See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/232563559

The Therapeutic Bond Scales: Psychometric Characteristics and

Relationship to Treatment Effectiveness

Article  in  Psychological Assessment · December 1989

DOI: 10.1037/1040-3590.1.4.323

CITATIONS READS

93 1,675

3 authors, including:

Stephen M Saunders David Orlinsky

Marquette University University of Chicago
63 PUBLICATIONS   1,622 CITATIONS    154 PUBLICATIONS   4,939 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

The Psychotherapeutic Profession in International Perspective View project

(1) Personal Self and Private Life of Psychotherapists (2) Development of Psychotherapists in Trainingg View project

All content following this page was uploaded by Stephen M Saunders on 27 May 2014.

The user has requested enhancement of the downloaded file.

 Psychological Assessment: Copyright 1989 by the American Psychological Association, Inc.
A Journal of Consulting and Clinical Psychology 1040-3590/89/$00.75
1989, Vol. 1, No. 4, 323-330

The Therapeutic Bond Scales: Psychometric Characteristics and

Relationship to Treatment Effectiveness
Stephen M. Saunders and Kenneth I. Howard
Northwestern University
David E. Orlinsky
University of Chicago

The Therapeutic Bond Scales assess the quality of the therapeutic relationship from the patient's
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

perspective. The therapeutic bond is composed of 3 aspects: working alliance, empathic resonance,
This document is copyrighted by the American Psychological Association or one of its allied publishers.

and mutual affirmation. Scales were developed to measure these aspects and the therapeutic bond as
a whole. The correlations between these scales and 2 measures of outcome (session quality assessed
by the patient and termination outcome evaluated by nonparticipant raters) were examined. All
scales were significantly correlated with session quality. Therapeutic bond was significantly corre-
lated with termination outcome in both a linear and a curvilinear fashion, suggesting that, at least
in the initial phase of therapy, the therapeutic bond can be too high as well as too low.

A substantial accumulation of empirical findings exists relat- assessment, by the patient, of the overall quality of the session
ing psychotherapy process variables to treatment outcome (cf. and a measure of outcome at termination, rated by nonpartici-
Orlinsky & Howard, 1986a). One of the most consistent find- pant judges.
ings in the psychotherapy research literature is that the quality
of the relationship between the patient and the therapist is a Conceptualization of the Therapeutic Bond
major determinant of psychotherapeutic effectiveness (e.g., the
therapeutic alliance literature: Alexander & Luborsky, 1986; Orlinsky and Howard's (1978, 1986a, 1987) definition of the
Marmar, Horowitz, Weiss, & Marziali, 1986). Reviews (Gur- therapeutic bond goes beyond previous conceptions of the
man, 1977; Lambert, Shapiro, & Bergin, 1986; Orlinsky & "therapeutic alliance":
Howard, 1986a; Patterson, 1984) consistently demonstrate that
The therapeutic alliance is a compact between the patient and ther-
(a) a good therapeutic relationship is at least a major contribut- apist to cooperate in performing their respective roles. The thera-
ing factor, if not a necessary and sufficient condition (Rogers, peutic bond, on the other hand, extends beyond patient and thera-
1957), for successful treatment, and (b) that the patient's per- pist roles to include certain personal qualities of the relationship
ception of the quality of the relationship is most consistently that forms—or fails to form—between the participants. (Orlinsky
& Howard, 1987, p. 10)
positively related to outcome. Yet most systems for measuring
this aspect of the therapeutic relationship are based on the non- In this conceptualization, the therapeutic bond is composed of
participant observer perspective. three dimensions.
The primary goal of the present study was to develop a reli- The first dimension is the energy that each participant invests
able measure of the quality of the therapeutic relationship, from in the process of psychotherapy or, more precisely, in his or her
the patient's perspective, based on a theoretical model of the respective role in the process. Initially it was called "reciprocal
therapeutic bond. We also investigated the relationship between role-investment" and was contrasted with "role-distancing" in
the quality of the bond, measured in this way in the early stage which the individual desires to convey the impression that the
of therapy, and two measures of therapeutic effectiveness: an therapy behaviors are "not really me." The concept was ex-
panded to include other investment dimensions, such as moti-
vation, and was relabeled the working alliance. It reflects, above
This work was partially supported by Research Grant R 01 all, an investment of the self into the appropriate role by each
MH42901 from the National Institute of Mental Health. The present participant. For example, it is necessary that the patient be will-
article is a version of a paper presented at the 17th Annual Meeting of ing to assume the role of the one who is seeking help from psy-
the Society for Psychotherapy Research, Wellesley College, Wellesley, chotherapy and that he or she be the one who will bring con-
Massachusetts, June 1986. We gratefully acknowledge the comments cerns into the session. The working alliance is determined by
and support of the Psychotherapy Research Team at the Institute of
how genuine or self-congruent the patient and the therapist per-
Psychiatry and would like to thank Bruce Briscoe for his assistance in
the computer work. We also thank the three anonymous reviewers, who
ceive their respective role behavior to be. Therapist genuineness
provided many helpful comments about the original manuscript. and credibility are examples of this dimension, as are the activ-
Correspondence concerning this article should be addressed to Ste- ity level and motivation of the patient (for further discussion of
phen M. Saunders, Department of Psychology, Northwestern Univer- this and the other two bond dimensions, see Orlinsky & How-
sity, Evanston, Illinois 60208. ard, 1978, 1986a, 1987).
323
 324 S. SAUNDERS, K. HOWARD, AND D. ORLINSKY

Empathic resonance, the second dimension, is the partici- version of the Therapy Session Report (TSR) questionnaire (Orlinsky
pants' sense of being "on the same wavelength." It is the pa- & Howard, 1966). The TSR, which requires 10 to 15 minutes to com-
tient's and the therapist's sense of understanding and being un- plete after a session, is a 145-item structured-response instrument de-
derstood. When empathic resonance is present in the relation- signed as a general survey of the experiences that patients have in indi-
vidual psychotherapy (see Orlinsky & Howard, 1986b, for a detailed
ship, it is reflected in communications that are characterized presentation of the TSR).
by mutual trust and comfortableness. Therapist empathic re- The TSR presents the patient with items organized under the follow-
ceptivity and patient expressiveness have been researched ex- ing broad questions: What did you talk about during this session? What
tensively, but the present definition focuses on the quality of did you hope or want to get out of this session? What problems or issues
the relationship rather than on attributes of individuals. Hence, were you concerned with in this session? How did you act towards your
patient empathic receptivity and therapist expressiveness are therapist during this session? How did you react to yourself during this
equally important. session? What were your feelings during this session? What did you get
The third dimension is called mutual affirmation. It is con- out of this session? Patients are also asked to indicate their motivation
ceptually close to the Rogerian concept of unconditional posi- for coming to the session and for returning the next time, to rate the
tive regard because of its emphasis on the reciprocity of caring overall quality of the session, and to assess their current level of func-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tioning.
and on the recognition that caring, although normally experi-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

The TSR also includes items concerning the patient's perception of

enced as warmth and acceptance, can on occasion lead to con- the therapist during the session: How well did your therapist seem to
frontation and challenge. Mutual affirmation is "an interest and understand you? How helpful was your therapist to you? How did your
endorsement of (the other's) well-being, motivating behavior therapist act towards you? What did your therapist seem to be feeling?
that is essentially caring in quality" (Orlinsky & Howard,
1986a, p. 348). This dimension encompasses the patient's sense
Patients
of a friendly, accepting attitude on the part of the therapist. But,
it is worth repeating, this definition emphasizes mutuality, and TSRs were completed by 113 psychotherapy outpatients who under-
the therapist will also experience such an attitude on the part of went individual treatment at Northwestern Memorial Hospital's Insti-
the patient within a good therapeutic bond (e.g., Tovian, 1977). tute of Psychiatry. This study utilized TSRs obtained from each patient
Working alliance, empathic resonance, and mutual affirma- after the third or fourth session of psychotherapy (the third session was
tion are the dimensions comprising the therapeutic bond. As preferred). The number of sessions attended by these patients ranged
may be evident from the above conceptualizations, the three from 3 to 242, and the median number of sessions was 26. Table 1 shows
the demographic characteristics of the patients. The typical patient was
dimensions are closely related. Indeed, by definition, they are a single, white female between the ages of 25 and 35 with at least some
mutually dependent. It is hard to imagine, for example, that college education. Patients were self-referred and were treated for a vari-
either of the latter two dimensions could develop in a relation- ety of mild to moderate psychological disorders. The patient sample is
ship that was not also characterized by a high level of energy reasonably representative of the psychotherapy patient population (cf.
investment on the part of both participants (i.e., a good working Taube, Kessler, & Feuerberg, 1984).
alliance). It should be recognized that the bond is not a reflec- Via screening interview, all patients were determined to be appropri-
tion of the individuals but of their interaction—it emerges in ate for psychodynamically oriented, intensive, individual therapy. Par-
the patient's and therapist's interaction during therapy, and its ticipation in the research project was voluntary, informed consent was
quality is a reflection of this interaction and not of the partici- obtained, and confidentiality of responses was ensured.
pants themselves.
Therapists
Method
The consideration of observational perspective is a crucial method- There were about 80 therapists in the Institute's Psychotherapy Pro-
ological issue. One might choose to operationalize working alliance, for gram. The majority of these were in some stage of training—psychology
example, either from the patient's perspective, from the therapist's per- practicum students, psychology interns, and psychiatry residents—al-
spective, or from the perspective of a nonparticipant observer. Although though most had had considerable additional experience. Sixty percent
each of these alternatives is valid in its own right and yields information of the therapists were psychologists, 29% were psychiatrists, and 11%
not duplicated by the others, we focused on the patient's observational were social workers. Ninety-four percent were 20-39 years of age, 54%
perspective, utilizing ratings made by patients of themselves and of their were male, and 54% were married. Eighty-three percent of the therapists
therapists. The quality of the therapeutic bond, perhaps above and be- had had personal therapy. The orientation of the Institute is psychody-
yond any other process variable, is phenomenological in nature, and namic; supervisors espouse this therapeutic approach, case presenta-
we assumed that it would be best captured from this perspective. This tions follow this model, and attempts are made to conceptualize each
assumption is supported by other studies. Gurman (1977) concluded case from this perspective.
that ratings of the therapist-offered relationship made by nonpartici-
pant judges and by the therapists themselves have not been powerful
The Therapeutic Bond Scales
predictors of outcome, whereas ratings from the patients' perspective
have consistently yielded positive findings. Similarly, Orlinsky and How- The Therapeutic Bond Scales were extracted from the TSR, which,
ard's (1986a) exhaustive review of process-outcome studies noted that as noted above, was designed as a general survey of the patients' intrases-
the perspectives of participant observers have most consistently been sion experiences. The development of the bond scales involved three
found to be related to outcome. steps. First, all TSR items were evaluated for appropriateness of inclu-
sion on one of the scales. Upon consensus of the three authors that an
Instrument item reflected an aspect of the bond, it was retained for that particular
The material for constructing measures of working alliance, em- scale. Second, correlations between each item of the three scales and
pathic resonance, and mutual affirmation was drawn from the Patient each scale were computed. Any item that correlated less with its as-
 THERAPEUTIC BOND SCALES 325
Table 1 and they were then averaged to yield a single rating of symptom change.
Individual and Social Characteristics of the Patient Sample These scores were combined to yield a single outcome index, which
ranged from 2 to 14.
Characteristic N Ratings were completed for all of the 113 patients. The two compo-
nent parts of the outcome index were highly correlated across all raters
Age (average r = .90). Interrater reliability of the index was also high (r =
18-24 years 23 20.4 .91). With a cutoff of 10 (indicating that both outcome dimensions were
25-28 years 29 25.6 rated 5 [slightly improved] by both raters) as a conservative lower limit
29-34 years 30 26.5
35-59 years 31 27.4 of treatment success, 71 (62.8%) of the patients were rated as successful
Sex (see Tablet).
Female 69 61.1
Male 44 38.9 Procedure
Race
White 101 89.4 We were interested in two general research questions: (a) What are
Nonwhite 12 10.6 the patients' perceptions of the quality of the therapeutic bond early in
Marital status
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Single 66 58.4 treatment and are these perceptions related to their judgment of the
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Married 25 22.1 quality of the session? and (b) Is the perception of the quality of the
Divorced 15 13.3 bond early in treatment predictive of eventual success or failure? Be-
Other 7 6.2 cause some researchers have reported that third-session data are partic-
Level of education ularly useful in predicting variables such as duration of treatment and
Some high school or less 4 3.5 outcome (e.g., O'Malley, Suh, & Strupp, 1983; Saltzman, Leutgart,
High school degree 13 11.5 Roth, Creaser, & Howard, 1976), only TSRs obtained in either the third
Some college 30 26.5 (preferably) or the fourth session were utilized.
College degree 40 35.4 In an attempt to make the scale scores directly analogous to the TSR
Some graduate school or more 26 23.0 items (which the patient rates on a 0-1-2 scale), we transformed each
Employment status
Employed 81 71.7 bond scale score so that it ranged from 0 to 20. Thus, a score of 0 would
Unemployed 32 28.3 correspond to no experience of that dimension, a score of 10 would
Termination Outcome Ratings indicate some experience of that aspect of the bond, and a score of 20
Considerably worse througfr no change 42 37.2 would be equivalent to "a lot of" experience of that bond dimension.
Slightly through considerably improved 71 62.8 The Global Bond scale was an average of the three subscales.
Note.N= 113.
Analyses and Results
The Therapeutic Bond Scales
signed scale than with the other two scales was dropped. Finally, an The Working Alliance scale. The patient's experience of the
alpha reliability coefficient was calculated for each scale. Any TSR item
working alliance (W\) was reflected in the endorsement of the
that lowered a scale's internal consistency was dropped from the scale.
Thus, the bond scales were developed on a conceptual basis and then items shown in Table 2. Patient motivation for coming to the
subjected to psychometric revision in order to achieve maximum reli- session and patient motivation for returning to the next session
ability. were viewed as relevant items for the WA scale. Two patient
feelings ("determined" and "serious") and three patient behav-
Therapeutic Effectiveness iors (talking, focusing on important concerns, and taking initia-
tive) were also included in this scale. Five items on this scale
Two measures of therapeutic effectiveness were used in the present were endorsed under the question stem, "This session I hoped
study. Session quality was assessed using the patient's overall assessment or wanted to." The five can be categorized into three identifiable
of the session just completed. The first item on the TSR asks the patient therapeutic realizations that the patient hoped to attain: insight
to rate this on a 7-point scale ranging from perfect to very poor. With ("understand the reasons behind my feelings and behavior" and
respect to these scores, 79.7% of the patients indicated that their session
"find out what my feelings really are and what I really want");
was pretty good or better (very good, excellent, or perfect); 20.3% rated
their sessions a&fair or poorer (pretty poor or very poor). catharsis ("get a chance to let go and get things off my chest"
Termination outcome scores were calculated from ratings of the and "get relief from tensions or unpleasant feelings"); and
closed clinic files using the evaluation method developed by Tovian working on a particular problem ("work out a particular prob-
(1977). To be usable, clinic files had to contain, at the minimum, infor- lem that's been bothering me"). Only three items in the Patient
mation about the intake interview and disposition at the end of treat- version of the TSR reflected the therapist's state of role-invest-
ment (or treatment summary). Most files also contained therapist notes ment. All of these were therapist affects perceived by the patient.
concerning individual sessions. Each file was examined independently The 15 items reflecting the patient's experience of the work-
by two judges, graduate students in clinical psychology. Each judge ing alliance are shown in Table 2, which also shows the item-
made two ratings: the patient's overall condition at closing, and symp- total correlations between the individual items and the \\A
tom change, that is, the extent of change in the presenting problem(s).
scale. The scale had a reliability (alpha coefficient) of .72 and
Each of these ratings was scored on a 7-point scale ranging from consid-
erably worse (1) through no change (4) to considerably improved (7). an average score (endorsement) of 13.3 (SD = 2.8), equivalent
Raters received no special training with the exception of being told to to between some and a lot o/experience of it. The range of scores
emphasize that each presenting problem be rated separately. When was 5 to 19.
there was more than one presenting problem, each was rated separately The Empathic Resonance scale. Empathic resonance (ER)
 326 S. SAUNDERS, K. HOWARD, AND D. ORLINSKY

Table 2
The 15 Items Comprising the Working Alliance (WA) Scale
Corrected item-total
correlations

Item WA scale Global bond

1. How did you feel about coming to therapy this session? .46 .43
2. To what extent are you looking forward to your next session? .39 .39
This session I hoped or wanted to:
3. Get a chance to let go and get things off my chest. .38 .31
4. Get relief from tensions or unpleasant feelings. .38 .29
5. Understand the reasons behind my feelings and behavior. .33 .16
6. Find out what my feelings really are, and what I really want. .38 .28
7. Work out a particular problem that's been bothering me. .37 .22
During this session how much did you feel:
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

8. Determined .39 .48

This document is copyrighted by the American Psychological Association or one of its allied publishers.

9. Serious .32 .17

During this session, how much:
10. Were you able to focus on what was of real concern to you? .39 .44
11. Did you take initiative in bringing up the subjects that were talked
about? .32 .38
12. Did you talk? .24 .25
During this session my therapist seemed to feel:
13. Thoughtful .24 .38
14. Involved .16 .40
15. Confident .20 .44

Note. Readers interested in the conversion of these items into corresponding Therapy Session Report items
should contact Stephen M. Saunders.

refers to a quality of communication between patient and thera- from the patient's perceptions of interpersonal behaviors and
pist that depends on their compatibility in range and style of affective states and are presented in Table 4. Both patient and
expressiveness and understanding. The 17 items defining this therapist affirmation were represented. MA was reflected in re-
scale are listed in Table 3. One TSR item reflecting the patient's sponses to three questions concerning, generally, the patient's
receptivity during the session was the question "Were you atten- friendly, negative, and critical behavior toward the therapist.
tive to what your therapist was trying to get across to you?" Six Seven patient feelings also reflected affirmation toward the ther-
other items focused on aspects of the patient's expressiveness, apist.
most directly the question "How much did you talk about what The therapist's affirmation toward the patient was reflected
you were feeling?" but also responses to questions about the in the following items: "Was your therapist warm and friendly
patient's feelings such as not feeling "inhibited" and "with- towards you?" and "Did your therapist tend to accept or agree
drawn." Indirectly, patients were judged as likely to have been with your ideas and point of view?" The patient's perceptions
more expressive in their sessions when they reported feeling of various therapist "affirming" feelings were also included in
"confident" and not feeling "embarrassed," "frustrated," and this scale. Indirectly, therapist affirmation was reflected by the
other similar affects. patient's feeling "accepted," "likeable" and "pleased." Finally,
The patient's view of the therapist's empathy was reflected in
one item—the patient's satisfaction with having "more of a per-
six TSR items. Two of these were responses to the questions
son-to-person relationship" with the therapist—was taken as an
"Was your therapist attentive to what you were trying to get
indicator of affirmation on the part of both participants.
across?" and "How well did your therapist seem to understand
what you were feeling and thinking?" The remaining three The items comprising the MA scale and the item-total corre-
items were patient perceptions of the therapist's affect during lations between them and the MA scale are shown in Table 4.
the session. This scale had the highest reliability of the three (a = .87). En-
Item-total correlations between these items and the ER scale dorsement of the MA scale was considerably lower than that
are listed in Table 3. This bond subscale's reliability was .77. of the other two scales: The average score was 8.0 (SD = 3.4),
Average endorsement of the ER scale was 14.5 (SD = 2.8), equivalent to some experience of the dimension, and the range
which is also between "some" and "a lot of" experience of the of scores was 0 to 17.
dimension, and the range was 6 to 19. The Global Bond scale. The Global Bond scale is a composite
The Mutual Affirmation scale. The third constituent facet in of the three subscales. Calculating its internal reliability as a
our conception of the therapeutic bond is mutual affirmation function of the relationship between the three subscales (i.e.,
(MA), reflecting the care, respect, and commitment to the other N = 3 in the formula), we obtained an alpha of .62. The item-
person's welfare that the patient and the therapist may evoke in total correlations of each item with the Global Bond scale are
and feel for one another. The 18 items defining MA were drawn shown in Tables 2, 3, and 4. Endorsement of the Global Bond
 THERAPEUTIC BOND SCALES 327

Table 3
The 17 Items Comprising the Empathic Resonance (ER) Scale
Corrected item-total
correlations
Item ER scale Global bond
1. How well did your therapist seem to understand what you were feeling
and thinking? .36 .54
During this session I felt:
2. Frustrated (-)" .55 .32
3. Impatient (-) .52 .34
4. Withdrawn (-) .44 .29
5. Confused (-) .41 .20
6. Cautious (-) .41 .24
7. Strange (-) .40 .28
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

8. Embarrassed (-) .34 .23

This document is copyrighted by the American Psychological Association or one of its allied publishers.

9. Confident .33 .37

10. Inhibited (-) .32 .18
11. Helpless (-) .28 .16
During this session, how much:
12. Was your therapist attentive to what you were trying to get across? .35 .44
13. Were you attentive to what your therapist was trying to get across to you? .26 .40
14. Did you talk about what you were feeling? .22 .42
During this session my therapist seemed to feel:
15. Bored (-) .33 .34
16. Interested .28 .44
17. Alert .21 .49
" These items were reversed so that low endorsement corresponded to a better bond score.

Table 4
The 18 Items Comprising the Mutual Affirmation (MA) Scale
Corrected item-total
correlations
Item MA scale Global bond
This session I felt:
1. Grateful .61 .61
2. Accepted .59 .59
3. Hopeful .56 .58
4. Close .55 .52
5. Pleased .51 .56
6. Likeable .46 .49
7. Affectionate .46 .37
I feel that I got:
8. More of a person-to-person relationship with my therapist. .62 .57
During this session, how much:
9. Was your therapist friendly and warm towards you? .54 .52
10. Did you tend to accept or agree with what your therapist said? .42 .57
11. Did your therapist tend to accept or agree with your ideas and point
ofview? .37 .34
12. Friendliness or respect did you show your therapist? .31 .41
During this session, my therapist seemed to feel:
13. Pleased .61 .60
14. Close .57 .53
15. Cheerful .52 .49
16. Affectionate .49 .35
17. Optimistic .32 .35
18. Attracted .30 .31
 328 S. SAUNDERS, K. HOWARD, AND D. ORLINSKY

Table 5 and then the square of the Global Bond score was entered. The
Pearson Product-Moment Correlations Among resulting multiple correlation was .27 (p < .05), and both the
the Bond Dimensions and Between the Bond linear and curvilinear components were statistically significant.
Dimensions and Outcome Measures The beta for the square of the Global Bond score was negative,
indicating that the lower and the upper ranges of the bond scores
Dimensions and outcome were associated with poorer outcome.
measures WA ER MA Global bond
Bond dimensions Discussion
Working Alliance (WA) (.72)
Empathic Resonance (ER) .09 (.77) The results indicate that the TSR provides a psychometrically
Mutual Affirmation (MA) .51** .42** (.87)
Global Bond .70** .65** .88** (.62) sound basis for assessing the theoretical qualities of the thera-
Outcome measures peutic bond. The relative ease with which this instrument is
Session quality .34** .51** .50** .60** administered, completed, and analyzed underscores its poten-
Termination outcome3 .11 .16 .16 .19* tial usefulness.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

The patients in this study seemed to experience mutual

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Note. N= 113. The diagonal entries are the alphas of the scales.
a
Bonferroni correction to control for experiment-wise error rate would affirmation as being related to both of the other scales, whereas
requires = .0125. working alliance and empathic resonance were experienced as
*p<.05. **p<.001. separate entities. We assumed that the three dimensions would
be related, because they assess the same construct, but we hoped
that they would not be so highly correlated as to preclude the
scale indicated that the average score was 11.9 (SD = 2.3), and establishment of separate subscales. The ER and WA scales
the range of scores was 6 to 18. shared little variance (less than 1%), which suggests that they
are distinct, but both were highly correlated with the MA scale.
Correlations Among Bond Scores and Therapeutic Together, they accounted for over 40% of the latter scale's vari-
Effectiveness ance, which might indicate that the MA scale is somewhat re-
dundant. It might be argued that the MA scale should be col-
Table 5 shows the correlations among the three scales and the lapsed into these two and that WA and ER are truly distinct
Global Bond scale. The highest correlation was between the WA parts of the bond. This argument fits well with various other
and MA scales, and they shared about 26% of their variance. conceptualizations and research results. Bordin's (1979) con-
The MA and ER scales were also significantly correlated, shar- ceptualization of the therapeutic alliance states that a good rela-
ing 17% of their variance. The WA and ER scales shared less tional bond is necessary for, and combines with agreement on
than 1% of their variance. and implementation of particular tasks and goals of therapy to
We next investigated the relationship between the patients' become, the therapeutic alliance. (Bordin's conceptualization
bond scale scores and two measures of effectiveness. First, we also recognizes the importance of both patient and therapist
wanted to determine if the patient's perception of the quality of influences on the bond.) Translated into the terminology of the
the therapeutic bond was related to a measure of session quality generic model, a good therapeutic bond is partly personal in-
operationalized as the patient's rating of the overall quality of vestment of energy (i.e., the working alliance = tasks and goals)
the session. Table 5 shows the correlations between each of the and partly relationship variables (i.e., empathic resonance =
four scales (WA, ER, MA, and Global Bond) and session quality. relational bond). Horvath and Greenberg (1986) developed the
All of the correlations between session quality and the bond Working Alliance Inventory to test Bordin's assertions and have
measures were significant at beyond the .001 level. The ER scale shown that it is predictive of patient ratings of treatment effec-
had the highest correlation of the three subscales, but the Global tiveness.
Bond scale's correlation was highest of all. Patients generally reported experiencing the bond scales at
As stated, the two parts (i.e., patient's overall condition at high levels (the lowest endorsement being for MA at around
closing and symptom change) comprising the termination out- some experience of it). This could indicate a genuine experience
come ratings were highly correlated across all raters, and inter- of the offered relationship, or it could mean a reluctance to be
rater reliability was high. Table 5 shows the correlations be- critical at such an early stage of treatment. It is likely that expec-
tween termination outcome ratings and the bond scales. The tation and an early form of transference play a role in the pa-
correlations between termination outcome and the various tient's experience of the early-therapy bond because there is lit-
bond subscales did not achieve significance. The correlation be- tle actual information with which the patient can realistically
tween termination outcome and the Global Bond score was sig- evaluate it. Practitioners try to remain aware of transference
nificant, however (r = . 19, p < .05). and expectation issues and the problems that they sometimes
We conducted an additional analysis to investigate the possi- pose for therapy. The bond scales might prove useful as an indi-
bility of a curvilinear aspect to the relationship between the cator of these phenomena and may help to identify patients who
therapeutic bond and treatment effectiveness. Post hoc, we the- have excessive or unreasonable needs for and expectations from
orized that there might be an optimal quality level of the thera- the therapeutic relationship. For example, Daskovsky (1988)
peutic bond and that perceiving its quality as either too low or used these scales to investigate the relationship between pre-
too high would predict poorer outcome. In a hierarchical multi- therapy assets of patients (specifically, a measure of object relat-
ple regression analysis, the Global Bond score was entered first, edness) and patients' capacity to enter into the therapeutic rela-
 THERAPEUTIC BOND SCALES 329

tionship (early-session Global Bond scores). Patients' level of used to investigate the congruence or association between the
object relatedness was inversely correlated with the therapeutic patient's and the therapist's perception of the therapeutic bond.
bond. These results were interpreted to mean that patients with Another important distinction of the present study concerns
better object relatedness were more able to be critical or, per- the length of therapy and, consequently, the unusually long pe-
haps, realistic about the quality of the bond in this early stage riod of time between the assessment of process and outcome.
of therapy. For example, the Penn Helping Alliance Scales studies (which
All of the scales (including the Global Bond scale) were highly reported a correlation of .58 between process and outcome
correlated with session quality. The strength of these corre- measures) were conducted on therapy that lasted between 3 and
lations with session quality was probably somewhat inflated be- 24 sessions, with a mean of about 12 sessions (Luborsky et al.,
cause both the session bond and session quality measures were 1985). In comparison, the range of sessions for patients in our
derived from the same instrument. Even so, the result is consis- sample was between 3 and 242, and over half of the patients
tent with other research reporting a significant correlation be- attended more than 26. Other studies that have reported high
tween the quality of the patient-therapist relationship and the correlations between the quality of the relationship and out-
rated quality of the session as a whole (e.g., Marziali, Marmar, & come also had relatively short therapy durations; for example,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Krupnick, 1981;O'Malleyetal., 1983;Sachs, 1983). In general, Marziali et al. (1981) studied brief (12-session) dynamic ther-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

patients who perceived a better quality bond rated the quality apy, and the Vanderbilt Psychotherapy Research Project pa-
of the session to be better. tients attended at most 25 sessions (Suh et al., 1986).
We based the present analyses on data drawn from the third It seems possible and perhaps likely that the various compo-
or fourth session in order to see whether early manifestations of nents of the therapeutic bond develop at different rates during
the bond were predictive of treatment effectiveness. Although therapy. Both the WA and the ER scales were endorsed at fairly
the therapeutic bond subscales were not predictive of treatment high levels: The average patient experience of these dimensions
outcome, the composite of these subscales, the Global Bond at this early stage was about halfway between some and a lot. It
scale, was significantly correlated with treatment outcome. In appears that patients generally felt that they and their therapists
addition, the hierarchical multiple regression analysis suggested were genuinely invested in the treatment process and that they
that there is a curvilinear component to the bond: Both the expressed themselves well and were well understood by their
lower and upper bounds of the Global Bond scale were associ- therapists. Relative to the other subscales, MA was endorsed
ated with poorer outcome. This is an important finding for both at the lowest level, slightly less than some. Theoretically, it is
researchers and practitioners. It suggests that patients can over- reasonable that mutual affirmation would develop gradually in
estimate the quality of the bond early in treatment and, presum- the dyad only after association between patient and therapist
ably, be disappointed with its development and consequently had begun to prove fruitful. Similar to what Frieswyk et al.
have poorer outcomes. Again, the Daskovsky (1988) finding (1986) suggested, the change in mutual affirmation across ther-
apy might be a useful indicator of its effectiveness.
that patients' level of object relatedness was inversely correlated
with the therapeutic bond is relevant. It is reasonable to suggest,
References
based on clinical experience and these findings, that some pa-
tients enter therapy with excessive "need" of the therapeutic Alexander, L. B., & Luborsky, L. (1986). The Penn Helping Alliance
bond and that this need may be detrimental to the progress of Scales. In L. S. Greenberg & W. M. Pinsof (Eds.), The psychothera-
therapy. peutic process: A research handbook (pp. 325-366). New York: Guil-
ford Press.
The results of our study are consistent with those of other Bordin, E. S. (1979). The generalizability of the psychoanalytic concept
studies that have shown a positive correlation between the qual- of the working alliance. Psychotherapy: Theory, Research, and Prac-
ity of the bond and treatment effectiveness. The size of the asso- tice, 16, 252-260.
ciation was not as large as that reported in other studies, how- Daskovsky, D. (1988). The patient's assets for psychotherapy. Unpub-
ever (e.g., Luborsky, McLellan, Woody, O'Brien, & Auerbach, lished doctoral dissertation, Northwestern University.
1985; Suh, Strupp, & O'Malley, 1986). It is possible that the Frieswyk, S. H., Allen, J. G., Colson, D. B., Coyne, L., Gabbard, G. Q,
relatively low correlation reported in this study was the result Horwitz, L., & Newsom, G. (1986). Therapeutic alliance: Its place as
a process and outcome variable in dynamic psychotherapy research.
of analyzing the correlation between a process instrument rated
Journal of Consulting and Clinical Psychology, 54, 32-39.
from the patients' perspective and an outcome instrument rated Gurman, A. S. (1977). The patient's perception of the therapeutic rela-
by nonparticipant judges. Orlinsky and Howard (1986a) re- tionship. In A. S. Gurman & A. M. Razin (Eds.), Effective psychother-
ported that congruent process-outcome perspectives consis- apy: A handbook of research (pp. 503-543). New York: Pergamon.
tently yielded the greatest proportion of positive findings. The Horvath, A. Q, & Greenberg, L. (1986). The development of the Work-
highest percentage of significant findings occurred in the sub- ing Alliance Inventory. In L. S. Greenberg & W. M. Pinsof (Eds.),
category that used therapists' process reports and therapists' The psychotherapeutic process: A research handbook (pp. 529-556).
outcome evaluations, followed by studies based on the patients' New York: Guilford Press.
Lambert, M. J., Shapiro, D. A., & Bergin, A. E. (1986). The effective-
perspective of both process and outcome (Orlinsky & Howard,
ness of psychotherapy. In S. L. Garfield & A. E. Bergin (Eds.), Hand-
1986a, p. 371). The relationship between the various perspec- book of psychotherapy and behavior change (3rd ed., pp. 157-211).
tives that are utilizing the same instrument has yet to be delin- New York: Wiley.
eated adequately, and this seems an important area on which to Luborsky, L., McLellan, X, Woody, G., O'Brien, C, & Auerbach, A.
concentrate future research effort (cf. Alexander & Luborsky, (1985). Therapist success and its determinants. Archives of General
1986). For example, the Therapist version of the TSR might be Psychiatry, 42,602-611.
 330 S. SAUNDERS, K. HOWARD, AND D. ORLINSKY

Marmar, C. R-, Horowitz, M. J., Weiss, D. S., & Marziali, E. (1986). Patterson, C. H. (1984). Empathy, warmth and genuineness in psycho-
The development of the Therapeutic Alliance Rating System. In L. S. therapy: A review of reviews. Psychotherapy, 21, 431-438.
Greenberg & W. M. Pinsof (Eds.), The psychotherapeutic process: A Rogers, C. R. (1957). The necessary and sufficient condition of thera-
research handbook (pp. 367-390). New York: Guilford Press. peutic personality change. Journal of Consulting Psychology, 21, 95-
Marziali, E., Marmar, C., & Krupnick, J. (1981). Therapeutic alliance 103.
scales: Development and relationship to psychotherapy outcome. Sachs, J. S. (1983). Negative factors in brief psychotherapy: An empiri-
American Journal of Psychiatry, 138, 361-364. cal assessment. Journal of Consulting and Clinical Psychology, 51,
O'Malley, S. S., Suh, C. S., & Strupp, H. H. (1983). The Vanderbilt 557-564.
Psychotherapy Process Scale: A report on the scale development and Saltzman, C., Leutgert, M. M., Roth, C. H., Creaser, J., & Howard, L.
a process-outcome study. Journal of Consulting and Clinical Psychol- (1976). Formation of a therapeutic relationship: Experiences during
ogy, 57,581-586. the initial phase of psychotherapy as predictors of treatment duration
Orlinsky, D. E., & Howard, K. I. (1966). Psychotherapy Session Report, and outcome. Journal of Consulting and Clinical Psychology, 44,
Form P and Form T. Chicago: Institute of Juvenile Research. 546-555.
Orlinsky, D. E., & Howard, K. I. (1978). The relation of process to out- Suh, C. S., Strupp, H. H., & O'Malley, S. S. (1986). The Vanderbilt
come in psychotherapy. In S. L. Garfield & A. E. Bergin (Eds.), Hand- Process Measures: The Psychotherapy Process Scale (VPPS) and the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

book of psychotherapy and behavior change (2nd ed., pp. 283-329). Negative Indicators Scale. In L. S. Greenberg & W. M. Pinsof (Eds.),
This document is copyrighted by the American Psychological Association or one of its allied publishers.

New York: Wiley. The psychotherapeutic process: A research handbook (pp. 285-323).
New York: Guilford Press.
Orlinsky, D. E., & Howard, K. I. (1986a). Process and outcome in psy-
Taube, C. A., Kessler, L., & Feuerberg, M. (1984). Utilization and ex-
chotherapy. In S. L. Garfield & A. E. Bergin (Eds.), Handbook of psy-
penditures for ambulatory mental health care during 1980. In Na-
chotherapy and behavior change (3rd ed., pp. 311-381). New York:
tional medical care utilization and expenditure survey (Data Report
Wiley.
5). Washington, DC: U.S. Department of Health and Human Ser-
Orlinsky, D. E., & Howard, K. E. (1986b). The psychological interior
vices.
of psychotherapy: Explorations with the Therapy Session Report Tovian, S. M. (1977). Patient experiences and psychotherapy outcome.
questionnaires. In L. S. Greenberg & W. M. Pinsof (Eds.), The psy- Unpublished doctoral dissertation, Northwestern University.
chotherapeutic process: A research handbook (pp. 477-502). New
York: Guilford Press. Received December 13, 1988
Orlinsky, D. E., & Howard, K. E. (1987). A generic model of psycho- Revision received May 19, 1989
therapy. Journal of Integrative and Eclectic Psychotherapy, 6, 6-27. Accepted June 1, 1989 •

View publication stats