journal of Consulting and Clinical Psychology Copyright 1992 by the American Psychological Association, Inc.

1992. Vol. 60, No. 6, 962-969 0022-006X/92/S3.00

Comparison of Integrated Systemic and Emotionally Focused Approaches

to Couples Therapy
A. Goldman L. Greenberg
Independent Practice \brk University
Vancouver, British Columbia, Canada North York, Ontario, Canada

This study compared couples receiving two marital therapy approaches and a control group over a
10-week treatment period. Integrated systemic therapy (1ST) and an emotionally focused approach
(EFT) were both found to be superior to the control and to be equally effective in alleviating
marital distress, facilitating conflict resolution and goal attainment, and reducing target com-
plaints at termination. 1ST couples, however, showed greater maintenance of gains from termina-
tion to 4-month follow-up on marital satisfaction and goal attainment. Clients' perceptions of how
change occurred and issues related to the use of a team of observers in 1ST are discussed.

Marital therapy researchers have recently begun to compare self-perpetuating: It comes to exist independently of its reason
the effectiveness of different approaches to treatment (Johnson for originally existing. A number of couples patterns such as
& Greenberg, 1985b; Snyder & Wills, 1989). An emotionally pursue-distance, demand-withdraw, and attack-attack have
focused approach to marital therapy has been shown to be been delineated (Greenberg & Johnson, 1984; Guerin, 1982;
equal to a behavioral marital treatment on some outcome indi- Wile, 1981). A number of therapeutic approaches target this
ces and superior on others (Johnson & Greenberg, 1985a). Con- negative cycle for intervention.
tinued research on emotionally focused couples therapy A key principle followed by a group of interactional thera-
(James, 1991; Johnson & Greenberg, 1985b) has demonstrated pists (Fisch et al., 1982; Sluzki, 1978, 1983) is that negative
that this approach shows promise. interaction patterns represent clients' "attempted solutions" to
Another promising set of approaches to marital therapy problems and that it is these attempted solutions that contrib-
come out of the systemic-interactional tradition (Fisch, Weak- ute to the problem's maintenance or escalation. If problem for-
land, & Segal, 1982; Sluzki, 1978,1983). Clinicians experienced mation and maintenance are seen as part of a circular process
in systemic marital therapies have attested to positive outcomes in which well-intended solution-behaviors maintain the prob-
resulting from their use (Coyne, 1984; Steinglass, 1978). How- lem, then alteration of these behaviors should interrupt the
ever, until recently little or no empirical evidence has supported cycle and initiate resolution of the problem. The primary aim
these claims. To determine the validity of present claims about of treatment, then, need not be resolution of all difficulties but,
the effectiveness of both systemic and emotionally focused instead, the initiation of a reversal in the interaction. In this
marital therapy, we undertook a study comparing their effec- type of systemic approach to marital therapy, a team of thera-
tiveness. pists is often used behind a one-way mirror to observe the
Because of the close link between marital dissatisfaction and interaction, to consult with the therapist in the room, and to
conflict, one of the main goals of marital therapy is to facilitate send messages to the couple designed to change the interaction.
the resolution of conflict between spouses. Conflict between Therapeutic interventions devised by the team and delivered by
marital partners is often associated with escalating interac- the therapist generally involve reframing or positively connot-
tional cycles that have evolved systematically. In a typical cycle ing the couple's symptom or the functioning of the system or
each partner has a specific role. For example, the wife may both. This is tantamount to "prescribing the symptom," which
engage in nagging criticism while the husband engages in pas- in marital therapy is the negative interactional pattern. The
sive withdrawal, which produces more nagging, leading to team, as well as developing the interventions, is also seen as
more withdrawal and thus keeping the cycle alive (Guerin, enhancing the persuasive impact of the interventions.
1982; Steinglass, 1978). Negative interaction cycles in couples Proponents of the emotionally focused approach, although
have been observed to evolve repeatedly when dealing with is- agreeing that the negative cycle is maintained by the interac-
sues concerning closeness versus distance and dominance ver- tion, view the interaction as resulting from the partners' feel-
sus submission. Once a negative interaction cycle is in place, it is ings of lack of entitlement to their adult unmet needs. The
therapeutic change process in this view is seen as involving the
accessing and expressing of primary feelings and needs that
This study was partially funded by Social Sciences and Humanities have not been previously expressed. The expression of these
Research Council Grant 410840135 to L. Greenberg. emotions and their acceptance by the partner are seen as lead-
Correspondence concerning this article should be addressed to L. ing to a change in the interaction. Thus if the nagging spouse
Greenberg, Department of Psychology, York University, 4700 Keele expresses an underlying loneliness or fear of abandonment and
Street, North York, Ontario, Canada M3J 1P3. the withdrawn partner assertively expresses the resentment or
962
 COMPARISONS IN COUPLES THERAPY 963

fear of being overwhelmed, this will restructure the interaction. that included 3 years of marital therapy implemented within the frame-
The interventions in this approach to marital therapy in gen- work of the model they used in this study. All of the therapists ex-
eral involve facilitating the expression and acceptance of under- pressed a commitment to the therapeutic orientations of their respec-
lying feelings in order to restructure the interaction. tive groups; those in the EFT group had at least 24 hr of supervised
training in experimental therapy beyond graduate training, and those
The aim of the present study was to compare the effects of
in the 1ST group had no less than 30 hr of training in structural-strate-
emotionally focused couples therapy (EFT) implemented ac- gic systemic therapy.
cording to a treatment manual (Greenberg & Johnson, 1984, Each group of therapists received, in addition, 12 hr of instruction in
1988) with the effects of integrated systemic marital therapy implementing their respective therapy's manual. Two hours of group
(1ST) implemented according to a treatment manual (Green- supervision of the therapists took place once weekly. The 1ST treat-
berg & Goldman, 1985). Although the emotionally focused ther- ment team consisted of three trained therapists behind a one-way
apy had been empirically tested in Johnson and Greenberg's mirror during the therapy sessions who helped formulate interventions
(1985a, 1985b) studies, the integrated systemic marital therapy and communicated with the therapist using a call-in phone. During
had never been empirically tested. the 1ST weekly supervision these consulting therapists viewed video-
tapes of the sessions and discussed reframes and prescriptions with the
couple's primary therapist. Both EFT and 1ST therapist groups re-
Method ceived an equivalent amount of weekly supervision.
In this study, 42 couples seeking help for problems in conflictual
relationships were randomly assigned to one of the two treatment Measures
groups (EFT or 1ST) and one of the seven therapists implementing
The Couples Therapy Alliance Scale (CTAS; Pinsof & Catherall,
each treatment or to the wait-list control group. All couples were tested
1983). This instrument measures the client's view of the therapeutic
at the inception of the project. The treatment couples received 10 ses-
relationship. The measure has three parts: agreement between client
sions of conjoint marital therapy and were tested after therapy and at a
and therapist on therapeutic goals, the existence of personal bonds
4-month follow-up. The control group was tested after a 10-week wait-
between therapist and client, and the development of tasks that are
ing period and then were treated; there was therefore no 4-month fol-
perceived by the client as relevant to his or her concerns. These were
low-up measure on the controls.
identified by Bordin (1979) as the key components of the therapeutic
alliance. The measure comprises 28 items (11 relate to self, 11 relate to
Subjects partner, and 6 relate to the relationship) that the subject responds to on
a Likert-ty pe 7-point scale. The instrument was intended to control for
Couples participating in the study were seeking help with problems the overall quality of the alliance.
in conflictual relationships. They had initially responded to a newspa- The Dyadic Adjustment Scale (DAS; Spanier, 1976). This self-report
per article that described the research project as evaluating the effec- measure was used for screening and as a general measure of outcome at
tiveness of marital counseling techniques in helping couples resolve posttest and at follow-up. The DAS can be used as a measure of global
their problems. Those couples who met the requirements of the project marital adjustment and comprises four subscales measuring dyadic
were offered 10 free sessions, 1 hr each week. To be included, couples satisfaction (10 items), consensus (13 items), cohesion (5 items), and
had to be living together for 18 months, have no plans for separation, affectional expression (94 items). It is at present the preferred instru-
not have received psychiatric treatment within the 2-year period pre- ment for the assessment of marital adjustment, on the basis of validity
vious to the treatment, and not be suffering from clinical depression or and reliability (Cronbach's coefficient alpha = .96).
any addiction. In addition, at least one partner in each dyad had to Target Complaints (TC; Battle et al., 1966). This measure was filled
score within the "distressed" range on the Dyadic Adjustment Scale out by each partner in conjunction with their therapist during the
(DAS; Spanier, 1976), that is, to have a score below 95. Of the 132 initial interview and was also given at posttest and at the 4-month
couples that responded to the newspaper article, 87 met these criteria. follow-up. The TC comprises three 5-point scales on which the client is
Of these, 42 were randomly selected and randomly assigned to one of asked to rate the amount of change on three different complaints. Bat-
the treatments or to the control group. tle et al. (1966) gave evidence as to the validity and reliability of this
Of the 42 couples who participated in the study, there were 14 in each measure and reported a test-retest reliability of .68.
treatment group and 14 in the control group. The mean score on the Goal Attainment Scaling (GAS; Kiresuk & Sherman, 1968). With
DAS was 84.2. The sample consisted of Caucasian couples with an this measure, clients were asked to set specific behavioral goals in
average age for men of 39.4 years (range = 26 to 57 years) and for relation to their main presenting concern in the assessment interview;
women of 37.6 years (range = 23 to 52 years). The average educational the attainment of these goals was measured at posttest and at the 4-
level of the spouses was 1 year of postsecondary schooling, and the month follow-up. Five levels of attainment—"worse than expected re-
mean annual income was $38,500. The average length of the partner- sults," "less than expected results," "expected results," "somewhat bet-
ships was 11.26 years (range = 1 to 27 years), there were an average of ter than expected results," and "much better than expected results"—
1.36 children per family (range = 0 to 5 children), and 13% of the were specified during assessment in terms of three observable
spouses had been married previously. Five couples had received pre- behaviors and one emotional response.
vious marital therapy. When these demographic variables were ana- Conflict Resolution Scale (CRS; Fournier, Olson, & Druckman,
lyzed, no significant differences on them were found among the three 1983). This measure was completed by each client in the assessment,
groups nor between the men and women. at posttest, and at the 4-month follow-up. The couple scores were de-
signed to tap the four main dimensions of differences or disagree-
Therapists ments, negative agreements, indecisive responses, and agreements.
Seven therapists (three men and four women) administered each Interview
treatment. All 14 of the therapists had at least a master's degree in
counseling psychology, clinical psychology, or social work. Therapists A posttreatment interview was conducted by independent inter-
in both groups possessed an average of 5 years of clinical experience viewers blind to the hypothesis of the study and unaware of which
 964 A. GOLDMAN AND L. GREENBERG

treatment group members they were interviewing. The purpose of the Implementation Checks
interview was to gather descriptive data on how couples experienced
the process of therapy. Two trained observers, unaware of which treatment conditions they
were observing, evaluated segments of audiotapes on a 25-item imple-
mentation checklist composed of interventions typical of each ther-
Procedure apy. A total of 93 sessions, approximately three of each couple's ses-
sions, were chosen at random and observed by the raters. Two seg-
Following screening, participating couples were informed of partici- ments of 10 min each were taken from the middle and final thirds of
pation procedures and given pretests. Couples in the treatment group these 93 sessions. Of the 2,268 interventions coded, only 64 (2.8%) were
were randomly assigned to treatments and therapists and were seen coded as inappropriate to the treatment condition being observed. A
weekly for 10 1-hr sessions. All sessions were videotaped and audio- Cohen's (1960) Kappa interrater reliability of .95 was obtained on rat-
taped. Couples completed the CTAS after the third session. At the end ings of 624 interventions. These results suggest that the treatments
of the 10 weeks, couples were given posttests; treatment couples were were implemented according to the manuals and that both the treat-
advised of follow-up procedures, and treatment was made available for ments and the interventions were reliably differentiated.
wait-list couples. Treatment couples were called by phone 16 weeks
after posttest, and follow-up questionnaires were mailed to them. Two
couples in each group did not complete follow-up information because Results
they had separated, and one couple in each group had moved and did
The three main questions guiding the analyses were as fol-
not return the questionnaires; follow-up results are therefore based on
11 couples in each treatment group. lows: How did the groups differ from one another? How did the
groups change over time? How did the group differences
change over time, that is, what were the patterns of interactions?
Experimental Conditions A repeated measures design was used with three treatment
groups (control, 1ST, and EFT) and three occasions (pretest,
The emotionally focused treatment used in this study is an inte- posttest, and follow-up). The unit of measurement was the cou-
grated affective-systemic approach. Because it is assumed that change ple's average score on the four dependent measures (DAS, CRS,
occurs within the individual as well as within the context of the rela- TC, and GAS).
tionship, the therapy is directed toward interpersonal relationship
changes as well as change within each partner. Psychological symp-
toms are seen as emanating from the deprivation of unmet adult needs. Differences Between the Groups
This view holds that it is not individuals' feelings and wants that cause
problems but rather the not allowing or disowning of these feelings Equivalence measures. The three groups were compared for
that leads to ineffective communication and escalating interactional, equivalence on demographic data and on therapeutic alliance.
cycles (Greenberg & Johnson, 1984,1988). A multivariate analysis of variance (MANOVA) on 11 demo-
The following nine steps from the manual (Greenberg & Johnson, graphic variables suggested no significant differences at the .05
1988) describe the EFT process: (1) Define issue as presented. (2) Iden- level (corrected by the Bonferroni procedure), nor were there
tify negative interactional cycle. (3) Facilitate clients in accessing and
any treatment group differences in the strength of alliances
accepting previously unacknowledged emotions underlying the cycle.
(4) Redefine the problem cycle in terms of these new emotions and the between therapists and their couples on the CTAS or any of its
clients' interacting sensitivities. (5) Encourage identification with previ- subscales. Results also showed that there were no differential
ously unacknowledged aspects of experience by enactment of rede- therapist effects in performance on the four outcome measures
fined cycle. (6) Facilitate acceptance of partners' positions. (7) Encour- at pretest, posttest, and follow-up using a repeated measures
age clients to state needs and wants arising from their new emotional analysis of variance (ANOVA) with couples nested in therapists,
synthesis. (8) Facilitate new solutions. (9) Help clients to solidify new nested in treatments. A preliminary MANOVA, Wilks's
relationship positions and ways of achieving intimacy. lambda procedure, approximate F(4, 76) = 0.160, p = .183,
Integrated systemic therapy is a team approach integrating a number followed by univariate ANOVAs corrected by the Bonferroni
of procedures from family and marital systems therapy (Sluzki, 1983). procedure, did not yield statistically significant differences be-
Therapy is directed primarily toward change at the interactional level.
tween the three groups on pretest DAS and CRS scores. See
Changing repetitive, self-perpetuating negative interactional cycles
(the "fight cycle") is believed to lead to second-order change in the Table 1 for means and standard deviations.
system: change in behaviors as well as in rules governing interactions. Posttest. Following a MANOVA, Wilks's lambda procedure,
Partners' awareness of internal processes is not seen as essential to the approximate F(8, 72) = 5.80, p = .000, that indicated statisti-
therapeutic process. cally significant differences, univariate ANOVAs suggested
In 1ST the primary aim is to initiate a reversal in the fight cycle by that there were significant differences between the 1ST, EFT,
changing the meaning attributed to the situation. The mechanisms of and control groups on all four outcome variables (DAS, CRS,
change stem from both structural (Minuchin & Fishman, 1981) and TC, and GAS). To avoid an escalating Type I error rate, the
strategic and interactional views (Fisch et al., 1982). As noted, a team of Bonferroni procedure was used to calculate the critical signifi-
therapists behind a mirror is used to consult with the therapist in the cance level for each univariate test (a = .05/4 = .0125). Tukey's
room, enhancing the persuasive impact of these interventions. procedure, post hoc t tests, yielded the following results: At
The treatment manual (Greenberg & Goldman, 1985) specifies the
posttest, both treatment groups were significantly higher than
following seven steps: (1) Define the issue presented. (2) Identify the
negative interactional cycle. (3) Attempt restructuring. (4) Reframe the the controls on the four outcome variables, but there were no
problem using positive connotation followed by prescribing of the significantly different treatment effects noted between the 1ST
symptom. (5) Restrain, using "go slow" and dangers of improvement and EFT groups. This was confirmed by / tests, corrected by
(Fisch et al., 1982). (6) Consolidate the frame. (7) Prescribe a relapse. the Bonferroni procedure, of all possible comparisons between
 COMPARISONS IN COUPLES THERAPY 965

Table 1
Means and Standard Deviations on Dependent Measures
Pretest Posttest Follow-up

Group M SD M SD M SD

Dyadic Adjustment Scale

1ST 83.86 8.81 96.75 13.12 101.00 8.64
EFT 86.32 8.25 100.14 14.24 92.05 14.7
Control 82.50 7.11 80.86 9.93
Conflict Resolution Scale
1ST 28.89 2.93 33.29 4.35 33.23 4.69
EFT 26.07 2.54 32.14 28.03 31.68 5.18
Control 27.54 4.13 28.03 3.31
Target Complaints
1ST 3.72 0.684 3.69 0.73
EFT — 4.02 0.700 3.46 1.05
Control 2.46 0.709
Goal Attainment Scale
1ST 5.25 0.893 5.50 0.921
EFT — — 5.71 0.699 5.00 1.02
Control 4.00 0.554
Note. Dashes indicate data are not available. 1ST = integrated systemic therapy; EFT = emotionally
focused therapy.

pairs, which showed that both treatments were more effective els of the treatment groups were maintained at follow-up, a
than the wait-list control. MANOVA was conducted. Statistically significant changes
Follow-up. Sixteen weeks after posttest, 11 of the 14 couples were found between the two groups between posttest and fol-
in each treatment group completed follow-up data on the four low-up, Wilks's lambda procedure, approximate F(4, 17) =
outcome measures. A MANOVA conducted between the two 7.10, p= .001. Repeated measures two-way ANOVAs showed
groups over the four outcome measures, Wilks's lambda proce- significant differences within the groups from posttest to fol-
dure, approximate F(4,17) = 0.79, p = .547, suggested no statis- low-up on the CRS. Dependent t tests corrected by the Bon-
tically significant differences. Univariate tests corrected by the ferroni procedure (a = .05/4) indicated that although there were
Bonferroni procedure (a = .05/4) and / tests suggested no signifi- no significant differences within the 1ST group between post-
cant differences on any of the four measures. test and follow-up on the four outcome measures, there were
significant differences in the EFT group between posttest and
Differences Between Times follow-up on the DAS, TC, and GAS dependent measures but
not on the CRS. Inspection of the means (see Table 1) indicated
Pretest to posttest. The second, more powerful statistical a reduction in scores in the EFT group. These comparisons
test that addressed how the groups changed respectively over suggest the following: (a) Both treatment groups made signifi-
time was a 3 (group: 1ST, EFT, control) X 2 (times: pretest, cant gains from pretest to posttest on the DAS and the CRS.
posttest) repeated measures analysis on the DAS and the CRS Posttest levels on both measures were maintained by the 1ST
scores, using repeated measures MANOVAs, then univariate group at follow-up; however, although the EFT group did
tests, followed by dependent t tests (corrected by the Bonferroni maintain posttest levels on the CRS at follow-up, they did not
procedure), which compared times for individual groups. Be- maintain these levels on the DAS. (b) Whereas the 1ST group
cause the TC and the GAS do not yield pre-post change scores, maintained levels on the TC and the GAS from posttest to
the pre-post analysis was conducted only on the DAS and the follow-up, the EFT group did not maintain these levels on ei-
CRS. To analyze posttest changes on the TC and the GAS, ther the TC or the GAS from posttest to follow-up.
posttest and follow-up scores were used.
The pre-post MANOVA, Wilks's lambda procedure, approx-
imate F(2, 38) = 16.74, p = .000 (see Table 2), and the one-way
Group X Time Interactions
ANOVA, followed by dependent t tests corrected by the Bon- The last part of the analysis of treatment effects for the main
ferroni procedure (a = .05/2), confirmed results from the first experiment addressed the question of how the group differ-
between-groups test that both treatment groups made signifi- ences changed over time: What were the patterns of interac-
cant gains from pretest to posttest on the DAS and the CRS. tions? It was analyzed as a 2 (group: 1ST, EFT) X 2 (occasion:
Posttest to follow-up. To determine whether the posttest lev- posttest, follow-up) repeated measures experiment, using re-
966 A. GOLDMAN AND L. GREENBERG

Table 2
Summary Multivariate Analyses of Variance for Dependent Measures
Wilks's
Source lambda Approximate F Approximate df
a
Pretest, posttest, follow-up for DAS, CRS
Group .82944 1.95 2, 19 .169
Time .19642 17.39 4, 17 .000
Time X Group .53436 3.70 4, 17 .024
Pretest, posttest for DAS, CRS"
Group .69136 4.91 4,76 .001
Time .53159 16.74 2,38 .000
Time X Group .63858 4.78 4,76 .002

Posttest, follow-up for DAS, CRS, TC, and GAS3

Group .83738 .83 4, 17 .527
Time .37459 7.10 4, 17 .001
Time X Group .49230 4.38 4, 17 .013
Note. DAS = Dyadic Adjustment Scale; CRS = Conflict Resolution Scale; TC = Target Complaints;
GAS = Goal Attainment Scale.
°N=22. *N=42.

peated measures MANOVAs, then univariate tests followed by results suggested that there were no differential effects of 1ST
dependent t tests, to test the significance of individual interac- and EFT for men and women over time.
tions.
Briefly, the pattern on pretest versus posttest, posttest versus Clinically Significant Change
follow-up, and pretest versus posttest versus follow-up gener-
ally confirms the patterns from the previous section and sup- According to Jacobson, Folette, and Elwood (1984), whose
ports the conclusion that both treatment groups made signifi- criterion of a nondistressed relationship is that spouses score
cant gains from pretest to posttest on the DAS and the CRS within the nondistressed range on one or more preselected
compared to the controls, who did not. These findings also measures of marital functioning, 67% of all treatment couples
confirm the previously mentioned comparisons of posttest in this study who finished therapy with relationships rated in
with follow-up and pinpoint the discrepancy between the two the nondistressed range (above 100 on the DAS) could be con-
treatment groups in the maintenance of gains made on the DAS sidered successful. To consider those 33% in the moderate to
and the GAS. severe range as treatment failures, however, overlooks the fact
Although the results pertaining to the DAS and the GAS on that they made substantial gains, showing either significant or
this interactions test are similar to those of the previous test of marked improvement during the course of therapy.
differences over time, this is not true for results on the TC. For With regard to deterioration, we used a criterion of 1.96 SE
the TC, the p value for 1ST versus EFT for the posttest versus (19.99 score points) on the DAS, and none of the treatment
follow-up period is .088 (see Table 3), which fails to reach signifi- couples' scores decreased significantly either at posttest or at
cance at the .05 level. It seems that on this more powerful test follow-up. Deterioration was less than in other research, where
the levels attained by the EFT group on the TC were main- approximately 5% of treated couples deteriorated (Jacobson et
tained. Although the previous test suggested that the EFT al, 1984).
group had slipped back on the TC and the GAS at follow-up,
results on the TC are not confirmed by the interactions and are Descriptive Data
therefore not conclusive.
The resolution of such an apparent contradiction in findings In the structured posttest interviews, couples in both groups
lies in the nature of statistical testing. Findings from the first shared common perceptions about the dynamics of their
test that suggested no significant differences between treat- change process. An equal number in each group (10 couples)
ment groups at follow-up were based on a test that compares indicated that the treatments had made an impact on their
groups with one another; this is not as powerful a test as the test awareness of communication patterns. Eleven couples in each
of differences over time or the even more powerful test of inter- group also reported an increase in a sense of trust and safety,
actions. mutual support, and an ability to talk calmly about issues and
In considering whether there were significant interaction ef- negotiate differently with each other. Surprising]^ even though
fects between gender and treatment over time, a repeated mea- the treatments differentially emphasize thoughts and feelings,
sures MANOVft. was conducted that included the two treat- couples in both groups (9 in 1ST and 8 in EFT) emphasized
ment groups and the four outcome measures over time. The changes in their ability to talk about both thoughts and feelings
 COMPARISONS IN COUPLES THERAPY 967

Table 3
Dependent F Tests: Significance (p) of Individual Interactions
Pretest Pretest Posttest Pretest vs.
vs. VS. vs. posttest vs.
Scale and interaction posttest follow-up follow-up follow-up
Dyadic Adjustment Scale
Control vs. 1ST .001
Control vs. EFT .001
1ST vs. EFT .823 .034 .015 .022
Conflict Resolution Scale
Control vs. 1ST .044
Control vs. EFT .002
1ST vs. EFT .401 .615 .616 .816
Target Complaints
1ST vs. EFT .088
Goal Attainment Scale
1ST vs. EFT .040
Note. 1ST = integrated systemic therapy; EFT = emotionally focused therapy.

about their relationships and greater awareness of their own and focuses almost exclusively on changing current interac-
and their partner's thoughts and feelings. It seems that changes tions, refraining patterns of behavior, and prescribing symp-
in emotional experiencing in EFT may lead to clients' changing toms.
their thoughts and beliefs and that changes in the meaning In interpreting these results it is important to address the
attached to situations (refraining) in 1ST may lead to clients' impact of the team in the 1ST treatment. It is likely that couples
changing their feelings. exposed to the 1ST therapy derived the benefit of the collective
Although couples shared many common perceptions, they efforts of trained professionals, a type of "think tank" that
also reported being affected by the treatments in different devoted considerable time and effort toward discussing the cou-
ways. In an open-ended question about the effects of therapy, 10 ple's relationship and interactional patterns. As noted, 1ST cou-
EFT couples referred positively either to their own emotional ples indicated at posttest that they were influenced by the
responses, such as feeling an "emotional release," or to becom- team's provocative messages, expertise, and neutrality. The fact
ing more aware of their partner's sensitivities and vulnerabili- that the message represents the consensus of the team, a re-
ties. In contrast, 8 1ST couples cited the team's expertise and spected and credible group, probably lent credence and legiti-
neutrality and the provocative messages of "go slow" and "don't macy to the message. The presence of the team is an integral
change" as leading to changes in their interaction patterns. part of the treatment and suggests that a host of social psycho-
Seven 1ST couples stated that their awareness of the team's pres- logical influencing variables might be an important part of
ence made them more confident of the therapeutic suggestions. how 1ST therapy comes to have an impact. It could be that the
1ST couples generally saw their therapists as neutral, whereas team was a major factor in making the 1ST treatment more
EFT couples generally saw their therapists as empathic and lasting. The use of the team also raises questions about the
caring. cost-effectiveness of this treatment. 1ST marital therapy, as de-
With regard to their own attributions of change, 6 couples in livered in this study, was less cost-effective. Research into more
the 1ST group cited the team's positive connotation of the func- efficient methods of delivery of this treatment, such as reducing
tion of the cycle and of their reasons for engaging in the cycle. the number of therapist-session hours or testing the 1ST ther-
Eleven EFT couples, compared with only 6 1ST couples, cred- apy without a team, would be useful.
ited changes in feelings and emotions. These differences in With regard to maintenance of gains, the statistical tests did
posttest responses suggested that the groups did experience show that at follow-up, posttest levels were maintained by the
qualitatively different responses to the two treatments. 1ST group on all four outcome measures. Although the EFT
group did maintain their posttest changes on the CRS, they
slipped back on the DAS, the GAS, and, less conclusively, on
the TC. However, in reviewing the data in terms of the clinical
Discussion significance of these findings, it seems important to note that at
follow-up, EFT couples continued to show target complaint
The results of this study at posttest suggest that two different improvement and reach pretest goals. Examination of the
treatments, stemming from two very different theoretical mean scores of EFT couples on the TC at follow-up shows that
frameworks, are both effective in helping couples alleviate mar- they ranged from "slightly improved" to "somewhat improved"
ital distress and resolve conflict but that 1ST may be somewhat on target complaints elicited before therapy. Examination of
more self-sustaining at follow-up. One treatment, EFT, focuses EFT couples' mean scores on the GAS at follow-up shows that
on encouraging the accessing, expression, and acceptance of they achieved "expected results" on the goals set out before
affective experiences in partners. The other, 1ST, uses a team therapy. At follow-up, 1ST couples' scores on the TC were
968 A. GOLDMAN AND L. GREENBERG

slightly higher than those of EFT couples but also ranged be- mentarity. A combination of experiential and systemic
tween "slightly improved" and "somewhat improved," and on approaches is an alternative rich with possibilities. Couples of-
the GAS, 1ST couples' scores were higher, with the overall aver- ten have complex problems that are not fully addressed by one
age falling between "expected results" and "better than ex- therapy to the exclusion of the other. Using 1ST and EFT inter-
pected results." ventions in some sequential or integrated fashion (Greenberg &
This study used Caucasian, low-to-middle-income, middle- Johnson, 1988) seems promising for dealing with the complex-
class, help-seeking couples drawn from a population with a ity of marital problems. Repetitive fight cycles that have be-
mean on the DAS of 84.2. The relationship scores of the couples come so entrenched that couples are resistant to an affective
treated ranged from moderately to severely distressed (scores experiential approach are often a major problem. Integrated
between 96 and 71.5 on the DAS). Fifty-seven percent of all systemic interventions would be useful for reframing those cy-
treated couples had relationships classified as severely dis- cles in order to interrupt and alleviate them. Similarly, an ac-
tressed at pretest. The results of this study can therefore be companying problem for couples is their inability to share feel-
generalized to a population of such severely distressed couples ings and develop intimacy. The emotionally focused approach
who seek help. The fact that the couples were in the severely would provide therapeutic opportunities for partners to experi-
distressed range seems relevant when considering the differ- ence and express such emotions and increase intimacy.
ence between the findings at follow-up of this study and those
of Johnson and Greenberg (1985a), whose EFT group did not
show signs of relapse at follow-up. The couples in Johnson and References
Greenberg's study were less distressed at the outset than those
in the present study and were in the nondistressed range at Battle, J., Imber, S., Hoehn-Saric, R., Stone, A. R., Nash, E. R., & Frank,
termination, and this could have contributed substantially to J. D. (1966). Target complaints as criteria of improvement. American
the maintenance of gains at follow-up. Journal of Psychotherapy, 20, 184-192.
Follow-up results of this study suggest that exposure of more Bordin, E. (1979). The generalizability of the psychoanalytic concept
severely distressed couples to a brief, time-limited course of of the working alliance. Psychotherapy: Theory, Research and Prac-
EFT did not lead to maintenance of posttreatment gains on the tice, 16(3), 222-229.
DAS and on goal attainment (the GAS). One interpretation is Cohen, J. (1960). A coefficient of agreement for nominal scales. Educa-
that a brief, emotionally focused therapy administered to se- tional and Psychological Measurement, 20, 37-46.
verely distressed couples, although leading to conflict resolu- Coyne, J. (1984). Strategic therapy with depressed married persons:
tion and target complaint reduction, does not create sufficient Initial agenda, themes and interventions. Journal of Marital and
Family Therapy, 10(1), 53-62.
intimacy or strong enough relationship satisfaction at termina- Fisch, R., Weakland, L., & Segal, L. (1982). Tactics of change: Doing
tion to engender maintenance of gains in overall marital adjust- therapy briefly. San Francisco: Jossey-Bass.
ment. Fournier, D. G, Olson, D. H., & Druckman, J. M. (1983). Assessing
Because of the intense negative feelings and negative views marital and premarital relationships. In E. E. Filsinger (Ed.),
prevalent among distressed couples, an approach such as 1ST, Marriage and family assessment (pp. 67-75). Beverly Hills, CA:
which offers them an alternative, positive way of thinking about Sage.
their interaction when they are still in a marginally distressed Greenberg, L. S., & Goldman, A. (1985). Integrated systemic couples'
range, as these couples were, may be necessary for more endur- therapy: A treatment manual. Unpublished manuscript, University
ing change. With 1ST therapy, partners seem to change their of British Columbia, Vancouver, Canada.
perceptions about their relationship. When couples see the posi- Greenberg, L., & Johnson, S. M. (1984). Emotionally focused couples
tive function of their negative patterns they are able to be more therapy: An integrated affective systemic approach. In N. S. Jacob-
son & A. S. Gurman (Eds.), The clinical handbook of marital therapy
self-accepting as well as accepting of their partners. This view (pp. 253-276). New York: Guilford Press.
likely provides couples with an ongoing positive framework for Greenberg, L., & Johnson, S. (1988). Emotionally focused therapy for
viewing their relationship. Systemic and interactional thera- couples. New \fork: Guilford Press.
pists (Fisch et al., 1982) have postulated that systemic therapy Guerin, P. J. (1982). The stages of marital conflict. The Family, 10(1),
works by initiating a reversal in negative interactional patterns 15-26.
—by changing the interaction governing rules—and that this Jacobson, N. S., Folette, W C, & Elwood, R. C. (1984). Outcome re-
reversal is enough to lead to enduring change even with dis- search on behavioral marital therapy: A methodological and con-
tressed clients. The results of the present study produce some ceptual reappraisal. In K. Hahlweg & N. S. Jacobson (Eds.), Marital
support for this notion. However, this study was limited to a interaction: Analysis and modification (pp. 113-129). New \brk:
4-month follow-up period, and further research and replication Guilford Press.
James, P. (1991). Effects of a communication training component
are necessary to determine if the difference between treat- added to an emotionally focused couples therapy. Journal of Marital
ments remains over a longer period of time. and Family Therapy, 3, 263-275.
Questions raised by this study relate to which specific aspects Johnson, S. M., & Greenberg, L. S. (1985a). Differential effects of expe-
of each treatment led to change and the processes involved in riential and problem-solving interventions in resolving marital con-
such change. The question of how emotional or systemic pro- flict. Journal of Consulting and Clinical Psychology, 53,175-184.
cesses lead to change should be addressed by research that fo- Johnson, S. M., & Greenberg, L. S. (1985b). Emotionally focused cou-
cuses specifically on the process of change. In practical applica- ples therapy: An outcome study. Journal of Marital and Family Ther-
tion the EFT and 1ST therapies seem to have strong comple- apy, 11, 313-317.
 COMPARISONS IN COUPLES THERAPY 969

Kiresuk, T. J, & Sherman, R. E. (1968). Goal attainment scaling: Gen- Snyder, D., & Wills, R. (1989). Behavioral versus insight-oriented mari-
eral method for evaluating comprehensive community mental tal therapy: Effects on individual and interspousal functioning.
health programs. Community Mental Health Journal, 38, 443-453. Journal of Consulting and Clinical Psychology, 57, 39-46.
Minuchin, S., & Fishman, H. (1981). Family therapy techniques. Cam- Spanier, G. (1976). Measuring dyadic adjustment. Journal of Marriage
bridge, MA: Harvard University Press. and the Family, 38,15-28.
Pinsof, W, & Catherall, D. (1983). The Couples Therapy Alliance Scale Steinglass, P. (1978). The conceptualization of marriage from a systems
manual. Chicago: Chicago Center for Family Studies, Northwestern theory perspective. In T. J. Paolino & B. S. McCrody (Eds.), Marriage
University. and marital therapy (pp. 298-367). New 'York: Brunner/Mazel.
Sluzki, C. E. (1978). Marital therapy from a systems theory perspec- Wile, D. E. (1981). Couples therapy: A non-traditional approach. New
tive. In T. Paolino & B. McCrady (Eds.), Marriage and marital ther- York: Wiley.
apy. New York: Brunner/Mazel.
Sluzki, C. E. (1983). Process, structure and world views: Toward an Received November 6,1990
integrated view of systemic models in family therapy. Family Pro- Revision received February 7,1992
cess, 22, 469-476. Accepted February 17,1992 •