The role of the therapeutic alliance in Psychotherapy

Abstract

The article traces the development of the concept of the therapeutic working alliance from its psychodynamic origins to current pantheoretical formulations. Research on the alliance is reviewed under four headings: the relation between a positive alliance and success in therapy, the path of the alliance over time, the examination of variables that predispose individuals to develop a strong alliance, and the exploration of the in-therapy factors that influence the development of a positive alliance. Important areas for further research are also noted.

Full text

Journal
of
Consulting
and
Clinical
Psychology
1993,
Vol.
61,
No.
4,561-573
Copyright
1993
by the
American Psychological
Association,
Inc.
0022-006X/93/S3.00
The
Role
of the
Therapeutic Alliance
in
Psychotherapy
Adam
O.
Horvath
and
Lester
Luborsky
The
article
traces
the
development
of the
concept
of the
therapeutic
working
alliance
from
its
psychodynamic
origins
to
current
pantheoretical
formulations.
Research
on the
alliance
is re-
viewed
under
four
headings:
the
relation
between
a
positive
alliance
and
success
in
therapy,
the
path
of
the
alliance
over
time,
the
examination
of
variables
that
predispose
individuals
to
develop
a
strong
alliance,
and the
exploration
of the
in-therapy
factors
that
influence
the
development
of a
positive
alliance.
Important
areas
for
further
research
are
also
noted.
Over
the
last
3
decades,
the
dynamic concept
of
the
therapeu-
tic
alliance (Freud, 1912; Greenson, 1967; Sterba, 1934; Zetzel,
1956)
has
taken root
in
other models
of
psychotherapy,
leading
to
several related explanations
of the
role
of the
alliance within
the
therapeutic process (Bordin,
1976;
Bowlby,
1988; Luborsky,
1976;
Rogers, 1957; Strong, 1968).
The
central
aim of
this article
was
to
review
the
theoretical issues
and the
research results
relevant
to
each
of
four
aspects
of the
role
of the
alliance
in
psychotherapy:
The
relation between
a
positive alliance
and
success
in
therapy;
the
path
of the
alliance over time;
the
vari-
ables
that predispose individuals
to
develop
a
strong alliance;
and the
in-therapy factors that
influence
the
development
of a
positive
alliance.
The
terms
therapeutic
alliance,
working
alliance,
and
helping
alliance
have been used
in the
past
to
refer
to
specific
aspects
of
the
alliance
or as
synonyms
for the
alliance
as a
whole. Because
the
use of
these terms
has not
been consistent,
the
term
the
alliance
is
used here generically
to
refer
to all of the
aforemen-
tioned
constructs, unless otherwise specified.
Development
of the
Concept
of the
Alliance
Psychodynamic
Origins
In
The
Dynamics
ofTransference,
Freud
(1912)
discussed
the
value
of the
analyst's maintaining "serious
interest"
in and
"sympathetic understanding"
of the
client
to
permit
the
healthy
part
of the
client's self
to
form
a
positive attachment
to
the
analyst. Later, Freud
(1913)
speculated that,
as a
result
of
the
supportive attitude
of the
analyst,
the
patient would uncon-
sciously link
the
therapist with
the
"images
of
people
by
whom
he
was
accustomed
to be
treated
by
affection"
(pp.
139-140).
Two
slightly
different
conceptualizations
of the
therapist-
Adam
O.
Horvath,
Faculty
of
Education
and
Department
of
Psy-
chology,
Simon
Eraser
University,
Burnaby,
British
Columbia,
Canada;
Lester
Luborsky,
Department
of
Psychiatry,
University
of
Pennsyl-
vania.
Some
of the
material
presented
in
this
article
will
appear
as a
chapter
by A. O.
Horvath,
L.
Gaston,
and L.
Luborsky
in the
Handbook
of
Psychodynamic
Psychotherapy:
Theory
and
Research.
Correspondence
concerning
this
article
should
be
addressed
to
Adam
O.
Horvath,
Faculty
of
Education,
Simon
Fraser
University,
Bur-
naby,
British
Columbia
V5A
1S6.
client
relationship
are
evident
in
Freud's writings:
In his
early
papers, this attachment
is
described
as a
form
of
beneficial
(positive)
transference. This
form
of
transference
had the
effect
of
"[the
client's] clothing
the
doctor
with
authority" (Freud,
1913,
pp.
99-108).
Freud
saw
this aspect
of the
relationship
as
the
source
of the
client's "belief
in his
[analyst's] communica-
tions
and
explanations" (Freud, 1913,
pp.
99-108).
Positive
transference
involves
a
distortion
of the
real relationship and,
like
its
negative counterpart, needs
to be to
interpreted
by the
analyst. However,
in his
later papers, Freud appeared
to
have
modified
this positive transference
view
of the
therapeutic
rela-
tionship
to
include
the
possibility
of a
beneficial
client-thera-
pist attachment grounded
in
reality.
The
implication
of
this
later perspective
is
that, although interpretation
of the
client's
projections
or
unresolved prior experiences
is
central
to
ther-
apy,
the
ability
of the
intact portion
of the
client's conscious,
reality-based
self
to
develop
a
covenant with
the
"real"
thera-
pist
makes
it
possible
to
undertake
the
task
of
healing.
Greenson
(1965)
elaborated
on
this
concept
of a
reality-based
collaboration
between therapist
and
client
and
coined
the
term
working
alliance.
He
proposed
a
model that
has
three compo-
nents:
transference,
the
working alliance,
and the
real relation-
ship
(Gelso
&
Carter, 1985). Zetzel
(1956)
clarified
some
of the
distinctions between transference
and
alliance, suggesting that,
in
effect,
the
nonneurotic component
of the
client-therapist
relationship (the alliance) permits
the
client
to
step back
and
use
the
therapist's
interpretations
to
better distinguish between
remnants
of
past
relationships
and the
real association between
himself
or
herself
and the
therapist.
She
argued that,
in
success-
ful
analysis,
the
client oscillates between
periods
when
the
rela-
tionship
is
dominated
by
transference
and
periods dominated
by
the
working alliance.
A
related conceptualization
of the
therapeutic alliance
was
offered
by
Bibring (1937),
who
suggested that
the
therapeutic
situation represents
a
"new-object relationship." This position
was
further
developed
by
Gitleson
(1962)
and
Horwitz
(1974)
and, more
recently,
by
Bowlby
(1988).
In
essence,
the
object-re-
lationists propose that
the
client,
as
part
of the
therapy process,
develops
the
capacity
to
form
a
positive, need-gratifying rela-
tionship with
the
therapist. This attachment
is
qualitatively
dif-
ferent
from
those based
on
early childhood experiences
and
thus
represents
a new
class
of
events.
The
therapist's task
is to
maintain
a
positive, reality-grounded stance toward
the
patient
561

562
ADAM
O.
HORVATH
AND
LESTER
LUBORSKY
and to
provide
an
opportunity
for the
client
to
reflect
on the
discrepancies between
the
distorted
and
reality-based aspects
of
the
relationship (Frieswyk
et
al.,
in
press).
These theorists argue that
the
alliance
and
transference
are
distinct
constructs,
but
there
are
others
who
maintain
that
all
aspects
of the
therapist-client relationship
are
manifestations
of
the
transference neurosis
and
should
be
interpreted
as
such.
They
claim that
the
client's alliance
fulfills
an
unconscious
wish
to
gain
the
therapist's approval
as a
parental
figure or,
alternatively,
the
apparent alignment
or
collaboration
is
actu-
ally
a
form
of
covert competition. This logic leads some
(e.g.,
Brenner,
1979;
Curtis, 1979)
to
conclude that using
the
alliance
construct
may be
counterproductive because
it
will
result
in a
watering
down
of the
central thrust
of the
analytic
work—the
interpretation
of
transference.
Whether
transference distortions
are
part
and
parcel
of the
alliance
has
been
one of the
ongoing controversies
in
dynamic
therapy.
The
central practical issue
is the
degree
to
which
the
alliance
is
dependent
on the
here-and-now motivation, skill,
and
"fit"
of the
client
and the
therapist,
as
opposed
to the
concept that
the
quality
of the
therapeutic relationship
is (at
least initially) predestined
by the
client's unconscious projec-
tions
based
on
past experiences
(Gelso
&
Carter, 1985; Gut-
freund,
1992).
The
alliance-as-transference
perspective implies that emo-
tions
and
thoughts associated with unresolved relationships
with significant
others
are
bound
to be
displaced
(transferred)
onto
the
relationship with
the
therapist. Thus,
the
relationship
"entails
a
misperception
or
misinterpretation
of the
therapist
.
. .
[and]
is an
unreal
relationship
in
this sense" (Gelso
&
Carter,
1985,
p.
170).
Accordingly, authors with this perspective
(e.g.,
Brenner, 1979)
see the
alliance
as
aim-softened transfer-
ence and, therefore,
not a
relationship properly speaking.
Others
argue that
all
relationships,
in the
broadest sense,
are
prejudiced
by
previous interpersonal experiences
and
that
the
alliance
is not a
separate
or
distinct variable
but an
alternative
perspective
on the
same phenomenon
as the
positive transfer-
ence
(Hatcher, 1990). Proponents
of
this view tend
to
maintain
that
the
therapeutic alliance
is
based
on the
quality
of the
current interpersonal synergy
of
therapist
and
client
yet it is
also
a
reflection
of the
client's previous unresolved relation-
ships.
The key
issue debated among researchers sympathetic
to
this position (Piper, Azim, Joyce,
&
McCallum,
199la)
is the
degree
to
which
the
client's past relationships
influence
the
alliance.
More
recently,
the
impact
of
countertransference (i.e.,
the in-
fluence
of
the
therapist's early relationships
or
relational capaci-
ties)
on the
formation
of a
viable alliance
has
also been raised.
The
theoretical
and
empirical parameters
of
this question
are
currently
not
clear.
If
countertransference
is a
factor influenc-
ing
the
development
of
the
alliance,
is it
most
usefully
conceptu-
alized
as a
therapist variable (i.e.,
as a
relational capacity that
is
somewhat constant across clients)
or
with
a
more
traditional
definition
of
countertransference (i.e.,
as an
interactional phe-
nomenon)?
If the
latter
is the
correct
view,
then
its
influence
on
the
alliance will
be
apparent only
if
specific
client behaviors
or
characteristics
are
present.
In
either case,
a
clear theoretical
model
is
needed
to
relate countertransference
to the
develop-
ment
of the
alliance
and
thus facilitate
the
empirical investiga-
tion
of the
possible impact
of
therapist distortions. Research
on
the
role
of
therapist distortions
is in its
early stages; data
currently
available
on
this topic
is
presented later
in
this article
(see Client
and
Therapist Factors
Influencing
the
Development
of
the
Alliance).
Despite
the
differences
of
view
on the
role
of
transference,
it
seems that
we are
witnessing
the
development
of a
growing
consensus among dynamic theorists: Most
feel
that
a
compre-
hensive
definition
of the
alliance needs
to
take account
of the
influence
of
past experiences
(we
include transference under
this heading) and,
at the
same time, delineate
the
alliance
as a
distinct aspect
of the
current relationship (Gaston,
1990).
Conceptual work
on the
alliance
as an
independent
(i.e.,
non-
transference)
factor
has
taken place both within
and
outside
the
analytic
framework.
The
main contributions
to
this position
within
the
psychodynamic perspective were reviewed earlier.
To put the
nonpsychodynamic position
in
proper light,
we
will
first
present
a
brief review
of
some alternative conceptualiza-
tions
of the
therapeutic relationship.
The
Client-Centered Concept
of
the
Therapeutic
Relationship
Rogers
(1951,1957)
asserted that
the
therapist's
ability
to be
empathic
and
congruent
and to
accept
the
client uncondition-
ally
were
not
only essential
but
sufficient
conditions
for
thera-
peutic gains. Although client-centered theory
is
often regarded
as
a
quintessentially
humanistic
position
emphasizing
the "I
and
Thou" aspect
of
therapy, Rogers' propositions
do not ad-
dress
the
possibility
of
variations
in the
clients' ability
and mo-
tivation
to
respond
to the
offer
of
such
a
relationship. There
is a
presumption
of a
fated
response
to the
correct
attitude
of the
therapist.
During
the
past
3
decades, many investigations were
launched
to
explore
the
effects
of the
Rogerian
Therapist-Of-
fered
Conditions (TOC).
The
initial
findings
were strongly sup-
portive
of the
original hypotheses: Therapists
who
provided
high
levels
of
TOCs were more successful than those
who
pro-
vided
less
of
these conditions (Barrett-Lennard, 1985; Rogers,
Gendlin, Kiesler,
&
Truax,
1967).
Some
of the
later reviews
of
the
research results (e.g., Mitchell, Bozart,
&
Krauft,
1977;
Or-
linsky
&
Howard, 1986), however, noted that
the
relation
be-
tween
the
TOCs
and
outcome were
not
uniform across
differ-
ent
therapy modalities. Moreover,
the
majority
of findings
indi-
cate
that
it is the
client's perception
of the
therapist
as an
empathic individual, rather than
the
actual therapist behavior,
that yielded
the
most robust correlation with outcome.
Although
there
are
clear theoretical
differences
between
the
TOC and the
alliance, research results indicate
a
moderate-to-
strong correlation between client-perceived empathy
and
some
aspects
of the
alliance, particularly
in
early stages
of
therapy
(Greenberg
&
Adler,
1989;
Horvath,
1981;
Jones,
1988;
Moseley,
1983).
However,
in
each
of
these investigations,
the
alliance
was
more predictive
of
outcome
than
empathy.
In a
more recent
study,
the
relation between
the
alliance
and all
four
TOC di-
mensions were examined during
the
mature
phase
(Session
20)
of
therapy (Salvio, Beutler, Wood,
&
Engle,
1992).
The
results
of
this investigation confirmed
the
close association among
the
TOC and
alliance components.
The
researchers used these data
to
develop
a
composite Strength
of
Therapeutic
Alliance score
using
principle-component analysis.
The
factor loading
on
this

SPECIAL
SECTION: THERAPEUTIC ALLIANCE
IN
PSYCHOTHERAPY
563
index
for the
four
TOC
components (regard, empathy, uncondi-
tionally,
and
congruency) ranged
from
.82 to
.72, whereas
the
values
for
the
alliance dimensions (task,
bond,
and
goal) ranged
from
.96 to
.92. Some interpret
the
close relationship between
these concepts
by
noting that
the
TOCs,
and
empathy
in
partic-
ular,
may be
instanciations
of the
alliance
(e.g.,
Watson
&
Greenberg,
in
press);
others
suggest that
the
TOCs
may be im-
portant preconditions
for
alliance development (e.g., Horvath,
in
press).
A
direct examination
of the
temporal
and
causal rela-
tion between these variables
is an
important
task; either path
analytic
or
structural modeling approaches could reveal impor-
tant causal
or
temporal connections among these constructs
and
help
us to
develop better therapeutic strategies.
The
Social
Influence Concept
of
the
Therapeutic
Relationship
The
mediating
effect
of the
client's judgments
of the
thera-
pist's attributes
was
explored
by
LaCrosse (1980)
and
Strong
(1968).
Their theory extended
the
work
of
social psychologists
Hovland
(Hovland, Janis,
&
Kelley,
1953)
and
Cartwright
(1965).
The
central hypothesis
of
this group
was
that
the
client's
impressions
of the
therapist
as
expert, trustworthy,
and
attrac-
tive
provides
the
helper with leverage (social
influence
[SI])
to
promote change. Because
the
strength
of the
therapist's
influ-
ence
is
proportional
to
these client attributions, these
beliefs
are
directly related
to the
benefits
the
client
is
likely
to
accrue
from
therapy.
This body
of
work
may be
seen,
in
part,
as a
reaction
to
Rogers' exclusive emphasis
on the
therapist's
role
in the
rela-
tionship. During
the
1970s
and the
1980s,
a
substantial corpus
of
empirical research accumulated relating
the SI
dimensions
of
attractiveness,
expertness,
and
trustworthiness
to
positive
therapy
outcome. Most
of the
early studies
reported
in the
liter-
ature involved
the
artificial
manipulation
of the
three
SI
vari-
ables
in
analog settings.
The
results
of
these investigations sug-
gested
a
reliable relation between
the SI
variables
and
outcome
(LaCrosse, 1980). However,
the
results
of
subsequent clinical
trials
did not
corroborate
the
analog studies, which indicates
that
the
impact
of
these variables
on
therapeutic gains
may be
quite
modest
and
inconsistent across therapy modalities
(Greenberg
&
Adler,
1989;
Horvath,
1981).
The
Pantheoretical
Concept
of
the
Alliance
A
major
discovery
in
psychotherapy research
of the
past
2
decades
has
been
the
consistent
finding
that, broadly speaking,
different
therapies produce similar amounts
of
therapeutic
gains
(Luborsky,
Singer,
&
Luborsky,
1975;
Smith
&
Glass,
1977;
Stiles, Shapiro,
&
Elliott,
1986).
In
spite
of
some
of the
criticism
that raised
the
possibility
of
methodological shortcomings
in
some
of
these research syntheses (Luborsky,
in
press; Shadish
&
Sweeney,
1991)
and
suggestions that systematic
differences
among therapies
may be
just
lurking
below
the
surface
of
these
broad general conclusions (Beutler, 1979), many researchers
have
interpreted these results
as an
indication that variables
common
to all
forms
of
psychotherapy
may be
responsible
for
a
large
part
of a
client's improvement. This renewed interest
in
the
generic elements common
to all
forms
of
therapy refocused
interest
on the
alliance
as a
pantheoretical
factor
that
may be
responsible
for a
significant proportion
of
this common vari-
ance.
One of the
early refinements
of the
concept
of the
alliance,
based
on
transcripts
from
the
Penn Psychotherapy Project,
was
developed
by
Luborsky
(1976).
He
suggested that
the
alliance
is
a
dynamic rather than
a
static entity responsive
to the
changing
demands
of
different
phases
of
therapy.
Two
types
of
helping
alliances were
identified:
Type
1,
more evident
in the
beginning
of
therapy,
and
Type
2,
more typical
of
later phases
of
treat-
ment.
Type
1
alliance
is "A
therapeutic alliance based
on the
patient's experiencing
the
therapist
as
supportive
and
helpful
with
himself
as a
recipient"
(p.
94);
Type
2
alliance
is
"...
a
sense
of
working together
in a
joint struggle against what
is
impeding
the
patient...
on
shared responsibility
for
working
out
treatment
goals...
a
sense
of'we-ness'"
(p.
94). Luborsky
(1976)
observed
that
the
strength
of
both
Type
1 and
Type
2
alliances were associated with
the
likelihood
of
improvement
in
psychodynamic therapy.
In
a
series
of
pioneering articles,
Bordin(1976,1980,1989,
in
press) proposed
a
broader
definition
of the
working alliance.
His
definition
builds
on
Greenson's
(1967)
earlier work
and
further
clarifies
the
distinctions between
the
unconscious pro-
jections
of the
client (i.e., transference)
and
what
he
defines
as
the
working alliance. This pantheoretical formulation empha-
sizes
the
client's positive collaboration with
the
therapist
against
the
common
foe
of
pain
and
self-defeating
behavior.
He
identified
three components
of the
alliance: tasks, bonds,
and
goals.
The
definitions
of
these components
may be
summa-
rized thus:
Tasks
refer
to the
in-counseling behaviors
and
cogni-
tions that
form
the
substance
of the
counseling process.
In a
well-functioning
relationship, both parties must perceive
these
tasks
as
relevant
and
efficacious;
furthermore, each must
ac-
cept
the
responsibility
to
perform these acts.
A
strong working
alliance
is
characterized
by the
therapist
and the
client mutu-
ally
endorsing
and
valuing
the
goals
(outcomes) that
are the
target
of the
intervention.
The
concept
of
bonds embraces
the
network
of
positive personal attachments between client
and
therapist that includes issues such
as
mutual
trust,
acceptance
and
confidence.
Bordin
also theorizes
on how the
alliance enhances
the
effi-
cacy
of
therapy (Bordin,
1989;
in
press). Positive alliance
is
not,
in
and of
itself,
curative; rather,
the
working alliance
is
seen
as
the
ingredient that
".
. .
makes
it
possible
for the
patient
to
accept
and
follow
treatment
faithfully"
(Bordin, 1980,
p. 2).
Bordin's
formulations
also
offer
an
alternative
to the
traditional
dichotomous perspective
of
technical
and
process
factors
in
therapy.
He
proposes that these
two
aspects
are not
separate
but
interdependent
and
that
the
positive developments
in
each pro-
vide
a
necessary
facilitative
base
for the
growth
of the
other
(Bordin,
in
press). This synergetic aspect
of the
model implies
that
the
client
forms
attachments
to the
therapist,
in
part,
based
on his or her
assessment
of the
relevance
and
potency
of
the
interventions
offered.
This contrasts with Rogers' sugges-
tion that clients
in
therapy respond automatically
to a
thera-
pist's positive attitude.
Recent research
on
clients' cognitions during therapy seem
to
support
the
idea that clients' assessments
of
therapy
are
based,
in
part,
on
their
own
expectations
of
therapy
and are
ultimately
interactive rather than merely responsive
to
therapist
factors
(Hill
&
O'Grady, 1985;
Horvath,
Marx,
&
Kamann,
1990; Mar-

564
ADAM
O.
HORVATH
AND
LESTER LUBORSKY
tin,
Martin,
&
Slemon, 1989).
The
results
of
these investiga-
tions,
however,
do not
imply that approval
of a
therapist's
style
will
necessarily result
in a
stronger alliance.
It
appears that
the
therapist
has to (a)
communicate
to the
client
the
important
links
between therapy-specific tasks
and the
overall goals
of
treatment
and (b)
maintain
an
awareness
of the
client's commit-
ment
to
these activities
and
effectively
intervene
if
resistance
is
present.
In
a
parallel
fashion,
the
client's assessment
of the
therapy
tasks
are
partially predicated
on a
sense
of
agreement
on
what
are
accepted
as
reasonable goals
of the
therapy: Although
it is
likely
that therapists
and
clients almost invariably have some
agreement
on the
global, long-term goals,
it has
been argued
that
the
short-
and
medium-term expectations
of
client
and
therapist
may be
quite
different:
"Clients seek speedy relief
from
the
pain
that
brought
them
to
therapy, whereas
the
thera-
pist perceives treatment
as a
process which
will
lead
to the
eventual
but not
necessarily immediate relief
of the
client's suf-
fering"
(Horvath
&
Symonds,
1991,
p.
620).
The
therapist's
goals
are
informed
by
theory
and may not be
readily available
to the
client.
To
establish
a
good alliance,
it is
important
for the
therapist
to
negotiate these immediate
and
medium-term
ex-
pectations
and
link these
to the
client's wish
to
obtain lasting
relief
from
suffering.
By
developing such linkages,
the
therapist
can
obtain
the
client's active consent
to
pursue these objectives
(i.e.,
an
alliance).
The
presence
of a
strong alliance helps
the
patient
to
deal with
the
immediate discomforts associated with
the
unearthing
of
painful
issues
in
therapy
and
makes
it
possi-
ble
to
postpone immediate gratification
by
using
both cogni-
tive
(i.e.,
endorsements
of the
tasks
of
therapy)
and
affective
(i.e.,
personal bonds) components
of the
relationship.
Luborsky's
research suggests that
the
client's initial response
to the
therapist might
well
be
dominated
by a
judgment con-
cerning whether
the
helper seems caring, sensitive, sympa-
thetic,
and
helpful
(i.e., Type
1
helping alliance; Luborsky,
1976).
Such judgments
may be
influenced
by
levels
of
per-
ceived
warmth
and
care (e.g., empathy), external features (e.g.,
attractiveness),
contextual information
(e.g.,
expertness),
and
the
client's past experiences
in
similar relationships (LaCrosse,
1980).
Subsequently,
the
influence
of
these initial impressions
are
augmented
or
even supplanted
by the
more cognitive (evalua-
tive
or
collaborative) components
of the
alliance
and the
capac-
ity
to
form
a
reciprocal relationship (Luborsky
et
al.,
1985).
Research Results
The
current significance
of the
alliance concept rests,
in
part,
on the
growing body
of
research linking
the
quality
of the
alliance with therapeutic outcome
and on the
more recent
find-
ings
relating
the
alliance
to
specific change processes.
The re-
sults
of
these empirical investigations
are
summarized
and
orga-
nized topically below.
Horvath
and
Symonds
(1991)
identified
five
clusters
of
related
measures used
in the
majority
of the
investigations:
the
Califor-
nia
Psychotherapy Alliance Scales,
(CALPAS/CALTRAS
1
;
Gaston
&
Ring, 1992; Marmar, Weiss,
&
Gaston,
1989);
the
Penn Helping Alliance scales
(Penn/HAQ/HA
cs
/HA
r
;
Alex-
ander
&
Luborsky,
1987;
Luborsky,
1976);
the
Therapeutic
Alli-
ance
Scale (TAS; Marziali, 1984a);
the
Vanderbilt Therapeutic
Alliance
Scale (VPPS/VTAS; Hartley
&
Strupp,
1983);
and the
Working
Alliance Inventory
(WAI;
Horvath,
1981,1982).
Each
of
these instruments
is
available
in
several versions,
and
most
have
been adapted
for use as an
observer's rating scale
or as a
self-report
measure.
In
addition
to
assessing
the
global
level
of
alliance,
these measures also
yield
scores
on a
variety
of
alli-
ance
components. Overall, these scales appear
to
have accept-
able
psychometric properties. Horvath
and
Symonds
(1991)
have
reported
that
the
therapist-based
measures
are the
most
stable
(r
«
.93),
and
observers' ratings
of the
alliance
are the
least
reliable
(r
«
.82).
Differences
Among Alliance Assessment Methods
Most
of
these alliance measures were developed indepen-
dently
and
each progressed through several stages
of
refine-
ments. Given
the
relatively large number
of
instruments,
it is
not
surprising that they measure related,
but not
identical,
un-
derlying
constructs.
In the
next section,
we
will
examine
the
following
questions:
How
closely related
are the
constructs
un-
dergirding these measures?
How
much variance
is
common
among
the
measures?
Do
these instruments predict outcome
equivalently?
It
seems that
two
core aspects
of the
alliance—personal
at-
tachments
and
collaboration
or
willingness
to
invest
in the
ther-
apy
process—are
common elements among each
of the
instru-
ments. Beyond these core components,
the
following
con-
structs
are
monitored
by two or
more
of the
instruments:
therapists'
and
clients' positive
or
negative contributions (e.g.,
CALPAS,
TAS), shared
or
agreed goals
for the
therapy (e.g.,
CALTRAS,
WAI, Penn), capacity
to
form
a
relationship (e.g.,
Penn,
VTAS, CALTRAS), acceptance
or
endorsement
of
ther-
apy
tasks (e.g., CALTRAS, WAI),
and
active participation
in
therapy
(e.g., Penn, CALTRAS, VTAS).
The
weight
or
emphasis
given
to
these
compohents,
however, varies among measures
(Hansell,
1990; Hartley, 1985; Horvath,
in
press; Marziali,
1984b).
Tichenor
and
Hill
(1989)
investigated
the
relation between
four
observer-rated instruments (Penn, VTAS, CALPAS,
and
WAI
0
).
They observed that there
was
a
12%-71
%
overlap among
the
scales (Tichenor
&
Hill, 1989).
In
another study
(Safran
&
Wallner,
1991),
two
self-report scales were compared
(WAI
and
CALPAS).
The
overall shared variance
was
reported
as
76%.
However,
at the
subscale level, overlap ranged
from
0%-67%,
Measurement
of
the
Alliance
and
Operational
Definitions
The
recent increase
in
studies
of the
alliance
was
supported
by
the
development
of
reliable methods
of
measuring
the
con-
cept. Because these measures operationally define
the
alliance,
we
begin this discussion
by
reviewing these
instruments.
Currently,
there
are at
least
11
alliance
assessment
methods.
1
CALPAS
and
CALTRAS
both
refer
to the
California Psychother-
apy
Alliance
Scales.
The two
versions
of the
instrument
are
quite
simi-
lar but
yield slightly
different
subscales.
To
avoid
confusion,
we
gener-
ally
use the
abbreviation
CALTRAS,
except
where
a
specific study
or
result
is
referenced;
in
those
cases,
the
abbreviation
reported
by the
original
authors
is
used.

SPECIAL SECTION: THERAPEUTIC ALLIANCE
IN
PSYCHOTHERAPY
565
with
an
average value
of
37%.
Information available
on
another
group
of
self-report
measures (Penn, VPPS
2
,
and
TAS) suggests
that
there
is a
wide range
of
commonalities
at the
subscale level
(0%-69%),
with
an
average overlap
of
approximately
21% for
client-rated instruments
and of 18% for the
therapists' scales
(Bachelor,
1991).
The
data
suggest
that,
at the
global level,
a
generous overlap
exists between measures. This
may be
interpreted
as
proof that
each
is
assessing
a
similar underlying
process.
However, most
instruments also attempt
to
capture distinct components
of the
alliance with subscales embedded
in the
instrument.
Low
sub-
scale correlations within instruments indicate independent
un-
derlying
factors
and are
cited
as
evidence
of the
scales'
validity
(Gaston
&
Ring, 1992).
It can be
argued, therefore, that
low
correlations across subscales
of
different
instruments
are to be
expected
and
that
a
more proper test
of
similarity would entail
comparison
of
parallel subscales
only.
Unfortunately, there
has
been
no
agreement across instruments
as to the
underlying
structure
of the
alliance; each instrument
has
different
sub-
scales, making such comparison impossible
at
this time.
Theoretical diversity
is
also
reflected
in the
instruments:
The
Penn scales arose
from
a
psychodynamic perspective
and are
influenced
by
Luborsky's
work
on the
alliance (Luborsky,
McLellan,
Woody,
O'Brien,
&
Auerbach, 1985); VPPS repre-
sents
a
blend
of
dynamic
and
eclectic
frameworks
that
has
been
influenced
by the
work
of
Strupp
and his
colleagues
at
Vander-
bilt
University (Henry
&
Strupp,
in
press;
Strupp, 1974);
the
TAS
and
CALPAS
scales
are
influenced
by
both traditional
psychodynamic
concepts
of the
alliance
and the
subsequent
work
of
Bordin (Gaston
&
Ring,
1992);
the WAI was
designed
to
capture Bordin's pantheoretical perspective
and was
vali-
dated using
his
definitions
of the
alliance dimensions
of
goals,
tasks,
and
bonds (Horvath
&
Greenberg,
1989).
Two of the
scales (Penn
and
WAI)
are
theoretically homogeneous,
and the
remaining
instruments represent
a
more eclectic position.
The
homogeneous scales tend
to
have higher subscale correlations,
whereas blended scales (particularly
the
CALTRAS) appear
to
have
more independent subscales (Gaston,
&
Ring,
1992;
Hor-
vath
&
Symonds,
1991).
Across
all
instruments, therapists' alliance scales have pro-
vided
significantly
poorer predictions
of all
types
of
outcomes
than clients'
or
observers' assessments (Horvath
&
Symonds,
1991).
This appears
to be
true irrespective
of the
outcome indi-
ces
predicted. There
may be
several interacting
reasons
for
this.
The
fault
could
lie in the
fact
that
the
therapists'
scales
are
direct rewordings
of
client instruments; thus far,
no
effort
has
been made
to
investigate
the
specific
impressions
and
experi-
ences that therapists associate with
the
clients'
experience
of
positive
alliance. Another reason
for the
poor predictive power
of
therapists' alliance scores
may be
that some therapists
signifi-
cantly
misjudge
the
relationship. Countertransference could
also account
for a
significant
portion
of
these
errors.
Also,
the
therapists'
own
early object relations
may
bias their judgment
of
the
alliance (Henry
&
Strupp,
in
press). Whatever
the
reason
for
the
apparent discrepancy between clients'
and
therapists'
evaluations
of the
relationship,
a
better understanding
of the
factors
influencing
the
therapist's (mis)perception
of the
alli-
ance
is
essential
for
more
effective
use of the
alliance construct
to
improve
therapy.
The
Relation
Between
the
Quality
of
the
Alliance
and
Outcome
in
Psychotherapy
During
the
past
2
decades, there
has
been
a
marked increase
in
both
the
number
of
empirical investigations
of the
alliance
and
the
breadth
of
issues covered
by
these
studies.
There
have
been several recent reviews
of
this research (Gaston,
1990;
Hor-
vath
&
Symonds,
1991;
Luborsky,
in
press).
The following is a
summary
of the
most salient
findings.
Horvath
and
Symonds
(1991)
used
meta-analytic
techniques
to
synthesize
the
quantitative research that links
the
relation-
ship between alliance
and
outcome.
On the
basis
of the 24
studies
included
3
in the
review,
they
found
an
average
effect
size
(ES),
which linked quality
of
alliance
to
therapy,
of r =
.26.
However,
the
actual
influence
of the
alliance
may be
greater
because
the
calculation
was
based
on the
cautious assumption
that
all
relations observed
but not
reported
by
researchers,
or
reported
as not
significant, were actually
r= .0.
This
is
likely
an
overly
pessimistic assumption, resulting
in an
underestimation
of
the
actual
ES.
Research
confirming
the
relation between good alliance
and
positive outcome
has
underlined
the
importance
of the
con-
struct,
but in
order
to
offer
a
more focused perspective
on the
function
of the
alliance,
we
need
to
examine more closely
the
factors
that
influence
the
relation between
the
alliance
and
outcome.
Factors
Influencing
Alliance-Outcome
Relations
Type
of
treatment.
Alliance research covers
a
broad variety
of
treatments
and
client problems.
The
impact
of the
alliance
has
been examined
in the
context
of
behavioral therapy (DeRu-
beis
&
Feeley,
1991;
Krupnick, Stotsky, Simmens,
&
Moyer,
1992),
cognitive therapy (Krupnick
et
al.,
1992;
Raue, Castun-
guay,
&
Goldfried,
1991;
Rounsaville
et
al.,
1987;
Svartberg
&
Stiles,
1992),
gestalt therapy (Horvath
&
Greenberg, 1989),
and
psychodynamic therapy (Eaton, Abeles,
&
Gutfreund, 1988;
Horowitz
&
Marmar, 1985; Krupnick
et
al.,
1992; Luborsky,
1976;
Luborsky
&
Auerbach, 1985; Marmar, Horowitz, Weiss,
&
Marziali,
1986;
Marziali,
Marmar,
&
Krupnick,
1981;
Piper,
DeCarufel,
&
Szkrumelack, 1985;
Saunders,
Howard,
& Or-
linsky,
1989;
Windholz
&
Silbershatz,
1988).
A
strong alliance
appears
to
make
a
positive contribution
in all of
these therapies
(Horvath
&
Symonds,
1991).
Moreover, neither
meta-analysis
of
cross-study
differences
(Horvath
&
Symonds,
1991)
nor
most
studies that directly contrasted alliance-outcome relations
across
different
treatment approaches (e.g., Krupnick
et
al.,
1992)
could
identify
significant
differences
in
alliance-out-
come relations among treatments.
Variety
of
outcomes.
Alliance measurements have been used
to
predict many
different
kinds
of
therapy outcomes ranging
2
The
VPPS
was
originally developed
as an
observer-rated
instru-
ment. However,
in the
study referenced,
it was
modified
for use as a
self-report
scale.
3
The five
inclusion criteria were
(a) the
investigator
identifying
the
independent variable
as
"therapeutic,"
"working,"
or
"helping"
alli-
ance;
(b) a
time lapse between
the
measurement
of
alliance
and the
outcome;
(c)
analog studies were excluded;
(d) a
minimum
of five
sub-
jects
in the
study;
and (e)
only individual treatments were included.

566
ADAM
O.
HORVATH
AND
LESTER LUBORSKY
from
drug
use
(Luborsky
et
al.,
1985)
to
social adjustment
(Rounsaville
et
al.,
1987)
to
clients'
and
therapists' subjective
ratings
of
global improvement (Eaton
et
al.,
1988).
In
general,
the
alliance measures
appear
to be
more
efficient
at
predicting
outcomes tailored
to the
individual client such
as the
Target
Complaints (TC; Battle
et
al.,
1966)
than broad-range symptom-
atic change such
as the
Symptom Distress Check
List—90
(SCL-90;
Derogatis, Rickels,
&
Rock,
1976).
The
average corre-
lation
of
alliance rating
and
outcome rated
on the TC was .30
(TV
of
8
studies);
the
same correlation
for the
SCL-90
was .09 (N of
6
studies) (Horvath, Gaston,
&
Luborsky,
in
press).
At first
glance,
the
self-rated outcome
may
appear
to be a
softer index
of
improvement
in
that
the
clients rate both process
and
out-
come,
and
thus
a
bias toward convergence
may be
part
of the
design. Inspection
of the
research results suggest, however, that
this
is not the
case:
The
average
relation
between
observers'
alliance rating
and
therapists'
judgment
of
outcome
is as
strong
as
the
relation obtained when both measures
are
rated
by
clients,
and the
relation between clients' alliance rating
and
observers' outcome
is
nearly
as
large (Horvath
&
Symonds,
1991).
4
Treatment
length.
The
impact
of the
alliance
has
been dem-
onstrated
in
treatments ranging
from
4 to
over
50
sessions.
The
length
of
treatment, however, does
not
appear
to
influence
the
relation between
the
quality
of the
alliance
and
therapy out-
come (Horvath
&
Symonds,
1991).
Early
versus
late alliance.
Gaston
et al.
(1992)
report that
the
strength
of
alliance remains relatively stable over time when
the
measures
are
averaged across cases. However, there
is
evi-
dence that
the
strength
of the
alliance
fluctuates
systematically
within
individual treatment dyads (Horvath
&
Marx,
1991).
Others (e.g.,
Adler,
1988) report
a
positive slope
of the
alliance
over
time,
but the
incremental gain
in
alliance strength
is
appar-
ently
not
related
to
outcome.
Horvath
and
Symonds
(1991)
reported
that early alliance
is a
slightly
more powerful prognosticator
of
outcome
(ES = .3)
than alliance measures averaged
across
sessions
or
taken
to-
ward
the
middle
of
treatment
(ES =
.2). Although
the
differ-
ences between these values
are
small, subsequent investigations
(i.e.,
DeRubeis
&
Feeley,
1991;
Piper
et
al.,
1991
a;
Piper,
Azim,
Joyce,
&
McCallum,
1991b;
Plotnicov,
1990)
appear
to
confirm
this
trend.
The
apparent anomaly that
we can
better predict
success
on the
basis
of
early alliance than
by
using later assess-
ments
may be
explained
by
examining
the
course
of the
alli-
ance over time. Longitudinal analyses
of the
levels
of the
alli-
ance
in
more
and
less successful therapy (e.g., Horvath
&
Marx,
1991;
Safran,
Crocker, McMain,
&
Murray, 1990; Safran,
Muran,
&
Wallner Samstag, 1992) appear
to
confirm
a
rup-
ture-repair cycle
in
successful therapies predicted
by
Zetzel
(1956)
and
Bordin
(1989).
The
alliance
in
this phase
of
therapy
may
be, of
necessity, quite labile between
or
even within ses-
sions. This
fluctuation in
alliance level
may
account
for the
modest correlation between outcome
and
alliance averaged
across sessions.
The
significant index
of
success
in the
later
phases
of
therapy
may be
more related
to the
degree
of
success
in
resolving these ruptures. Thus,
it may be
interesting
to ex-
plore
the
relation between
the
overall level
of the
alliance,
the
relative
strengths
of the
components
at
different
points
in the
therapy,
and the final
outcome within individual dyads.
The
relation between alliance
and
early
benefits
of
therapy.
One of the
important
debates
concerning
the
validity
and
use-
fulness
of
relationship variables
in
general,
and the
alliance
in
particular,
is
whether there
is a
clear distinction between early
session-level
benefits
and the
alliance.
Are
what
we
currently
identify
as
indications
of
good alliance simply aspects
of
early
therapeutic progress?
If
we
were
to
discover that
the
alliance
is a
by-product
of
therapeutic gains, this would seriously challenge both
the
theo-
retical
and
practical usefulness
of the
concept (Horvath,
1991).
This scenario
of
"alliance
as an
artifact"
can be
tested
by
exam-
ining
the
following
hypotheses:
(a) If the
alliance
is a
by-product
of
successful
therapy,
its
development ought
to
follow
therapeu-
tic
progress
(i.e.,
in an
idealized successful treatment situation,
its
quality ought
to
grow
from
poor
to
better);
(b)
because thera-
peutic gains accumulate
and
stabilize over time, early measures
of
the
alliance
ought
to be
less
efficient
predictors
of
outcome
than measures obtained later
in
therapy
(Gelso
&
Carter,
1985);
and (c) the
alliance's unique contribution
to
prediction
of the
final
therapeutic success
may be
more directly examined
if the
variance associated with early therapy gains were statistically
controlled (partialed out) before alliance-outcome relations
were
computed.
There
is
emerging evidence mitigating against
the
theory
of
alliance
as an
artifact.
The
relevant data will
be
reviewed
in the
same order
as the
hypotheses were listed earlier. Safran
and his
colleagues
have
found
that
positive outcome
was
most closely
associated with
successful
repairs
of
alliance ruptures (break-
down
of the
relationship) rather than with
a
linear
or
parallel
development throughout therapy (Safran, Crocker,
et
al.,
1990;
Safran
et al,
1992). Horvath
and
Marx
(1991)
likewise
de-
scribed
the
course
of the
alliance
in
successful therapy
as a
series
of
developments, decays,
and
repairs.
Kokotovic
and
Tracey
(1990)
and
Plotnicov
(1990)
reported
that
very
early (first-session) alliance measures were predictive
of
therapy dropouts. Horvath
and
Symonds
(1991)
found
that
the
magnitude
of the
alliance-outcome relation
was not a
direct
function
of
time: Both
the
early
and
late alliance measures pre-
dict outcome better than alliance assessments obtained
in the
middle phase
of
therapy.
Gaston
and her
colleagues
(1991)
computed partial correla-
tions (controlling
for
therapy gains) associated with alliance
and
posttherapy outcome.
She
found
that alliance predicted
36%
to 57% of the
variance
in
posttherapy outcome beyond
short-term improvements. Similar results were
found
in a re-
cent investigation
of the
impact
of
psychodynamic interven-
tions
by
Barber, Crits-Cristoph,
and
Luborsky (1992).
Client
and
Therapist Factors Influencing
the
Development
of
the
Alliance
It
is
likely
that
both clients'
and
therapists'
personal
histories
have
some influence
on the
capacity
to
develop
a
good
thera-
peutic
alliance.
Moreover, some
of
these
client-therapist
quali-
ties could interact
to
produce particularly propitious
or
poor
alliance
patterns.
Horvath (1991) summarized
11
studies
re-
porting
on the
impact
of
client characteristics
on the
alliance.
4
For a
full
discussion
of the
possible
"halo
effect"
in
alliance-
outcome
relations,
see
Horvath
and
Symonds
(1991,
pp.
143-144).

SPECIAL
SECTION: THERAPEUTIC ALLIANCE
IN
PSYCHOTHERAPY
567
To
make
the
interpretation
of a
broad spectrum
of
potential
client variables easier, Horvath sorted factors into three catego-
ries: interpersonal capacities
or
skills,
intrapersonal
dynamics,
and
diagnostic features.
The
interpersonal capacities category
included
measures
of the
quality
of the
clients' social relation-
ships
(e.g.,
Moras
&
Strupp,
1982)
and
family
relations
(e.g.,
Ko-
kotovic
&
Tracey,
1990),
and
indices
of
stressful
life
events
(e.g.,
Luborsky,
Crits-Christoph, Alexander, Margolis,
&
Cohen,
1983).
The
intrapersonal dynamics category subsumes indices
of
clients' motivation (e.g.,
Marmar
et
al,
1989),
psychological
status
(e.g.,
Ryan
&
Cicchetti, 1985), quality
of
object relations
(e.g.,
Piper
et
al.,
199
Ib),
and
attitudes (e.g., Kokotovic
&
Tracey,
1990).
Diagnostic features
refers
to the
severity
of the
client's
symptoms
in the
beginning
of
treatment
(e.g.,
Luborsky
et
al.,
1983)
or to
prognostic indices
(e.g.,
Klee,
Abeles,
&
Muller,
1990).
It
appears that both intrapersonal
and
interpersonal client
variables
have similar
and
significant
effects
on the
alliance.
The
average correlation coefficients
(rs,
weighted
by
sample
size)
between these variables
and
alliance were
.3 and
.32,
re-
spectively.
Thus, clients
who
have
difficulty
maintaining social
relationships
(e.g.,
Moras
&
Strupp, 1982)
or
have poor
family
relationships
(e.g.,
Kokotovic
&
Tracey,
1990)
are
less likely
to
develop strong alliances.
Similarly,
patients with little hope
for
success
(e.g.,
Ryan
&
Cicchetti, 1985),
who
have
poor
object
relations
(e.g.,
Piper
et
al.,
1991b),
who are
defensive
(e.g.,
Gas-
ton, Marmar, Thompson,
&
Gallagher,
1988),
and who are not
psychologically
minded (e.g.,
Ryan
&
Cicchetti, 1985)
are
often
associated
with
poor alliance
in
therapy. Severity
of
symptoms,
on
the
other hand,
had but a
small impact
on the
ability
to
develop
a
good therapeutic relationship.
In
a
recent
study,
Piper
and his
colleagues (Piper
et
al.,
1991
a)
investigated
the
relations among clients' quality
of
object rela-
tions
(QOR),
six
aspects
of
current interpersonal
functioning,
the
level
of the
alliance,
and
three posttreatment
and
follow-up
outcome indicators. Their results support
the
link between
the
client's early object relations
and the
ability
to
develop
a
positive
alliance—QOR
had a
statistically
significant
relation with alli-
ance
as
well
as
outcome (these correlations were similar
in
mag-
nitude,
r
KK
.23)—but
they
found
no
significant relations
be-
tween
the
quality
of the
client's current relationships
and the
alliance.
However,
the
alliance
was a
superior predictor
of
out-
come, compared with
the
QOR, suggesting that
the
quality
of
the
alliance
is
influenced
but not
determined
by
early relation-
ship
experiences. Unfortunately,
no
information
was
provided
on
the
partial correlations among QOR, alliance,
and
outcome
to
help
us
decide
the
important question
of
whether
QOR and
alliance
were predicting overlapping
or
distinct portions
of the
outcome.
Also,
the
alliance measure used
in
this investigation
was
specifically
developed
for the
project; thus,
the
results
may
not
be
generalizable.
An
important
but
unanswered question with respect
to the
impact
of the
client pretreatment factors remains:
Do
these
variables
affect
the
alliance only
at the
beginning phase
of
treatment
or
throughout therapy? This question
has
crucial
practical implications. Therapists' strategies would depend
on
whether
the
deleterious
effects
of
poor
object relations
affect
only
the
initial building
of the
alliance
or its
subsequent mainte-
nance also.
Information
on the
impact
of the
therapists'
pretreatment
characteristics
is
just emerging: Henry
and
Strupp
(in
press)
analyzed therapist-client transactions using
a
circumplex
model (the Structural Analysis
of
Social
Behavior
[SASB];
Ben-
jamin,
1974)
and
concluded that "the central theoretical mecha-
nism
which underlies
an
interpersonal definition
of the
alli-
ance
is
clearly
the
process
of
introjection whereby [therapists]
intrapsychically
represent
past
interpersonal relationships.
In-
ternal
representations
of
others guide
not
only actions toward
the
self,
but
also tend
to
re-create
the
original interpersonal
patterns
in
current relationships."
In a
study currently
in
pro-
gress,
the
relation between therapist attachment styles
and the
ability
to
develop
the
bond aspect
of the
alliance will
be di-
rectly
assessed
for the first
time
(G. M.
McKee, personal com-
munication, June 1992).
Another
gap in our
current knowledge
is the
effect
of
specific
therapist-client variable combinations
on the
development
of
the
alliance.
An
examination
of the
interaction
of
some
of
these
client-therapist variables
may
throw light
on the
configura-
tions that result
in the
therapist's misjudging
the
true status
of
the
alliance.
Do
some
specific
combinations
of
factors make
dropping
out
early
from
therapy more likely?
We
have some
evidence indicating that
successful
alliance
may not be a
matter
of
direct matching
of
therapist-client characteristics
but a
prod-
uct
of
complementarity (Blumenthal, Jones,
&
Krupnick,
1985;
Goren,
1991;
Kiesler
&
Watkins,
1989;
Vervaeke
&
Vertommen,
1991).
The
investigation
of
these therapist
and
therapist-client
factors
(including
the
investigation
of the
possible impact
of
countertransference)
represent
an
important
new
challenge
in
alliance research.
Clinical
Implications
Critical
Alliance
Phases
Current
findings
lead
us to
speculate that there might
be two
important alliance phases.
The first is the
initial development
of
the
alliance, which takes place within
the first five
sessions
and
probably peaks during Session
3
(Horvath,
1981;
Saltzman,
Leutgert,
Roth, Creaser,
&
Howard,
1976).
During this phase,
satisfactory
levels
of
collaboration
and
trust must
be
estab-
lished;
the
client needs
to
join
the
therapist
as a
participant
in
the
therapeutic journey, agree
on
what needs
to be
accom-
plished,
and
develop
faith
in the
procedures
that
provide
the
framework
of the
therapy.
The
second critical phase occurs
as
the
therapist begins
to
challenge
old
neurotic patterns.
The
client
may
experience
the
therapist's more active interventions
as
reduction
of
sympathy
and
support; this could reactivate
the
client's
past
dysfunctional
relational
beliefs
and
behaviors, thus
weakening
or
rupturing
the
alliance. Such deterioration
of the
relationship must
be
repaired
if
therapy
is to
continue success-
fully
(Crits-Cristoph, Barber,
&
Kurcias,
1991;
Safran
et
al.,
1992;
Safran, Muran,
&
Wallner Samstag,
in
press).
This model implies that alliance
may be
damaged
at
differ-
ent
times
for
different
reasons.
Depending
on
when such failure
occurs, therapy
will
be
affected
in
diverse
ways.
Difficulties
in
developing
a
supportive relationship
and
problems with con-
sensual endorsement
of the
therapeutic procedures
can
occur
in
the
beginning
of
therapy.
In
such cases,
it is
unlikely that
the
client
will
adhere
to the
therapeutic regime,
and
premature
termination
is
probable (DeClericq,
Goffinet,
Hoyois,
&
Brus-

568
ADAM
O.
HORVATH
AND
LESTER
LUBORSKY
selmans,
1991;
Horvath,
1991;
Kokotovic
&
Tracey, 1990).
Later,
failure
to
experience interruptions
in the
alliance
may be
a
sign either that
the
therapy
is
"coasting"
(i.e.,
dysfunctional
thoughts,
affects,
or
behavior patterns
are not
challenged)
or
that
the
client
is
responding
to the
therapist
in an
unrealistic,
idealized manner
and is
failing
to use the
therapeutic
en-
counter
to
deal with important issues (i.e., there
is
positive trans-
ference
but no
alliance present).
In
these cases, therapy itself
becomes cyclical insofar
as the
in-therapy
events have
a
regular
repetitive
structure that
may
mirror past unresolved conflicts
(Henry
&
Strupp,
in
press). Therapy
may
also
fail
if the
chal-
lenges
to the
alliance
in the
second phase
are too
severe
or not
attended
to
properly and,
as a
result,
the
alliance rupture does
not
heal
(Safran
et
al,
1990,
in
press).
The
model
we are
proposing draws
on the
concepts proposed
by
Luborsky
(1976),
Bordin
(1975,1976),
Horvath
(1991),
and
Safran
et al.
(1990).
There appears
to be
some evidence support-
ing
the
presence
of the first
alliance phase: Some researchers
have
found
that
failure
to
develop these initial levels
of the
alliance
has a
deleterious impact
on
outcome
and may
disrupt
therapy
(i.e.,
DeClericq
et
al.,
1991;
Frank
&
Gunderson, 1990;
Kokotovic
&
Tracey, 1990; Plotnicov, 1990; Saunders
et
al.,
1989).
Empirical
support
for the
second critical alliance phase
is
emerging.
Safran
and his
colleagues (1990) suggested that
"most
therapy cases
. . . are
characterized
by at
least
one or
more ruptures
in the
therapeutic alliance over
the
course
of
therapy"
(p.
154).
Luborsky (1976) found that
the
strength
of
the
Type
2
alliance
was
related
to the
quality
of
outcome
in
dynamic
therapy. Similarly, Foreman
and
Marmar
(1985)
re-
ported that therapists were able
to
improve
the
level
of
alliance
by
addressing
the
conflicted in-therapy relationships. Gaston
and
her
colleagues
(1991)
found that therapists
who
focused
on
a
client's problematic relationship improved
the
alliance,
in
contrast
with
those
who
focused
on
problem content.
Reandeau
and
Wampold
(1991)
examined verbal transactions
of
high-
and
low-alliance therapist-client dyads
and found
that
high-alliance
clients responded
to
therapist-challenging inter-
ventions
with nonreactive high-involvement statements,
whereas
low-alliance clients used avoidance (low-involvement)
responses
in
these situations. Although some
of the
studies
cited here
are
based
on
small samples,
the findings
appear
to be
consistent:
The
therapist's
focus
on the
conflictual
relationship
patterns
in
therapy
and the
client's ability
to
respond
to
these
challenges
by
involvement,
as
opposed
to
avoidance, contribute
to
better alliance.
Therapist
Interventions
Most
studies
of the
alliance deal with
the
impact
of the
qual-
ity
of the
alliance
on
therapy outcome;
we
have much less
re-
search data
on the
specific therapist techniques
that
improve
the
relationship (Horvath
&
Symonds,
1991).
This
state
of af-
fairs
is
hardly surprising. Therapy involves such
a
multitude
of
contextual contingencies
that
isolating therapy
techniques
that
reliably
affect
the
alliance across
this
broad
range
of
situations
is
very
difficult
(Butler
&
Strupp,
1986).
Nonetheless, research
is
unlikely
to
provide guidance
to
clinical
practice
unless
the
relations between clearly defined therapist
actions
in
specific
contexts
and the
effect
of
these interventions
on
process
or
outcome
can be
demonstrated. Although this inquiry
is in its
early
stage,
we
will
present some preliminary
findings
that
ad-
dress
these issues.
Collaboration
There
is
empirical
support
of
Freud's original proposition
that
a
friendly,
sympathetic attitude toward
the
client
is
benefi-
cial
for the
initial development
of the
alliance (Greenberg
&
Adler,
1989; Horvath,
1981;
Jones, 1988; Kokotovic
&
Tracey,
1990).
This proposition
is
further confirmed
by the finding
that
early
alliance measures
are
significantly correlated with mea-
sures
of
empathy
as
well
as
with
outcome
(Adler,
1988;
Horvath
&
Greenberg, 1989; Moseley, 1983). However, some
of the re-
search
on
early alliance also suggests
that
the
endorsements
of
the
tasks involved
in
therapy
and a
sense
of
collaboration with
the
therapist
are the
factors most closely associated with posi-
tive
outcome (Adler, 1988; Horvath
&
Greenberg, 1989). These
results seem
to
indicate that
the
clients' perception
of the
thera-
pists
as
accepting
and
supporting
is
closely linked with
the
clients' perception
of the
appropriateness
of
technical aspects
of
treatment
and
with
the
therapists'
apparent willingness
to
negotiate treatment goals.
On one
hand,
the
clients "clothe
the
therapist with authority" (Freud,
1913,
pp.
99-108),
but on the
other hand,
they
also need
to
feel
that this power
and
authority
is
shared. This sense
of
collaboration
and
participation
may
importantly
contribute
to a
sense
of
safety that
is
essential
for
the
development
of
trust between therapist
and
client
and
deep
commitment
to the
therapeutic journey.
Using
the
Therapy
Relationship
Safran
and his
colleagues (Safran
et
al.,
in
press) developed
a
theoretical model
of the
function
of the
alliance during therapy.
Their schema
is
congruent with Luborsky's
(1977)
work
on the
Core Conflictual Relationship
and
also with Bordin's (1976)
model
of the
alliance.
Briefly,
they suggest
that
the
client brings
dysfunctional
interpersonal relationship
schemas
into
the
thera-
peutic
situation that
are
reactivated during therapy.
If the
thera-
pist responds
in a
manner that confirms
the
schema,
the
cycle
is
maintained
or
even exacerbated.
If, on the
other
hand,
the
pattern
is
recognized
and the
client's negative
feelings
toward
the
therapist
are
examined,
it is
possible
to
disrupt
the
cycle
and
assist
the
client
to
gain
a
better grasp
of his or her
patho-
genic
ideas.
There
is
some empirical support
for
this model: Reandeau
and
Wampold
(1991)
found
that clients
who
responded with
high
engagement
to
strong therapist intervention developed
better alliances than those patients
who
withdrew
or
used avoid-
ance maneuvers. Kivlighan
and
Schmitz
(1992)
described thera-
pists'
equivalents
of
these
findings:
Therapists
who
were more
challenging
and
oriented toward addressing
the
therapeutic
re-
lationship were more likely
to
improve
the
alliance than those
who
did not
address
the
current relationship
or who
were less
focused.
There
is
some support
for the
view
that
the
alliance
is
not
a
steady
state
or
linear phenomenon. Such regular varia-
tions
in
alliance levels were observed
in a
small-scale study
(Horvath
&
Marx,
1991).
Other investigators, however,
did not
report such cycles over time
(Gomes-Swartz,
1978; Hartley
&
Strupp,
1983; Krupnick, 1990; Marmar
et
al.,
1989; Morgan,

SPECIAL SECTION: THERAPEUTIC ALLIANCE
IN
PSYCHOTHERAPY
569
Luborsky,
Crits-Cristoph,
Curtis,
&
Solomon, 1982).
The
fail-
ure
to
observe variations over time
in
these investigations
may
be due to the
fact
that, whereas
Horvath
and
Marx
(1991)
moni-
tored individual alliance scores over time,
the
other investiga-
tors averaged scores across individuals, possibly masking
asynchronous fluctuations.
The
clinical implications
of
this model
are
worth noting:
It
follows
from
these principles that negative client sentiments,
avoidance,
or
even high levels
of
compliance
may be
signs
of
disruption
in the
alliance.
Safran
et
al.
(in
press) argue that
the
therapist should attend
carefully
to
these signals
and
provide
support
and
empathy
for the
client
in
order
for him or her to
bring these conflicted
feelings
into
full
awareness.
The
proposi-
tion
that direct attention
to the
vicissitudes
of the
in-therapy
relationship
has a
salutary impact
fits
well with Bordin's
(in
press)
notion
that
the
alliance
directly
taps
into
the
client's
past
pathogenic
relationships. There
is
also preliminary empirical
evidence,
provided
by
Foreman
and
Marmar
(1985),
that thera-
pists
may be
able
to
strengthen
the
alliance
by
focusing
on the
therapy
relationship directly. These investigators
found
that ini-
tially
poor alliances improved
as a
result
of
such action.
Al-
though
they
interpret their
findings
with caution, noting that
the
clients
in the
study
had
broadly heterogeneous symptoms
before
therapy, other researchers
(i.e.,
Coady, 1988;
Marziali,
1984a)
also noted that
therapists'
attention
to
negative in-ther-
apy
experiences improved
the
alliance.
Interpretation
The
research
on the
impact
of
interpretation
on the
alliance
presents
a
complex picture. Gaston (manuscript
in
preparation)
investigated
the
effect
of
frequency
of
interpretation
on the
alliance
and
failed
to find
evidence
of
significant benefits.
Crits-Cristoph
et al.
(1991)
researched
the
impact
of the
accu-
racy
of
interpretations
and
observed that there
was a
relation
between
a
specific
type
of
interpretation (involving
the
ele-
ments
of
wish
and
response)
and the
quality
of the
alliance
but
other interpretations appeared unrelated
to the
alliance. Also,
Piper
(1991
b)
discovered
a
curvilinear relationship between
fre-
quency
of
interpretations
and the
quality
of the
alliance.
It
seems
likely
that
the
type
of
interpretation, timing,
and the
client's response potentials
are
important factors
in
determin-
ing
whether this particular therapist intervention will
strengthen
the
alliance.
At
least
one
recent study appears
to
verify
elements
of
such reciprocity (Hentschel
&
Bijleved,
1991).
Some
of the
research results suggest that
the
relevance
of the
interpretation
to the
current relationship between
the
therapist
and
client
is a
factor
in its
efficacy:
Addressing current stress
in
the
therapeutic relationship
is
more likely
to
improve
the
alli-
ance (repair
the
rupture)
and
result
in
therapeutic gains than
interpretations addressing out-of-therapy (past) events (Fore-
man &
Marmar, 1985; Reandeau
&
Wampold,
1991).
Additional
evidence supporting
the
intricacy
of the
relation
between
therapist activity
and the
alliance
is
provided
by
Mal-
linckrodt
and
Nelson
(1991),
who
studied
the
alliance ratings
of
therapists with more advanced
and
less advanced levels
of
training.
The
therapists with more advanced levels
of
training
had
higher ratings
on the
Task
and
Goal
scales
of the
Working
Alliance Inventory
but
were
not
rated
differentially
on the
Bond scale. These results suggest that more
and
less trained
therapists were equally liked
and
trusted;
however,
the
better
trained,
more experienced therapists who, presumably, were
more
adept
at the
timing
and
pacing
of
their interventions, were
better able
to
engage
the
clients
as
collaborators
and as
capable
of
selecting appropriate goals.
Training
Implications
Although
there
is
strong evidence that development
and
maintenance
of a
strong alliance
is
helpful
in
keeping clients
in
therapy
as
well
as
being
a
likely contributor
to
positive out-
come,
the
most
effective
curriculum
and
method
of
training
therapists
to use
this information
in
clinical practice
is not yet
clear.
Our
increasing knowledge
of
therapists'
actions that
are
likely
to
contribute
to
alliance improvements
do not
necessar-
ily
translate directly
to
effective
training paradigms.
For
exam-
ple, Henry
and
Strupp
(in
press) have
reported
that training
specifically
designed
to
sensitize
therapists
to the
importance
of
maintaining
a
positive alliance
did not
result
in the
kinds
of
therapist's verbal behavior that,
as
they predicted, would lead
to
better alliance. Similar
findings
were also reported
by
Babin
(1991).
Henry
and
Strupp's data,
in
particular, seem
to
suggest
that
the
therapist's ability
to
respond accurately
to the
demands
of
the
therapeutic relationship
may be
blocked
by her or his
own
relational issues. Whether these therapist factors
are
uni-
form
across clients
or
specifically
reactive
to
specific client
is-
sues
is
currently unclear.
Conclusions
Current
Status
of
Alliance
Research
Recent developments
of the
alliance concept made
it
possible
to use
this construct
in a
variety
of
theoretical frameworks.
Some
of the
current alliance formulations also
facilitate
the
bridging
of the
traditional distinctions between
the
relation-
ship
and
technique
aspects
of
therapy
and
thus help
us to
better
understand
the
interactive nature
of all
aspects
of
treatment.
Much
of the
impetus
for
alliance research
can be
traced
to
the
development
of
measures that operationalized
the
alliance.
These measures show evidence
of
reliability,
and
there
are
some
data supporting their
validity.
Most
of the
support
for
these
instruments'
validity,
however, comes
from
the
positive relation
between
the
measures
and
outcomes
of
psychotherapy. Less
evidence
is
available
on how
well
these instruments represent
the
various theoretical
definitions.
A
variety
of
alliance compo-
nents have also been operationalized
by the
development
of
these instruments,
and
there
is
emerging evidence that these
components
may be of
different
importance
at
different
points
of
therapy. However, there
is
limited agreement across instru-
ments
on the
identity
of
these components; thus,
the
evidence
has yet to
converge
on
what elements
of the
alliance
play
an
important role
at
particular stages
of
different
therapies.
The
positive relation between good alliance
and
successful
therapy outcome
is
reasonably
well
documented across
a
vari-
ety
of
different
therapies.
We are
adequately certain that alli-
ance
in the
beginning
of the
therapy
is a
good prognosticator
of
premature
termination. There
is
also data
to
support
the hy-
pothesis
that
good
alliance
is
supported
by
somewhat
different

570
ADAM
O.
HORVATH
AND
LESTER
LUBORSKY
factors
in the
beginning
and
later stages
of
therapy.
Moreover,
fluctuations
in
the
alliance, particularly
in the
middle phase,
appear
to
reflect
the
reemergence
of the
client's
dysfunctional
relationship patterns
and the
therapist's
skill
in
recognizing
and
resolving these issues.
A
related
finding is
that
the
client's
(and
perhaps
the
therapist's) pretherapy interaction patterns
or
capacities
influence,
but do not
determine,
the
course
of
alli-
ance
development.
Future Directions
for
Alliance
Research
Although
the
past
15
years
of
research
has
verified
the
useful-
ness
of the
alliance construct,
we
need
to
further
develop
the
broader theoretical
framework
of
therapeutic change
in
which
this concept
may be
fruitfully
embedded. Models describing
specific
alliance-related client experiences leading
to
predict-
able change
patterns
are
needed.
Also,
to
make
the
alliance
more
useful
in
clinical practice,
we
must explore
not
only
the
pretherapy
variables that
influence
each participant's potential
to
forge
an
alliance
but
also
the
specific
therapist activities that
address
the
different
aspects
of the
alliance
at
different
phases
of
therapy.
The
most urgent priorities
to
achieve these goals
are (a) a
better understanding
of the
overlap
and
differences
among
the
various
measures
of the
alliance
and the
alliance components;
(b)
more
fine-grained
studies aimed
at
identifying
specific
ther-
apist actions that facilitate alliance development
or the
repair
of a
disrupted alliance
in
specific
treatments
and at
different
stages
of
therapy;
(c)
investigation
of the
differences
between
the
therapist's
and the
client's assessment
of the
alliance
to
identify
the
factors
that lead
to
therapist misjudgment
of the
quality
of the
alliance;
(d)
investigation
of the
relation between
the
clients'
and
therapists' relational issues
and
development
of
a
good alliance;
and (e)
examination
of the
relation between
therapeutic
gains
and
subsequent changes
in the
level
of the
alliance
throughout
the
course
of
therapy.
References
Adler,
J. V
(1988).
A
study
of
the
working alliance
in
psychotherapy.
Unpublished
doctoral
dissertation,
University
of
British
Columbia.
Alexander,
L.
B.,
&
Luborsky,
L.
(1987).
The
Penn Helping Alliance
Scales.
In L. S.
Greenberg&
W
M.
Pinsof
(Eds.),
Thepsychotherapeu-
tic
process:
A
research handbook (pp.
325-366).
New
'York:
Guilford
Press.
Babin,
R. J.
(1991). Basic counselor skills scoring
method
as
a
predictor
of
the
counselor trainee's
effectiveness
in
establishing
a
working alli-
ance. Unpublished
master's
thesis, Nicholls
State
University,
Thibo-
daux,
LA.
Bachelor,
A.
(1991).
Comparison
and
relationship
to
outcome
of di-
verse dimensions
of the
helping
alliance
as
seen
by
client
and
thera-
pist.
Psychotherapy: Theory Research
and
Practice,
28,
534-539.
Barber,
J.,
Crits-Cristoph,
P., &
Luborsky,
L.
(1992, June).
"Purified"
helping
alliance,
early
symptomatic improvement
and
outcome
in dy-
namic
psychotherapy
for
depression.
Paper
presented
at the
annual
meeting
of the
Society
for
Psychotherapy
Research,
Berkeley,
CA.
Barrett-Lennard,
G. T.
(1985).
The
helping
relationship:
Crisis
and ad-
vance
in
theory
and
research.
The
Counseling Psychologist,
13,
278-
294.
Battle,
L.
C.,
Imber,
S.
D.,
Hoen-Saric,
R.,
Stone,
A.
R.,
Nash,
E. H., &
Frank,
J. D.
(1966).
Target
complaints
as
criteria
of
improvement.
American
Journal
of
Psychotherapy,
20,184-192.
Benjamin,
L. S.
(1974).
Structural
analysis
of
social
behavior.
Psychol-
ogical Review,
81,
392-425.
Beutler,
L. E.
(1979).
Toward specific
psychological
therapies
for
spe-
cific
conditions. Journal
of
Consulting
and
Clinical Psychology,
47,
882-897.
Bibring,
E.
(1937).
On the
theory
of the
results
of
psychoanalysis. Inter-
national Journal
of'Psycho-Analysis,
18,
170-189.
Blumenthal,
S.
J.,
Jones,
E.
E.,
&
Krupnick,
J. L.
(1985).
The
influence
of
gender
and
race
on the
therapeutic alliance. Psychiatry
Update
Annual
Review,
4,
586-601.
Bordin,
E. S.
(1976,
September).
The
working alliance: Basis
for a
gen-
eral
theory
of
psychotherapy.
Paper
presented
at the
Annual Conven-
tion
of the
American
Psychological
Association, Washington,
DC.
Bordin,
E. S.
(1976).
The
generalizability
of the
psychoanalytic
con-
cept
of the
working alliance. Psychotherapy: Theory Research
and
Practice,
16,
252-260.
Bordin,
E. S.
(1980, June).
Of
human
bonds that bind
or
free.
Paper
presented
at the
annual meeting
of the
Society
for
Psychotherapy
Research,
Pacific
Grove,
CA.
Bordin,
E. S.
(1989,
April). Building therapeutic alliances:
The
base
for
integration.
Paper
presented
at the
annual
meeting
of the
Society
for
the
Exploration
of
Psychotherapy
Integration,
Berkeley,
CA.
Bordin,
E. S. (in
press).
Theory
and
research
on the
therapeutic work-
ing
alliance:
New
directions.
In A. O.
Horvath
& L. S.
Greenberg
(Eds.),
The
working alliance: Theory, research,
and
practice.
New
York:
Wiley.
Bowlby,
J.
(1988).
A
secure base: Clinical applications
of
attachment
theory.
London: Routledge
&
Kegan
Paul.
Brenner,
C.
(1979).
Working alliance,
therapeutic
alliance,
and
transfer-
ence.
Journal
of
the
American
Psychoanalytic Association,
27,
136-
158.
Butler,
S. E, &
Strupp,
H. H.
(1986).
Specific
and
nonspecific factors
in
psychotherapy:
A
problematic
paradigm
for
psychotherapy
re-
search.
Psychotherapy: Theory Research
and
Practice,
23,
30-40.
Cartwright,
D. S.
(1965). Influence
leadership
and
control.
In J. G.
March
(Ed.),
Handbook
of
organizations (pp.
1-47).
Chicago:
Rand
McNally.
Coady,
N.
(1988, June). Prediction
of
outcome
from interpersonal pro-
cess:
A
study
of
the
worker-client relationship.
Paper
presented
at the
annual meeting
of the
Society
for
Psychotherapy
Research,
Santa
Fe,
NM.
Crits-Cristoph,
P.,
Barber,
J.,
&
Kurcias,
J. S.
(1991).
The
accuracy
of
therapists'interpretations
and the
development
of
the
therapeutic alli-
ance. Manuscript
submitted
for
publication.
Curtis,
H. C.
(1979).
The
concept
of the
therapeutic
alliance:
Implica-
tions
for the
"widening
of
scope."
Journal
of
the
American
Psychoan-
alytic
Association,
27,
159-192.
DeClericq,
M.,
Goflmet,
S.,
Hoyois,
P., &
Brusselmans,
M.
(1991,
June). Crisis intervention
in
emergency:
Project
design
and
prelimi-
nary outcome data.
Paper
presented
at the
annual
meeting
of the
Society
for
Psychotherapy
Research,
Lyon,
France.
Derogatis,
L.
R.,
Rickels,
K., &
Rock,
A. F.
(1976).
The
SCL-90
and the
MMPI:
A
step
in the
validation
of a new
self-report
scale.
British
Journal
of
Psychiatry,
128,
280-289.
DeRubeis,
R.
J.,
&
Feeley,
M.
(1991).
Determinants
of
change
in
cogni-
tive
therapy
for
depression.
Cognitive Therapy
and
Research.
Eaton,
T. T,
Abeles,
N.,
&
Gutfreund,
M. J.
(1988).
Therapeutic
alli-
ance
and
outcome:
Impact
of
treatment
length
and
pretreatment
symptomology. Psychotherapy: Research
and
Practice,
25,
536-542.
Foreman,
S.,
&
Marmar,
R. C.
(1985).
Therapist
actions
that
address
initially
poor
therapeutic
alliances
in
psychotherapy.
American
Journal
of
Psychiatry,
922-926.
Frank,
A. E, &
Gunderson,
J. G.
(1990).
The
role
of the
therapeutic
alliance
in the
treatment
of
schizophrenia.
Archives
of
General
Psy-
chiatry,
47,
228-236.
Freud,
S.
(1958).
The
dynamics
of
transference.
In J.
Starchey
(Ed.),
The

SPECIAL
SECTION: THERAPEUTIC ALLIANCE
IN
PSYCHOTHERAPY
571
standard
edition
of the
complete psychological
works
of
Sigmund
Freud
(pp.
99-108).
London: Hogarth
Press.
(Original work pub-
lished
1912)
Freud,
S.
(1913).
On the
beginning
of
treatment: Further recommenda-
tions
on the
technique
of
psychoanalysis.
In J.
Strachey
(Ed.
and
Trans.),
The
standard edition
of
the
complete
psychological
works
of
Sigmund
Freud(pp.
122-144).
London: Hogarth
Press.
Frieswyk,
S. H.,
Gabbard,
G. O.,
Horwitz,
L.
Allen,
J.
G,
Colson,
D. B.,
Newsom,
G. E., &
Coyne,
L. (in
press).
The
role
of the
working
alliance
in
psychoanalytic therapy with borderline patients.
In A. O.
Horvath
& L. S.
Greenberg
(Eds.),
The
working
alliance:
Theory,
re-
search
and
practice.
New
York:
Wiley.
Gaston,
L.
(1990).
The
concept
of the
alliance
and its
role
in
psycho-
therapy:
Theoretical
and
empirical considerations.
Psychotherapy,
27,
143-153.
Gaston,
L.
(1991).
Reliability
and
criterion-related validity
of the
Cali-
fornia
Psychotherapy Alliance
Scales—Patient
version.
Psychologi-
cal
Assessment,
3,
68-74.
Gaston,
L.,
Marmar,
C. R.,
Thompson,
L. W, &
Gallager,
D.
(1988).
Relation
of
patient pretreatment characteristics
to the
therapeutic
alliance
in
diverse psychotherapies.
Journal
of
Consulting
and
Clini-
cal
Psychology,
56,
483-489.
Gaston,
L.,
Marmar,
C.
R.,
Thompson,
L. W, &
Gallager,
D.
(1991).
Alliance
prediction
of
outcome beyond
in-treatment
symptomatic
change
as
psychotherapy progresses.
Psychotherapy
Research,
1,
104-112.
Gaston,
L.,
&
Ring,
J. M.
(1992).
Preliminary results
on the
Inventory
of
Therapeutic Strategies. Journal
of
Psychotherapy
Research
and
Practice,
1,1-13.
Gelso,
C.
J.,
&
Carter,
J. A.
(1985).
The
relationship
in
counseling
and
psychotherapy:
Components, consequences,
and
theoretical ante-
cedents.
The
Counseling
Psychologist,
2,155-243.
Gitleson,
M.
(1962).
The
curative functions
in
psychotherapy.
Interna-
tional
Journal
of
Psychoanalysis,
43,194-205.
Gomes-Swartz,
B.
(1978).
Effective
ingredients
in
psychotherapy: Pre-
diction
of
outcome
from
process variables.
Journal
of
Consulting
and
Clinical
Psychology,
46,1023-1035.
Goren,
L.
(1991).
The
relationship
of
counselor
androgyny
to
the
working
alliance.
Unpublished doctoral dissertation, University
of
Southern
California.
Greenberg,
L.
S.,
&
Adler,
J.
(June
1989).
The
working alliance
and
outcome:
A
client
report
study.
Paper presented
at the
annual meeting
of
the
Society
for
Psychotherapy Research. Toronto, Ontario.
Greenson,
R. R.
(1965).
The
working alliance
and the
transference
neuroses.
Psychoanalysis
Quarterly,
34,
155-181.
Greenson,
R. R.
(1967).
Technique
and
practice
of
psychoanalysis.
New
York:
International University Press.
Gutfreund,
M. J.
(1992,
June).
A
close-up
look
at the
alliance
in a
single
case.
Paper presented
at the
annual meeting
of the
Society
for
Psycho-
therapy
Research, Berkeley,
CA.
Hansell,
J.
(1990,
June).
The
relationship
of
the
California
Psychotherapy
Alliance
Scales
to
other measures
of
the
alliance.
Paper
presented
at
the
annual meeting
of the
Society
for
Psychotherapy Research, Win-
tergreen,
VA.
Hartley,
D.
(1985).
Research
on the
therapeutic alliance
in
psychother-
apy.
In R.
Hales
& A.
Frances
(Eds.),
Psychiatry
update
annual
review
(pp.
532-549).
Washington,
DC:
American Psychiatric Press.
Hartley,
D.
E.,
&
Strupp,
H. H.
(1983).
The
therapeutic alliance:
Its
relationship
to
outcome
in
brief psychotherapy.
In J.
Masling
(Ed.),
Empirical
studies
in
analytic
theories (pp.
1-37).
Hillside,
NJ:
Erl-
baum.
Hatcher,
R.
(1990,
June).
Transference
and the
therapeutic alliance.
Paper presented
at the
annual meeting
of the
Society
for
Psychother-
apy
Research, Wintergreen,
VA.
Henry,
W P., &
Strupp,
H. H. (in
press).
The
therapeutic alliance
as
interpersonal
process.
In A. O.
Horvath
& L. S.
Greenberg
(Eds.),
The
working
alliance:
Theory,
research
and
practice
New
York:
Wiley.
Hentschel,
U,
&
Bijleved,
C. J. H.
(1991).
It
takes
two to
therapy:
On
differential
aspects
in the
formation
of
therapeutic
alliance (Leiden
Psychological Reports: Personality Psychology
No.
03-91).
Leiden
University,
the
Netherlands.
Hill,
C.
E.,
&
O'Grady,
K. E.
(1985). List
of
therapist
intentions illus-
trated
in a
case study
and
with therapists
of
varying
theoretical
orientations.
Journal
of
Counseling
Psychology,
32,
3-22.
Horowitz,
M.,
&
Marmar,
C.
(1985).
The
therapeutic alliance with
difficult
patients.
In R.
Hales
& A.
Frances
(Eds.),
Psychiatry
Update
Annual Review
(pp.
573-584). Washington,
DC:
American Psychiat-
ric
Press.
Horvath,
A. O.
(1981).
An
exploratory
study
of
the
working
alliance:
Its
measurement
and
relationship
to
outcome. Unpublished doctoral dis-
sertation, University
of
British Columbia.
Horvath,
A. O.
(1982).
Working Alliance Inventory (Revised). Instruc-
tional
Psychology Research Group,
82
(1).
Simon
Fraser
University,
Burnaby,
British Columbia, Canada.
Horvath,
A. O.
(1991).
What
do we
know about
the
alliance
and
what
do
we
still have
to find
out? Paper presented
at the
annual meeting
of the
Society
for
Psychotherapy Research,
Lyon,
France.
Horvath,
A. O. (in
press). Empirical research
on the
alliance.
In A. O.
Horvath
& L. S.
Greenberg
(Eds.),
The
working alliance:
Theory,
re-
search
and
practice.
New
York:
Wiley.
Horvath,
A.
O.,
Gaston,
L.,
&
Luborsky,
L. (in
press).
The
role
of the
alliance
in
psychotherapy.
In N.
Miller,
L.
Luborsky,
J.
Barber,
& J.
Docherty
(Eds.),
Handbook
of
psychodynamic
psychotherapy:
Theory
and
research.
New
York:
Basic
Books.
Horvath,
A.
O.,
&
Greenberg,
L. S.
(1989).
Development
and
validation
of the
Working
Alliance
Inventory.
Journal
of
Counseling
Psychology,
36,
223-233.
Horvath,
A.
O.,
&
Marx,
R. W
(1991).
The
development
and
decay
of
the
Working Alliance during time-limited counselling. Canadian
Journal
of
Counselling,
24,
240-259.
Horvath,
A. O,
Marx,
R. W, &
Kamann,
A. M.
(1990).
Thinking about
thinking
in
therapy:
An
examination
of
clients' understanding
of
their therapists' intentions. Journal
of
Consulting
and
Clinical Psy-
chology,
55,614-621.
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.
Horwitz,
L.
(1974).
Clinical
prediction
in
psychotherapy.
Northvale,
NJ:
Jason Aronson.
Hovland,
C. I,
Janis,
I.
L.,
&
Kelley,
H. H.
(1953).
Communication
and
persuasion:
Psychological
studies
of
opinion
change.
New
Haven,
CT:
Yale
University Press.
Jones,
S. S.
(1988).
An
exploration
of
the
relationship
between
client
ex-
pectations
and the
working
alliance. Unpublished doctoral
disserta-
tion,
University
of
Mississippi.
Kiesler,
D.
J.,
&
Watkins,
L. M.
(1989).
Interpersonal complementarity
and the
therapeutic alliance:
A
study
in
relationship
in
psychother-
apy.
Psychotherapy:
Research
and
Practice,
26,
183-194.
Kivlighan,
D.
M.,
Jr.,
&
Schmitz,
P. J.
(1992).
Counselor technical activ-
ity
in
cases with improving working alliances
and
continuing-poor
working
alliances. Journal
of
Counseling
Psychology,
39,
32-38.
Klee,
M. R,
Abeles,
N.,
&
Muller,
R. T.
(1990).
Therapeutic alliance:
Early
indicators, course,
and
outcome.
Psychotherapy:
Research
and
Practice,
27,166-174.
Kokotovic,
A.
M.,
&
Tracey,
T. J.
(1990).
Working alliance
in the
early
phase
of
counseling. Journal
of
Counseling
Psychology,
37,
16-21.
Krupnick,
J.
(1990,
June).
The
relationship
of
therapeutic
alliance
and
outcome
in
interpersonal
psychotherapy
and
cognitive
behavioral
ther-
apy.
Paper presented
at the
annual meeting
of the
Society
for
Psycho-
therapy
Research, Wintergreen,
VA.

572
ADAM
O.
HORVATH
AND
LESTER
LUBORSKY
Krupnick,
J.,
Stotsky,
S.,
Simmens,
S., &
Moyer,
J.
(1992,
June).
The
role
of
therapeutic
alliance
in
psychotherapy
and
pharmacotherapy
out-
come:
Findings
in the
NIMH
Treatment
of
Depression
Collaborative
Research
Program.
Paper
presented
at the
annual meeting
of the
Society
for
Psychotherapy Research, Berkeley,
CA.
LaCrosse,
M. B.
(1980). Perceived counselor social
influence
and
coun-
seling outcomes:
Validity
of the
Counselor Rating Form. Journal
of
Counseling
Psychology,
27,
320-327.
Luborsky,
L.
(1976).
Helping alliances
in
psychotherapy.
In J. L.
Clegh-
horn
(Ed.),
Successful
psychotherapy
(pp.
92-116).
New
York:
Brun-
ner/Mazel.
Luborsky,
L.
(1977).
Measuring
a
persuasive psychic structure
in
psy-
chotherapy:
The
core
conflictual
relationship theme.
In N.
Freed-
man & S.
Grand
(Eds.),
Communicative structures
and
psychic struc-
tures
(pp.
367-395).
New
York:
Plenum.
Luborsky,
L. (in
press). Therapeutic alliance measures
as
predictors
of
future
benefits
of
psychotherapy.
In A. O.
Horvath
& L. S.
Greenberg
(Eds.),
The
working
alliance:
Theory,
research
and
practice.
New
York:
Wiley.
Luborsky,
L.,
&
Auerbach,
A.
(1985).
The
therapeutic relationship
in
psychodynamic
psychotherapy:
The
research evidence
and its
meaning
for
practice.
In R.
Hales
& A.
Frances
(Eds.),
Psychiatry
update
annual
review
(pp.
550-561).
Washington,
DC:
American
Psychiatric Association.
Luborsky,
L.,
Crits-Cristoph,
P.,
Alexander,
L.,
Margolis,
M.,
&
Cohen,
M.
(1983).
Two
helping alliance methods
for
predicting outcomes
of
psychotherapy:
A
counting signs
vs. a
global rating method. Journal
of
Nervous
and
Mental Disease, 171,
480-491.
Luborsky,
L.,
McLellan,
A.
T,
Woody,
G. E.,
O'Brien,
C. P., &
Auerbach,
A.
(1985). Therapist success
and its
determinants. Archives
of
Gen-
eral
Psychiatry,
42,
602-611.
Luborsky,
L.,
Singer,
B.,
&
Luborsky,
L.
(1975).
Comparative studies
of
psychotherapies;
"Is it
true that everybody
has won and all
must
have
prizes?" Archives
of
General
Psychiatry,
32,
995-1008.
Mallinckrodt,
B.,
&
Nelson,
M. L.
(1991).
Counselor training level
and
the
formation
of the
psychotherapeutic working alliance. Journal
of
Counseling
Psychology,
38,133-138.
Marmar,
C.
R.,
Horowitz,
M.
J.,
Weiss,
D.
S.,
&
Marziali,
E.
(1986).
The
development
of the
therapeutic alliance rating system.
In L. S.
Greenberg
& W M.
Pinsof
(Eds.),
The
psychotherapeutic
process:
A
research
handbook (pp.
367-390).
New
York:
Guilford Press.
Marmar,
C.,
Weiss,
D.
S.,
&
Gaston,
L.
(1989). Toward
the
validation
of
the
California Therapeutic Alliance Rating System. Psychological
Assessment:
A
Journal
of
Consulting
and
Clinical
Psychology,
1, 46-
52.
Martin,
J.,
Martin,
W, &
Slemon,
A. G.
(1989).
Cognitive-mediational
models
of
action-act
sequences
in
counseling. Journal
of
Counseling
Psychology,
36,
8-16.
Marziali,
E.
(1984a). Prediction
of
outcome
of
brief psychotherapy
from
therapist interpretive interventions.
Archives
of
General
Psychi-
atry,
41,
301-305.
Marziali,
E.
(1984b). Three viewpoints
on the
therapeutic alliance
scales similarities,
differences
and
associations with psychotherapy
outcome. Journal
of
Nervous
and
Mental Disease, 172,
417-423.
Marziali,
E.,
Marmar,
C, &
Krupnick,
J.
(1981).
Therapeutic
alliance
scales: Development
and
relationship
to
psychotherapy
outcome.
American Journal
of
Psychiatry,
138,
361-364.
Mitchell,
K.
M.,
Bozart,
J. D, &
Krauft,
C. C.
(1977). Reappraisal
of
the
therapeutic effectiveness
of
accurate empathy,
non-possessive
warmth,
and
genuineness.
In A. S.
Gurman
& A. M.
Razin
(Eds.),
Effective
psychotherapy
(pp.
482-502).
New
York:
Pergamon
Press.
Moras,
K.,
&
Strupp,
H. H.
(1982).
Pretherapy interpersonal relations,
patients' alliance
and the
outcome
in
brief therapy. Archives
of
Gen-
eral
Psychiatry,
405-409.
Morgan,
R.,
Luborsky,
L.,
Crits-Cristoph,
P.,
Curtis,
H.,
&
Solomon,
J.
(1982).
Predicting
the
outcomes
of
psychotherapy
by the
Penn Help-
ing
Alliance Rating Method. Archives
of
General
Psychiatry,
39,397-
402.
Moseley,
D.
(1983).
The
therapeutic
relationship
and its
association with
outcome.
Unpublished master's thesis, University
of
British Colum-
bia, Vancouver, Canada.
Orlinsky,
D.
E.,
&
Howard,
K. I.
(1986).
The
psychological interior
of
psychotherapy: Explorations with
the
Therapy Session Report
Questionnaires.
In L. S.
Greenberg
& W M.
Pinsof
(Eds.),
The
psy-
chotherapeutic
process:
A
research
handbook.
New
York:
Guilford
Press.
Piper,
W
E.,
Azim,
H. F.
A.,
Joyce,
A.
S.,
&
McCallum,
M.
(199la).
Transference interpretations, therapeutic alliance
and
outcome
in
short-term individual psychotherapy. Archives
of
General
Psychiatry,
48,
946-953.
Piper,
W
E.,
Azim,
H. F.
A.,
Joyce,
A.
S.,
McCallum,
M.,
Nixon,
G. W
H.,
&
Segal,
P. S.
(1991b). Quality
of
object relations
vs.
inter-
personal
functioning
as
predictor
of
therapeutic alliance
and
psycho-
therapy outcome. Journal
of
Nervous
and
Mental Disease, 179, 432-
438.
Piper,
W
E.,
DeCarufel,
F
L.,
&
Szkrumelack,
N.
(1985).
Patient pre-
dictors
of
process
and
outcome
in
short-term individual psychother-
apy.
Journal
of
Nervous
and
Mental Disease,
173,
726-733.
Plotnicov,
K. H.
(1990).
Early
termination from counseling:
The
client's
perspective.
Unpublished doctoral
dissertation,
University
of
Pitts-
burgh,
PA.
Raue,
P.
J.,
Castunguay,
L.
G.,
&
Goldfried,
M. R.
(1991,
July).
The
working
alliance:
A
comparison
of
two
therapies.
Paper presented
at
the
annual meeting
of the
Society
for
Psychotherapy Integration,
London.
Reandeau,
S.
G.,
&
Wampold,
B. E.
(1991).
Relationship
of
power
and
involvement
to
working alliance:
A
multiple-case
sequential analysis
of
brief therapy. Journal
of
Counseling
Psychology,
38,
107-114.
Rogers,
C. R.
(1951).
Client-centered
therapy.
Cambridge,
MA:
River-
side Press.
Rogers,
C. R.
(1957).
The
necessary
and
sufficient
conditions
of
thera-
peutic
personality
change.
Journal
of
Consulting
andClinical
Psychol-
ogy,
22,
95-103.
Rogers,
C.
R.,
Gendlin,
G. T,
Kiesler,
D. V, &
Truax,
L. B.
(1967).
The
therapeutic
relationship
and its
impact:
A
study
of
psychotherapy
with
schizophrenics.
Madison: University
of
Wisconsin Press.
Rounsaville,
B.
J.,
Chevron,
E.
S.,
Prusoff,
B.
A.,
Elkin,
L,
Imber,
S.,
Sotsky,
S.,
&
Watkins,
J.
(1987).
The
relation between specific
and
general
dimensions
of the
psychotherapy
process
in
interpersonal
psychotherapy
of
depression. Journal
of
Consulting
and
Clinical
Psy-
chology,
55,
379-384.
Ryan,
E.
R.,
&
Cicchetti,
D.
V(1985).
Predicting quality
of
alliance
in
the
initial psychotherapy interview. Journal
of
Nervous
and
Mental
Disease,
717-725.
Safran,
J. D,
Crocker,
P.,
McMain,
S.,
&
Murray,
P.
(1990).
The
therapeu-
tic
alliance rupture
as a
therapy event
for
empirical investigation.
Psychotherapy,
27,154-165.
Safran,
J. D,
Muran,
C.
J.,
&
Wallner Samstag,
L.
(1992,
June).
A
com-
parison
of
therapeutic
alliance
rupture
resolution
and
nonresolution
events.
Paper presented
at the
annual meeting
of the
Society
for
Psychotherapy Research, Berkeley,
CA.
Safran,
J. D,
Muran,
J.
C.,
&
Wallner Samstag,
L. (in
press). Resolving
therapeutic ruptures:
A
task analytic investigation.
In A. O.
Horvath
& L. S.
Greenberg
(Eds.),
The
working alliance:
Theory,
research,
and
practice.
New
\brk:
Wiley.
Safran,
J.
D.,
&
Wallner,
L. K.
(1991).
The
relative predictive validity
of
two
therapeutic alliance measures
in
cognitive therapy.
Psychologi-
cal
Assessment:
A
Journal
of
Consulting
and
Clinical
Psychology,
3,
188-195.
Saltzman,
C.,
Leutgert,
M.
J.,
Roth,
C.
H.,
Creaser,
J.,
&
Howard,
L.
(1976).
Formation
of a
therapeutic relationship: Experiences
during

SPECIAL
SECTION:
THERAPEUTIC ALLIANCE
IN
PSYCHOTHERAPY
573
the
initial phase
of
psychotherapy
as
predictors
of
treatment dura-
tion
and
outcome.
Journal
of
Consulting
and
Clinical
Psychology,
44,
546-555.
Salvio,
M.
A.,
Beutler,
L. E.,
Wood,
J.
M.,
&
Engle,
D.
(1992).
The
strength
of the
therapeutic alliance
in
three treatments
for
depres-
sion.
Psychotherapy
Research,
2,
31-36.
Saunders,
S.
M.,
Howard,
K.
I.,
&
Orlinsky,
D. E.
(1989).
The
therapeu-
tic
bond scales: Psychometric characteristics
and
relationship
to
treatment
effectiveness.
Psychological
Assessment:
A
Journal
of
Con-
sulting
and
Clinical
Psychology,
1,
323-330.
Shadish,
W. R.,
Jr.,
&
Sweeney,
R. B.
(1991).
Mediators
and
moderators
in
meta-analysis:
There's
a
reason
we
don't
let
dodo birds tell
us
which
psychotherapies
should have prizes.
Journal
of
Consulting
and
Clinical
Psychology,
59,
883-893.
Smith,
M.
L.,
&
Glass,
G. V
(1977).
Meta-analysis
of
psychotherapy
outcome studies. American
Psychologist,
32,
752-760.
Sterba,
R. F.
(1934).
The
fate
of the ego in
analytic therapy.
Interna-
tional
Journal
of
Psychoanalysis,
115,
117-126.
Stiles,
W
B.,
Shapiro,
D.,
&
Elliott,
R.
(1986).
Are all
psychotherapies
equivalent?
American
Psychologist,
41,165-180.
Strong,
S. R.
(1968).
Counseling:
An
interpersonal
influence
process.
Journal
of
Counseling
Psychology,
15,
215-224.
Strupp,
H. H.
(1974).
On the
basic ingredients
of
psychotherapy.
Psy-
chotherapy
and
Psychosomatics,
24,
249-260.
Svartberg,
M.,
&
Stiles,
T. C.
(1992,
June).
Therapeutic
alliance
and
treatment
competence:
Their
relations
to
patient change
in
short-term
anxiety
provoking
psychotherapy.
Paper presented
at the
annual
meeting
of the
Society
for
Psychotherapy Research, Berkeley,
CA.
Tichenor,
Y, &
Hill,
C. E.
(1989).
A
comparison
of six
measures
of
working
alliance.
Psychotherapy,
26,
195-199.
Vervaeke,
G.,
&
Vertommen,
H.
(1991,
July).
Similarity,
acceptability
and
convergence:
Specific
and
non-specific
factors.
Paper presented
at
the
annual meeting
of the
Society
for
the
Exploration
of
Psychother-
apy
Integration, London.
Watson,
J.
C.,
&
Greenberg,
L. S. (in
press).
The
alliance
in
experiential
therapy: Enacting
the
relationship conditions.
In A. O.
Horvath
&
L. S.
Greenberg
(Eds.),
The
working
alliance:
Theory
and
research
New
York:
Wiley.
Windholz,
M. I, &
Silbershatz,
G.
(1988). Vanderbilt psychotherapy
process scale:
A
replication with adult outpatients.
Journal
of
Con-
sulting
and
Clinical
Psychology,
56,
56-60.
Zetzel,
E. R.
(1956).
Current concepts
of
transference. International
Journal
of
Psychoanalysis,
37,
369-376.
Received
April
18,1992
Revision
received December
16,1992
Accepted February
26,1993
•
Low
Publication
Prices
for APA
Members
and
Affiliates
Keeping
You
Up-to-Date:
All
APA
members (Fellows; Members;
Associates,
and
Student
Affiliates)
receive—as
part
of
their annual
dues—subscriptions
to the
American
Psychologist
and
APA
Monitor.
High School Teacher
and
International
Affiliates
receive
subscriptions
to the APA
Monitor,
and
they
can
subscribe
to the
American
Psychologist
at a
significantly reduced rate.
In
addition,
all
members
and
affiliates
are
eligible
for
savings
of up to 60%
(plus
a
journal
credit)
on all
other
APA
journals,
as
well
as
significant
discounts
on
subscriptions
from
coop-
erating societies
and
publishers
(e.g.,
the
American Association
for
Counseling
and
Develop-
ment, Academic
Press,
and
Human Sciences
Press).
Essential
Resources:
APA
members
and
affiliates
receive special rates
for
purchases
of
APA
books, including
the
Publication
Manual
of
the
APA,
the
Master
Lectures,
and
Journals
in
Psychol-
ogy:
A
Resource
Listing
for
Authors.
Other
Benefits
of
Membership: Membership
in APA
also provides eligibility
for
low-cost
insurance plans covering
life,
income protection,
office
overhead, accident protection, health
care, hospital indemnity, professional liability, research/academic professional liability, stu-
dent/school
liability,
and
student health.
For
more
information,
write
to
American
Psychological
Association,
Membership
Services,
750
First
Street,
NE,
Washington,
DC
20002-4242,
USA