Psychotherapy
Volume 38/Summer 2001/Number 2
CHANGES IN CLIENTS' ATTACHMENT STYLES OVER THE
COURSE OF TIME-LIMITED DYNAMIC PSYCHOTHERAPY
LINDA A. TRAVIS
Program in Geriatrics and Neuropsychiatry
Department of Psychiatry at University of
Rochester Medical Center
Rochester, New York
JEFFREY L. BINDER
Georgia School of Professional Psychology
Atlanta, Georgia
NANCY G. BLIWISE
Department of Psychology at Emory
University, Atlanta, Georgia
H. LYNN HORNE-MOYER
Georgia School of Professional Psychology
Atlanta, Georgia
This study utilized a theory-specific
measure to examine client relational
change over the course of time-limited
dynamic psychotherapy. Specifically,
this is the first empirical investigation
with a clinical sample to measure change
and stability in clients' attachment styles.
Categorical and dimensional ratings of
attachment styles were obtained.
Pretreatment and posttreatment
measures of attachment styles were then
examined in association with Global
Assessment Scale (GAS) scores and
symptoms. At posttreatment, a significant
number of clients were evaluated as
having changed from an insecure to a
secure attachment style. Additionally, the
sample as a whole demonstrated
significant changes toward increased
secure attachment. Significant
relationships were also found among
changes in attachment style, GAS scores,
and symptom levels. Implications for
psychodynamid'interpersonal
psychotherapy research and practice
are discussed.
This research was completed as part of Dr. Travis' Clinical
Research Project in the doctoral program in clinical psychology at the Georgia School of Professional Psychology.
Correspondence regarding this article should be addressed to
Linda A. Travis, Psy.D., Program in Geriatrics and Neuropsychiatry, Department of Psychiatry, University of Rochester Medical Center, 300 Crittenden Boulevard, Rochester, New York
14642. E-mail: [email protected]
Contemporary psychodynamic/interpersonal researchers have called for theoretically relevant
outcome measures to evaluate intrapersonal and
interpersonal changes in psychotherapy (Barber,
1994; Crits-Christoph, 1992). Theory-specific
psychotherapy investigations are characterized by
congruence among problem formulation, conception of change, outcome goals, and assessment
(Schacht, Strupp, & Henry, 1988). Psychotherapy research conducted from a theory-specific
framework provides unique opportunities to examine change in theoretically relevant psychological variables over the course of therapy and to
better understand the links among psychotherapy
process and outcome treatment components.
Psychodynamic/interpersonal therapists highlight the central role of interpersonal interactions
in the onset, maintenance, and treatment of symptoms and psychopathology (Anchin & Kiesler,
1982; Horowitz & Vitkus, 1986; Sullivan, 1953,
1954). Therefore, standard measures of symptoms and global functioning are unlikely to capture the areas most targeted for client change.
Furthermore, changes in relational patterns and
styles are theoretically very relevant to psychodynamic/interpersonal models of treatment. Objectrelations psychotherapy measures (Azim, Piper,
149
L. A. Travis et al.
Segal, Nixon, & Duncan, 1991; Hoglend, 1988,
1993; Hoglend, Sorlie, Heyerdahl, Sorbye, &
Amlo, 1993; Piper et al., 1991) and interpersonal
psychotherapy measures (Anchin & Kiesler, 1982;
Horowitz, Rosenberg, Baer, Ureno, & Villasenor, 1988; Horowitz, Rosenberg, & Kalehzan,
1992) have been developed to assess variations in
the qualities of clients' pretreatment intrapersonal
and interpersonal patterns of relating.
Two relational measures merit further discussion. The Quality of Object Relations (QOR)
scale developed by Azim et al. (1991) offers an
important contribution by employing a dimensional rather than a categorical approach to assessments of clients' intrapersonal relational structures. Nonetheless, as Piper et al. (1991)
acknowledged, the QOR scale requires raters to
have a high level of clinical sophistication. The
Structural Analysis of Social Behavior (SASB;
Benjamin, 1982, 1993) is an interpersonal measure yielding ratings for the relational domains of
self, other, and introject. SASB has enhanced
understanding of the therapy interpersonal process. However, Strupp (1996) stated that SASB
may miss nuances of the interaction, particularly
pejorative communication as experienced by the
client. Safran (1992) commented that the exclusive use of interpersonal measures misses key
mediating intrapsychic mechanisms that are crucial to understanding clients' perceptions of transactions. In order to build on the strengths of previous relational measures and to address their
limitations, a theory-based measure is needed that
is dimensional and adept in linking intrapersonal
and interpersonal functioning.
Attachment theory (Bowlby, 1969, 1973,
1980) provides a useful framework for understanding individual differences in relationship
structures. Attachment styles are relatively enduring characteristics that continue to pattern the
quality of adult interpersonal interactions throughout the life span. Bowlby (1988) proposed that
an internal working model is the mechanism by
which one's view of the self and others remains
steady over time. The concept of an internal
working model is also the cornerstone for adult
attachment theory and for various relational structures in adult attachment styles (Pietromonaco &
Barrett, 2000). Attachment theory posits three
attachment styles; secure, insecure-avoidant, and
insecure-ambivalent (Hazan & Shaver, 1987).
Particular patterns of functioning are empirically
linked with each style. For example, clinical in-
150
vestigations assessing clients' pretreatment attachment styles in relation to posttreatment
Global Assessment of Functioning (GAP; American Psychiatric Association, 1987) scores have
revealed distinct psychotherapy outcome patterns
of functioning for individuals with different attachment styles. Across a range of settings, posttreatment GAP scores are consistently higher for
secure clients (Dolan, 1992;Fonagy et al., 1996),
consistently lower for ambivalent clients (Dolan,
1992; Fonagy et al., 1996; Pilkonis, Heape, &
Proietti, 1993), and consistently midway between
these two groups for avoidant clients (Dolan,
1992; Fonagy etal., 1996). Regarding symptoms,
clients with ambivalent attachment entered and
terminated treatment with the highest number of
symptoms, clients with avoidant attachment entered and terminated treatment with the least
amount of symptoms, and clients with secure attachment entered and terminated treatment with
symptom levels between the two other groups
(Dolan, 1992). Attachment styles are clearly associated with therapy outcome and warrant additional investigation.
Recent developments in attachment theory
have further characterized differences among internal working models for each attachment style.
The Bartholomew and Horowitz (1991) model of
attachment distinguishes between differences in
how individuals see themselves (positive or negative) and others (positive or negative). In the fourcategory model of attachment, Bartholomew
(1990) proposed that the avoidant category is
more accurately divided into two separate styles,
fearful and dismissing. Fearful attachment is
characterized by negative views of self and other
and is conceptually most similar to avoidant attachment in Hazan and Shaver's model (1987).
Dismissing attachment is typified by a positive
view of self and negative view of other, and it is
seen as a variant of avoidant attachment. Preoccupied attachment is typified by a negative view
of self and positive view of others, and is conceptually most similar to ambivalent attachment in
Hazan and Shaver's model. Preoccupied, fearful,
and dismissive styles are insecure attachments.
Secure attachment is characterized by positive
views of self and other.
Bartholomew and Horowitz (1991) developed
an attachment rating system in order to empirically validate the four-category model of adult
attachment and the corresponding internal working models of self and other for each attachment
Attachment Styles
style. The Bartholomew and Horowitz rating system yields prototypical and dimensional measures
of attachment styles. The prototype method rates
an individual's attachment interview and narrative on its degree of likeness to each attachment
style, thereby yielding four ratings that reflect the
level of each attachment theme for each individual. The highest of these prototype ratings is
considered to be the individual's dominant categorical attachment style. The inclusion of dimensional and categorical ratings of attachment styles
provides intricate analysis of relational functioning. Griffin and Bartholomew (1994) argued that
attachment measures must capture the interpersonal and intrapersonal nuances that exist for persons of the same dominant category. For example, someone with a fearful-dominant style yet
high-secure dimension is different from someone
with a fearful-dominant style yet high-dismissing
dimension. Given the dimensional and categorical
ratings of the Bartholomew and Horowitz system,
it is possible that this attachment rating system
may offer a more precise measure of relational
change than has been achieved through object
relations or interpersonal measures.
There is considerable evidence for the fourcategory model of attachment (Diehl, Elnick,
Bourbeau, & Labouvie-Vief, 1998). The Bartholomew and Horowitz (1991) attachment rating
system has been used in nonclinical investigations
(Baldwin & Meunier, 1999; Bliwise, 1992) as
well as in clinical investigations (Horowitz, Rosenberg, & Bartholomew, 1993; Kalehzan, 1993).
Horowitz and colleagues illustrated that the four
attachment styles were associated with four different areas of interpersonal distress as measured
by the Inventory of Interpersonal Problems (IIP:
Horowiz et al., 1988). Nonetheless, Horowitz et
al. (1993) did not obtain standard outcome data
regarding symptoms and global functioning
scores. Furthermore, neither Horowitz nor the
above-mentioned attachment studies (Dolan,
1992; Fonagy et al., 1996; Pilkonis, Heape, &
Proietti, 1993) explored change in attachment
styles over the course of treatment. Given the
significant findings regarding clients' pretreatment attachment styles being associated with
areas of interpersonal distress and differential
patterns of posttreatment symptom and global
functioning scores, the next natural step in clinical attachment investigations is to measure
changes in attachment styles over the course
of therapy.
Adult attachment theory is increasingly being
referenced as a framework by which therapists
may conceptualize and facilitate change in clients' intrapersonal and interpersonal relationships
(Dolan, Arnkoff, & Glass, 1993; Horowitz et al.,
1993; Liotti, 1991; Sperling & Berman, 1994;
West & Sheldon-Keller, 1994). Although many
clinicians and researchers have discussed the viability of change in attachment styles, to date we
are not aware of any studies examining this
change. Changes in attachment style may be especially associated with significant love relationships, major life transitions such as parenting or
retirement, or involvement in psychotherapy
(Bowlby, 1988; Ricks, 1985). Bowlby proposed
a therapy outline to address change in attachment
styles. He noted that his outline had much in
common with other therapy models, including
time-limited dynamic psychotherapy (TLDP;
Strupp & Binder, 1984).
TLDP explicitly targets clients' maladaptive
interpersonal patterns as enacted in the clienttherapist relationship. TLDP also formulates the
possibility of client internalization of the clienttherapist interpersonal activity (Strupp & Binder,
1984). TLDP research has elucidated client internalization of the interpersonal process in therapy
(Harrist, Quintana, Strupp, & Henry, 1994), offered evidence of therapists' contributions to positive introject change for clients (Harrist et al.,
1994; Henry, Schacht, & Strupp, 1986), and offered evidence of therapists' contributions to negative introject change for clients (Henry et al.,
1986; Henry, Schacht, & Strupp, 1990). Thus,
it is possible that TLDP may facilitate dimensional and categorical changes in attachment
styles, a theoretically relevant definition of successful outcome in psychodynamic/interpersonal
therapies.
Two additional features of the present investigation are particularly noteworthy. First, TLDP
and attachment theory share similar assumptions
regarding treatment. These assumptions include
the notion that intrapersonal and interpersonal
change is possible, that change is an integration
of affective, interpersonal, and cognitive components, and that therapists have an active, not passive, stance in the interpersonal world of the client
(Bowlby, 1988; Strupp & Binder, 1984). In addition, both attachment theory and TLDP propose
that clients not only construct the therapy relationship from internal working models but are also
affected by the ways in which the therapist is
151
L. A. Travis et al.
perceived as further contributing to, or refuting,
these models of self and others. Second, the present client and therapist sample is drawn from the
Vanderbilt II Study (Strupp, 1993) designed to
examine the effects of a one-year manualized
training program in TLDP on psychotherapy process and outcome. The Vanderbilt II investigation
was made up of three separate cohorts of clients,
all of whom were selected on the basis of significant interpersonal problems. Cohort I consisted
of those clients who therapists treated before receiving TLDP training, Cohort II consisted of
those clients who therapists treated while receiving training, and Cohort III consisted of those
clients who therapists treated after receiving training. Cohort III was used for the present study and
provides a rich data base for exploring questions
about changes in attachment style, psychotherapy
process, and psychotherapy outcome.
The primary goal for the present theory-specific
investigation is to build upon previous studies
by examining whether systematic and significant
changes in clients' categorical and dimensional
attachment styles are possible in TLDP. A secondary goal is to better understand the associations between attachment styles and standard
change measures by examining pretreatment and
posttreatment measures of attachment styles,
symptoms, and global functioning scores.
Methods
Clients
The Vanderbilt II Study (Strupp, 1993) provided most of the data for this investigation. The
Vanderbilt II study was designed to examine the
effects of a 1-year therapist training program in
TLDP (Strupp & Binder, 1984) on psychotherapy
process and outcome. (See Henry, Strupp, Butler,
Schacht, & Binder [1993] for a complete description of the Vanderbilt II study, patients, selection
procedures, and training procedures for psychotherapists.)
The present study was comprised of the cohort
of clients (A' = 29) seen after therapists completed TLDP training. A total of 84 clients, across
three cohorts, were accepted into the study (19%
of the potential client pool). Of these initial 84
clients, 77% were female and 96% were White.
The sample ranged from high school to graduate
levels of education. Subjects were 24-64 years
old (A/ = 41.0 years, SD = 10.4 years). Seventyone percent participated in one or two previous
psychotherapy sessions over 10 years.
152
All accepted clients had genuine and significant
interpersonal problems and demonstrated no initial indications for alternative treatments, such
as medication or hospitalization. In addition, all
clients satisfied minimal requirements for outpatient psychotherapy. These requirements included
receiving at least one Axis I or Axis II diagnosis
from the third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III,
American Psychiatric Association, 1980). From
the sample, 87% of clients received an Axis I
diagnosis and 67% received an Axis II diagnosis.
The mean Global Severity Index (GSI) T-score
from the Symptom Check List 90-Revised (SCL90-R; Derogatis, 1983) was 48.1 (SD = 5.8),
with no client having a GSI score less than 1 SD
below the outpatient normative means. Clients
who left the project following four or fewer sessions were considered dropouts and were replaced. Clients who completed five or more sessions were retained for regular analyses (A' =
80). A total of 84 clients began the study, and
four dropped out (5%). Disregarding dropouts,
the mean number of sessions was 21.4 (range =
5.25, SD = 6.1). A total of 29 clients comprised
the Cohort III sample for the present study.
Therapists
Psychotherapists participating in the study
were licensed clinical psychologists and psychiatrists and had been recommended by senior colleagues who typically were their teachers. All
therapists had at least 2 years of full-time postinternship or postresidency clinical experience and
were in private practice (years of experience
ranged from 2 to 14 years; M = 5.3 years with
a SD = 3.7 for psychiatrists and SD = 2.3 for
psychologists). There were 10 men and 6 women
(6 male psychiatrists, 4 male psychologists, 2
female psychiatrists, and 4 female psychologists).
All therapists were White and the majority considered themselves to be psychodynamically
oriented.
Procedures
The principal procedure involved rating clients' pretherapy and posttherapy attachment
styles from intake and termination videotaped interviews conducted for the Vanderbilt II Study.
Attachment style ratings were applied to videotapes of semistructured interviews about interpersonal relations conducted at entry into the Vanderbilt II Study and at the end of treatment. These
Attachment Styles
interviews were conducted by independent clinicians, not by therapists treating the patients. As
the Bartholomew and Horowitz (1991) system
can be applied to any type of attachment narrative
as long as the narrative provides sufficient detail
about interpersonal relations, it was appropriate
and useful in this study. Also, the Bartholomew
and Horowitz rating system provides dimensional
and categorical ratings of attachment styles.
Raters were undergraduate psychology majors
recruited from a southeastern college and were
blind to the purpose of this study. The raters were
trained by one of the authors of this study (NGB)
who received training in the Bartholomew attachment rating system. During the 24 hours of training delivered over 3 separate days, raters received
direct instruction and a training manual, which
included a description of attachment theory and
specifically, the Bartholomew model for rating
attachment styles. The manual included specific
criteria for listening to client narratives and rating
each of the attachment styles. The 15 raters were
trained to a criterion level of reliability (0.85) on
standardized written protocols unrelated to the
present study. If the desired criterion level was
not achieved after 24 hours of structured training,
raters received additional training. Six raters received additional training, with five raters reaching the desired criterion level and one rater released from the project unable to reach a minimal
level of reliability. Another rater elected to leave
the project.
Two other pre-post measures were also utilized
and collected at the intake and termination interview sessions. First, self-reports of clients' symptoms were obtained from SCL-90-R scores. Second, Global Assessment Scale (GAS) scores were
obtained from independent clinicians at the intake
and termination interviews. The three pre and
post measures (attachment styles, SCL-90-R, and
GAS) were then analyzed for changes over the
course of TLDP.
Measures
Bartholomew Attachment Rating Scale. The
interview-based attachment-style rating system of
Bartholomew and Horowitz (1991) was used to
rate clients' attachment styles. From the videotaped intake and termination interviews, two independent raters assessed each client on four
9-point scales describing the client's degree of
correspondence with each of the four attachment prototypes.
. Two different procedures were used to obtain
clients' pretreatment and posttreatment attachment ratings. The first procedure examined the
extent to which a client's narrative, as evaluated
from videotaped intake and termination interviews, matched each of the four attachment style
prototypes as described by Bartholomew and Horowitz (1991). Each prototype received an ordinal
prototype rating from 1 (no correspondence) to 9
(complete correspondence) based on the narrative's degree of correspondence to each attachment prototype. This procedure yielded four ratings that reflected the level of each attachment
theme present in the narrative. Next, a dominant
attachment style was assigned to each narrative
based on the highest prototype rating across all
four prototypes.
Different ratings by two raters on a particular
client's narrative were resolved by a third independent rater. Final ratings were based on ratings
from all raters. The prototype ratings across all
four styles were examined for each subject. The
highest rating was used as an index of each client's "dominant" attachment style. Final reliability assessment of the raters revealed adequate reliability across styles (secure x = .76; preoccupied
x = .75; fearful x = .70; dismissing x = .74).
Although higher levels of reliability were preferred, the figures obtained may represent the
complexity of the attachment style-rating task.
Symptom Checklist-90-Revised. (SCL-90-R;
Derogatis, 1983). The SCL-90-R is a 90-item
self-report inventory with nine symptom dimensions and three global indices. Four symptom dimension scores were selected for the present
study. Anxiety and depression were chosen as
they were widely reported by the TLDP sample.
Hostility and interpersonal sensitivity were the
two other symptoms selected for the present investigation because they represent interpersonal
symptoms. Lambert and Hill (1994) discussed
that the SCL-90-R is particularly useful for detecting change in brief therapy and for a client
sample with heterogeneous diagnoses.
Global Assessment Scale. (GAS; Endicott,
Spitzer, Fleiss, & Cohen, 1976). The GAS is a
simplified version of the Health-Sickness Rating
Scale (HSRS; Luborsky, 1962) and was used in
the Diagnostic and Statistical Manual of Mental
Disorders-Ill (American Psychiatric Association,
1980) at the time when Vanderbilt II was conducted. The GAS is a scale ranging from 1-100
that is used to evaluate clients' overall level of
153
L. A. Travis et al.
functioning. The GAS is anchored at each 10point interval with a clinical description that includes level of occupational and social functioning, as well as subjective distress. GAS ratings
obtained from independent clinicians, not clients'
therapists, are utilized for analysis.
Results
Hypothesis 1
There will be a significant increase, from pretreatment to posttreatment, in the number of clients rated as having a dominant-secure attachment
style. In addition, there will be a lower rate of
stability of attachment styles over the course of
therapy that has been found in test-retest reliability studies.
Pretreatment categorical attachment ratings revealed there were 0 secure clients and 29 insecure
clients. Pretreatment insecure clients were categorized as follows: 11 preoccupied clients (38%),
16 fearful clients (55%), and 2 dismissive clients
(7%). Posttreatment analysis revealed 7 secure
clients (24%), 10 preoccupied clients (34%), 8
fearful clients (28%), and 4 dismissive clients
(14%). Ten clients (34%) retained the same attachment style over the course of this investigation. This 34% is much lower than the 75-80%
test-retest reliability reported by Scharfe and
Bartholomew (1994) in an 8-month study and the
78-95% test-retest reliability reported by
Bakersman-Kranenburg and van-Ijzendoorn
(1993) in a 2-month study. With 19 of the clients
(66%) changing attachment styles, it was important to inquire if the changes were systematic
or random. A McNemar test indicated that there
was a significant difference between the number
of pretreatment secure/insecure clients and the
number of posttreatment secure/insecure clients
(X- = .Q\,p< .05).
The 19 clients changing dominant attachment
styles were distributed as follows. For those 11
clients with pretreatment preoccupied attachment
styles, 3 (27%) changed to secure, 4 (40%) remained preoccupied, 3 (27%) changed to fearful,
and 1 (9%) changed to dismissive. For those 16
clients with pretreatment fearful attachment
styles, 3 (19%) changed to secure, 6 (60%)
changed to preoccupied, 5 (31 %) remained fearful, and 2 (12%) changed to dismissive. For those
2 clients with pretreatment dismissive attachment
styles, 1 (50%) became secure and 1 (50%) remained dismissive. Although it was hoped that
154
the majority of clients would change to a secure
style, a chi-square goodness-of-fit test indicated
that the pattern of change was consistent with
uniform distribution, \2 (3, N = 19) = 2.66,
p > .05. However, in the absence of base-rate
information, it is not known what rate and pattern
of change in attachment styles is reasonable to
expect for a 25-session format of therapy. Thus,
this study demonstrated that significant categorical changes from insecure to secure attachment
styles occurred over the course of treatment. It is
now important to examine if dimensional changes
in attachment styles also occurred over the course
of treatment.
Hypothesis 2
TLDP will significantly reduce levels of insecure attachment themes and significantly increase
levels of secure attachment themes.
A t test for paired samples revealed that there
was a significant increase in secure attachment
themes from pretherapy to posttherapy in clients'
relationship narratives, r(28) = -2.90, p< .05.
There was a significant decrease of fearful attachment themes from pretherapy to posttherapy
across clients' narratives, t(28) = 3.03, p < .05.
The observed decrease of preoccupied attachment
themes from pretherapy to posttherapy was not
statistically significant, r(28) = 1.30, p > .05.
Dismissive attachment themes remained the same
from pretherapy to posttherapy, t(2S) = —.38,
p > .05 (see Table 1).
Thus, the Bartholomew and Horowitz (1991)
attachment measure allowed the tracking of dimensional ratings of insecure and secure attachment themes over the course of therapy. In summary, analysis employing this dimensional measure
indicated that there was significant movement for
the group of clients toward secure attachment.
Hypothesis 3
Clients with posttreatment dominant-secure attachment styles will show more positive change
on SCL-90-R scales of depression, anxiety, hostility, and interpersonal sensitivity than those with
posttreatment dominant-insecure attachment styles.
Repeated measures ANOVA revealed no significant interaction between attachment security
and the route of change in symptoms of anxiety,
depression, interpersonal sensitivity, and hostility. However, main effects for time and attachment security were observed for most of these
symptom subscales. The main effects for time
Attachment Styles
TABLE 1. Change in Attachment Themes
Attachment Themes
Secure
Preoccupied
Fearful
Dismissive
Pretreatment
Posttreatment
M
SD
M
SD
t value
2.09
5.10
1.01
1.79
2.02
1.71
3.41
4.45
4.04
3.08
2.18
2.21
1.77
2.01
-2.90*
1.30
3.03*
-.38
5.43
2.95
Note. * p < .05.
reveal that symptoms declined consistently from
pretreatment to posttreatment for anxiety, F(l,
27) = 7.15, p < .01; for depression, F(l,
27) = 10.80,p< .01; for interpersonal sensitivity,F(l,27) = 10.38, p< .01; and for hostility,
F(l, 27) = 6.04, p < .025. Main effects for
attachment security were found for anxiety, F(l,
27) = 6.16, p < .025; for depression, F(l,
27) = 5.85,p< .025; and for interpersonal sensitivity, F(l, 27) = 6.73, p < .02. Therefore,
clients who changed to secure attachment had
consistently lower symptoms scores. Taken together, these main effects reveal that although
those who changed to secure status entered the
study with lower levels of symptoms, their symptoms reduced at the same rate as those who entered and terminated therapy with insecure attachment styles.
Hypothesis 4
At posttreatment, GAS scores will be significantly higher for clients with dominant-secure attachment styles as compared to clients with
dominant-insecure attachment styles.
ANOVA with planned comparisons revealed
that clients with secure attachment styles had significantly higher GAS scores than did the three
other groups of clients with insecure attachment
styles, f(28) = 3.6, p < .05 (see Table 2).
TABLE 2. Posttreatment Dominant Attachment Styles with
Posttreatment GAS Scores
Posttreatment Attachment Styles
Secure
Preoccupied
Fearful
Dismissive
Note. * p < .05.
Mean GAS Scores
79.1*
62.4*
66.3
67.0
Hypothesis 5
At posttreatment, GAS scores will be significantly lower for preoccupied clients as compared
to clients with other insecure or secure styles.
ANOVA with planned comparisons revealed
that there was a significant difference in the expected direction between those clients with dominant posttreatment preoccupied attachment styles
and all other client attachment groups, f(28) =
-2.48, p < .05 (see Table 2).
Discussion
This investigation demonstrated the importance
of using theory-specific measures to examine the
nature of change in psychodynamic/interpersonal
psychotherapy. Successful outcomes for psychodynamic/interpersonal therapies reflect changes
in clients' relational structures. In the present investigation, systematic and significant changes in
clients' attachment styles and themes were evidenced over the course of TLDP treatment. TLDP
is a treatment that formulates change in clients'
maladaptive relationship patterns as its principal
goal (Strupp & Binder, 1984) and is thereby wellsuited for an investigation examining change in
attachment styles with a client sample identified
as having significant interpersonal problems.
However, it is recognized that this investigation's
design does not offer confirmation that TLDP
treatment facilitated change in clients' attachment
styles and themes. Nonetheless, the present study
has provided a range of data consistent with the
theory and process of change in the attachment
literature and in TLDP, and it is plausible that
TLDP may have facilitated changes in clients'
attachment styles and themes. Furthermore, this
investigation provided meaningful information
regarding associations between relational and
standard measures of client change.
Changes in clients' dominant pretreatment and
posttreatment attachment styles were tracked and
155
L. A. Travis et al.
evaluated in this study. Only 34% of clients retained the same attachment classification, and this
percentage was much lower than rates of attachment stability found in other studies (BakersmanKranenburg & van Ijzendoorn, 1993; Scharfe &
Bartholomew, 1994). Moreover, the changes in
dominant attachment styles were not random, as
illustrated by the significant difference between
pretreatment categories (0 clients with secure attachment styles and 29 clients with insecure attachment styles) and posttreatment categories (7
clients with secure attachment styles and 22 clients with insecure attachment styles). To date,
there are no other clinical studies examining
changes in attachment styles. Thus, the rate and
pattern of change in attachment styles with clinical populations is simply not known.
Although it was important to examine if clients
changed from insecure to secure attachment
styles, it was also essential to understand if the
client group, as a whole, became more secure.
Using the Bartholomew and Horowitz (1991) dimensional measure of attachment, significant
changes were identified from pretreatment to
posttreatment levels of attachment themes across
the client sample. This finding suggests some
working through of relationship problems and indicated the client group as a whole was becoming
more secure and less insecure. Specifically, there
was a significant increase in secure attachment
themes from pretherapy to posttherapy across clients' relationship narratives and a significant decrease in fearful attachment themes across clients'
narratives from pretreatment to posttreatment.
There was a decrease, although not significant,
in preoccupied attachment themes and a small,
though not significant, increase in dismissive
themes. Other researchers have documented similar results regarding areas and rates change of
interpersonal change in brief dynamic/interpersonal therapy. Horowitz et al. (1993) found a
relatively rapid improvement in the interpersonal
problems associated with secure styles (e.g.,
overly nurturant) or fearful styles (e.g., exploitable or socially introverted) and a relatively
slower rate of change for interpersonal problems
associated with preoccupied styles (e.g., intrusion) or dismissive styles (e.g., hostile dominance). Alternatively, Bein et al. (2000) found
that many therapists in the Vanderbilt studies did
not achieve complete competence in providing
TLDP. It is possible that more skilled therapists
would be better able to modify difficult areas of
156
interpersonal interactions. In order to maximize
possibilities for client change, therapists need to
be cognizant that change for preoccupied themes
may occur at a slower rate. Also, change for
dismissive clients may rely on therapists' skills
in learning how to effectively manage hostility or
negative therapy process (Binder & Strupp, 1997).
In order to understand the relationship between
theory-specific and standard measures of change,
clients' symptoms and GAS scores were examined in relation to attachment styles. Regarding
symptoms, there was a significant decrease in the
pre-post measures of anxiety, depression, interpersonal sensitivity, and hostility for all clients.
Furthermore, there were not different rates of
symptom change for clients with posttreatment
insecure or secure attachment styles. However,
posttreatment-secure clients actually entered the
study with lower levels of symptoms than did
posttreatment-insecure clients. Individuals with
secure attachment are more likely than individuals
with insecure attachment to recognize their competence and vulnerability and thereby appropriately seek help when under distress (Cole-Detke
& Kobak, 1996). It is possible that this study's
posttreatment-secure clients sought help at an earlier stage than the other clients. Regarding GAS,
clients with posttreatment dominant-secure attachment styles were found to have significantly
high posttreatment GAS scores than were clients
with posttreatment-insecure attachment styles. In
addition, clients with posttreatment dominantpreoccupied attachment styles were found to have
significantly lower posttreatment GAS scores
than were clients of other attachment styles.
These two findings parallel previous clinical studies examining posttreatment GAP scores among
attachment styles (Dolan, 1992; Fonagy et al.,
1996). Consistent with their hyperactivating strategy (making contact with others through portraying themselves in extreme distress and in need
of sympathy and attention), it is quickly apparent
why preoccupied clients would be rated with
lower GAS and GAP scores by independent clinicians. Attachment styles become particularly evident in times of stress. Research with attachment
styles has continually demonstrated that preoccupied individuals report and display high levels of
distress and emotion-based coping across a range
of interpersonal, war-related, and health-related
stressors (Mikulincer & Florian, 1998). Thus, attachment theory provides a rich and clinically
meaningful framework from which to link inter-
Attachment Styles
personal coping strategies with differential symptom and global functioning patterns for each attachment style.
Two features of the sample merit comment.
First, the client sample was heavily represented
by White (96%) female (76%) individuals with a
high level of education (mean of 3 years of college). Second, 71% of the clients entered the
study with previous exposure to psychotherapy
that ranged from one to two sessions to over 10
years of treatment. Such homogeneity in sample
characteristics limits the generalizability of this
study's findings.
It is important that the present study is replicated with a larger and more representative sample of clients and therapists in order to attain
further evidence regarding changes in attachment
styles over the course of therapy. In addition,
learning more about clients who prematurely
leave therapy would be beneficial to researchers
and clinicians. Kazdin (1994) noted that it is crucial to obtain data on clients who leave therapy
prematurely, as these clients are likely to differ,
at least to some degree, from clients who stay in
therapy. Investigating those clients who terminate
therapy early may offer a distinctive view of qualitative differences within insecure and secure attachment categories.
In summary, psychodynamic/interpersonal
psychotherapy research conducted from a theoryspecific framework provides unique opportunities
to examine change in theoretically relevant relational structures over the course of therapy. Adult
attachment theory offers a meaningful research
and clinical framework for understanding and
measuring client relational change, as well as hypothesized change processes in TLDP. This investigation provided evidence for systematic and
significant changes in clients' categorical and dimensional attachment styles and distinct patterns
of symptom and global functioning changes for
each attachment style. A comprehensive picture
of the effectiveness of brief psychodynamic therapy may only be obtained from research that examines the interplay between theory-specific and
standard measures of change. Such research contributes a rich understanding of the nature of
change, as well as the change processes, over the
course of therapy. It is hoped that future psychodynamic/interpersonal psychotherapy investigations will continue to utilize theory-specific measures of change to better understand the process
and outcome of psychotherapy.
References
American Psychiatric Association. (1980). Diagnostic and
statistical manual of mental disorders (3rd ed.). Washington, DC: Author.
American Psychiatric Association. (1987). Diagnostic and
statistical manual of mental disorders (3rd ed.-rev.). Washington, DC: Author.
ANCHIN, J. C., & KIESLER, D. J. (Eds.). (1982). Handbook of
interpersonal psychotherapy. New York: Pergamon Press.
AZIM, H. F., PIPER, W. E., SEGAL, P. M., NIXON, G. W.,
& DUNCAN , S. (1991). The quality of object relations scale.
Bulletin of the Menninger Clinic, 55, 323-343.
BAKERSMAN-KRANENBURG, M. J., & VAN UZENDOORN, M. H.
(1993). A psychometric study of the Adult Attachment
Interview: Reliability and discriminant validity. Developmental Psychology, 29(5), 870-879.
BALDWIN, M. W., & MEUNIER, J. (1999). The cued activation
of attachment relational schemas. Social Cognition, 17(2),
209-227.
BARBER,). P. (1994). Efficacy of short-term dynamic psychotherapy. Journal of Psychotherapy Practice and Research,
3(2), 108-121.
BARTHOLOMEW, K. (1990). Avoidance of intimacy: An attachment perspective. Journal of Social and Personal Relationships, 7, 147-178.
BARTHOLOMEW, K., & HOROWIZ, L. (1991). Attachment
styles among young adults: A test of a four-category model.
Journal of Personality and Social Psychology, 61, 226244.
BEIN, E., ANDERSON, T., STRUPP, H. H., HENRY, W. P.,
SCHACHT, T. E., BINDER, J. L., & BUTLER, S. F. (2000).
The effects of training in time-limited dynamic psychotherapy: Changes in therapeutic outcome. Psychotherapy Research, 10(2), 119-132.
BENJAMIN, L. S. (1982). Use of the Structural Analysis of
Social Behavior (SASB) to guide interventions in psychotherapy. In J. C. Anchin & D. J. Kiesler (Eds.), Handbook
of interpersonal psychotherapy (pp. 190-212). New York:
Pergamon Press.
BENJAMIN, L. S. (1993). Interpersonal diagnosis and treatment of personality disorders. New York: Guilford.
BINDER, J. L., & STRUPP, H. H. (1997). "Negative Process":
A recurrently discovered and underestimated facet of therapeutic process and outcome in the individual psychotherapy
of adults. Clinical Psychology: Science and Practice, 4(2),
121-159.
BLIWISE, N. G. (1992, November). Attachment styles and
interpersonal relations in adulthood. Paper presented at the
Gerontological Society of America, Washington, DC.
BOWLBY, J. (1969). Attachment and loss: Vol. 1. Attachment.
New York: Basic Books.
BOWLBY, J. (1973). Attachment and loss: Vol. 2. Separation:
Anxiety and anger. New York: Basic Books.
BOWLBY, J. (1980). Attachment and loss: Vol. 3. Loss: Sadness and depression. New York: Basic Books.
BOWLBY, J. (1988). A secure base: Parent-child attachment
and healthy human development. New York: Basic
Books.
COLE-DETKE, H., & KOBAK, R. (1996). Attachment processes
in eating disorders and depression. Journal of Counseling
and Clinical Psychology, 64(2), 282-290.
CRITS-CHRISTOPH, P. (1992). The efficacy of brief dynamic
psychotherapy: A meta-analysis. American Journal of Psychiatry, 149(2), 151-158.
157
L. A. Travis et al.
DEROGATIS, L. (1983). SCL-90-R: Administration, scoring,
and procedures manual for the revised version. Baltimore,
MD: Clinical Psychometric Research.
DIEHL, M., ELNICK, A. B., BOURBEAU, L. S., & LABOUVIEVIEF.G. (1998). Adult attachment styles: Their relations to
family context and personality. Journal of Personality and
Social Psychology, 74(6), 1656-1669.
DOLAN, R. T. (1992). Attachment style and the quality of the
therapeutic alliance in psychotherapy. Unpublished doctoral dissertation, Catholic University of America, Washington, DC.
DOLAN, R. T., ARNKOFK, D., & GLASS, C. (1993). Client
attachment style and the psychotherapist's interpersonal
stance. Psychotherapy, 30(3), 408-412.
ENDICOTT, J., SPITZER, R., FLEISS, J., & COHEN, J. (1976).
The Global Assessment Scale: A procedure for measuring
overall severity of psychiatric disturbance. Archives of General Psychiatry^, 33, 761-771.
FONAGY, P., LEIGH, T., STEELE, M., STEELE, H., KENNEDY,
R., MATTON, G., TARGET, M., & GERBER, A. (1996). The
relation of attachment status, psychiatric classification, and
responses to psychotherapy. Journal of Clinical and Consulting Psychology, 64, 22-31.
GRIKFIN, D., & BARTHOLOMEW, K. (1994). Models of the self
and other: Fundamental dimensions underlying measures
of adult attachment. Journal of Personality and Social Psychology, 67(3), 430-435.
HARRIST, S., QUINTANA, S., STRUPP, H. H., & HENRY, W.
(1994). Intemalization of interpersonal process in timelimited dynamic psychotherapy. Psychotherapy, 31, 4957.
HAZAN, C., & SAVER, P. (1987). Conceptualizing romantic
love as an attachment process. Journal of Personality and
Social Psychology, 52, 511-524.
HENRY, W. P., SCHACHT, T. E., & STRUPP, H. H. (1986).
Structural analysis of social behavior: Application to a study
of interpersonal processes in differential therapy outcome.
Journal of Consulting and Clinical Psychology, 54, 2731.
HENRY, W. P., SCHACHT, T. E., & STRUPP, H. H. (1990).
Patient and therapist introject, interpersonal process, and
differential psychotherapy outcome. Journal of Consulting
and Clinical Psychology, 58(6), 768-774.
HENRY, W. P., STRUPP, H. H., BUTLER, S. F., SCHACHT,
T. E., & BINDER, J. L. (1993). Effects of training in timelimited dynamic psychotherapy: Changes in therapist behavior. Journal of Consulting and Clinical Psychology,
61(3), 434-440.
HOGLEND, P. (1988). Brief dynamic psychotherapy for less
well-adjusted patients. Psychother Psychosom, 49, 197204.
HOGLEND, P. (1993). Suitability for brief dynamic psychotherapy: Psychodynamic variables as predictors of outcome. Ada Psychiatrica Scandinavica, 88, 104-110.
HOGLEND, P., SORLIE, T., HEYERDAHL, P., SORBYE, O., &
AMLO, S. (1993). Brief dynamic psychotherapy: Patient
suitability, treatment length, and outcome. Journal of Psychotherapy, Practice and Research, 2(3), 230-241.
HOROWITZ, L. M., ROSENBERG, S., BAER, B., URENO, G.,
& VILLASENOR, V. A. (1988). Inventory of Interpersonal
Problems: Psychometric properties and clinical applications. Journal of Consulting and Clinical Psychology,
56, 885-892.
HOROWITZ, L. M., ROSENBERG, S., & BARTHOLOMEW, K.
(1993). Interpersonal problems, attachment styles, and
158
brief dynamic psychotherapy. Journal of Consulting and
Clinical Psychology, 61(4), 529-560.
HOROWITZ, L. M., ROSENBERG, S., & KALEHZAN, B. M.
(1992). The capacity to describe other people clearly: A
predictor of interpersonal problems in brief dynamic psychotherapy. Psychotherapy Research, 2, 37-51.
HOROWITZ, L. M., & VITKUS, J. (1986). The interpersonal
basis of psychiatric symptoms. Clinical Psychology Review, 6, 443-469.
KALEHZAN, B. (1993). The relationship between adult attachment style, interpersonal problems, and the manifestation
of symptoms in clinical depression. Unpublished doctoral
dissertation, Pacific School of Psychology, Palo Alto, CA.
KAZDIN, A. (1994). Methodology, design, and evaluation in
psychotherapy research. In A. E. Bergin & S. L. Garfield
(Eds.), Handbook of psvchotherap\ and behavior change
(4th ed., pp. 19-71). New York: John Wiley.
LAMBERT, M. J., & HILL, C. E. (1994). Assessing psychotherapy outcomes and processes. In A. E. Bergin & S. L.
Garfield (Eds.), Handbook of psychotherapy and behavior
change (4th ed., pp. 72-113). New York: John Wiley.
LIOTTI, G. (1991). Patterns of attachments and the assessment
of interpersonal schemata: Understanding and changing difficult client-therapist relationships in cognitive psychotherapy. Journal of Cognitive Psychotherapy, 5(2), 105-114.
LUBORSKY, L. (1962). Clinician's judgments of mental health: A
proposed scale. Archives of General Psychiatry, 7, 407-417.
MIKULINCER, M., & FLORIAN, V. (1998). The relationship
between adult attachment styles and emotional and cognitive reactions to stressful events. In J. A. Simpson &
W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 143-165). New York: Guilford.
PIETROMONACO, P. R., & BARRETT, L. F. (2000). The internal
working models concept: What do we really know about
the self in relation to others? Review of General Ps\cholog\,
4(2), 155-175.
PILKONIS, P. A., HEAPE, C. L., & PROIETTI, J. M. (1993).
Adult attachment classification styles, personality disorder,
and treatment outcome in depression. Unpublished manuscript, Western Psychiatric Institute and Clinic, Pittsburgh, PA.
PIPER, W. E., AZIM, H. F., JOYCE, A. S., MCCALLUM, M.,
NIXON, G. W., & SEGAL, P. S. (1991). Quality of object
relations versus interpersonal functioning as predictors of
therapeutic alliance and psychotherapy outcome. Journal
of Nervous and Mental Disease, 179(7), 432-438.
RICKS, M. (1985). The social transmission of parental behavior: Attachment across generations. In I. Bretherton & E.
Waters (Eds.), Growing points in attachment theory and
research, Monographs of the Society for Research in Child
Development, 50 (pp. 211-230).
SAFRAN, J. D. (1992). Extending the pantheoretical applications of interpersonal inventories. Journal of Psychotherapy
Integration, 2(2), 101-105.
SCHACHT, T. E., STRUPP, H., & HENRY, W. P. (1988).
Problem-treatment-outcome-congruence: A principle
whose time has come. In H. Dahl & H. Kaechele (Eds.),
Psychoanalytic process research strategies (pp. 1-14).
New York: Springer.
SCHARFE, E., & BARTHOLOMEW, K. (1994). Reliability and
stability of adult attachment patterns. Personal Relationships, I , 23-43.
SPERLING, M. B., & BERMAN, W. H. (1994). (Eds.), Attachment in adults: Clinical and developmental perspectives.
New York: Guilford.
Attachment Styles
STRUPP, H. H. (1993). The Vanderbilt psychotherapy studies:
Synopsis. Journal of Consulting and Clinical Psychology,
SULLIVAN, H. S. (1953). The interpersonal theory ofpsychiatry. New York: W. W. Norton.
6/(3), 431-433.
STRUPP, H. H. (1996). Some salient lessons from research
and practice. Psychotherapy, 33, 135-138.
SULLIVAN, H. S. (1954). The psychiatric interview. New
York: W. W. Norton.
WEST, M., & SHELDON-KELLER, A. (1994). Patterns ofrelat-
STRUPP, H. H., & BINDER, J. L. (1984). Psychotherapy in a
ing: An adult attachment perspective. New York: Guilford.
new key. New York: Basic Books.
159