Content uploaded by Jonathan Cook
Author content
All content in this area was uploaded by Jonathan Cook
Content may be subject to copyright.
Working Alliance in Online Therapy as Compared
to Face-to-Face Therapy: Preliminary Results
JONATHAN E. COOK, M.S.,
1
and CAROL DOYLE, Ph.D.
2
ABSTRACT
Online therapy, defined as the provision of mental health services through the Internet, is a
growing field that has sparked an abundance of interest and controversy. A primary concern
in the practice of online therapy is whether a working alliance, considered a central compo-
nent of successful therapy, can develop when participants are geographically separated.
Working alliance scores were compared between a small, primarily female sample of online
therapy consumers and a representative sample of traditional face-to-face therapy clients. Re-
sults revealed significantly higher means on the goal subscale and composite score of the
Working Alliance Inventory in the online sample, suggesting that a working alliance can be
adequately established in therapy delivered online. No significant differences in the level of
working alliance were found within the online therapy sample with respect to modality of
communication, client presenting problem, or therapist. Themes from comments suggest the
importance for participants of the disinhibiting effects of the medium.
95
CYBERPSYC HOLOGY & BEHAVIOR
Volume 5, Number 2, 2002
© Mary Ann Liebert, Inc.
INTRODUCTION
R
ECENT RESEARCH
3
indicates that more than
14% of American adults with Internet ac-
cess go online to find mental health informa-
tion. Since at least 1995,
4
many have also used
the Internet as a medium through which to re-
ceive mental health services. Although empiri-
cal research regarding the effectiveness of
online therapy is nascent, preliminary results
are encouraging, with several published studies
finding similar outcomes between traditional,
face-to-face therapy and online therapy.
5–7
Posi-
tive outcomes in both audio/video and text-
based computer-mediated therapy have been
found for a number of specific presenting prob-
lems, including anxiety,
5
posttraumatic stress
disorder,
8
eating disorders,
9
and panic disorder
with agoraphobia.
10
While most of these studies
have focused on individual counseling, im-
provements have also been noted in online
group settings.
7
Despite growing evidence of effectiveness,
potential advantages and disadvantages to on-
line therapy continue to be vigorously debated
and discussed
11–13
and include a range of ethi-
cal, legal, and clinical issues. One of the
primary clinical concerns is the possible diffi-
culty, or even inability, for therapists and
clients to establish a strong therapeutic rela-
tionship in the absence of nonverbal cues.
This emphasis on the relationship as an im-
portant outcome variable is grounded in a
well-established body of research. Lambert’s
common factors research
14
suggests that rela-
tionship factors are the single largest variable
over which therapists have some control, ac-
counting for approximately 30% of therapeutic
1
Graduate School of Social Work, Regional Reseach Institute for Human Services, Portland State University, Port-
land, Oregon.
2
Department of Counseling Psychology, Lewis & Clark College, Portland, Oregon.
outcome across therapist theoretical orienta-
tions. Although the therapeutic relationship
has been operationalized in a variety of ways,
most research in the past 20 years has favored
the working alliance construct, which can
be broadly defined as collaboration between
therapeutic participants to facilitate healing.
15
Horvath and Symonds
16
conducted a meta-
analysis of 24 studies relating the quality of
working alliance to outcome. Across a variety
of studies using different treatments, therapist
theoretical orientations, and alliance measure-
ment devices, they found a combined effect
size (Pearson’s
r
) of 0.26, showing a small to
moderate association between alliance and
outcome.
Working alliance can be measured from the
perspective of clients, therapists, and indepen-
dent observers, but research
16–19
has consis-
tently shown the strongest correlation between
clients’ view of working alliance and outcome.
While it appears that type and severity of
clients’ pretherapy symptomatology is unre-
lated to establishment of alliance,
20,21
quality of
past relationships does seem to have an im-
pact, as clients who have had a history of diffi-
culties in interpersonal relationships are less
likely to develop strong alliances.
21,22
Similarly,
therapists’ ability to foster a strong emotional
bond in therapy is mediated by the extent to
which their own interpersonal relationships
and skills are well established.
23
In addition,
highly motivated clients, who enter counsel-
ing with the expectation that they will need to
assume personal responsibility for doing the
work of therapy, are more likely to form a
strong working alliance.
24
Research is mixed,
but it appears that experienced therapists may
be slightly more skilled at cultivating high lev-
els of working alliance.
23,25
Given its importance, it is understandable
that there is concern about how well alliance
can be established online. Indeed, if the ab-
sence of nonverbal cues impairs relationship
development, it is likely that the future of the
Internet as a medium for delivery of psycho-
logical services will be limited to information
provision, at least until videoconferencing is
more widely available. Yet both research and
anecdotal data suggest that strong and lasting
relationships are being formed online.
For instance, in one of the earliest studies of
online relationships, Walther and Burgoon
26
found that relationship development in a
computer-mediated sample approximated that
of a face-to-face sample. More recently, Parks
and Roberts
27
conducted a study of relationship
development in on-line, real-time, text-based
virtual environments called MOOs (Multi-User
Dimensions, Object Oriented). They found that
93.6% of their sample had formed ongoing per-
sonal relationships that they identified as either
close friendships, friendships, or romantic rela-
tionships. A comparison of members’ off-line
relationships to their on-line relationships re-
vealed no differences in terms of the levels of
breadth and depth that were achieved.
Miller and Gergen,
28
in a study of exchanges
occurring on an electronic bulletin board de-
voted to the topic of suicide, found the most
frequent forms of discourse (in order of fre-
quency) to be empathic understanding, sup-
port, and gratitude for responses. Similarly
supportive communities have also been found
in a sports injury support group
29
and an
online women’s mailing list.
30
Finn
31
studied
online self-help groups for individuals with
disabilities and found that participants consis-
tently provided supportive, empathic, and
therapeutic responses, enabling members to
connect and form a community that would not
have otherwise been possible.
In one of the most comprehensive research
efforts studying relationship formation on the
Internet, McKenna
32
conducted four experi-
mental and descriptive studies and consis-
tently found evidence of intimate relationship
development online. In one project, McKenna
randomly selected ten newsgroups on the In-
ternet and solicited participants for in-depth
interviews. Of the 30 participants (57% fe-
male), 51% of males and 60% of females tran-
scended their exclusively online relationships
and met face-to-face, 23% of males and 21% of
females had an affair, and 3.8% of males and
7.5% of females eventually married partners
they initially met online, demonstrating that
significant relationship formation occurred
in this sample. Like Parks and Roberts,
27
McKenna found no significant differences in
the depth or breadth of interactions between
participants’ online and offline friendships.
96 COOK AND DOYLE
Although a thorough description is beyond
the scope of this paper, McKenna
32
hypothe-
sized that the lack of gating features (i.e., nor-
mal barriers to friendship due to a variety of
elements such as physical appearance and geo-
graphic location), accompanied by anonymity
of participants, allow many Internet-based re-
lationships to form very quickly and become
closer and deeper than “real life” friendships.
Anonymity, or perceived anonymity, may fos-
ter intimacy by increasing the amount of per-
sonal, self-disclosure in friendships on the
Internet, where the fear of rejection that may
prevent disclosure in face-to-face relationships
does not exist. In short, people are often more
frank when they feel anonymous, as many do
over the Internet, and this leads indirectly to
greater intimacy.
In an intriguing finding that may provide a
clue to the temperament of some who seek
counseling online, McKenna provides evi-
dence to counter the claim that Internet use in-
creases loneliness and isolation. Instead, she
found that those who are already lonely and
socially anxious and who have difficulties es-
tablishing face-to-face relationships, are more
likely to develop relationships on the Internet.
Rather than causing loneliness and isolation,
the Internet seems to provide an outlet that al-
lows those who are lonely and socially anxious
to connect with others in a safe, controlled en-
vironment. In addition, the friendships formed
over the Internet are frequently integrated into
users’ real lives, decreasing loneliness and so-
cial anxiety and enhancing social skills.
The cumulative results of research on Inter-
net relationship development indicate that re-
lationships are frequently formed online and
that they are as strong and viable as their off-
line counterparts. In addition, many, if not
most, of the relationships developed on the In-
ternet occur through text-based mediums,
often considered the least expressive mode of
communication online. It is likely that many of
these same elements (e.g., empathy, support,
and a readiness to self-disclose) that develop
informally and foster deep and intimate rela-
tionship formation in online environments,
will enable therapists and clients to establish a
robust working alliance in therapy. Conse-
quently, we hypothesized that working al-
liance levels in a sample of participants who
were receiving counseling online would not be
significantly different from a representative
sample of face-to-face therapy recipients.
MATERIALS AND METHODS
Participant recruitment
Participants were recruited with the assis-
tance of online therapists in one of two ways:
through a link posted on therapists’ web pages
or through an e-mail announcement sent by
therapists. Therapists were found by browsing
lists of online therapists, performing Internet
searches, and through referrals. One hundred
individual therapists and group practices were
contacted by e-mail and asked to inform
clients about this research opportunity. In ad-
dition, several posts explaining the research
project and asking therapists’ help in recruit-
ing clients were made to e-mail lists, including
the general and members-only groups hosted
by the International Society for Mental Health
Online (ISMHO). A total of 16 therapists
agreed to inform their clients, but participant
responses were limited to one male and four
female therapists (one Ph.D., three master’s
level, and one master’s student) operating
from a range of theoretical orientations. Clients
recruited for participation were given a link to
the home page for the research project
(www.lclark.edu/
,
jecook) and a user name
and password unique to their therapist.
Participants
Participants were one male and 14 female
recipients of online therapy ranging in age
from 19 to 80 years (
M
= 41.40;
SD
= 15.99).
While the majority were United States citizens
(53.3%), one-third (33.3%) were Canadian, due
to the presence of a Canadian counselor. One
participant listed the Philippines as the coun-
try of citizenship, and one did not indicate citi-
zenship other than not being an American
citizen.
Participants were primarily white and
tended to be well-educated; all participants
had at least a high school diploma or GED,
WORKING ALLIANCE IN ONLINE THERAPY 97
and the majority had a college or graduate de-
gree. Participant income was spread across all
response choices with most reporting an an-
nual income of $25,000–39,999. Weekly time
spent online ranged from less than 1 hour to
more than 20 hours, with the majority (60%)
indicating that they spent between 1 and 10
hours online per week.
All participants in this sample received indi-
vidual counseling and communicated with
therapists through a text-based medium—
either e-mail or chat. A third of the sample also
reported communicating with therapists
through a second modality, three using e-mail,
one using chat, and one using audio confer-
encing. All participants who indicated that
chat was their primary mode of communica-
tion also used e-mail as a secondary method.
Most participants were seeking counseling for
relationship issues (40%) and depression (20%;
Table 1).
Clients were asked to complete the WAI im-
mediately after the third session, but prelimi-
nary results indicated a ceiling effect for
number of sessions, with all clients answering
“more than three,” the highest possible answer
choice. As a result, we added answer choices
to this question after data collection had
begun, making it impossible to accurately de-
termine the average number of sessions clients
had received, although the range was from
one session to more than five.
Measures
Working Alliance Inventory.
Although there
are a variety of alliance measurement instru-
ments, only the Working Alliance Inventory
1,2
(WAI) was designed to avoid any specific the-
oretical bias and to apply across theoretical
orientations. Based on the work of Bordin,
33
the WAI consists of a measure of overall al-
liance, as well as three subscales: tasks, bonds,
and goals. The tasks dimension refers to the
collaboration between therapeutic partners on
specific, technical, in-session behaviors and
techniques (e.g., guided imagery, role plays,
exposure therapy). Goals refers to the degree
that therapist and client agree on the desired
outcomes of therapy. The concept of bonds
is similar to the empathy construct. It refers
to the human relationship between therapist
and client in which trust and attachment are
formed so that the intimate work of therapy
can progress.
The WAI is a 36-item self-report question-
naire with 12 questions in each subscale that
respondents answer on a fully anchored seven-
point Likert scale from never to always. Using
a multitrait-multimethod analysis, Horvath
and Greenberg
1
found good construct validity
and high internal consistency on the compos-
98 COOK AND DOYLE
T
ABLE
1. P
ARTICIPANT
C
HARACTERISTICS
n %
Sex
Male 1 6.7
Female 14 93.3
Race
White 14 93.3
Filipino 1 6.7
Nationality
United States 8 53.3
Canada 5 33.3
Philippines 1 6.7
Other 1 6.7
Primary modality used
E-mail 12 80.0
Chat 3 20.0
Secondary modality used
None 10 66.7
E-mail 3 20.0
Chat 1 6.7
Audio conference 1 6.7
Presenting problem
Depression 3 20.0
Anxiety 1 6.7
Relationship issues 6 40.0
Family issues 1 6.7
School issues 1 6.7
Grief/bereavement issues 1 6.7
Other 2 13.3
Weekly time spent online
Less than 1 hour 1 6.7
1–5 hours 5 33.3
6–10 hours 4 26.7
11–20 hours 3 20.0
More than 20 hours 2 13.3
Education
High school diploma/GED 1 6.7
Some college 5 33.3
College degree 6 40.0
Some graduate school 1 6.7
Graduate degree 2 13.3
Income
Under $25,000 3 20.0
$25,000–$39,999 5 33.3
$40,000–$54,999 4 26.7
$55,000–$70,000 2 13.3
Over $70,000 1 6.7
ite score (0.93 on the client form) as well as
good internal consistency estimates for the
subscales (0.85–0.88 on the client form). Koko-
tovic and Tracey
21
found even higher subscale
reliability estimates in their sample ranging
from 0.88 to 0.91 on the client version. Reliabil-
ity coefficients from the current study were
slightly lower than those found elsewhere,
with a composite alpha of 0.86 and subscale
scores of 0.59, 0.70, and 0.76 for tasks, bonds,
and goals respectively. Studies that have com-
pared the WAI with other working alliance
instruments
17,19,34
have found high intercorre-
lations among the measures, providing addi-
tional support for its construct validity.
Although the WAI attempts to measure
three distinct constructs (tasks, bonds, and
goals) as hypothesized by Bordin,
33
factor
analysis has not yielded clear evidence of dis-
crete delineations between the subscales. Both
confirmatory
35
and exploratory
36
factor analy-
sis have revealed that the composite score is
the most salient measurement of working al-
liance. In particular, Hatcher and Barends
36
found that the goals and tasks subscales
grouped together, indicating a large degree of
overlap between these two constructs. This re-
search suggests that interpretation of the WAI
yields more meaningful results from the com-
posite score than from the individual subscale
scores.
In a meta-analysis of eight studies that
used the WAI to investigate the relationship
between working alliance and outcome, Hor-
vath
34
found an average weighted effect size
of 0.33 for the client-based measures, but es-
timated that the actual size may be closer to
the upper bound of the 95% confidence inter-
val (0.43) due to the use of a conservative
procedure to estimate effect size. These fig-
ures suggest a moderate link between clients’
perception of the working alliance as mea-
sured by the WAI and therapeutic outcome,
similar to alliance results obtained else-
where.
14,16,36
Meta-analysis has also revealed that early
alliance is slightly more predictive of outcome
than alliance measures taken as an average
across all sessions or taken towards the middle
of treatment. Eaton et al.
37
found that level of
alliance, regardless of the length of therapy,
was established within the first three sessions.
Accordingly, most alliance research has used
the completion of the third session as the
time to have participants complete alliance
measurements.
Demographics questionnaire.
In addition to the
WAI, participants completed 11 demographic-
type questions in order to obtain general,
descriptive information. Examples of these
questions included participants’ sex, age, in-
come level, education level, presenting problem,
and method of communication with therapists
(e.g., chat, e-mail, videoconferencing).
Additional comments.
Upon completing the
demographics questionnaire and the WAI,
participants were given the opportunity to
type comments about their experience of re-
ceiving counseling online.
Procedure
At the home page, participants had the
choice of beginning the study, or viewing sev-
eral other pages with background and contact
information, as well as a link where results of
the research were posted when the study was
complete. Those who chose to participate were
directed to the informed consent page and
provided with general information about the
nature of the study, a definition of working al-
liance, and a description of the questionnaires.
Information in the informed consent ensured
participants of their anonymity, unless they
entered an optional $50 raffle, in which case
they were asked to provide an e-mail address
that was treated confidentially and deleted
upon completion of the study. Participants
were given information about potential risks
and benefits to participating in the study and
informed of their freedom to withdraw con-
sent at any time and immediately exit the
study by using a link at the bottom right of
each page. Disagreeing to the informed con-
sent returned individuals to the home page,
while agreeing allowed participants to con-
tinue, redirecting them to a Microsoft Win-
dows 2000 Professional system using an
Internet Information Systems 5 web server. A
security dialogue box (based on Windows NT
WORKING ALLIANCE IN ONLINE THERAPY 99
File System [NTFS] technology) would now
open, asking participants for their user name
and password. The site was password pro-
tected both to reduce the possibility of errone-
ous responses from individuals not eligible to
participate and to track participant data based
on referring therapist. After entering a correct
user name and password, the demographics
questionnaire was displayed and participants
could begin. In order to ensure complete re-
sponses, a JavaScript application was used on
this page and on the 3 pages of the WAI to in-
form participants if they had left any questions
unanswered. Unless they chose to exit the
study, participants could not continue without
answering all the questions.
Upon completion of the demographics page,
participants were directed to the instructions
for the Working Alliance Inventory. Consistent
with the suggestions of Rosenfeld and col-
leagues,
38
the instructions and WAI questions
were transcribed online as accurately as possi-
ble from the paper and pencil form. Partici-
pants were presented with the 36 questions of
the WAI across three web pages. When they
had finished, participants were able to enter
comments about online therapy in a text box,
although completion of the comments page
was optional and users were not prompted to
enter information if they left it blank.
In order to reduce or eliminate data entry er-
rors, participant responses were automatically
coded into number values (where appropriate)
and entered into a Microsoft Access 2000 data-
base, using Active Server Page (ASP) tech-
nology. Participants’ Internet Protocol (IP)
address and the date and time of data entry
were also entered into the database in order to
check for multiple submissions by the same re-
spondent, a possible disadvantage to conduct-
ing research over the Internet noted by several
researchers.
39–41
Upon completion of the study,
results were imported from Access into the
Statistical Package for the Social Sciences
(SPSS 10) for statistical analysis.
RESULTS
The face-to-face comparison group was the
small sample (
N
= 25) on which the WAI was
initially validated.
1,2
To assess the representa-
tiveness of the scores from this sample, we
computed the standard error of measurement
(SEM) and created a confidence interval with
which to compare scores from four other stud-
ies that used the WAI (combined
N
of 182).
Comparing the results of the four studies to the
initial validation sample revealed that 11 of
the 12 subscale scores fell within one SEM of the
initial validation scores. The other subscale,
from the study with the smallest sample
(
N
= 22), was slightly outside the 68% confi-
dence interval, but was within the 95% confi-
dence interval, or approximately two SEMs.
These results indicate that despite the lack of a
large sample, scores on the initial validation of
the WAI nevertheless seem to be representative.
Single sample
t
tests were used to conduct
the primary analysis, in which working al-
liance composite and subscale scores from the
online sample were compared to scores from
the representative sample of face-to-face ther-
apy clients (for scores, see Table 2). Subscale
and composite scores were all higher for the
online therapy sample, although only the goal
(
t
(14)
= 3.039;
p
< 0.01) and composite (
t
(14)
=
2.307,
p
< 0.05) scores were significantly differ-
ent from the scores in the face-to-face sample.
Given the small sample and corresponding
lack of statistical power, the effect sizes for
the significant
t
scores were relatively large at
d
= 0.79 and
d
= 0.60 for the goal and compos-
ite scales respectively. Post-hoc power analy-
ses were conducted on the scales that did not
reach significance, using G*Power.
42
These re-
sults ranged from 0.22 to 0.33, indicating a
large potential for type II error.
100 COOK AND DOYLE
T
ABLE
2. S
AMPLE
M
EANS AND
T T
EST
C
OMPARISONS
BETWEEN
O
NLINE
T
HERAPY AND
F
ACE
-
TO
-F
ACE
S
AMPLES
Sample means
Online Face-
therapy to-face
WAI scale (N = 15) (N = 25) t
a
p
Task 70.33 68.6 1.260 0.228
Bond 72.47 69.6 1.627 0.126
Goal 72.27 67.3 3.039 0.009
Composite 215.07 205.5 2.307 0.037
a
df = 14.
Additional comparisons were conducted to
determine whether there were differences in
working alliance based on modality of com-
munication. Although participants who used
chat for the primary mode of communication
reported consistently higher means on the
WAI, no significant differences were found.
Results among those who used a secondary
modality also revealed a trend of higher (but
again nonsignificant) means on all WAI sub-
scales and the composite score (Table 3).
Given the small numbers in each category,
presenting problems were combined into three
groups for statistical analysis: depression and
anxiety, relationship issues, and other (school
issues, family issues, grief/bereavement, and
dependency issues). One-way analysis of vari-
ance revealed no significant differences or
trends on WAI subscale or composite scores
based on type of presenting problem (Table 4).
Additional analyses, conducted to test for
differences based on other variables, revealed
no significant differences in working alliance
between therapists, participant weekly time
online, participant education, participant in-
come, or participants’ nationality.
Comments
Nine of the 14 participants entered addi-
tional comments at the end of the question-
naire. Seven themes emerged from an analysis
of the comments. These themes are described
in greater detail below along with the number
of participants making that type of comment
(in parentheses): viability (six), disinhibition
(five), cost (three), travel (three), client/coun-
selor relationship (three), advantages of writ-
ten communication (three), and convenience
and flexibility (two).
Viability.
This theme refers to participants’
feeling that online therapy is a viable method
for receiving mental health assistance. Six par-
ticipants explicitly stated that they believe on-
line therapy to be effective.
Disinhibition.
This theme was discussed in
the most depth. Participants described the
sense of freedom they felt to express them-
selves online without embarrassment or fear
of judgment from therapists. Many expressed
the stress they typically feel in a face-to-face
therapy situation and indicated that, for the
first time, they were able to be completely hon-
est and open with a therapist.
Cost.
Participants who discussed cost were
uniformly pleased with the affordability of on-
line therapy.
Travel.
These comments focused on the ad-
vantage of online therapy for those who have
mobility challenges or who live in an isolated
region.
Client/counselor relationship.
Two participants
discussed the strength of the relationship with
their therapists, noting the care and respect
their therapists demonstrated. Another partic-
ipant indicated the importance of establishing
an honest relationship with the therapist.
Advantages of written communication.
These
comments focused on participants’ percep-
tions of the unique advantages to text-based
WORKING ALLIANCE IN ONLINE THERAPY 101
T
ABLE
3. W
ORKING
A
LLIANCE
M
EANS AND
S
TANDARD
D
EVIATIONS
B
ASED ON
M
ODALITY
Primary modality Secondary modality
E-mail Chat None Other
a
(n = 12) (n = 3) (n = 10) (n = 5)
WAI scale Mean SD Mean SD Mean SD Mean SD
Task 69.75 4.94 72.67 7.37 68.80 4.59 73.40 5.86
Bond 70.92 6.52 78.67 4.51 70.50 7.11 76.40 4.51
Goal 71.00 5.89 77.33 6.43 70.70 6.11 75.40 6.15
Composite 211.67 14.47 228.67 17.50 210.00 14.82 225.20 14.70
a
Other consists of e-mail (n = 3), chat (n = 1), and audio conferencing (n = 1).
therapy. Participants appreciated the ability to
re-read messages from therapists and noted
that this allowed for greater cognitive process-
ing than verbal communication. Participants
also described the greater ease they felt to ex-
press thoughts and feelings through writing.
Convenience and flexibility.
Two participants
described the convenience and flexibility af-
forded by online therapy, including the ab-
sence of scheduling issues, lack of travel time
and parking challenges, and not needing to be
concerned with personal grooming. One par-
ticipant noted the advantage to being able to
communicate intimate thoughts when most
personally appropriate.
DISCUSSION
Results from this small, self-selected sample
of online therapy consumers should be inter-
preted and generalized cautiously. Despite
limitations, however, working alliance levels
demonstrate that participants felt a collabora-
tive, bonding relationship with therapists, and
comments overwhelmingly indicated partici-
pants’ belief that online therapy was a positive
experience with unique advantages over face-
to-face counseling.
Based on previous research documenting
the depth and breadth of online relationship
development,
27,30,32
we predicted that working
alliance levels from the online sample would
be comparable to the face-to-face population.
If this was incorrect, and alliance is not able to
be adequately established online, we would
have expected WAI scores from the online
sample to be lower than those from the face-to-
face sample. In fact, results revealed that all WAI
subscales and the composite score were higher
in the online sample. Exceeding expectations
and despite low power, results on the composite
scale and goal subscale of the WAI were signifi-
cantly higher than those from the representative
sample of face-to-face clients. These findings
strongly suggest that, at least among those who
find online therapy to be an appealing method
of receiving mental health assistance, working
alliance, and perhaps most importantly, an em-
pathic relationship, can be strongly established
regardless of modality of communication.
While all WAI subscales were higher in the
online sample, only the therapeutic goals sub-
scale was significantly higher than the face-to-
face comparison group. Although subscale
scores should be interpreted with caution, it
may be that online therapy can be a particularly
effective method for therapeutic participants to
clearly establish and agree upon the outcomes
of therapy. Upon reflection this result may not
be surprising, given that the easily accessible,
written record of agreed-upon goals in text-
based therapy leaves less room for ambiguity
than a verbal agreement in traditional therapy.
Online participants’ relatively low rating of the
task subscale in comparison to the other sub-
scales from the online sample (although still
higher than the face-to-face sample) may be ex-
plained by the smaller role of traditional in-ses-
sion tasks in asynchronous therapy.
Despite the absence of significant differ-
ences in working alliance level between com-
munication modalities, a trend was observed
indicating that participants who corresponded
with therapists using more than one modality
102 COOK AND DOYLE
T
ABLE
4. W
ORKING
A
LLIANCE
M
EANS AND
S
TANDARD
D
EVIATIONS
B
ASED ON
C
LIENT
P
RESENTING
P
ROBLEM
Depression and anxiety Relationship issues Other
a
(n = 4) (n = 6) (n = 5)
WAI scale Mean SD Mean SD Mean SD
Task 74.25 8.42 69.17 3.54 68.60 3.05
Bond 74.00 8.29 74.00 2.28 69.40 9.34
Goal 76.00 4.69 72.33 6.95 69.20 6.14
Composite 224.25 20.79 215.50 11.64 207.20 15.72
a
Other includes school issues, family issues, grief/bereavement, and dependency
issues.
consistently had higher composite and sub-
scale scores on the WAI. It is possible that the
limited sample size in this study masked sig-
nificant effects about modality that may be
found in a larger sample. The impact on rela-
tionship development of communicating in
multiple formats, especially of a text-based na-
ture, has the potential to be an important area
for future research.
As expected, the type of problems for which
participants sought help did not significantly
impact working alliance scores. This finding is
similar to previous research on working al-
liance development in face-to-face therapy
showing that severity and type of client pre-
senting problems are unrelated to working
alliance.
20,21
The overwhelming majority of participants
in this study were female, consistent with previ-
ous research that has found that more women
than men use the Internet for mental health in-
formation and services.
4,43,44
Interestingly, par-
ticipants were well-represented at all income
levels, in contrast to surveys that have shown
those with higher incomes to be over repre-
sented in the Internet population.
3
Participants
did, however, tend to be highly educated.
Lack of control presented difficulties in this
study that may have affected sample selection
and results. For instance, many participants
had received more than three sessions when
they completed the questionnaires, even though
they were asked to complete them after the
third. However, Horvath and Symonds’
16
find-
ing that alliance is fairly stable when taken as
an average over the course of therapy, suggests
that results may not have been greatly affected.
It is also unclear how systematically participat-
ing therapists informed clients of this research
project. Bias may have been introduced by the
method therapists used to notify clients, al-
though this was probably less of a factor for
those who posted notifications on their web
sites.
Online therapy is not for everyone and many
will continue to prefer and need face-to-face
counseling. Previous research has indicated
that individuals with certain personality char-
acteristics, such as introversion,
45
and particular
problems, such as panic disorder and agora-
phobia,
10
may prefer the perceived anonymity
and sense of control the Internet offers. These
findings are supported by participants’ com-
ments that disinhibition was a major attraction
to online therapy. The importance for partici-
pants of the disinhibiting effects of online ther-
apy also suggests that the population who
seeks online therapy may differ from the gen-
eral therapy-seeking public. Those who go on-
line for therapy may, in fact, share more in
common with the population described by
McKenna.
32
Many in McKenna’s sample tended
to be uncomfortable in traditional social situa-
tions and turned to the Internet to connect with
others. For the individuals in the current study,
it is conceivable that online therapy, in addition
to frequently being more affordable, is more
conducive to strong therapeutic relationship
development than face-to-face therapy. The in-
teraction between online relationship develop-
ment and personality type is a fertile area for
future research.
Large-scale research projects examining
both working alliance and treatment outcomes
are necessary to increase the validity of re-
search into online counseling. It will also be
important for future research to obtain a wider
cross section of therapists willing to partici-
pate. Nevertheless, this project adds to a
rapidly developing body of research indicat-
ing that the delivery of mental health services
through the Internet is both plausible and ef-
fective and that the empathic bond, central to
working alliance and successful outcome, can
occur when therapeutic participants are geo-
graphically separated.
ACKNOWLEDGMENTS
We would like to acknowledge and thank the
online therapists who informed their clients of
this research project. Without their help, this
study could not have been conducted.
REFERENCES
1. Horvath, A.O., & Greenberg, L.S. (1986). The devel-
opment of the Working Alliance Inventory. In: Pin-
sof, W.M., ed. The psychotherapeutic process: A research
handbook. New York: Guilford Press, pp. 529–556.
WORKING ALLIANCE IN ONLINE THERAPY 103
2. Horvath, A.O., & Greenberg, L.S. (1989). Develop-
ment and validation of the Working Alliance Inven-
tory. Journal of Counseling Psychology, 36:223–233.
3. Rainie, L., Packel, D., Fox, S., et al. (2001). More on-
line, doing more: 16 million newcomers gain Internet
access in the last half of 2000 as women, minorities,
and families with modest incomes continue to surge
online [On-line]. Available: www.pewinternet.org/
reports/pdfs/PIP Changing Population.pdf.
4. Powell, T. (1998). Online counseling: a profile and
descriptive analysis [On-line]. Available: http://
netpsych.com/Powell.htm.
5. Cohen, G.E., & Kerr, B.A. (1998). Computer-
mediated counseling: an empirical study of a new
mental health treatment. Computers in Human Ser-
vices, 15: 13–26.
6. Day, S.X., & Schneider, P. (2000). The subjective expe-
riences of therapists in face-to-face, video, and audio
sessions. In: Walz, G.R., ed. Cybercounseling and cy-
berlearning: strategies and resources for the millennium.
Alexandria, VA: American Counseling Association,
pp. 203–218.
7. Barak, A., & Wander-Schwartz, M. (2000). Empirical
evaluation of brief group therapy conducted in an
Internet chat room. Journal of Virtual Environments.
Available: www.brandeis.edu/pubs/jove/html/v5/
v5n1jove.html.
8. Lange, A., Schrieken, B., van de Ven, J.-P., et al.
(2000). “Interapy”: the effects of a short protocolled
treatment of posttraumatic stress and pathological
grief through the Internet. Behavioural & Cognitive
Psychotherapy, 28:175–192.
9. Celio, A.A., Winzelberg, A.J., Wilfley, D.E., et al.
(2000). Reducing risk factors for eating disorders:
comparison of an Internet- and a classroom-
delivered psychoeducational program. Journal of
Consulting & Clinical Psychology, 68:650–657.
10. Bouchard, S., Payeur, R., Rivard, V., et al. (2000).
Cognitive behavior therapy for panic disorder with
agoraphobia in videoconference: preliminary re-
sults. Cyberpsychology & Behavior, 3:999–1007.
11. Fenichel, M. (2001). Online psychotherapy: technical
difficulties, formulations and processes [On-line].
Available: www.fenichel.com/technical.shtml.
12. Childress, C. (1998). Potential risks and benefits of
online psychotherapeutic interventions [On-line].
Available: www.ismho.org/issues/9801.htm.
13. Barak, A. (1999). Psychological applications on the
internet: a discipline on the threshold of a new mil-
lennium. Applied & Preventive Psychology, 8:231–245.
14. Lambert, M.J. (1992). Psychotherapy outcome re-
search: implications for integrative and eclectical
therapists. In: Goldfried, M.R., ed. Handbook of psy-
chotherapy integration. New York: Basic Books, pp.
94–129.
15. Bachelor, A., & Horvath, A. (1999). The therapeutic
relationship. In: Duncan, B.L., ed. The heart and soul of
change: what works in therapy. Washington, DC: Amer-
ican Psychological Association, pp. 133–178.
16. Horvath, A.O., & Symonds, B.D. (1991). Relation be-
tween working alliance and outcome in psychother-
apy: a meta-analysis. Journal of Counseling Psychology,
38:139–149.
17. Tichenor, V., & Hill, C.E. (1989). A comparison of six
measures of working alliance. Psychotherapy, 26:
195–199.
18. Hatcher, R.L., Barends, A., Hansell, J., et al. (1995).
Patients’ and therapists’ shared and unique views of
the therapeutic alliance: an investigation using con-
firmatory factor analysis in a nested design. Journal
of Consulting & Clinical Psychology, 63:636–643.
19. Safran, J.D., & Wallner, L.K. (1991). The relative pre-
dictive validity of two therapeutic alliance measures
in cognitive therapy. Psychological Assessment, 3:
188–195.
20. Horvath, A.O. (1995). The therapeutic relationship:
from transference to alliance. In Session—Psychother-
apy in Practice, 1:7–17.
21. Kokotovic, A.M., & Tracey, T.J. (1990). Working al-
liance in the early phase of counseling. Journal of
Counseling Psychology, 37:16–21.
22. Horvath, A.O., & Luborsky, L. (1993). The role of the
therapeutic alliance in psychotherapy. Journal of Con-
sulting & Clinical Psychology, 61:561–573.
23. Dunkle, J.H. (1996). Contribution of therapist experi-
ence and personal characteristics to the working al-
liance. Journal of Counseling Psychology, 43:456–460.
24. Tokar, D.M., Hardin, S.I., Adams, E.M., et al. (1996).
Clients’ expectations about counseling and percep-
tions of the working alliance. Journal of College Stu-
dent Psychotherapy, 11:9–26.
25. Mallinckrodt, B., & Nelson, M.L. (1991). Counselor
training level and the formation of the psychothera-
peutic working alliance. Journal of Counseling Psy-
chology, 38:133–138.
26. Walther, J.B., & Burgoon, J.K. (1992). Relational
communication in computer-mediated interaction.
Human Communication Research, 19:50–88.
27. Parks, M.R., & Roberts, L.D. (1998). “Making
MOOsic”: the development of personal relationships
on line and a comparison to their off-line counter-
parts. Journal of Social & Personal Relationships, 15:
517–537.
28. Miller, J.K., & Gergen, K.J. (1998). Life on the line: the
therapeutic potentials of computer-mediated con-
versation. Journal of Marital & Family Therapy, 24:
189–202.
29. Preece, J. (1999). Empathic communities: balancing
emotional and factual communication. Interacting
with Computers, 12:63–77.
30. Adams-Price, C.E., & Chandler, S. (2000). The star
fleet ladies auxiliary: evolution of an online women’s
mailing list. Cyberpsychology & Behavior, 3:811–816.
31. Finn, J. (1999). An exploration of helping processes
in an online self-help group focusing on issues of
disability. Health & Social Work, 24:220–231.
32. McKenna, K.Y.A. (1998). The computers that bind:
relationship formation on the Internet [Doctoral dis-
sertation]. Ohio University.
33. Bordin, E.S. (1979). The generalizability of the psy-
choanalytic concept of the working alliance. Psycho-
therapy: Theory, Research & Practice, 16:252–260.
104 COOK AND DOYLE
34. Horvath, A.O. (1994). Empirical validation of Bor-
din’s pantheoretical model of the alliance: the Work-
ing Alliance Inventory perspective. In: Greenberg,
L.S., ed. The working alliance: theory, research, and prac-
tice. New York: John Wiley & Sons, pp. 109–128.
35. Tracey, T.J., & Kokotovic, A.M. (1989). Factor struc-
ture of the Working Alliance Inventory. Psychological
Assessment, 1:207–210.
36. Hatcher, R.L., & Barends, A.W. (1996). Patients’ view
of the alliance in psychotherapy: exploratory factor
analysis of three alliance measures. Journal of Con-
sulting & Clinical Psychology, 64:1326–1336.
37. Eaton, T.T., Abeles, N., & Gutfreund, M.J. (1988).
Therapeutic alliance and outcome: impact of treat-
ment length and pretreatment symptomatology. Psy-
chotherapy, 25:536–542.
38. Rosenfeld, P., Booth-Kewley, S., & Edwards, J.E.
(1993). Computer-administered surveys in organiza-
tional settings: alternatives, advantages, and appli-
cations. American Behavioral Scientist, 36:485–511.
39. Hewson, C.M., Laurent, D., & Vogel, C.M. (1996).
Proper methodologies for psychological and socio-
logical studies conducted via the Internet. Behavior
Research Methods, Instruments, & Computers, 28:
186–191.
40. Pasveer, K.A., & Ellard, J.H. (1998). The making of a
personality inventory: help from the WWW. Behavior
Research Methods, Instruments, & Computers, 30:
309–313.
41. Schmidt, W.C. (1997). World-Wide Web survey re-
search: benefits, potential problems, and solutions.
Behavior Research Methods, Instruments, & Computers,
29:274–279.
42. Erdfelder, E., Faul, F., & Buchner, A. (1996). GPOWER:
a general power analysis program. Behavior Research
Methods, Instruments, & Computers, 28: 1–11.
43. Fox, S., Rainie, L., Horrigan, J., et al. (2000). The
online health care revolution: how the web helps
Americans take better care of themselves [On-line].
Available: www.pewinternet.org/reports/pdfs/
PIP_Health Report.pdf.
44. Stubbs, P. (2000). Mental health care online [On-line].
Available: www.geocities.com/online form/.
45. Hamburger, Y.A., & Ben-Artzi, E. (2000). The rela-
tionship between extraversion and neuroticism and
the different uses of the Internet. Computers in
Human Behavior, 16:441–449.
Address reprint requests to:
Jonathan E. Cook, M.S.
Regional Institute for Human Services
Portland State University
P.O. Box 751
Portland, OR 97207–0751
E-mail:
cookj@rri.pdx.edu
WORKING ALLIANCE IN ONLINE THERAPY 105
