Resolving Attachment Injuries in Couples Using Emotionally Focused Therapy: Steps Toward Forgiveness and Reconciliation

Abstract

The goal of this study was to use task analysis to verify that the attachment injury resolution model described in this article discriminates resolved from nonresolved couples. Twenty-four couples with an attachment injury received, on average, 13 sessions of emotionally focused therapy (EFT). At the end of treatment, 15 of the 24 couples were identified as resolved. Segments of best sessions for all couples were transcribed and rated on 2 process measures. Resolved couples were found to be significantly more affiliative and achieved deeper levels of experiencing than nonresolved couples. They also showed significant improvements in dyadic satisfaction and forgiveness than nonresolved couples. The results support the attachment injury resolution model and suggest that resolution during EFT is beneficial to couples.

Full text

Resolving Attachment Injuries in Couples Using Emotionally Focused
Therapy: Steps Toward Forgiveness and Reconciliation
Judy A. Makinen
University of Ottawa
Susan M. Johnson
University of Ottawa, Alliant University, and Ottawa Couple &
Family Institute
The goal of this study was to use task analysis to verify that the attachment injury resolution model
described in this article discriminates resolved from nonresolved couples. Twenty-four couples with an
attachment injury received, on average, 13 sessions of emotionally focused therapy (EFT). At the end of
treatment, 15 of the 24 couples were identified as resolved. Segments of best sessions for all couples were
transcribed and rated on 2 process measures. Resolved couples were found to be significantly more
affiliative and achieved deeper levels of experiencing than nonresolved couples. They also showed
significant improvements in dyadic satisfaction and forgiveness than nonresolved couples. The results
support the attachment injury resolution model and suggest that resolution during EFT is beneficial to
couples.
Keywords: attachment injury, emotionally focused therapy, forgiveness, impasses, reconciliation
Many couples encounter situations or life events that may lead
to emotional distress. However, attachment-related incidents can
have deleterious effects on the relationship bond. These incidents
have been called attachment injuries (Johnson, 1998) and have
been characterized as a perceived abandonment, betrayal, or
breach of trust in a critical moment of need for support expected of
attachment figures (Johnson, Makinen, & Millikin, 2001). The
incident becomes a clinically recurring theme, is often used as a
standard for the dependability of the other partner, and creates an
impasse that blocks relationship repair. As such, the couple be-
comes stuck in a rigid, negative, interactional cycle (e.g., attack–
defend, pursue– distance), which may then escalate into severe
marital distress.
The attachment injury concept emanated from the practice of
emotionally focused therapy (EFT). EFT is an empirically vali-
dated approach to marital therapy, and it is recognized as one of
the most effective approaches in resolving relationship distress
(Baucom, Shoham, Mueser, Daiuto, & Stickle, 1998). It has dem-
onstrated a very large effect size of 1.3, with recovery rates
between 70% and 73% (Johnson, Hunsley, Greenberg, & Schin-
dler, 1999).
EFT, although not developed from attachment theory, is suited
to address important aspects of relationship functioning described
by attachment theory. Adult attachment theory is one of the most
promising theories of adult romantic relationships (Hazan &
Shaver, 1987; Johnson & Whiffen, 2003). Attachment theory
emphasizes the propensity for human beings to make and maintain
powerful affectional bonds with significant others. In couples, a
secure attachment bond is an active, affectionate, reciprocal rela-
tionship in which partners mutually derive and provide closeness,
comfort, and security. These bonds are based on a “profound
psychological and physiological interdependence” and, therefore,
have an impact on psychological well-being (Hazan & Zeifman,
1999, p. 351). Self-reports of a secure romantic attachment are
linked with positive aspects of relationship functioning, including
high levels of trust, commitment, interdependence, and higher
dyadic satisfaction (e.g., Kirkpatrick & Davis, 1994; Mikulincer,
1998). Supportive relationships encourage the creation of optimal
experiences for both partners. Attachment theorists have pointed
out that, perhaps because of this interdependence, incidents in
which one partner fails to respond at times of urgent need seem to
disproportionately influence the quality of an attachment relation-
ship (Simpson & Rholes, 1994).
Attachment theory has also been referred to as a theory of
trauma (Atkinson, 1997, p. 3). When people are without physical
or emotional support, they are most vulnerable and have difficulty
regulating their emotions. Disturbances of affect are central to all
descriptions of traumatic stress and its sequelae (Stone, 1996). It
has been argued that wounds to attachment relationships resulting
from emotional inaccessibility by one partner may be equated to
trauma with a small “t” (Johnson, 2002). Following traumatic
abandonment, the injured partner may exhibit symptoms charac-
teristic of posttraumatic stress disorder. As Abrahms-Spring
(1997) pointed out, disturbing memories, vivid images, and sen-
sations puncture an injured partner’s concentration and sleep.
When awake, they ruminate excessively, and hypervigilance be-
comes the norm. Avoidance and numbing, which are both natural
and self-protective strategies against the barrage of intrusive symp-
toms, interfere with emotional engagement and attunement be-
tween partners. Couples dealing with trauma incurred by a partner
Judy A. Makinen, Department of Psychology, University of Ottawa,
Ottawa, Ontario, Canada; Susan M. Johnson, Department of Psychology,
University of Ottawa; Alliant University; and Ottawa Couple and Family
Institute, Ottawa, Ontario, Canada.
Correspondence concerning this article should be addressed to Judy A.
Makinen, who is now at the Royal Ottawa Hospital, 1145 Carling Avenue,
Ottawa, Ontario K1Z 7K4, Canada; and the Ottawa Couple & Family
Institute, 1869 Carling Avenue, Ottawa, Ontario K2A 1E6, Canada. E-
mail: jumakine@rohcg.on.ca
Journal of Consulting and Clinical Psychology Copyright 2006 by the American Psychological Association
2006, Vol. 74, No. 6, 1055–1064 0022-006X/06/$12.00 DOI: 10.1037/0022-006X.74.6.1055
1055

tend to express more intense negative affect than typical distressed
couples (Johnson & Williams-Keeler, 1998), and clinical experi-
ence shows that this is true also for couples dealing with relation-
ship traumas or attachment injuries. As attachment theorists have
pointed out, situations in which an attachment figure is both the
source of and the solution to emotional pain are inherently difficult
to tolerate and result in a fundamental disorganization of the
attachment system (Main & Hesse, 1990).
Recently, couples therapists have attempted to address relation-
ship traumas that make the resolution process difficult to achieve.
The forgiveness literature is particularly relevant (e.g., Enright &
Fitzgibbons, 2000; Hargrave, 2004) as couples that have recovered
from the impact of a relationship betrayal often allude to the role
of forgiveness in the resolution process (Gordon, Baucom, &
Snyder, 2000). Forgiveness occurs in response to an interpersonal
violation and involves mending emotional wounds, restoring trust,
and repairing the relationship bond. Rowe et al. (1989) were
among the first researchers to acknowledge that forgiveness has a
positive impact on both the injured and the offending parties.
Gordon and Baucom (1998) concurred that forgiveness involves a
“complex interaction including the person who is forgiving, the
person who is being forgiven, and the dyadic interaction between
these two people” (p. 426). Thus, although not directly related to
EFT or attachment theory, forgiveness may be an important step in
the process of resolving attachment traumas among couples.
The goal of this study was to describe how attachment injuries
may resolve during EFT, according to the attachment injury res-
olution model. The attachment injury resolution model emanated
from task analysis of change process in EFT (Johnson & Green-
berg, 1988), clinical experience (Johnson, 2004; Johnson et al.,
2001), and empirical research (Millikin, 2000; Naaman, Pappas,
Makinen, Zuccarini, & Johnson, 2005). The task analytic approach
to studying the process of change involves two phases. The first
phase, which is discovery-oriented, consists of six steps beginning
with task selection and proceeding to a model of change. The
second phase, which is referred to as the verification phase and is
the focus of this study, consists of two steps (Greenberg, 1984;
Greenberg & Foerster, 1996). The first step involves comparing
couples that successfully complete the change process with those
that remain at an impasse to verify the markers of successful
performance. The second step involves beginning to link success-
ful task performance to outcome.
In preliminary exploratory studies, Millikin (2000) used task
analysis to study the process of change in mildly to moderately
distressed couples that successfully resolved the attachment injury
in 10 to 15 sessions of EFT. The findings showed that, although
attachment injuries often emerge at the assessment stage, a thera-
pist could not effectively intervene until the negative interactional
pattern between partners had de-escalated. The attachment injury
comes alive again when the injured partner is invited to risk
emotionally engaging with his or her partner with the hope of
creating a new positive emotional experience. This is the optimal
time for therapeutic intervention and the resolution process. The
steps of the attachment injury resolution model (which act as a
guide for therapists working with such couples), along with the
process measure descriptions on the Structural Analysis of Social
Behaviour (SASB; Benjamin, 1974; Benjamin, Foster, Roberto, &
Estroff, 1986) and the Experiencing Scale (ES; Klein, Mathieu-
Coughlin, & Kiesler, 1986) are outlined below. The SASB and ES
assess key components of what are believed to be the indicators of
change during the resolution process. Identifying these key events
in therapy is essential in task analysis as it scales down data into
manageable amounts for coding and analysis.
Attachment Injury Resolution Model
Attachment Injury Marker
1. In a highly emotional manner, the injured partner de-
scribes the incident in which he or she experienced a
violation of trust that damaged his or her belief in the
relationship. The incident is alive and present rather than
a calm recollection (SASB ⫽ 6 [blame] and 7 [attack];
ES ⫽ 3 [describes events/reactive]).
2. Offending partner discounts, denies, or minimizes the
incident and his or her partner’s pain and moves into a
defensive stance. (SASB ⫽ 7 [defend and withdraw];
ES ⫽ 3 [describes events/reactive]).
Differentiation of Affect
3. Injured partner stays in touch with the injury and begins
to articulate its impact and attachment significance
(SASB ⫽ 6 [blame] and 1 [assert]; ES ⫽ 2 [intellectual
description of event]).
4. Offending partner begins to hear and understand the
significance of the injurious event (SASB ⫽ 7 [defend
and withdraw]; ES ⫽ 2 [intellectual description of
event]).
Reengagement
5. Injured partner tentatively moves toward a more inte-
grated articulation of the injury and allows the other to
witness his or her vulnerability by expressing grief and
fear concerning the specific loss of the attachment bond
(SASB ⫽ 2 [disclose and express]; ES ⫽ 4 to 5 [increas-
ing emotional involvement]).
6. Offending partner becomes more emotionally engaged
and acknowledges responsibility for his or her part and
expresses empathy, regret, and remorse (SASB ⫽ 2 [af-
firm and understand]; ES ⫽ 4 to 5 [increasing emotional
involvement]).
Forgiveness and Reconciliation
7. Injured partner then risks asking for comfort and caring
that was unavailable at the time of the injury (SASB ⫽ 4
[trust and rely]; ES ⫽ 6 to 7 [express feelings/awareness
of present feelings]).
8. Offending partner responds in a caring manner that acts
as an antidote to the traumatic experience (SASB ⫽ 3
[nurture and comfort]; ES ⫽ 6 to 7 [express feelings/
awareness of present feelings]).
1056
MAKINEN AND JOHNSON

In this study, we employed a quasi-experimental, task analytic
research methodology to build on previous research by verifying
the resolution model. A strict adherence to the EFT treatment
manual assisted in treating the therapists’ behavior as a controlled
variable (Johnson & Greenberg, 1988), thus examining couple
processes and describing how change occurs within the couple
system. The SASB and ES were used to test whether the attach-
ment injury resolution model discriminated resolved from nonre-
solved couples. We predicted that resolved couples would exhibit
(a) significantly more affiliative responses (i.e., disclose and ex-
press needs, affirm and understand) and fewer hostile or distant
responses (i.e., belittle and blame, protest and withdraw) and (b)
deeper levels of experiencing in “best sessions” than nonresolved
couples. In addition, we predicted that, at the end of treatment,
resolved couples would exhibit (a) significant decreases in marital
distress, (b) significant gains in the level of relationship trust, (c)
significantly less avoidant and anxious attachment, and (d) signif-
icantly more forgiveness and less emotional pain than nonresolved
couples.
Method
Participants
Of the couples meeting inclusion criteria (see Procedure section), 30
couples identified an event that altered the quality of their relationship
bond and expressed a desire for closure. Of these, 3 couples withdrew prior
to the onset of therapy. One couple dropped out of the study after the initial
session. Two couples terminated their relationship once therapy com-
menced. Therefore, 24 couples received, on average, 13 sessions of therapy
and completed the pre- and posttreatment questionnaire packages. The
couples (the “completers”) were included in analyses because the focus
was on describing processes among couples actually undergoing the ther-
apy. Of these couples, there were 19 injured wives and 5 injured husbands.
The types of attachment injuries reported were an actual abandonment (n ⫽
3), perceived abandonment following a miscarriage (n ⫽ 2), infidelity (n ⫽
10), flirtation (n ⫽ 4), exotic massage (n ⫽ 1), Internet relationship (n ⫽
1), friendship with opposite sex (n ⫽ 1), insulting remark (n ⫽ 1), and
financial deception/loss (n ⫽ 1). The couples were moderately distressed,
with the mean couple dyadic adjustment of 84 (range ⫽ 67–100). The
majority of couples were Caucasian with 1 East Indian and 1 couple of
European descent. Their ages ranged from 25 to 52 years (M ⫽ 39.79
years, SD ⫽ 7.87), they had been married for 1 to 30 years (M ⫽ 13.53
years, SD ⫽ 8.99), and had on average 2 children. Of the 48 individuals,
33 (69%) had a postsecondary college diploma or university degree. The
gross family income for the entire sample ranged from $40K to $100K
(M ⫽ 73.33, SD ⫽ 23.90).
Process Measures
The ES and SASB were used to code best sessions.
ES. The ES (Klein et al., 1986) is a 7-point rating scale used to rate
partner’s level of emotional involvement in therapy. Stage 1 reflects a low
level of emotional experiencing. The content is impersonal, abstract, and
general. Moving up the scale, there is a gradual progression from super-
ficial external events to an internal experience and a synthesis of new
feelings and meanings. At Stage 7, there is an expanding of awareness of
present feelings in internal processes. The validity of this scale has been
supported by its correlation with client variables, such as introspection and
cognitive complexity, and has been used to predict client change (Orlinsky
& Howard, 1986).
SASB. The SASB (Benjamin, 1974; Benjamin et al., 1986), a coding
system designed to measure the changing quality of interaction between
partners, comprises a two-dimensional grid consisting of a two axes. The
horizontal axis describes degrees of affiliation (i.e., unfriendly to friendly),
and the vertical axis describes degrees of interdependence (i.e., autono-
mous to submissive/controlling). Social interactions may be coded at vary-
ing levels of complexity. At the simplest level, the model is divided into 4
quadrants with Affiliative in the top right-hand corner, and Distant, Hostile,
and Friendly proceeding in a counterclockwise order. The model may be
further divided into eight clusters (i.e., 1 ⫽ assert and separate/free and
forget, 2 ⫽ disclose and express/affirm and understand, 3 ⫽ approach and
enjoy/nurture and comfort, 4 ⫽ trust and rely/help and protect, 5 ⫽ defer
and submit/watch and manage, 6 ⫽ sulk and appease/belittle and blame,
7 ⫽ defend and withdraw/attack and reject, 8 ⫽ wall off and avoid/ignore
and neglect). Cluster 1 falls at the top pole of the vertical axis, and the
remaining clusters proceed in a clockwise order with Clusters 3, 5, and 7
falling on the remaining three poles. The most complex level involves
coding statements as belonging to 1 of 36 tracks. In this study, statements
are characterized as belonging to one of the quadrants and clusters. Coded
data are nominal and, therefore, require nonparametric statistical analysis.
Raters and reliability. Two graduate students who were unaware of the
hypotheses and the resolved status of the couples were trained on both
process measures to a satisfactory reliability (i.e., Cohen’s kappa level of
.70). Training consisted of coding transcripts from the training manuals
prior to carrying ratings on the transcripts from the study. To arrive at a
single rating for each partner for each session, the peak rating (deepest
level in the segment) was used for the ES (Klein et al., 1986), and the
modal rating (most frequent cluster and quadrant rating) was used for the
SASB (Benjamin et al., 1986). Interrater reliability was high, yielding
kappas of .83 ( p ⬍ .001) for the ES and .84 ( p ⬍ .001) for the SASB
quadrant and .73 ( p ⬍ .001) for the SASB cluster.
Self-Report Measures
Attachment Injury Measure (AIM). The AIM (Millikin, 2000) is a
modification of the single-item Target Complaints Discomfort Box Scale
(Battle et al., 1966). The measure was expanded to four items designed to
measure the current severity of the injury using a 5-point scale ranging
from 1 (severe)to5(negligible). For example, it asks the couple to rate the
significance of the event that injured the relationship bond. Reliability
coefficients for the injured and offending partners at pre- and posttreatment
ranged from .67 to .90.
Post-Session Resolution Questionnaire (PSRQ). The PSRQ (Orlinsky
& Howard, 1975) is a four-item measure designed to assess the amount of
in-session change and to identify best sessions. It consists of three 5-point
and one 7-point session evaluation items. For example, one item asks the
couple how resolved they feel at the end of the session regarding the issue
that brought them into counseling. High scores are indicative of no change,
and low scores are indicative of much change. This instrument has face
validity only, but it has been used in previous studies to successfully
identify best sessions (e.g., Greenberg & Foerster, 1996).
Couples Therapy Alliance Scale. The Couples Therapy Alliance Scale
(Pinsof & Catherall, 1986), a 28-item measure designed to assess the
perception of the therapeutic alliance, was administered to each partner to
ensure that differences between resolved and nonresolved groups could not
be attributed to the therapeutic alliance. Partners rate each item using a
7-point scale ranging from 7 (completely agree)to1(completely disagree).
High scores reflect a higher quality of the alliance between the couple and
therapist. This measure was administered once after the third session.
Cronbach’s alpha was .96 for the injured and offending partners.
Dyadic Adjustment Scale (DAS). The DAS (Spanier, 1976) is a widely
used, 32-item self-report measure of marital satisfaction. Both partners
completed the DAS. All items were answered using a Likert-type response
format. Scores can range from 0 to 151, with higher scores (97 and higher)
indicative of greater marital satisfaction. Reliability coefficients for the
1057
SPECIAL SECTION: RESOLVING ATTACHMENT INJURIES

injured and offending partners at pre- and posttreatment ranged from .74 to
.95.
Relationship Trust Scale (RTS). The RTS (Holmes, Boon, & Adams,
1990) is a 30-item inventory designed to assess interpersonal trust. Both
partners respond to items on a 7-point scale, ranging from 1 (strongly
disagree)to7(strongly agree). A high overall score indicates a higher
level of trust. Reliability coefficients for the injured and offending partners
at pre- and posttreatment ranged from .92 to .96.
Experiences in Close Relationships (ECR). The ECR (Brennan, Clark,
& Shaver, 1998) is a 36-item measure used to assess individual differences
with respect to attachment-related anxiety (i.e., the extent to which people
are insecure vs. secure regarding the availability and responsiveness of
romantic partners) and attachment-related avoidance (i.e., the extent to
which people are uncomfortable being close to and depending on others).
Both partners respond to items on a 7-point scale ranging from 1 (disagree
strongly)to7(agree strongly). The four attachment-style categories may
be derived from the two 18-item dimensional subscales and the classifica-
tion coefficients provided by the authors (Brennan et al., 1998). Internal
consistency for the injured and offending partners at pre- and posttreatment
ranged from .87 to .90.
Interpersonal Relationship Resolution Scale (IRRS). The IRRS (Har-
grave & Sells, 1997) is 44-item measure consisting of two scales designed
to assess the extent to which a person continues to feel pain as a result of
the offense and has forgiven his or her partner for the offense. Injured
partners were asked to respond with a yes or no to each of the items.
Summing the weights of each response yields a score for the two scales.
High scores indicate emotional healing and progress in the work of for-
giveness. Internal consistency for the forgiveness scale was .76 at pretreat-
ment and .87 at posttreatment. For the pain scale, internal consistency at
pre- and posttreatment was .58 and .73, respectively.
Procedure
All couples (N ⫽ 107) that responded to media advertisements that
described the University of Ottawa Research Ethics Board-approved study
were assessed for suitability using a telephone screening interview. Cou-
ples were probed for the presence of an attachment injury and had to meet
eligibility criteria, including living together for at least 1 year, no history of
sexual abuse, no reported psychiatric history, no problems with drugs and
alcohol, and no physical violence in the relationship. Suitable couples (n ⫽
43) were invited to the university, at which time informed consent was
obtained and a questionnaire package consisting of demographic informa-
tion, the AIM, DAS, RTS, ECR, and IRRS was completed. The DAS was
scored and used to further identify prospective couples. To be included in
the study, the injured partner’s DAS score ideally fell within the mild to
moderately distressed range (i.e., 80 and 97). Although mild to moderate
distress was preferred, more severely distressed couples were included in
the study providing both partners were committed to the relationship (as
measured by a score of 3 to 5 on DAS Item 32). Couples included in the
study were randomly assigned to 1 of 13 trained EFT therapists and offered
12 free sessions of couples therapy. Those that did not meet criteria were
referred to another provider of psychological services.
At the end of each therapy session, therapists asked both partners to
complete the PSRQ. At the end of the third session, therapists asked each
partner to complete the Couples Therapy Alliance Scale. At the final
session, each partner completed the PSRQ, AIM, DAS, RTS, ECR, and
IRRS. At the end of the study, couples had the option of continuing with
their therapist or being referred to another therapist.
Selection of resolved and nonresolved couples. At the end of the
treatment, couples were identified as either resolved or nonresolved. To be
identified as a resolved couple, three criteria had to be met: (a) Both
partners had to have a mean score greater than 10 on the AIM, (b) the
therapist had to indicate that the couple had completed the resolution steps
at the end of therapy, and (c) a clinical judge had to indicate that the couple
moved beyond the impasse and was making progress toward resolution. To
facilitate this, researchers selected “best sessions” for each couple based on
highest PSRQ ratings by the couple during the study. The clinical judge,
trained in EFT and the attachment injury resolution model, listened to best
sessions. Note that all best sessions occurred later in the process of therapy
(i.e., approximately Session 10 for both groups). There was no difficulty
identifying the 15 resolved couples. These couples had AIM ratings be-
tween 11 and 18, were selected by the therapist as having resolved the
injury, and were selected by the clinical judge as having made progress
toward forgiveness and reconciliation. Only 1 of the nonresolved couples
had an AIM rating higher than 10 but did not meet the other two criteria.
Selection of transcripts for coding. Audiotapes of first sessions were
queued by the therapist to the place that marked the telling of the attach-
ment injury. Best sessions were queued by the therapist to the place that
marked the most advanced step in the model. As in previous research, these
were used as a starting point for a 10-min segment that was transcribed for
process coding (Johnson & Talitman, 1997; Millikin, 2000). Although both
the designation of resolved status and the examination of change process
variables focused on best sessions, the designation of resolved status
resulted only in part from clients’ cognitive assessment of movement
beyond their impasse at the end of the best sessions, whereas the coding of
process variables focused on client behaviors during these sessions. The
designation of resolved status also required the unanimous judgment of
movement beyond the impasse to be made by a clinical judge and couples
therapist.
Implementation check. To verify that treatment was implemented ac-
cording to the EFT manual, several verification checks were carried out.
First, segments of therapy sessions were played during weekly group
supervision to ensure proper implementation. Second, the researcher au-
dited therapy sessions, randomly selected during the course of the study,
and judged that treatment was more than adequate. Third, two independent,
trained raters coded therapists’ statements in 10-min segments of each
couple’s 1st and 10th therapy sessions (randomly queued) against a check-
list of 16 (8 EFT and 8 non-EFT) interventions (Dandeneau & Johnson,
1994; Johnson & Talitman, 1997). Interrater reliabilities (Cohen’s kappa)
for total therapists’ interventions, therapists’ interventions for the resolved
group, and therapists’ interventions for the nonresolved group were .80,
.79, and .84, respectively. The mean rater percentages of EFT statements
for the resolved and nonresolved groups were 96% and 90%, which were
not significant, t(22) ⫽ –1.6, p ⬎ .05.
Results
At the end of treatment, 15 couples were identified as having
resolved the attachment injury, and 9 couples were identified as
nonresolved. The data file was examined for accuracy of data
entry. Data were missing for 3 couples on the ECR and the IRRS.
In addition, 2 couples did not complete the Couples Therapy
Alliance Scale. To avoid losing data because of casewise deletion,
we used mean substitution as it did not produce any systematic
bias to the results. Univariate normality was assessed by an ex-
amination of skewness, as well as an examination of the histo-
grams. The scores were within reasonable limits. No univariate or
multivariate outliers were found. Multicollinearity and singularity
were assessed using SPSS collinearity diagnostics and by exam-
ining correlation coefficients of the dependent variables. No mul-
ticollinearity or singularity was evident. Finally, analysis of Box’s
M test of homogeneity of variance–covariance matrices within
each cell for the two groups at pre- and posttreatment for the
injured and offending partners using a significance level of .001
showed no violation of the assumption. Levene’s test of equality of
error variances for the univariate between-groups analyses showed
no violation at p ⫽ .001.
1058
MAKINEN AND JOHNSON

Group Differences
No significant group differences were found on any of the
demographic variables, including age, gender, relationship length,
number of children, family income, level of education, previous
marriage, previous couples therapy, type of injury, and onset of
injury. No group differences were found in the therapeutic alliance
for the injured partner, offending partner, and the couple. There
were no pretreatment group differences on the severity of the
attachment injury, dyadic adjustment, the attachment dimensions
(i.e., avoidance and anxiety), and emotional pain and forgiveness.
However, significant pretreatment group differences were found
for the injured partners’ level of trust, F(1, 22) ⫽ 9.29, p ⫽ .006,
␩
p
2
⫽ .30. The injured partners in the nonresolved group reported
significantly lower levels of trust at pretreatment than the injured
partners in the resolved group. Pretreatment trust was used as a
covariate in the parametric analyses.
The Process of Change
To test the hypotheses related to the process of change, we used
the SASB and the ES. Chi-square analyses were conducted on first
session coded process data to see whether the resolved and non-
resolved groups differed on the following process variables: belit-
tle and blame, defend and withdraw, disclose and express needs,
affirm and understand, and depth of experiencing. In analyses in
which the expected frequency was less than 5 in a cell, Fisher’s
exact test was used. The results showed that the frequency of the
partners’ responses was similar for both groups.
For best sessions, chi-square analyses showed that resolved
couples had significantly more affiliative responses (i.e., disclose
and express, affirm and understand) than hostile/distant responses
(i.e., belittle and blame, defend and withdraw). Resolved partners
openly disclosed and expressed their needs and were affirming and
understanding, whereas nonresolved partners remained in a blame/
withdraw stance. In terms of level of experiencing, both partners
had significantly deeper levels of experiencing than the nonre-
solved group. All of the resolved couples exhibited deeper levels
of experiencing, ranging from a self-description of feelings (Stage
4) to expanding awareness of feelings (Stage 7). All but one of the
nonresolved couples had significantly lower levels of experienc-
ing, ranging from totally detached/superficial (Stage 1) to reactive
responding (Stage 3). The frequencies of response categories are
presented in Table 1.
Outcome
To test the hypotheses that resolved couples would exhibit
significant improvements on the outcome variables compared with
nonresolved couples, we conducted repeated-measures analyses
with resolved and nonresolved groups as the between-subjects
independent variable and partner type (i.e., injured vs. offending)
and time (i.e., pre- vs. posttreatment) as the within-subjects inde-
pendent variables. Separate analyses were conducted on each
dependent variable (i.e., dyadic adjustment, attachment avoidance,
and attachment anxiety). Given that the IRRS forgiveness and
emotional pain scores were available for only the injured partners,
we performed a doubly multivariate analysis (i.e., MANOVA with
a repeated-measures component) on forgiveness and emotional
pain. Trust was used as a covariate in all of the analyses.
The first set of repeated-measures analyses involved the depen-
dent variable dyadic adjustment. There was no three-way interac-
tion for partner type by time by group, F(1, 21) ⫽ 2.68, p ⬎ .01,
␩
p
2
⫽ .11. There were no two-way interaction effects for partner
type by group, F(1, 21) ⫽ 1.97, p ⬎ .01, ␩
p
2
⫽ .09, or partner type
by time, F(1, 21) ⫽ .00, p ⬎ .01, ␩
p
2
⫽ .00. However, there was
a significant two-way interaction for time by group, F(1, 21) ⫽
36.19, p ⫽ .000, ␩
p
2
⫽ .63, which is a large effect. There was a
main effect for group, F(1, 21) ⫽ 27.35, p ⫽ .000, ␩
p
2
⫽ .57, and
for time, F(1, 21) ⫽ 12.28, p ⫽ .002, ␩
p
2
⫽ .37. There was also a
main effect for partner type, F(1, 21) ⫽ 13.42, p ⫽ .001, ␩
p
2
⫽ .39,
suggesting that the injured partners were significantly more dis-
tressed than the offending partners. The main effects for group and
time are best interpreted in terms of their interaction. Figure 1
shows the mean DAS scores for the resolved and nonresolved
couples at pre- and posttreatment. Paired-samples t tests showed
no significant improvement in dyadic adjustment over time, t(8) ⫽
.58, p ⬎ .05, for the nonresolved group. However, the resolved
group showed a significant improvement in dyadic adjustment
from pre- to posttreatment, t(14) ⫽ –6.76, p ⫽ .000. According to
the three criteria outlined by Jacobson and Truax (1991), the
22-point increase in dyadic adjustment is a clinically significant
improvement. Of the 15 resolved couples, 47% of the injured
partners and 27% of the offending partners scored in the clinically
significant change range. See Table 2 for the means and standard
deviations.
The second repeated-measures analysis involved the anxiety
attachment dimension as the dependent variable. There were no
three-way interaction effects for partner type by time by group
interaction, F(1, 21) ⫽ 3.73, p ⬎ .01, ␩
p
2
⫽ .003. There were no
two-way interaction effects for partner type by group, F(1, 21) ⫽
1.30, p ⬎ .01, ␩
p
2
⫽ .06, for partner type by time, F(1, 21) ⫽ 4.45,
p ⬎ .01, ␩
p
2
⫽ .18, or for time by group, F(1, 21) ⫽ .31, p ⬎ .01,
Table 1
Response Frequencies in Best Sessions for the Resolved and
Nonresolved Groups on the Process Measures
Response category Resolved Nonresolved
Disclose and express needs (injured)
Present 14 1
Absent 1 8
Affirm and understand (offender)
Present 15 0
Absent 0 9
Belittle and blame (injured)
Present 0 7
Absent 15 2
Defend and withdraw (offender)
Present 0 7
Absent 15 2
Injured high experiencing
Present 15 0
Absent 0 9
Offender high experiencing
Present 15 1
Absent 0 8
Note. All ps ⫽ .000.
1059
SPECIAL SECTION: RESOLVING ATTACHMENT INJURIES

␩
p
2
⫽.02. There were no main effects for group, F(1, 21) ⫽ .001,
p ⬎ .01, ␩
p
2
⫽ .00, partner type, F(1, 21) ⫽ .16, p ⬎ .01, ␩
p
2
⫽ .01,
or time, F(1, 21) ⫽ 2.53, p ⫽ .13, ␩
p
2
⫽ .11.
Similarly, the avoidance of closeness dimension revealed no
three-way interaction effects for partner type by time by group,
F(1, 21) ⫽ 4.60, p ⬎ .01, ␩
p
2
⫽ .18. There were no two-way
interaction effects for partner type by group, F(1, 21) ⫽ 1.75, p ⬎
.01, ␩
p
2
⫽ .08, for partner type by time, F(1, 21) ⫽ .005, p ⬎ .01,
␩
p
2
⫽ .000, or for time by group, F(1, 21) ⫽ .02, p ⬎ .01, ␩
p
2
⫽
.001. There were no main effects for group, F(1, 21) ⫽ 5.95, p ⬎
.01, ␩
p
2
⫽ .22, partner type, F(1, 21) ⫽ .000, p ⬎ .01, ␩
p
2
⫽ .000,
or time, F(1, 21) ⫽ .75, p ⬎ .01, ␩
p
2
⫽ .03.
To investigate whether scores on the IRRS differed for the
resolved and nonresolved groups over time, we performed a dou-
bly multivariate analysis on the dependent variables emotional
pain and forgiveness. Pretreatment trust was used as a covariate.
Using Wilks’s criterion, we found that the combined dependent
variables revealed a significant group by time interaction, F(2,
20) ⫽ 8.31, p ⫽ .002, ␩
p
2
⫽ .45. There were no main effects by
group, F(2, 20) ⫽ 2.46, p ⬎ .01, ␩
p
2
⫽ .20, but there was a main
effect for time, F(2, 20) ⫽ 5.10, p ⫽ .01, ␩
p
2
⫽ .34.
Univariate analyses of emotional pain revealed no group by time
interaction effects, F(1, 21) ⫽ 3.78, p ⬎ .01, ␩
p
2
⫽ .15. There was
no main effect for group, F(1, 21) ⫽ 2.20, p ⬎ .01, ␩
p
2
⫽ .10.
There was, however, a significant main effect for time, F(1, 21) ⫽
6.53, p ⫽ .01, ␩
p
2
⫽ .24, suggesting that both groups benefited
from EFT. This difference was not considered clinically significant
because it was less than 2 standard deviations from the sample
mean (Jacobson & Truax, 1991).
For the forgiveness factor, univariate analyses revealed a sig-
nificant group by time interaction, F(1, 21) ⫽ 17.38, p ⫽ .000, ␩
p
2
⫽ .45, with significant main effects for time, F(1, 21) ⫽ 29.46,
p ⫽ .006, ␩
p
2
⫽ .31, but not for group, F(1, 21) ⫽ 4.89, p ⬎ .01,
␩
p
2
⫽ .19. Figure 2
shows the interaction for the resolved and
nonresolved groups at pre- and posttreatment. Paired-samples t
tests revealed no significant improvement in the work toward
forgiveness over time for the nonresolved couples, t(8) ⫽ .62, p ⬎
.05. However, the resolved group showed a significant improve-
Pretreatment Posttreatment
Estimated Marginal Means
120
110
100
90
80
70
Groups
Nonresolve
d
Resolved
Figure 1. Mean dyadic adjustment as a function of time with pretreat-
ment trust as a covariate.
Table 2
Injured and Offending Partners Dyadic Adjustment, by Group and Time
Partner type
Pretreatment Posttreatment
Bonferroni tMSD M SD
Resolved group (n ⫽ 15)
Injured 85.47 8.23 110.13 12.32 ⫺7.11
*
Offending 90.07 10.19 108.53 10.93 ⫺5.80
*
Nonresolved group (n ⫽ 9)
Injured 78.89 13.38 75.67 9.98 .96
Offending 80.89 6.70 81.00 11.20 .03
*
p ⬍ .001.
Pretreatment Posttreatment
Estimated Marginal Means
42
40
38
36
34
32
30
Groups
Nonresolved
Resolved
Figure 2. Mean forgiveness as a function of time with pretreatment trust
as a covariate.
1060
MAKINEN AND JOHNSON

ment in forgiveness from pre- to posttreatment, t(14) ⫽ –9.92, p ⫽
.000. This improvement is considered clinically significant (Jacob-
son & Truax, 1991). See Table 3 for the means and standard
deviations for the resolved and nonresolved groups on emotional
pain and forgiveness at pre- and posttreatment.
Discussion
The purpose of this study was to build on previous attachment
injury process research by describing the resolution process among
couples treated with EFT. Task analysis was employed to inves-
tigate whether the resolution model discriminated resolved from
nonresolved couples and to explore differences in outcome. The
goal was to validate the attachment injury resolution model so as
to better explain therapeutic change using EFT.
At the outset, the resolved and nonresolved couples did not
differ on the frequency of hostile responses (i.e., belittle and
blame, defend and withdraw) or on their depth of experiencing.
These findings are consistent with the first two steps of the
attachment injury resolution model that mark the disclosure of the
event (Johnson et al., 2001; Millikin, 2000; Naaman et al., 2005).
When the injured partner speaks of the incident, there is often an
explosion of emotions, such as anger and rage. The offending
partner finds this aversive and moves into a defensive stance or
withdraws. Although a couple with an attachment injury may
appear similar to any other distressed couple (i.e., they get caught
in a pursue–withdraw interactional pattern), the defining feature in
attachment-related events is that they are not easily forgotten and
are often used to define relationship safety. Attachment injuries
often exacerbate conflicts and block risk taking and intimacy
(Johnson, 2004).
The prediction that resolved couples would be more affiliative and
exhibit deeper levels of experiencing as treatment progressed, com-
pared with nonresolved couples, was supported. As expected, the
frequency of affiliative responses, such as disclosing and expressing
needs and affirming and understanding, was significantly different for
the resolved and nonresolved couples. Furthermore, the frequency of
hostile responses, such as belittle and blame and defend and with-
draw, significantly differed, as did the frequency of injured and
offending partners’ level of experiencing. These robust findings are
consistent with the attachment injury resolution model that has shown
that deeper levels of experiencing are key ingredients of change
(Johnson et al., 2001; Millikin, 2000). Previous EFT process research
also has found that when “blamers” move from impersonal, self-
limiting emotional involvement to an exploration of underlying needs,
particularly when their partner is emotionally engaged, a softening
occurs. A softening is a shift from anxious, emotional reactivity to
showing vulnerability by expressing needs in a manner that pulls for
emotional connection. This softening changes the relationship by
initiating safe emotional engagement and connection (Johnson &
Greenberg, 1988).
To illustrate the process of change, a generic example of an
attachment-injured couple (John and Pat) is provided. John and
Pat, both successful professionals, had been married for 30 years
when Pat found out that John had had an affair 3 years before when
he was away on business. Pat felt betrayed, and her trust for John
was shattered. Over the years, Pat and John never discussed the
incident, although John complained that Pat repeatedly reminded
him of his indiscretion. At the intake, Pat presented as angry and
hostile toward John, and he was very defensive. Pat was hyper-
vigilant whenever John went out. She wanted to know where he
was going, whom he was meeting, and when he was coming home.
When he was late, she would become distraught, and the cycle
would escalate immediately upon his return. John reported feeling
controlled and refused to comply with her unrealistic demands.
When John was home, he often retreated to his music studio and
listened to music with headphones, which incited Pat’s anger.
Therapist: Help me here, John; from what you are telling me,
the incident that brought you into therapy is still
very much in the foreground.
John: Well, I don’t know if it is in the foreground, but it
definitely is still there for me. I kind of blocked it
out for the longest time. The more she would throw
it at me, the more I built a wall around it and
myself. Now, I feel ashamed about what I did.
Therapist: I’m wondering if it would help if you talked about
your shame?
John: I don’t know. I’m scared to bring up all the emotions.
Table 3
Injured Partners for the Two Interpersonal Relationship Resolution Scale Variables, by Group
and Time
Variable
Pretreatment Posttreatment
Bonferroni tMSDMSD
Resolved group (n ⫽ 15)
Pain 36.72 2.88 39.68 1.32 ⫺4.24
Forgiveness 32.02 3.31 40.75 1.84 ⫺9.92
*
Nonresolved group (n ⫽ 9)
Pain 35.56 3.01 37.11 4.81 ⫺0.62
Forgiveness 32.22 4.79 33.67 4.33 ⫺1.39
*
p ⬍ .001.
1061
SPECIAL SECTION: RESOLVING ATTACHMENT INJURIES

Therapist: You have compartmentalized your feelings, never
talked about your shame because when she got
scared she would remind you of it. Is that it?
John: Yeah, she would get angry, and it was like replay-
ing an old movie. I’d hear it over and over again . . .
I couldn’t escape. I knew that I’d have to deal with
it one day, but....
Therapist: So this is new for you to talk about your feelings,
yes?
John: Maybe it’s because I have a really hard time choos-
ing my words and how I would say it.
Therapist: It’s hard to take that in when there is so much
shame and regret about what happened. Can you try
to tell her now?
John (tearfully
turning to Pat): I am so, so sorry . . . I’m really sorry. I wasn’t
thinking, and it was a silly, selfish thing to do . . .
without any thought on my part of how it would
affect you.
Pat: Well . . . I accept your apology.
John (sobbing): Because I do love you so very much, and I do want
to grow old with you.
Pat: I can’t believe how I feel . . . since we’ve been
coming here. It is in the past, and I don’t carry it
with me anymore. My life doesn’t revolve around it
anymore. I don’t get up in the morning and think
about it. I am sad to hear that you are dealing with
it more now than before. I don’t want you to worry
about it.
Therapist: It saddens you to hear that he still carries this?
Pat: Uh hmm . . . because it is gone for me, I have
forgiven him. It is in the past. It is not even an issue
anymore. I will never again . . . I know . . . that I
will never worry about it. I just miss you. I need
you in my life. I too want to grow old with you.
John: It has helped talking about the affair and hearing
how you felt. When you told me how you felt
before, all I heard was your anger, which is now
gone. This makes such a big difference when we
talk now.
Although John was processing his shame, they both indicated
that the incident that brought them to therapy was clearly in the
past. They moved from the negative pattern of attack–defend to a
more affiliative stance. They were able to openly disclose and
respond to each other in an affirming and understanding manner.
In addition, they both shifted from reactive responding and de-
scribing the event to a greater awareness of primary feelings as
they emerged. In general, resolved couples were open to exploring
feelings related to the event. Injured partners were able to disclose
and express their needs once the offending partner took responsi-
bility for his or her behavior. Unresolved couples, however, re-
mained stuck in an attack-and-blame and defend-and-withdraw
stance. They were reactive and intellectualized the event.
Couples identified as having resolved the attachment injury also
showed significant gains on outcome. For the resolved couples, both
the injured and offending partners’ dyadic adjustment scores signifi-
cantly increased from moderately distressed to the nondistressed
range. Similarly, resolved couples made significant gains in their
reported level of forgiveness, whereas no change was noted for the
nonresolved couples. This clinically significant finding gives an in-
dication as to where the injured partners are in the process of forgive-
ness, which involves being able to recognize negative interactional
patterns, setting limits, restoring trust and building emotional bonds,
and taking responsibility (Hargrave & Sells, 1997). Despite this group
difference on forgiveness at posttreatment, there were no group dif-
ferences with respect to emotional pain. However, the injured partners
in the resolved and nonresolved groups showed a significant decrease
in emotional pain over time. Given that the injured partners are the
ones emotionally wounded by their partner, it makes sense clinically
and theoretically that the injured partners in both groups would
experience significant decreases in emotional pain as a result of EFT.
Although it can be argued that the research design used in this
study does not definitely rule out the possibility that change
occurred as a result of time, the average time that couples nursed
these wounds before seeking treatment was 5 years. This length of
time and clinical experience of the impasse that these violations of
trust create in relationships argue against the notion that changes
observed in this study were simply the result of the passing of time.
Other models of therapy may also be able to elicit a similar process
to the change described here; however, other models do not use an
attachment frame to understand relationship events and focus on
other change processes besides the ones described here (i.e., deep-
ening and sharing of reprocessed emotional experience).
The attachment dimensions, however, revealed no significant
group differences. It has been argued that these patterns, which have
also been referred to in the literature as styles, strategies, orientations,
and habitual forms of engagement (Bartholomew & Horowitz, 1991;
Brennan et al., 1998; Johnson, 2002), may be more enduring charac-
teristics that are not easily modified. Individuals with a secure style
have developed constructive coping strategies and may be better able
to talk openly with their partners in response to violations of trust
(Mikulincer, 1998), communicate their needs, and reach out for and
provide support (Simpson, 1990). As such, secure individuals may be
better equipped to work through attachment-related events. In this
study, very little is known about the attachment patterns prior to their
injurious events. Most of the injured partners were insecurely attached
at intake. In other words, the attachment injury may not have caused
the pretreatment attachment patterns. The injurious event may have
elicited attachment-related behaviors consistent with the various at-
tachment patterns. The old strategies for dealing with conflict may
become activated by the injury, thus maintaining the negative inter-
actional cycle.
It may also be that changing attachment as measured here is a
long-term process that cannot be captured within the time period of
the study. Traditional attachment theory states that working mod-
els of self and other have to be revised for change to occur. This
revision may require a number of positive experiences even after
an optimal “corrective” experience of new and reparative interac-
tions in therapy.
Another explanation may be a measurement issue. The attach-
ment measure used in this study instructs each partner to respond
to the 36 statements in terms of how he or she generally experi-
1062
MAKINEN AND JOHNSON

ences romantic relationships. Thus, the measure may not have
been sensitive enough to detect changes in attachment anxiety and
avoidance with the specific partner as a result of treatment. Per-
haps if the items were modified to capture how he or she experi-
ences the current partner, rather than the history of romantic
partners, a significant shift in attachment anxiety and avoidance
may have been detected for the resolved couples.
Overall, the results of this study have both theoretical and
clinical implications. They provide empirical support for the at-
tachment injury resolution model. Many couples with an attach-
ment injury seem unaware of the impact such events have on their
relationship bond. In addition, couples therapists may not attend to
seemingly minor incidents that can impede the therapeutic process.
In our working definition of attachment injuries, we focused on the
meaning of these events. Prototypical attachment-related injuries,
such as infidelity, may be more obvious to a therapist. However,
both the couple and the therapist may overlook the significance of
seemingly benign emotional wounds (e.g., insulting remark at a
key moment). Therefore, the couple may reach an impasse in
therapy (i.e., partners are unable to move beyond de-escalation and
relapse once therapy is terminated).
In this study, 63% of the sample successfully resolved the
attachment injury to the point of achieving a clinically significant
improvement in relationship distress and forgiveness. Although
this recovery rate is less that what has been found in the EFT
literature (Johnson et al., 1999), the results of the current study
were found to be stable at a 3-year follow-up (Schnare, Makinen,
& Johnson, 2006). It is important to note that over the course of
therapy, therapists discovered that seven of the nine nonresolved
couples had compound injurious events that may have contributed
to the lower levels of trust at the beginning of treatment. In some
cases, the offending partner repeatedly injured his or her spouse. In
other cases, partners hurt each other at two different points in time.
In either case, compound injuries may be potent enough to irre-
trievably shatter the trust in the relationship, thus making the
resolution process even more arduous.
An important prognostic indicator of outcome is trust, particu-
larly placing faith in one’s partner (Johnson & Talitman, 1997).
Faith in one’s partner curbs feelings of uncertainty, thus enabling
one to relax and surrender fears and insecurities about the rela-
tionship (Holmes et al., 1990). Given that attachment injuries have
been likened to trauma, restoring trust in couples with compound
injuries may require longer treatment than the limited number of
sessions offered in this study. The resolution process may involve
more extensive treatment to work through the emotional wounds
and rebuild trust. The clinical norm for working with traumatized
couples is 30 to 35 sessions (Johnson, 2002). But before the
therapist can even begin to address the injury, the therapist has to
develop a good working alliance with both partners and complete
the de-escalation stage of EFT. Depending on the couple, building
an alliance and then attaining de-escalation may be a long process.
Therapists working with couples that have compound attachment
injuries may need to extend the duration of treatment even further
to heal connections (Johnson, 2002).
This study has several methodological limitations that suggest
possible future research. First, regarding the process of change, the
study focused solely on client behaviors. There are now some
studies that use task analysis research to investigate therapists’
behaviors that facilitate change in couples that have undergone
EFT (e.g., Bradley, 2004). To further extend the theoretical un-
derstanding of the attachment injury resolution process and to
provide a more detailed clinical map for therapists working with
such couples, it would be beneficial to track and identify therapist
interventions. However, tracking client behaviors from session to
session may lead to the examination of the contribution of different
change factors in successful outcomes. For example, changes in
the injured partner’s attributions about the offending partner may
be less important in the overall process than changes in the injured
partner’s affect regulation and, thus, his or her ability to engage
with their spouse in bonding enactments.
Second, task analysis is a labor-intensive project that inevitably
requires realistic limits on the number of participants involved.
EFT has been shown to be highly effective at reducing general
marital distress (Johnson et al., 1999). The results discussed here
were consistent with subsequent regression analyses showing that
attainment of forgiveness in this small sample predicted dyadic
adjustment ( p ⬍ .002) at posttreatment. However, it is not possible
in task analysis to make the three-way link between in-session
process, resolution of a task, and treatment outcome. It is possible
that in-session change processes are not causal and that resolution
is the result of other factors. A large randomized control study with
a focus on outcome may provide more robust findings.
Finally, a key issue in evaluating psychotherapy research is the
extent to which the findings may be generalized. The motivation of
couples responding to media advertisements offering free couples’
therapy may differ from couples seeking therapy. Specifically,
couples in this study may not have been as distressed as couples
typically presenting with attachment-related injuries. Moreover,
many of the couples in this study had complex attachment injuries.
To be able to generalize these findings beyond couples with a
single attachment injury, research with a focus on complex attach-
ment injuries and the process involved in restoring trust, fostering
forgiveness, and facilitating attachment security is warranted. In
addition, it may be beneficial to look at the impact of a single event
(e.g., infidelity) versus more chronic incidents, such as continuous
minor flirtations.
These limitations notwithstanding, this study contributes to the
attachment literature and the field of couples therapy. To summa-
rize, we used task analysis to better understand couples with
attachment-related problems that often block relationship repair.
The study provided validation for the attachment injury resolution
model, as outlined in the EFT literature, by providing therapists
with a map of key responses and moments in the forgiveness
change process. It also showed that it is important for therapists to
attend to attachment injuries so that they can foster the growth of
trust and create positive cycles of bonding and reconciliation.
Moreover, it brings EFT into the realm of forgiveness and recon-
ciliation and thus allows the EFT therapist to address key impasses
in couples therapy and maximize success.
References
Abrahms-Spring, J. (1997). After the affair: Healing the pain and rebuild-
ing trust when a partner has been unfaithful. New York: HarperCollins.
Atkinson, L. (1997). Attachment and psychopathology: From laboratory to
clinic. In L. Atkinson & K. J. Zucker (Eds.), Attachment and psycho-
pathology (pp. 3–16). New York: Guilford Press.
Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among
1063
SPECIAL SECTION: RESOLVING ATTACHMENT INJURIES

young adults: A test of a four-category model. Journal of Personality
and Social Psychology, 61, 226 –244.
Battle, C. C., Imber, S. D., Hoehn-Saric, R., Stone, A. R., Nash, E. R., &
Frank, J. D. (1966). Target complaints as criteria of improvement.
American Journal of Psychotherapy, 20, 184–192.
Baucom, D., Shoham, V., Mueser, K., Daiuto, A., & Stickle, T. (1998).
Empirically supported couple and family interventions for marital dis-
tress and adult mental health problems. Journal of Consulting and
Clinical Psychology, 66, 53–88.
Benjamin, L. S. (1974). Structural analysis of social behavior. Psycholog-
ical Review, 81, 392–425.
Benjamin, L. S., Foster, S. W., Roberto, L. G., & Estroff, S. E. (1986).
Breaking the family code: Analysis of videotapes of family interactions
by Structural Analysis of Social Behavior (SASB). In L. S. Greenberg &
W. M. Pinsof (Eds.), The psychotherapeutic process: A research hand-
book (pp. 391–438). New York: Guilford Press.
Bradley, B. (2004). Toward a mini-theory of the blamer softening event:
Tracking the moment-by-moment process. Journal of Marital and Fam-
ily Therapy, 30, 233–246.
Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self-report measure-
ment of adult attachment. In. J. A. Simpson & W. S. Rholes (Eds.),
Attachment theory and close relationships (pp. 46–76). New York:
Guilford Press.
Dandeneau, M., & Johnson, S. M. (1994). Facilitating intimacy: Interven-
tions and effects. Journal of Marital and Family Therapy, 20, 17–33.
Enright, R. D., & Fitzgibbons, R. P. (2000). Helping clients forgive: An
empirical guide for resolving anger and restoring hope. Washington,
DC: American Psychological Association.
Gordon, K. C., & Baucom, D. H. (1998). Understanding betrayals in marriage:
A synthesized model of forgiveness. Family Process, 37, 425–449.
Gordon, K. C., Baucom, D. H., & Snyder, D. K. (2000). The use of
forgiveness in marital therapy. In M. E. McCullough, K. I. Pargament,
& C. E. Thoresen (Eds.), Forgiveness: Theory, research, and practice
(pp. 203–227). New York: Guilford Press.
Greenberg, L. S. (1984). Task analysis: The general approach. In L. N.
Rice & L. S. Greenberg (Eds.), Patterns of change: Intensive analysis of
psychotherapy process (pp. 124–148). New York: Guilford Press.
Greenberg, L. S., & Foerster, F. S. (1996). Task analysis exemplified: The
process of resolving unfinished business. Journal of Consulting and
Clinical Psychology, 64, 439– 446.
Hargrave, T. D. (2004). Families and forgiveness: A theoretical and ther-
apeutic framework. The Family Journal: Counseling and Therapy for
Couples and Families, 2, 339 –348.
Hargrave, T. D., & Sells, J. N. (1997). The development of a forgiveness
scale. Journal of Marital and Family Therapy, 23, 41–62.
Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an attach-
ment process. Journal of Personality and Social Psychology, 52, 511–524.
Hazan, C., & Zeifman, D. (1999). Pair bonds as attachments: Evaluating
the evidence. In J. Cassidy & P. Shaver (Eds.), Handbook of attachment:
Theory, research, and clinical applications (pp. 336–354). New York:
Guilford Press.
Holmes, J. G., Boon, S. D., & Adams, S. (1990). The Relationship Trust
Scale. Unpublished manuscript, University of Waterloo.
Jacobson, N. S., & Truax, P. (1991). Clinical significance: A statistical
approach to defining meaningful change in psychotherapy research.
Journal of Consulting and Clinical Psychology, 59, 12–19.
Johnson, S. M. (1998). Listening to the music: Emotion as a natural part of
systems theory. Journal of Systemic Therapies, 17, 1–17.
Johnson, S. M. (2002). Emotionally focused couple therapy with trauma
survivors: Strengthening attachment bonds. New York: Guilford Press.
Johnson, S. M. (2004). The practice of emotionally focused marital ther-
apy: Creating connections (2nd ed.). New York: Brunner/Mazel.
Johnson, S. M., & Greenberg, L. S. (1988). Relating process to outcome in
marital therapy. Journal of Marital and Family Therapy, 14, 175–183.
Johnson, S. M., Hunsley, J., Greenberg, L. S., & Schindler, D. (1999).
Emotionally focused couples therapy: Status and challenges. Clinical
Psychology: Science and Practice, 6, 67–79.
Johnson, S. M., Makinen, J. A., & Millikin, J. W. (2001). Attachment
injuries in couple relationships: A new perspective on impasses in
couples therapy. Journal of Marital and Family Therapy, 27, 145–155.
Johnson, S. M., & Talitman, E. (1997). Predictors of success in emotionally
focused marital therapy. Journal of Marital and Family Therapy, 23,
135–152.
Johnson, S. M., & Whiffen, V. E. (Eds.). (2003). Attachment processes in
couple and family therapy. New York: Guilford Press.
Johnson, S. M., & Williams-Keeler, L. (1998). Creating healing relation-
ships for couples dealing with trauma: The use of emotionally focused
couples therapy. Journal of Marital and Family Therapy, 24, 25–40.
Kirkpatrick, L. A., & Davis, K. E. (1994). Attachment style, gender, and
relationship stability: A longitudinal analysis. Journal of Personality
and Social Psychology, 66, 502–512.
Klein, M. H., Mathieu-Coughlan, P., & Kiesler, D. J. (1986). The experiencing
scales. In L. S. Greenberg & W. M. Pinsoff (Eds.), The psychotherapy
process: A research handbook (pp. 21–71). New York: Guilford Press.
Main, M., & Hesse, E. (1990). Parent’s unresolved traumatic experiences
are related to infant disorganized attachment status. In M. Greenberg &
D. Cicchetti (Eds.), Attachment in the pre-school years (pp. 152–176).
Chicago: University of Chicago Press.
Mikulincer, M. (1998). Attachment working models and the sense of trust:
An exploration of interaction goals and affect regulation. Journal of
Personality and Social Psychology, 74, 1209–1224.
Millikin, J. W. (2000). Resolving attachment injuries in couples using
emotionally focused therapy: A process study. Unpublished doctoral
dissertation, Virginia Polytechnic Institute and State University.
Naaman, S., Pappas, J. D., Makinen, J., Zuccarini, D., & Johnson, S.
(2005). Treating attachment injured couples with emotionally focused
therapy: A case study approach. Psychiatry: Interpersonal and Biolog-
ical Processes, 68, 55–77.
Orlinsky, D., & Howard, K. (1975). Varieties of psychotherapeutic expe-
rience. New York: Teachers College Press.
Orlinsky, D., & Howard, K. (1986). The psychological interior of psycho-
therapy: Explorations with the therapy session reports. In L. S. Green-
berg & W. M. Pinsoff (Eds.), The psychotherapy process: A research
handbook (pp. 477–502). New York: Guilford Press.
Pinsof, W. M., & Catherall, D. R. (1986). The integrative psychotherapy
alliance: Family, couple and individual therapy scales. Journal of Mar-
ital and Family Therapy, 12, 137–151.
Rowe, J. O., Halling, S., Davies, E., Leifer, M., Powers, D., & van
Bronkhorst, J. (1989). Exploring the breadth of human experience. In
R. S. Valle & S. Halling (Eds.), Existential–phenomenological perspec-
tives in psychology (pp. 233–244). New York: Plenum Press.
Schnare, R., Makinen, J. A., & Johnson, S. M. (2006). Resolving attach-
ment in couples: A three-year follow-up study. Unpublished manuscript,
University of Ottawa.
Simpson, J. A. (1990). The influence of attachment styles on romantic rela-
tionships. Journal of Personality and Social Psychology, 59, 971–980.
Simpson, J. A., & Rholes, W. S. (1994). Stress and secure base relation-
ships in adulthood. In K. Bartholomew & D. Perlman (Eds.), Attachment
processes in adulthood. Advances in personal relationships (pp. 181–
204). Philadelphia: Jessica Kingsley.
Spanier, G. (1976). Measuring dyadic adjustment. Journal of Marriage and
the Family, 38, 15–28.
Stone, A. (1996). Trauma and affect. In D. Nathanson (Ed.), Knowing and
feeling (pp. 288–302). New York: Norton.
Received July 5, 2005
Revision received September 5, 2006
Accepted September 6, 2006 䡲
1064
MAKINEN AND JOHNSON