFS ACCURATELY DISTINGUISH BETWEEN HIGH AND LOW FIDELITY
EFT?
Differences in EFT-TFS item scores between lower or higher fidelity EFT
sessions were analyzed by a multivariate analysis of variance (MANOVA) for
the 40 sessions that were rated by at least two raters. A mean total score
was computed using the average of the two raters’ total score and was
used in the analysis. A session is considered higher fidelity EFT if the sum
of the items is at least 39 points or above (Denton, 2007). Twenty-seven
sessions were classified as lower-fidelity EFT and 13 sessions were classified
as higher-fidelity EFT. Table 3 presents the classification results.
RQ4. WHICH OF THE EFT SKILLS (OF THE 13 SPECIFIC BEHAVIORS LISTED IN THE
EFT-TFS) ARE HIGHER PREDICTORS OF FIDELITY TO EFT?
A stepwise discriminant analysis was also conducted, and revealed one sig-
nificant discriminant function (Eigen value =3.74, Canonical Correlation =
.89, Wilks’ λ=.211, X2=48.99, df =13, p<.05, Effect size =0.79). Table 4
presents the mean scores for each item and the result of the MANOVA.
Downloaded by [] at 09:29 16 October 2015
TABLE 2 Communalities and Factor Loadings for Principal Axis Factoring
Initial Extraction Final Loadings
Items Communalities Communalities Loadings
1. Alliance .566 .107 .327
2. Validation .601 .155 .394
3. Reframe to Cycle .534 .447 .669
4. Management of Couple’s Interaction .621 .284 .533
5. Processing Emotion .693 .397 .630
6. Working with Primary Emotion .847 .723 .850
7. Emotion in Cycle .836 .739 .859
8. Enactments .457 .266 .516
9. Managing Defensive Responses .580 .319 .565
10. Session Focus on Emotion/Cycle/Attachment Needs .577 .481 .694
11. Framing Cycle/Emotions/Problem to Attachment Needs .638 .504 .710
12. Following Steps and Stages of EFT .616 .426 .653
13. Consolidation of Change and Development of New Narratives .405 .110 .332
Eigen Value 5.49
% of Variance 38.15
∗p<.05. ∗∗p<.01. ∗∗∗p<.001.
261
Downloaded by [] at 09:29 16 October 2015
262 J. G. Sandberg et al.
TABLE 3 Discriminant Classification Results
0=Low EFT, 1 =High EFT
Predicted Group
Membership Total
Original Count .00 1.00
.00 27 0 27
1.00 2 11 13
%.00 100 .0 100.0
1.00 15.484.6 100.0
Note. 95.0% or original grouped cases correctly classified.
As seen in Table 4, the means for Alliance and Consolidation of Change
were not significantly different between groups. Only EFT-TFS items that
significantly added to the separation of groups (high or low fidelity EFT) were
included in the discriminant function (alliance and consolidation of change
were not included). Items are listed based on their predictive power within
the discriminant function: management of couples’ interaction (Loading =
.487, F =33.61, p<.05), working with primary emotion (Loading =.458,
F=29.76, p<.05), managing defensive responses (Loading =.448, F =
28.52, p<.05), reframing the problem to cycle (Loading =.393, F =21.90,
p<.05), placing emerging emotion in the cycle (Loading =.385, F =21.01,
p<.05), maintaining session focus on emotion and cycle (Loading =.361,
F=18.50, p<.05), Framing Cycle/Emotions/Problem to Attachment Needs
(Loading =.351, F =17.46, p<.05), and processing emotion (Loading
=.327, F =15.20, p<.05) (only those items with loadings above .30).
Using this discriminant function, 95.0% of sessions were correctly classified
as either high or low fidelity EFT. Zero of the low fidelity EFT sessions and
only 2 of the 13 high fidelity EFT sessions were misclassified. Because the
base rate for a dichotomous classification is 50%, the discriminant function
increased correct classification by approximately 45%.
DISCUSSION
This pilot study sought to test the psychometric properties of the EFT-TFS
and to determine which therapist behaviors/skills are most predictive of high
fidelity EFT. The analysis provided preliminary support for the measure,
namely that it appears the scale can be used reliably (both inter and within-
rater), even when used by novice raters to score a heterogeneous group of
therapists. Results also suggest that the measure might be used to discriminate
effectively (95% correct classification) between low and high fidelity EFT.
And finally, the identified factor structure accounted for 38% of the variance
from the indicator variables, signifying that the other 62% of the variance
in fidelity to EFT was accounted for by other factors. Although additional
Downloaded by [] at 09:29 16 October 2015
TABLE 4 Means, F Scores, and pValues of EFT Versus Non-EFT Sessions
Items EFT (n =13) Non-EFT (n =27) Wilks’ λF
Alliance 4.94 4.65 .96 1.730
Validation 4.92 4.18 .84 7.33∗∗
Reframe to Cycle 4.27 2.70 .63 21.90∗∗∗
Management of Couple’s Interaction 4.02 2.28 .53 33.61∗∗∗
Processing Emotion 4.02 2.74 .71 15.20∗∗∗
Working with Primary Emotion 3.10 1.85 .56 29.76∗∗∗
Emotion in Cycle 2.94 1.83 .64 21.01∗∗∗
Enactments 2.58 1.57 .82 8.41∗∗
Managing Defensive Responses 3.08 1.81 .57 28.52∗∗∗
Session Focus on Emotion/Cycle/Attachment Needs 3.67 2.54 .67 18.50∗∗∗
Framing Cycle/Emotions/Problem to Attachment Needs 3.06 1.81 .69 17.46∗∗
Following Steps and Stages of EFT 3.50 2.67 .80 9.26∗∗
Consolidation of Change and Development of New Narratives 1.96 1.69 .98 .99
∗p<.05. ∗∗p<.01. ∗∗∗p<.001.
263
Downloaded by [] at 09:29 16 October 2015
264 J. G. Sandberg et al.
research is needed with a broader range of therapists and clients, the EFT-TFS
appears to be a psychometrically sound measure of fidelity to the Emotionally
Focused Couples Therapy model.
Of particular interest among the results is the finding that certain skills
seem to be more predictive of high fidelity EFT. Of the eleven EFT spe-
cific skills analyzed (therapeutic alliance and consolidation were excluded
because they were common factors in therapy regardless of model; see
Sprenkle, Davis, & Lebow, 2009), two of the three strongest predictors (man-
agement of couples’ interaction and managing defensive responses) relate
to the therapist’s ability to intervene in session when interactions become
defensive or distracting and refocus the process on primary emotion or at-
tachment needs (see EFT-TFS rating scale; Denton et al., 2009). This skill of
successfully managing the session focus and interaction may be important
because all other skills (which are aimed at developing an awareness of the
cycle of conflict, deepening and processing primary emotion, and identify-
ing and expressing attachment needs) hinge on keeping the session directed
at meeting basic attachment needs within the couple relationship (Johnson,
2004). This finding (that maintaining the session focus on primary emotion
and attachment needs is central to effective EFT) is supported by previous
EFT process research that suggests therapist facilitation and prompting are
keys to softening moments where one partner reaches out to and is then
reassured by her/his partner (Furrow, Bradley & Johnson, 2011). In addition,
these results also highlight that such behaviors are measurable and consistent
in high fidelity cases across differing sets of therapists and clients.
It is also important to note which skills were least influential in pre-
dicting high fidelity EFT. In this sample, the results suggest that validation,
enactments, and following steps and stages of EFT are less likely to predict
high fidelity EFT. The finding regarding enactments runs contrary to previous
theory and research regarding EFT (Bradley & Furrow, 2004; Tilley & Palmer,
2013). These previous papers have described how central enactments are in
successful EFT cases, specifically the positive effect that occurs when a ther-
apist can facilitate an individual reaching out to her/his partner in a soft
way and the partner accepting the reach and responding supportively. The
process of reaching out and responding in therapy is labeled an enactment
in EFT.
One possible explanation for this contrary finding may be that therapists,
on average, used enactments less frequently and effectively in this study
(mean score of 2 out of 5 on EFT-TFS scale). This explanation is plausible
because none of the therapists were licensed MFTs or officially certified EFT
therapists (see www.iceeft.com for a complete description of the certification
process). As a result, they may have lacked the confidence and skills to
effectively introduce and facilitate enactments. It may also be that enactments
are more common and effective in later stages of therapy once couples have
softened, which could not be assessed because the current study did not
track therapy beyond the 11th session.
Downloaded by [] at 09:29 16 October 2015
Measuring Fidelity in EFT 265
Implications for Practice and Training
The results can be used to inform EFT specific practice and training. For
example, this study suggests that fidelity can be effectively measured; there-
fore, supervisors and trainers may wish to use the EFT-TFS in supervision to
improve adherence and competence. This finding is of particular value con-
sidering the tremendous interest (clinical and research) in the model and the
number of clinicians in EFT training programs. Therefore, the EFT-TFS could
be used, in conjunction with other approaches, to help measure, evaluate,
and standardize EFT training and practice. In addition, the fidelity scale may
be used as a training tool by asking trainees to watch and rate their own
sessions. Such a practice requires clinicians to break down the micro inter-
actions of therapy and consider if and how their own in-session behavior is
related to the theory and skills outlined in EFT.
The results also suggest that supervisors, trainers, and clinicians may
wish to focus on ways to manage defensiveness and couple interaction in
session. Previous research has concurred that a therapist’s ability to structure
a session is key to successful outcomes in family therapy (Shields, Sprenkle,
& Constantine, 1991). Specific to EFT, trainers and trainees may need to
focus their efforts to structure sessions on emotion and attachment. When
therapists can guide couples to the vulnerability beneath defensiveness or
the longing for connection behind distractions, they will likely be much more
effective at fostering bonding moments in session (Johnson, et al., 2004).
Implications for Future Research
Although this first attempt to use the EFT-TFS to measure fidelity was suc-
cessful in many regards, additional research is needed to establish fully the
reliability and validity of this measure. Also, more investigation is required
to carefully distinguish which EFT skills are specific to this model and which
could be considered common factors (Sprenkle, Davis, & Lebow, 2009). For
example, is there an attachment specific method to validate or create al-
liance that uniquely contributes to success in EFT? More specificity is also
needed to determine how therapists manage sessions, focus on emotion,
and foster the expression of attachment needs. The current study measures
if these skills are present and highlights their importance, but does not mea-
sure exactly ‘how’ therapists demonstrate these skills in session. Perhaps a
mixed method approach, where researchers compare observationally coded
behavior in high vs. low fidelity cases, could provide a clearer description
of therapist behaviors that lead to positive outcomes.
From these results we have come to believe the EFT-TFS is a good
measurement tool for rating EFT sessions among student therapists. How-
ever, a larger sample size and more diversity among therapist experience
level might reveal a more complex and accurate factor structure than the
simple one in this study, giving a clearer picture of how the various items
Downloaded by [] at 09:29 16 October 2015
266 J. G. Sandberg et al.
of the EFT-TFS relate including potential subscales (Costello & Osbourne,
2005). For instance, alliance and consolidation consistently loaded on their
own factors during the analysis; theoretically these items could be con-
sidered common factors in therapy and play a distinct role in the overall
EFT process.
Limitations
As is common in preliminary studies of therapist behavior (Vallis, et al.,
1986), the current study has a number of limitations. First, the therapists
were unlicensed and in the midst of clinical training. Therefore, the results
may look quite different for a group with more clinical experience. Future
studies would be benefited by a broader range of therapists in terms of
experience, training, and practice setting. Second, the sample size was typical
of a preliminary study (16 couples, 4 therapists) and limits the generalizability
of the study. Third, although the fidelity scores represented a broad range
of adherence to the model (21–60), the majority of scores were near or
below the cutoff (39) for high fidelity EFT. As a result, it is not clear if this
measure would be psychometrically sound when used to measure certified
EFT therapists. And fourth, couples were compensated for participation in
the study which may have influenced their commitment to therapy.
CONCLUSION
The results of this study suggest that the EFT-TFS can be appropriately
used to measure fidelity to the EFT model. In addition, the results suggest
that the ability to manage defensiveness in session and maintain focus on
emotion and attachment needs predicts high fidelity EFT. Additional research
is needed to determine if high fidelity EFT is more predictive of therapeutic
outcomes.
REFERENCES
Barber, J. P., Gallop, R., Crits-Christoph, P., Frank, A., Thase, M. E., Weiss,
R. D., & Connolly Gibbons, M. B. (2006). The role of therapist adher-
ence, therapist competence, and alliance in predicting outcome of individ-
ual drug counseling: Results from the National Institute Drug Abuse Col-
laborative Cocaine Treatment Study. Psychotherapy Research,16(2), 229–240.
doi:10.1080/10503300500288951
Betz, N. E. (1987). Use of discriminant analysis in counseling psychology research.
Journal of Counseling Psychology,34, 393–403.
Cattell, R. B. (1966). The screen test for the number of factors. Multivariate Behav-
ioral Research,1, 245–276.
Downloaded by [] at 09:29 16 October 2015
Measuring Fidelity in EFT 267
Cerny, C. A., & Kaiser, H. F. (1977). A study of a measure of sampling adequacy
for factor-analytic correlation matrices. Multivariate Behavioral Research,12(1),
43–47.
Christensen, A., Atkins, D. C., Baucom, B., & Yi, J. (2010). Marital status and sat-
isfaction five years following a randomized clinical trial comparing traditional
versus integrative behavioral couple therapy. Journal of Consulting and Clinical
Psychology,78, 225–235.
Christensen, A., Atkins, D. C., Berns, S., Wheeler, J., Baucom, D. H., & Simpson, L. E.
(2004). Traditional versus integrative behavioral couple therapy for significantly
and chronically distressed married couples. Journal of Consulting and Clinical
Psychology,72, 176–219.
Costello, A. B., & Osborne, J. W. (2005). Best practices in exploratory factor anal-
ysis: Four recommendations for getting the most from your analysis. Practical
Assessment, Research & Evaluation,10(7), 1–9.
Denton, W. (2007).Manual for the Emotion-Focused Therapy Fidelity Scale (EFT-TFS).
Version 4. Unpublished manuscript.
Denton, W., Burleson, B., Clark, T., Rodriguez, C., & Hobbs, B. (2000). A randomized
trial of emotionally focused therapy for couples in a training clinic. Journal of
Marital and Family Therapy,26, 65–78.
Denton, W. H., Johnson, S. M., & Burleson, B. R. (2009). Emotion-Focused Therapy-
Therapist Fidelity Scale (EFT-TFS): Conceptual development and content validity.
New York, NY: Taylor and Francis Group.
Field, A. (2005). Discovering statistics using SPSS (2nd ed.). London, England: Sage.
Furrow, J. L., Johnson, S. M., & Bradley, B. A. (2011). Chapter 1. The emotionally
focused casebook: New directions in treating couples. New York, NY: Routledge.
Johnson, S. M. (2004). The practice of emotionally focused couple therapy: Creating
connection (2nd ed.). New York, NY: Brunner-Routledge.
Lahey, M., Downey, R., & Saal, F. (1983). Intraclass correlations: There is more than
meets the eye. Psychological Bulletin,93, 586–595.
Lebow, J. L., Chambers, A. L., Christensen, A., & Johnson, S. M. (2012). Research on
the treatment of couple distress. Journal of Marital and Family Therapy,38(1),
145–169.
MacNair, R. R., & Corazzini, J. G. (1994). Client factors influencing group therapy
dropout. Psychotherapy: Theory, Research, Practice, Training,31(2), 352–362.
doi:10.1037/h0090226
Makinen, J. A., & Johnson, S. (2006). Resolving attachment injuries in couples using
EFT: Steps towards forgiveness and reconciliation. Journal of Consulting and
Clinical Psychology,74(6), 1055–1064.
Markman, H. J., & Rhoades, G. K. (2012). Relationship education research: Current
status and future directions. Journal of Marital and Family Therapy,38(1),
169–200.
Shaw, B. F., Elkin, I., Yamaguchi, J., Olmsted, M., Vallis, T. M., Dobson, K. S., & Im-
ber, S. D. (1999). Therapist competence rating in relation to clinical outcome in
cognitive therapy of depression. Journal of Consulting and Clinical Psychology,
67(6), 837–846.
Shields, C. G., Sprenkle, D. H, & Constantine, J. A. (2001). Anatomy of an initial in-
terview: The importance of joining and structuring skills. The American Journal
of Family Therapy, 19, 3–18.
Downloaded by [] at 09:29 16 October 2015
268 J. G. Sandberg et al.
Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in assessing rater
reliability. Psychological Bulletin,86, 420–428.
Sprenkle, D. H., Davis, S. D., & Lebow, J. L. (2009). Common factors in couple and
family therapy: The overlooked foundation of effective practice. New York, NY:
Guilford Press.
Stevens, J. (1996). Applied multivariate statistics for the social sciences 3rd (ed.).
Mahwah, NJ: Lawrence Erlbaum Associates.
Tilley, D., & Palmer, G. (2013). Enactments in emotionally focused couple ther-
apy: Shaping moments of contact and change. Journal of Marital and Family
Therapy,39(3), 299–313.
Vallis, T. M., Shaw, B. F., & Dobson, K. S. (1986). The cognitive therapy scale:
Psychometric properties. Journal of Consulting and Clinical Psychology,54(3),
381–385.
Vivian, D., Hershenberg, R., Teachman, B. A., Drabick, D. A. G., Goldfried, M.
R., & Wolfe, B. (2012). A translational model of research-practice integration.
Psychotherapy,49(2), 143–151.
Webb, C. A., DeRubeis, R. J., & Barber, J. P. (2010). Therapist adherence/competence
and treatment outcome: A meta-analytic review. Journal of Consulting and
Clinical Psychology,78(2), 200–211. doi:10.1037/a0018912
Wood, N. D., Crane, D. R., Schaalji, G. B., & Law, D. D. (2005). What works for
whom: A meta-analytic review of marital and couples therapy in reference to
marital distress. The American Journal of Family Therapy,33, 273–287.
Downloaded by [] at 09:29 16 October 2015
