self-reported attachment styles and therapeutic orientation of therapists and their relationship...
TRANSCRIPT
Self-reported attachment styles and therapeutic
orientation of therapists and their relationship
with reported general alliance quality
and problems in therapy
Susanna Black1*, Gillian Hardy2, Graham Turpin2 and Glenys Parry3
1Sheffield Care NHS Trust, UK2Department of Clinical Psychology, University of Sheffield, UK3Sheffield Centre for Health and Related Research, University of Sheffield, UK
The aims of this study were to explore the relationship between therapists’ self-
reported attachment styles and therapeutic orientation with the self-reported general
therapeutic alliance and therapist-reported problems in psychological therapy.
A sample of 491 psychotherapists from differing therapeutic orientations responded to
a postal questionnaire. The questionnaire contained standardized measures of
therapeutic alliance quality, attachment behaviours, a checklist of problems in therapy,
and a brief personality inventory.
Therapist-reported attachment styles generally explained a significant additional
proportion of the variance in alliance and problems in therapy, over and above variance
explained by general personality variables. Self-reported secure attachment style was
significantly positively correlated with therapist-reported general good alliance.
Self-reported anxious attachment styles were significantly negatively correlated with
good alliance, and significantly positively correlated with the number of therapist-
reported problems in therapy. Therapeutic orientation independently predicted a small
but significant amount of the variance in reported general alliance quality in addition to
that explained by attachment behaviours.
It is now widely believed that comparable outcomes are achieved by different therapies
despite differences in their underlying assumptions about the development of
psychological dysfunction, and in therapeutic techniques employed (e.g. Horvath &
*Correspondence should be addressed to Dr Susanna Black (e-mail: [email protected]).
TheBritishPsychologicalSociety
363
Psychology and Psychotherapy: Theory, Research and Practice (2005), 78, 363–377
q 2005 The British Psychological Society
www.bpsjournals.co.uk
DOI:10.1348/147608305X43784
Greenberg, 1994; Luborsky, McClellan, Diguer, Woody, & Seligman, 1997). Horvath and
Greenberg (1994) have suggested that this raises fundamental questions about which
elements of therapy are responsible for change.
Psychotherapy research has shown that differences in both therapeutic approach,
and/or differences between client groups only account for a small proportion of the
outcome variance (Beutler, 2002). In contrast, one fairly consistent finding has been that
the quality of the therapeutic alliance, particularly in the initial stages of therapy, is
predictive of outcome (Martin, Garske, & Davis, 2000). Horvath and Symonds (1991),
for example, found that the alliance had a more robust link to outcome than other
relationship factors, and Stiles, Agnew-Davies, Hardy, Barkham, and Shapiro (1998)
found that all aspects of the therapeutic alliance were positively correlated with
outcome.
Research has tended to focus on which client characteristics impact on the
therapeutic alliance (e.g. Satterfield & Lyddon, 1995; Mallinckrodt, 1991). Horvath
(1994) suggested that pre-therapy variables such as clients’ motivation, social
relationships, and so forth, impact upon both the alliance and outcome. Luborsky
(1994) argued that in part, the positive relationship capacity of the client provides the
foundation for the alliance. Therefore, the capacity to form an alliance is certainly partly
a quality that the client brings to treatment.
However, it has been argued that therapists’ capacity to form relationships is equally
important in alliance formation. Beutler (1997) suggests that these variables should be
given at least as much attention as client factors and therapeutic models in
psychotherapy research. Interest in the factors affecting the therapeutic alliance has
recently begun to broaden to include therapist characteristics. Nevertheless, much of the
research has focused on issues of therapist competence and adherence to treatment
protocol (e.g. Kivlighan, Patton, & Foote, 1998). Furthermore, Elkin (1999) stated that
much of the literature on therapists has been carried out as a secondary factor in
outcome/alliance research. This has meant that often relatively small numbers of
therapists have been studied, and the generalizability of the findings is unclear.
Notwithstanding these limitations, the literature in this area indicates that the personal
qualities of the therapist, and, in particular, the ability to form a warm and supportive
relationship are important factors in the formation of the alliance (e.g. Orlinsky, Grawe, &
Parks, 1994; Roth & Fonagy, 1996).
Attachment theory describes characteristic ways that individuals interact and relate
to others. These secure or insecure attachment styles develop in childhood (Bowlby,
1988) and remain relatively stable in adulthood (Collins, 1996). Secure individuals are
comfortable with closeness and intimacy, and with depending on others and having
others dependent on them. Insecure attachment styles are characterized by either over-
or under-involvement in relationships. The primary features of discomfort with
closeness and need for independence are associated with an avoidant or dismissing
attachment style. Other aspects of insecure attachment styles include the very opposite
features; here individuals show a preoccupation with relationships and a need for
Susanna Black et al.364
approval. Such individuals are described as having an anxious-ambivalent or
preoccupied attachment style.
Bowlby (1988) states that whenever people are frightened, fatigued, or sick they
show attachment behaviours aimed to elicit care-giving behaviours in others. The type
of care-giving offered is dependent on the attachment styles of the both the distressed
person and the ‘caregiver’ (Bowlby, 1988). It could be argued that one of the defining
features of psychotherapy is that it occurs when the client is vulnerable, and therefore
the attachment style of both the client and therapist (as caregiver) is likely to have an
impact upon the formation of the therapeutic relationship. For example, Hardy et al.
(1999) found that client attachment styles were related to differences in therapists’
responses at significant moments in therapy.
Attachment can be measured by self-report or by interview. The adult attachment
interview (AAI; Main & Goldwyn, 1984) is generally thought to be the most accurate
measure of attachment style. However, a number of self-report measures have also been
developed. Some of these measures focus on romantic attachment (Hazan & Shaver,
1987), whilst others focus on attachment in relationships more generally (Bartholomew
& Horowitz, 1991). These behavioural dimensions of attachment styles have been
operationalized in a self-report measure: the Attachment Style Questionnaire (Feeney,
Noller, & Hanrahan, 1994). This questionnaire contains five scales derived from
principal component analysis: Confidence, Preoccupied with relationships, Need for
approval, Relationships as secondary, and Discomfort with closeness. It has been used in
other studies exploring non-romantic attachment style (Danov & Bucci, 2002; Troisi,
D’Argenio, Peracchio, & Petti, 2001).
Relatively little research has examined therapist attachment styles in relation to any
aspect of therapy. Those studies that have been conducted have tended to suggest that
the quality of the therapeutic relationship, issues that arise in therapy, and perceived
competence of the therapist are related to therapists’ attachment styles. At least one
study has found no relationship between counter-transference behaviours, attachment
styles and the therapeutic alliance (Ligiero & Gelso, 2002). Wheeler (2000) found that
personality constructs such as personable/aloof and open/closed (rather than
constructs related to learning, such as intelligent/limited, or to the role as a counsellor,
such as professionally skilled/incompetent) are most frequently used by trainers to
discriminate between good and bad counsellor trainees. A further study which explored
the therapists’ resolution of ruptures in the alliance, and the attachment style of both the
therapist and the client, found that more anxiously attached therapists tended to
respond less empathically, particularly with fearful and secure patients (Rubino, Barker,
Roth, & Fearon, 2000).
In another two studies Tyrell, Dozier, Teague, and Fallot (1999) and Dozier, Cue, and
Barnett (1994) found that there was an interaction between the attachment styles of
clinical case managers and their clients in terms of responsiveness to dependence needs,
client functioning, and the therapeutic relationship. They found that case managers
Attachment, alliance and problems in therapy 365
with secure attachments were more likely to challenge their clients’ internal model of
relationships than case managers with insecure attachments.
Research in related areas also has produced findings suggesting that factors
relating to attachment are important. Henry, Schact, and Strupp (1990) and Henry
and Strupp (1994) found that therapists’ internal representations of past relationships
(introjects) had a strong influence on their session process with clients, and on the
quality of the alliance with at least some clients. Therapists with hostile introjects had
poorer alliances and were more likely to engage in counter-therapeutic or
problematic therapy processes. Dunkle and Friedlander (1996) carried out a study
which looked at experience and personal characteristics of trainee therapists within a
university setting. They found a link between certain personality characteristics of the
therapist, such as comfort with closeness, and the clients’ ratings of the quality of the
alliance. In a review of therapist factors in the alliance, Ackerman and Hilsenroth
(2003) found a positive relationship between a number of therapist attributes and
alliance. These include warmth, openness, flexibility, honesty, trustworthiness,
respectfulness, and so forth. Many of these traits are similar to those used to describe
secure attachment.
Meyer and Pilkonis (2001), in a review of studies of attachment style and its role in
individual psychotherapy, suggested that although secure attachment plays an
important role in the therapist’s skill in handling difficulties in therapy, it is at least
as important for the alliance that the therapist is not anxiously attached. Secure and
avoidant therapists appear to perform better. If not secure, then it appears to be
important to have an opposite attachment style to the client.
The evidence comparing the alliance quality of therapies from different therapeutic
orientation is mixed. Most studies report little differences in alliance ratings across
therapies (e.g. Krunick et al., 1996; Marmar, Gaston, Gallagher, & Thompson, 1989).
A further study, though, did find a difference between therapies (Agnew-Davies, Stiles,
Hardy, Barkham, & Shapiro, 1998), when a more positive alliance was reported for
cognitive behaviour therapy than for psychodynamic-interpersonal therapy by clients,
although not by therapists. Other studies have found differences in the personality style
of therapists of different orientations. Arthur (1999) found that cognitive-behavioural
(CB) therapists tended to be more independent and experience less anxiety, whilst
psychodynamic/analytic (PI) therapists tended to experience more performance
anxiety and neurotic symptoms. Viney (1994) also found differences between therapists
of different orientations in the therapists’ reactions to client expressed emotion.
The aims of this study are to establish the extent to which self-reported attachment
styles of therapists are associated with reported general alliance quality and reported
problems in therapy, over and above their therapeutic orientation. More specifically, our
hypotheses are:
(1) Therapists who report more secure attachment relationships will report having
better general alliances with their clients.
Susanna Black et al.366
(2) Therapists who report more insecure attachment relationships will report having
more problems in therapy.
(3) Therapists’ attachment behaviours and therapeutic orientation will independently
affect the alliance and reported problems in therapy.
Method
Participants
Questionnaires were distributed to 1,400 psychotherapists who were selected from
three registers of accredited therapists. The lists were published by the United Kingdom
Council for Psychotherapists (UKCP), the British Association of Behavioural and
Cognitive Psychotherapists (BABCP), and the British Confederation of Psychotherapists
(BCP). Therapists working primarily with children and/or families were excluded.
Additionally, some therapists in the London area were excluded due to overlap with
a similar study being conducted there concurrently. All BABCP and BCP therapists who
were not excluded for the above reasons were contacted. From the UKCP listings, one
in three therapists were selected, with either the next or the previous therapist on the
list being selected if the third was excluded.
A total of 491 psychotherapists responded, giving a response rate of 36%;
19 psychotherapists failed to complete either one or both of the major measures within
the questionnaire, resulting in the different sample sizes for different analyses. This was
the expected response rate, but we have no data on non-responders and therefore
cannot say how representative the responders were of the whole sample.
Of the responses, 30% (N ¼ 146) of the psychotherapists were men and 70%
(N ¼ 345) were women. The mode age groupwas 51 þ years and the median was 46 to
50 years. More than half of therapists (54.8%) had over 10 years post-qualification
experience, and only eight therapists had less than 2 years; 79% of therapists had at least
1 year of formal post-qualification training, and 53.8% had more than 3 years formal
post-qualification training.
The therapeutic orientation of the sample was as follows: 34.8% (N ¼ 171) indicated
they were PI therapists; 24.2% (N ¼ 119) were CB therapists; 19.1% (N ¼ 94)
were cognitive-analytic (CAT)/integrative or eclectic therapists; 14.9% (N ¼ 73) were
humanistic therapists, and 6.9% (N ¼ 34) did not specify their orientation.
The therapists had a mixture of core training; 163 were UKCP trained
psychotherapists, 104 had a nursing background, 78 were chartered psychologists,
56 had a core training in social work, 27 were psychiatrists, 22 were accredited by the
British Association of Counsellors (BAC), and 41 were categorized as ‘other’.
Measures
Attachment behaviours were measured using the Attachment Style Questionnaire (ASQ;
Feeney et al., 1994). This is a dimensional questionnaire and was selected because it has
reasonable reliability and validity (Stein, Jacobs, Ferguson, Allen, & Fonagy, 1998).
Attachment, alliance and problems in therapy 367
In addition, it concerns people’s style of interacting in relationships in general, rather
than focusing specifically on romantic attachment.
The ASQ is a 40-item self-report instrument containing five dimensions. The
Confidence scale contains eight items relating to secure attachment behaviours, such as
trust in others and belief in self-worth. There are two scales measuring preoccupied
attachment behaviours: Preoccupation with relationships (8 items) is described as an
anxious reaching out to others in order to fulfil dependency needs, and Need for
approval (7 items) reflects the individual’s need for others’ acceptance and
confirmation. There are also two scales that measure dismissing attachment behaviours:
Relationships as secondary (7 items) contains items which describe the individual as
protecting themselves against hurt and vulnerability by emphasizing achievement and
independence. Discomfort with closeness (10 items) relates to feeling uncomfortable
with intimacy and closeness. Respondents are asked to rate items on a 6-point scale
ranging from totally agree to totally disagree.
In the original study, the scales had good internal and test-retest reliability, and the
items also loaded appropriately on Hazan and Shaver’s (1987) forced-choice attachment
measure. The alpha’s for this sample for the five scales were Confidence
(.78), Discomfort with closeness (.86), Need for approval (.77), Preoccupation with
relationships (.74), and Relationships as secondary (.72).
The therapeutic alliance was measured using the Agnew Relationship Measure
(ARM; Agnew-Davies et al., 1998). It is reported to have good reliability and validity
(Agnew-Davies et al., 1998) and also focuses specifically on the alliance without
reference to therapeutic techniques. There are both therapist and client versions of the
ARM; however, only the therapist form is used here.
The ARM is a self-report 28-item questionnaire. The wording of the questionnaire has
been modified in this study to allow its use as a generalized measure of therapeutic
alliance formation. Given that our survey design might potentially have involved NHS
therapists across a number of health authorities, we sought approval from the UK
Medical Research Ethics Committee (MREC). We were advised that individual consent
would be required for every client in the survey if the therapists completed the alliance
questionnaire with a specific client in mind. Although we disagreed with this advice, we
nevertheless were bound by it and it rendered the original study impossible. We
therefore chose to revise the alliance questionnaire. Respondents were instructed to
think of their clients in general when completing the questionnaire (e.g. ‘I feel warm
and friendly with my clients’, ‘my clients have confidence in me and my techniques’,
‘my clients feel free to express the things that worry them’). This was done for ethical
reasons. Asking therapists to think of any particular client would have meant obtaining
permission from each client. Therapists are asked to rate the items on a 7-point scale
from strongly agree to strongly disagree. The alpha coefficients were re-calculated for
the full questionnaire to ascertain the effect that re-wording had on the internal
consistency reliability of the scale. The alpha for the modified questionnaire was
.84, indicating an acceptable level of internal reliability.
Susanna Black et al.368
Problems in therapy were measured using the Therapist problem checklist (PCL;
Shroder, personal communication 1999), a measure of therapist-perceived problems.
The scale consists of seven items relating to problems encountered during therapy. The
questionnaire is organized with the following stem: ‘currently, how often do you
feel: : :?’, and the items include ‘lacking in confidence that you can have a beneficial
effect on the client’, and ‘unable to comprehend the essence of a client’s problems’. The
items are rated on a 6-point scale from always to never. High scores on this scale
correspond to more problems in therapy. The internal consistency reliability of this
scale was tested for this sample and was adequate (.79).
General personality features were measured using the Brief Eysenck Personality
Questionnaire (EPQ; Eysenck & Eysenck, 1969), an 8-item version of the full EPQ which
contains four items measuring neuroticism (sample a ¼ :67) and four items measuring
extroversion (sample a ¼ :70).
Results
Therapist-reported attachment styles and reported general therapeutic alliance
Our first hypothesis was that therapists who report more secure attachment
relationships would also report having better alliances with their clients. Correlations
between the variables show a significant positive correlation between the ASQ
Confidence scale and the mean ARM score, r ¼ :441, p , :001, but also significant
negative correlations between mean ARM and the subscales of the ASQ representing
insecure attachment behaviour; ASQ Discomfort with closeness, r ¼ 2:258, p , :001,
ASQ Relationships as secondary, r ¼ 2:182, p , :001, ASQ Need for approval,
r ¼ 2:278, p , :001, and ASQ Preoccupation with relationships, r ¼ 2:315, p , :001.
To test that these associations were not a function of more general personality
variables, and to examine the relative importance of insecure attachment behaviours, we
conducted amultiple regression analysis inwhich two general personality variables (EPQ
extroversion and EPQ neuroticism) were entered first, followed by the five ASQ
dimensions (ASQ Confidence, ASQ Need for approval, ASQ Preoccupation with
relationships, ASQ Discomfort with closeness, and ASQ Relationships as secondary).
EPQ extroversion and EPQ neuroticism explained 9.4% of the variance. This was
significant, adjusted R2 ¼ :090, Fð2; 380Þ ¼ 19:8, p , :001. With the second block the
regression explained 21.3% of the variance, adjusted R2 ¼ :199, which was significant,
Fð7; 375Þ ¼ 14:53, p , :001. The attachment dimensions explained a further 11.9% of
the variance in total alliance score. The R 2 change was tested and was highly significant,
Fð5; 382Þ ¼ 11:34, p , :001. The beta weights suggested that the predictors which
remained significant once all shared variance was removed were ASQ confidence,
b ¼ 5:18, p , :001, and ASQ Preoccupation with relationships, b ¼ 22:01, p , :05.
Neither of the personality variables reached significance in the regression when shared
variance was removed. This supports the hypothesis that self-reported secure
attachment (therapists who report greater confidence in their relationships) would
Attachment, alliance and problems in therapy 369
add significantly to the amount of the variance in total alliance quality explained by
personality measures alone.
In addition to our primary prediction, therapists who scored high on the ASQ
Preoccupation with relationships scale were associated with poorer alliance scores,
suggesting that therapists who reported a dependent attachment style reported having
poorer general therapeutic relationships than those who reported having a more secure
attachment style.
Self-reported therapist attachment style and reported problems in therapy
Our second hypothesis was that therapists who report more insecure attachment
relationships would also report having more problems in therapy. Again, correlations
between the PCL and the four insecure attachment scales from the ASQ show significant
correlations. In all cases, higher insecure scores correlated with higher problems in
therapy, ASQ Preoccupation with relationships, rð464Þ ¼ :322, p , :001, ASQ Need for
approval, rð464Þ ¼ :165, p , :001, ASQ Relationships as secondary, rð463Þ ¼ :165,
p , :001, and ASQ Discomfort with closeness, rð459Þ ¼ :252, p , :001.
To test that these associations were not a function of self-reported more general
personality variables or self-reported lack of secure attachment behaviours, we
conducted a multiple regression analysis in which the same seven variables were
entered into the multiple regression (using method enter) in two blocks (block one,
EPQ variables; block two, ASQ dimensions). EPQ extroversion and EPQ neuroticism
explained 15.8% of the variance in reported problems in therapy in the first block,
which was significant, adjusted R2 ¼ :154, Fð2; 426Þ ¼ 40:06, p , :001. When the five
attachment dimensions were added in the second block, the variables accounted for
23.3% of the variance, multiple R ¼ :483,R2 ¼ :233, adjustedR2 ¼ :220. This regression
was significant, Fð7; 429Þ ¼ 18:29, p , :001. The regression indicated that the
attachment dimensions explained an additional 7.5% of the variance in problems in
therapy over and above the EPQ variables. The R2 change was tested and found to be
highly significant, Fð5; 429Þ ¼ 8:22, p , :001. Examination of the beta weights
indicated that the predictors with the most significance when all shared variance was
removed were ASQ Need for approval, b ¼ 4:83, p , :001, and EPQ neuroticism,
b ¼ 4:38. p ¼ :001. These were both positive relationships meaning that increasing
reported neuroticism and/or need for approval predicted increased reported problems
in therapy. These analyses offer some support to the hypothesis that self-reported
insecure attachment styles explain a significant additional amount of variance in
self-reported problems in therapy over and above personality variables. However, only
one of the four insecure attachment styles, Need for approval, was associated with high
levels of reported problems in therapy. In addition, the general personality scale of
neuroticism remained significantly associated with high levels of reported problems in
therapy.
Susanna Black et al.370
Therapists’ therapeutic orientation, self-reported general alliance and problems
in therapy
To test our third hypothesis that therapists with different therapeutic orientations will
vary in their alliances with their clients, and in the frequency of problems in therapy that
they report, we conducted two analyses of covariance. Gender was used as a covariate
as there were proportionately more men than women in the CBT group, than in the
other three orientation groups, x2ð3Þ ¼ 13:56, p , :004.
In the first ANOVA we tested for differences in total alliance scores between
orientation groups. The main effect for orientation was significant, Fð3; 390Þ ¼ 12:00,
p , :001. Post hoc t tests using Bonferroni correction indicated that, as predicted, PI
therapists, M ¼ 5:38, SD ¼ 0:50, reported significantly lower alliance scores than
humanistic therapists, M ¼ 5:48, SD ¼ 0:53; tð194Þ ¼ 3:68, p , :001, CBT therapists,
M ¼ 5:51, SD ¼ 0:45; tð243Þ ¼ 25:17, p , :001, and CAT/integrative therapists
M ¼ 5:48, SD ¼ 0:47; tð217Þ ¼ 24:2, p , :001. The other group differences were not
significant.
The second ANOVA was completed to test differences in problems in therapy
between orientation groups. Again, gender was entered as a covariate. The effect of
orientation was significant, Fð3; 439Þ ¼ 4:18, p ¼ :006. Post hoc t tests using Bonferroni
correction showed that psychodynamic/analytic therapists, M ¼ 16:43, SD ¼ 3:66,
reported significantly more problems in therapy than both CBT therapist, M ¼ 15:03,
SD ¼ 3:6; tð273Þ ¼ 2:98, p ¼ :003, and CAT/integrative therapists, M ¼ 15:11,
SD ¼ 3:4; tð249Þ ¼ 2:83, p ¼ :005. No other pairs of groups were significantly different.
Separate effects of therapists’ reported attachment styles and therapeutic
orientation on reported general alliance and problems in therapy
Finally, the hypotheses that therapists’ reported attachment styles and therapeutic
orientation will independently affect the self-reported alliance and reported problems in
therapy were tested via multiple regression analyses.
The attachment dimensions and the therapist orientation dummy variables were
entered into a multiple regression of mean total alliance score (using method enter) in
two blocks (block one, attachment dimensions; block two, orientation group dummy
variables). The first block explained 20.4% of the variance in mean total ARM score,
which was significant, adjusted R2 ¼ :193, Fð3; 358Þ ¼ 18:34, p , :001. When the
orientation dummy variables were added in the second block, the variables accounted
for 24.2% of the variance, adjusted R2 ¼ :23. This regression was significant,
Fð8; 355Þ ¼ 14:20, p , :001. The regression indicated that the orientation variables
explained a further 3.8% of the variance in mean total ARM score over and above
attachment behaviours. The R2 change was tested and was found to be highly
significant, Fð3; 355Þ ¼ 6:01, p ¼ :001. Examination of the beta weights indicated that
the predictors with the most significance when all shared variance was removed were
ASQ confidence, b ¼ 0:37, p , :001, and Psychodynamic or not, b ¼ 20:24, p , :001.
Attachment, alliance and problems in therapy 371
The regression was repeated with the blocks entered in the opposite order, but the
relative variance explained by each block did not change significantly as a result of this.
The results support the hypothesis that reported attachment styles and therapeutic
orientation independently predict reported general alliance score.
The attachment dimensions and the therapist orientation dummy variables were
entered into a multiple regression of mean PCL score (using method enter) in two
blocks (block one, attachment dimensions; block two, orientation group dummy
variables). The first block explained 19.7% of the variance in mean PCL score, which
was significant, multiple R ¼ :44, R2 ¼ :197, adjusted R2 ¼ :19, Fð5; 401Þ ¼ 19:66,
p , :001. When the orientation dummy variables were added in the second block, the
variables accounted for 20.9% of the variance, multiple R ¼ :46, R2 ¼ :209, adjusted
R2 ¼ :19. This regression was significant, Fð8; 398Þ ¼ 13:14, p , :001. However, the
regression indicated that the orientation variables did not explain anything significant
over and above attachment behaviours, R 2 change ¼ .02, ns. Examination of the beta
weights indicated that the predictors with the most significance when all shared
variance was removed were ASQ Need for approval, b ¼ 5:35, p , :001, and
Psychodynamic or not, b ¼ 1:95, p , :05.
The regression was repeated with the blocks entered in the opposite order but the
relative variance explained by each block did not change significantly as a result of this.
However, the regression of first block containing the therapeutic orientation variables
did reach significance, multiple R ¼ :143, R2 ¼ :02, adjusted R2 ¼ :01,
Fð3; 403Þ ¼ 2:80, p ¼ :040. The results do not support the hypothesis that attachment
behaviours and therapeutic orientation independently predict mean PCL score.
Discussion
This study provides some tentative support that therapist characteristics are important
in the formation of the therapeutic alliance and on the therapists’ experiences of
problems in therapy. More specifically, the results indicate that the attachment styles
therapists self-report are significantly associated with their perceived quality of their
overall general therapeutic relationships with clients and reported problems in therapy.
As predicted, therapists who reported more secure attachments also reported having
better therapeutic general alliances with their clients. The results support the findings of
Henry et al. (1990) that therapists’ representations of their early relationships affect
their interactions with clients. Similarly, Peschken and Johnson (1997) found that level
of therapist trust in their clients (arguably a feature of secure attachment) led to an
increase in facilitative attitudes and correlated with client trust in the therapist. Dozier
and Tyrrell (1998) report that secure models of attachment result in an open,
flexible, and non-defensive approach to attachment related issues. This in turn translates
into approaches to others that are open, receptive and collaborative. These
characteristics are likely to facilitate a positive therapeutic alliance.
This study also found that therapists who reported higher levels of insecure
attachment significantly predicted poorer reported general therapeutic alliance.
Susanna Black et al.372
The finding supports the conclusion of the review by Meyer and Pilkonis (2001) that it is
at least as important not to have anxious attachment as it is to be securely attached. It is
also supported by other findings in attachment and alliance research such as Dozier et al.
(1994) who found that insecure clinicians intervened in more depth with hyper-
activating (anxious) clients, whereas secure therapists intervened in more depth with
deactivating (avoidant) clients. They concluded that insecure therapists intervened in
a way that failed to challenge their clients’ customary interpersonal strategies. Secure
clinicians appeared to be best able to provide clients with experiences that challenge
their working models. It is perhaps these experiences that are likely to result in a more
positive alliance and outcome.
Only some support was found for the hypothesis that self-reported insecure
attachments would predict reporting more problems in therapy: only one of the four
insecure attachment scales of the ASQ, Need for approval, was associated with problems
in therapy. It is possible that therapists who report more problems in therapy may do so
because they have a higher need for approval, and as a result set more perfectionistic
targets (Andersson & Perri, 2000). It is equally possible that those with a high need for
approval encounter more problems in therapy because they are less able to manage
difficulties when they arise. In addition the EPQ neuroticism was also significantly
associated with problems in therapy. This suggests that perhaps it is confidence in
general (rather than confidence in relationships), that is important in determining
problems in therapy.
The finding of differences in alliance and reported problems in therapy for different
therapeutic orientations was interesting. Psychodynamic therapists had the lowest
self-rated alliance score, with CBT therapists having the highest. This reflected previous
research (e.g. Arthur, 1999; Agnew-Davies et al., 1998). A multiple regression showed
that both orientation and attachment behaviours independently predicted a significant
amount of the variance in alliance. ASQ Confidence (linked to secure attachment) and
psychodynamic orientation were significant predictors of the quality of the alliance
once shared variance was removed. Psychodynamic orientation predicted a less positive
alliance.
There are a number of possible explanations for this finding. First, there may be
genuine differences between the orientation groups in the quality of the therapeutic
alliance. It may be that the alliance in psychodynamic therapy is weaker than the alliance
in the other therapies (Raue, Goldfried, & Barkham, 1997). Secondly, the theoretical
models informing the way in which different orientations think about relationships may
result in viewing the therapeutic relationship differently. It is possible that theoretical
orientation resulted in differences in the way in which the questions were reflected on
and answered.
The results showed that psychodynamic therapists also reported the most problems
in therapy, with the other three orientation groups reporting similar levels to each other.
Again, this may be as a consequence of the underlying theoretical models informing
responses to questions about problems in therapy. Psychodynamic therapy tends to
Attachment, alliance and problems in therapy 373
emphasize the client and focus on interpersonal and relationship issues, which could
conceivably make them more aware of or focused upon problems when they exist. It
should be noted that in all cases, the level of problems reported was very low.
There are a number of weaknesses in this study that should be taken into account
when interpreting the findings. First, the data are correlational and all the variables were
collected via the therapists’ own perceptions. Significant associations may, therefore, be
as a result of common method variance. Second, this sample is of experienced
psychotherapists, all of whom have an accredited training in psychotherapy. Therefore,
the findings may not generalize to the normal population of psychotherapists and
counsellors, since this sample is likely to have more training than is normal. Third, the
implications are constrained by the low response rate. A large number of therapists did
not return the questionnaires and a number were returned uncompleted. It was not
possible to gain information about the people who did not respond other than which
organization they belonged to.
Fourth, it is possible that the generalized measure of the alliance used in this study
may be measuring something quite different to the ARM as it was originally used. The
reduction of relationships with all clients down to one ‘average client’ is problematic
because relationships with clients can vary within one therapist. This may result, for
example, rather than measuring the quality of the alliance, in measuring the therapists’
confidence in their ability to form alliances with clients. Future research might wish to
compare responses to the ARM using different wording.
In conclusion, three hypotheses were posed at the start of these this study. The first
was that therapists who report more secure attachment relationships will report having
better alliances with their clients. This hypothesis was supported, and in addition there
was evidence that insecure attachment also negatively influenced alliance. The second
hypothesis that insecure therapists would report more problems in therapy was
supported, but a general measure of neuroticism was also important. Finally, there was
partial support for the third hypothesis that attachment behaviours and therapeutic
orientation would independently predict variance in the alliance and problems in
therapy. This was only true for predictions of the alliance.
Although this study indicates a relationship between self-reported attachment styles
and reported general alliance and problems in therapy, further study is needed to
establish the direction of causality and possible interactions between therapist
attachment and client characteristics. For example, Bernier and Dozier (2002) found
that when therapists and clients were more securely attached, both were more
receptive, and more skilled at providing a corrective emotional experience leading to
positive outcome. More detailed study is necessary, using both interviews and
qualitative methods, with therapist-client dyads, to explore how the attachment style of
both parties is important in the formation of the therapeutic relationship. It may be that
different factors are more or less important in the relationship between particular
clients and therapists.
Susanna Black et al.374
Acknowledgements
Research conducted in partial fulfilment of a Doctorate in Clinical Psychology degree at the
University of Sheffield. Financial support received from Community Health Sheffield Research and
Development Department.
References
Ackerman , S. J., & Hilsenroth, M. J. (2003). A review of therapist characteristics and techniques
positively impacting the therapeutic alliance. Clinical Psychology Review, 23, 1–33.
Agnew-Davies, R., Stiles, W. B., Hardy, G. E., Barkham, M., & Shapiro, D. A. (1998). Alliance
structure assessed by the agnew relationship measure (ARM). British Journal of Clinical
Psychology, 37, 155–172.
Andersson, P., & Perris, C. (2000). Attachment styles and dysfunctional assumptions in adults.
Clinical Psychology and Psychotherapy, 7, 47–53.
Arthur, A. R. (1999). Clinical psychologists, psychotherapists and orientation choice: Does
personality matter? Clinical Psychology Forum, 125, 33–37.
Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: A test of a four-
category model. Journal of personality and Social Psychology, 61, 226–244.
Bernier, A., & Dozier, M. (2002). The client-counsellor match and the corrective emotional
experience: Evidence from interpersonal and attachment research. Psychotherapy:
Theory/Research/Practice/Teaching, 39, 32–43.
Beutler, L. E. (1997). The psychotherapist as a neglected variable in psychotherapy research: An
illustration by reference to the role of therapist experience and training. Clinical Psychology:
Science and Practice, 4, 44–52.
Beutler, L. E. (2002). The Dodo bird is extinct. Clinical Psychology: Science and Practice, 9,
30–34.
Bowlby, J. (1988). A secure base: Clinical applications of attachment theory. London: Routledge.
Collins, N. (1996). Working models of attachment: Implications for explanation, emotion and
behaviour. Journal of Personality and Social Psychology, 58, 644–663.
Danov, R., & Bucci, W. (2002). Attachment processes among violence-prone minority youths and
their caretakers. NYS Psychologist, 14, 28–32.
Dozier, M., Cue, K. L., & Barnett, L. (1994). Clinicians as Caregivers: Role of attachment
organisation and treatment. Journal of Consulting and Clinical Psychology, 62, 793–800.
Dozier, M., & Tyrrell, C. (1998). The role of attachment in therapeutic relationships (Chap 9). In
Simpson, J.A., and Rholes, W.S. (Eds.) (1998). Attachment theory and close relationships.
New York: Guilford Press.
Dunkle, J. H., & Friedlander, M. L. (1996). Contribution of therapist experience and personal
characteristics to the working alliance. Journal of Counseling Psychology, 43, 456–460.
Elkin, I. (1999). A major dilemma in psychotherapy outcome research: Disentangling therapists
from therapies. Clinical Psychology: Science and Practice, 6, 10–32.
Eysenck, S. B., & Eysenck, H. J. (1969). Scores of three personality variables as a function of age,
sex, and social class. British Journal of Social and Clinical Psychology, 8, 69–76.
Attachment, alliance and problems in therapy 375
Feeney, J. A., Noller, P., & Hanrahan, M. (1994). Assessing adult attachment. In M. B. Sperling &
W. H. Berman (Eds.), Attachment in adults: Clinical and developmental perspectives
(pp. 128–152). New York: Guilford Press.
Hardy, G. E., Aldridge, J., Davidson, C., Rowe, C., Reilly, S., & Shapiro, D. A. (1999). Therapist
responsiveness to client attachment styles and issues observed in client-identified significant
events in psychodynamic interpersonal psychotherapy. Psychotherapy Research, 9, 36–53.
Hazan, C., & Shaver, P. R. (1987). Romantic love conceptualized as an attachment process. Journal
of Personality and Social Psychology, 52, 511–524.
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, 768–774.
Henry, W. P., & Strupp, H. H. (1994). The therapeutic alliance as interpersonal process (Chap 3). In
A. O. Horvath & L. S. Greenberg (Eds.), The working alliance: Theory research and practice.
New York: Wiley.
Horvath, A. O. (1994). Research on the alliance (Chap 11). In A. O. Horvath & L. S. Greenberg
(Eds.), The working alliance: Theory research and practice. New York: Wiley.
Horvath, A. O., & Greenberg, L. S. (1994). The working alliance: Theory research and practice.
New York: Wiley.
Horvath, A. O., & Symonds, B. D. (1991). Relation between working alliance and outcome in
psychotherapy: A meta-analysis. Journal of Counseling Psychology, 38, 139–149.
Kivlighan, D. M., Jr, Patton, M. J., & Foote, D. (1998). Moderating effects of client attachment on
the counselor experience-working alliance relationship. Journal of Counseling Psychology,
45, 274–278.
Krunick, J. L., Sotsky, S. M., Simmens, S., Moyer, J., Elkin, I., Watkins, J., & Pilkonis, P. A. (1996). The
role of the therapeutic alliance in psychotherapy and pharmacotherapy: Findings in the
national institute of mental health treatment for depression collaborative research program.
Journal of Consulting and Clinical Psychology, 64, 532–539.
Ligiero, D. P., & Gelso, C. J. (2002). Countertransference, attachment and the working alliance:
The therapist’s contributions. Psychotherapy: Theory/Research/Practice/Teaching, 39,
3–11.
Luborsky, L. (1994). Therapeutic alliances as predictors of psychotherapy outcomes: Factors
explaining the predictive success (Chap 2). In A. O. Horvath & L. S. Greenberg (Eds.), The
working alliance: Theory research and practice. New York: Wiley.
Luborsky, L., McLellan, A. T., Diguer, L., Woody, G., & Seligman, D. A. (1997). The psychotherapist
matters: Comparison of outcomes across twenty-two therapists and seven patient samples.
Clinical Psychology: Science and Practice, 4, 53–65.
Main, M. & Goldwyn, R. (1984). Predicting rejection of her infant from mother’s representation of
her own experience. Child Abuse and Neglect, 8, 203–217.
Mallinckrodt, B. (1991). Clients’ representations of childhood emotional bonds with parents,
social support and formation of the working alliance. Journal of Counseling Psychology, 38,
401–409.
Marmar, C. R., Gaston, L., Gallagher, D., & Thompson, L. W. (1989). Alliance and outcome in late
life depression. Journal of Nervous and Mental Disease, 177, 464–472.
Susanna Black et al.376
Martin, D. J., Garske, J. P., & Davis, M. K. (2000). Relationship of the therapeutic alliance with
outcome and other variables: a meta-analytic review. Journal Consulting and Clinical
Psychology, 68, 438–450.
Meyer, B., & Pilkonis, P. (2001). Attachment style. Psychotherapy, 38, 466–472.
Orlinsky, D. E., Grawe, K., & Parks, B. K. (1994). Process and outcome in psychotherapy-noch
einmal (Chap 8). In A. E. Bergin & S. L. Garfield (Eds.), Handbook of psychotherapy and
behavior change. New York: Wiley.
Peschken, W. E., & Johnson, M. E. (1997). Therapist and client trust in the therapeutic
relationship. Psychotherapy Research, 7, 439–447.
Raue, P. J. Goldfried, M. & Barkham, M. (1997). The therapeutic alliance in psychodynami-
c/interpersonal and cognitive behavioural therapy. Journal of Consulting ond Clinical
Psychology, 65, 582–587.
Roth, A., & Fonagy, P. (1996). What works for whom? A critical review of psychotherapy
research. London: Guilford Press.
Rubino, G., Barker, C., Roth, T., & Fearon, P. (2000). Therapist empathy and depth of interpretation
in response to potential alliance ruptures. Psychotherapy Research, 10, 408–420.
Satterfield, W. A., & Lyddon, W. J. (1995). Client attachment and perceptions of the working
alliance with counselor trainees. Journal of Counseling Psychology, 42, 187–189.
Schroder, T. A. (1999) Collaborative Research Network. International study of the development of
psychotherapists. Difficulties in Therapeutic Practice. Personal Communication.
Stein, H., Jacobs, N. J., Ferguson, K. S., Allen, J. G., & Fonagy, P. (1998). What do adult attachment
scales measure? Bulletin of the Menninger Clinic, 62, 33–82.
Stiles, W. B., Agnew-Davies, R., Hardy, G. E., Barkham, M., & Shapiro, D. A. (1998). Relations of the
alliance with psychotherapy outcome: Findings in the second Sheffield psychotherapy
project. Journal of Consulting and Clinical Psychology, 66, 791–802.
Troisi, A., D’Agenio, A., Perucchio, F., & Petti, P. (2001). Insecure attachment and alexithymia in
young men with mood symptoms. Journal of Nervous and Mental Diseases, 189, 311–316.
Tyrrell, D. L., Dozier, M., Teague, G. B., & Fallot, R. D. (1999). Effective treatment relationships for
persons with serious psychiatric disorders: The importance of attachment states of mind.
Journal of Consulting and Clinical Psychology, 67, 725–733.
Viney, L. L. (1994). Sequences of emotional distress expressed by clients and acknowledged by
therapists: Are they associated more with some therapists than others? British Journal of
Clinical Psychology, 33, 469–481.
Wheeler, S. (2000). What makes a good counsellor ? An analysis of ways in which counsellor
trainers construe good and bad counselling trainees. Counselling Psychology Quarterly, 13,
65–84.
Received 8 October 2003; revised version received 3 September 2004
Attachment, alliance and problems in therapy 377