Testing the Efficacy of Theoretically Derived Improvements in the
Treatment of Social Phobia
Ronald M. Rapee and Jonathan E. Gaston
Maree J. Abbott
University of Sydney
Recent theoretical models of social phobia suggest that targeting several specific cognitive factors in
treatment should enhance treatment efficacy over that of more traditional skills-based treatment pro-
grams. In the current study, 195 people with social phobia were randomly allocated to 1 of 3 treatments:
standard cognitive restructuring plus in vivo exposure, an “enhanced” treatment that augmented the
standard program with several additional treatment techniques (e.g., performance feedback, attention
retraining), and a nonspecific (stress management) treatment. The enhanced treatment demonstrated
significantly greater effects on diagnoses, diagnostic severity, and anxiety during a speech. The specific
treatments failed to differ significantly on self-report measures of social anxiety symptoms and life
interference, although they were both significantly better than the nonspecific treatment. The enhanced
treatment also showed significantly greater effects than standard treatment on 2 putative process
measures: cost of negative evaluation and negative views of one’s skills and appearance. Changes on
these process variables mediated differences between the treatments on changes in diagnostic severity.
Keywords: social phobia, treatment, cognitive processes
The majority of empirically validated treatment packages for
social phobia currently comprise a combination of exposure and
cognitive restructuring as their principal components (Hofmann,
2007; Rodebaugh, Holaway, & Heimberg, 2004). Several meta-
analyses have confirmed that exposure-based treatment packages,
either in isolation or in combination with other components, have
shown good efficacy (Fedoroff & Taylor, 2001; Feske & Chamb-
less, 1995; Gould, Buckminster, Pollack, Otto, & Yap, 1997;
Norton & Price, 2007). Overall, treatment packages comprising
exposure and cognitive restructuring have produced changes of
around 0.9 standard deviations at the end of treatment, and these
changes have been maintained over follow-up periods from 3 to 12
Current models of the maintenance of social phobia have begun
to provide fine-grained analyses of the cognitive processes in-
volved in the maintenance of social phobia (Clark & Wells, 1995;
Hofmann, 2007; Rapee & Heimberg, 1997). Some of the key
cognitive processes include an overestimate of the likelihood and
cost of negative evaluation, an excessively negative perception of
the self, attentional bias toward this negative self appraisal (or
toward external indicators of negative evaluation), and the use of
subtle behavioral strategies that reinforce the negative self percep-
tion. Empirical evidence has demonstrated support for the impor-
tance of these processes (e.g., Heinrichs & Hofmann, 2001; Musa
& Lepine, 2000; Rapee & Abbott, 2007). Of clinical relevance,
several studies have begun to demonstrate that social anxiety can
be reduced through direct manipulation of these psychological
factors (Bo ¨gels, 2006; Garcia-Palacios & Botella, 2003; Harvey,
Clark, Ehlers, & Rapee, 2000; Kim, 2005), although not all results
have been consistent (Rodebaugh & Chambless, 2002; Smits,
Powers, Buxkamper, & Telch, 2006). A more recent focus onto
schema-based interventions has also begun to be applied to the
management of social phobia (Pinto-Gouveia, Castilho, Galhardo,
& Cunha, 2006) and is consistent with the personality-like features
of the disorder (Rapee & Spence, 2004).
Incorporation of these principles into clinical management pro-
grams has received little empirical evaluation. The main exception
has been Clark’s group in the United Kingdom and their collabo-
rators who have reported a few trials over the past 5 years that
utilized what could be described as an “enhanced” cognitive–
behavioral therapy (CBT) model of social phobia treatment (Clark
et al., 2003, 2006; Stangier, Heidenreich, Peitz, Lauterbach, &
Clark, 2003). Effect sizes from this program have been impressive,
indicating pre-to posttreatment changes of around 2 standard de-
viations on self report measures of symptoms (Clark et al., 2003,
2006). However, some applications outside of the core research
team have produced considerably smaller effects (e.g., Stangier et
al., 2003; around 1.2 standard deviations for individual application
and 0.6 standard deviations for group application).
The theoretical considerations described above and the promis-
ing effects obtained by Clark’s group suggest that empirical com-
parison between treatments based on conventional or standard
CBT and a CBT package enhanced with these additional, theoret-
Ronald M. Rapee and Jonathan E. Gaston, Centre for Emotional Health,
Macquarie University, Sydney, Australia; Maree J. Abbott, Department of
Psychology, University of Sydney, Sydney, Australia.
This research was supported by National Health and Medical Research
Council Grant 192107 to Ronald M. Rapee. We would like to acknowledge
the assistance of Nicola Kemp and Anna Thompson in development of the
treatments and measures; Jennifer Mitchell and Lexine Stapinski for
project management; and Jennifer Allen, Leigh Carpenter, Amanda Gam-
ble, and Sarah Perini for research assistance. The many excellent therapists
included the above plus David Bonsor, Leah Campbell, Lissa Johnson,
Susan Kennedy, Natalie Robinson, and Julie Sposari.
Correspondence concerning this article should be addressed to Ronald
M. Rapee, Department of Psychology, Macquarie University, Sydney,
NSW, Australia, 2109. E-mail: Ron.Rapee@mq.edu.au
Journal of Consulting and Clinical Psychology
2009, Vol. 77, No. 2, 317–327
© 2009 American Psychological Association
ically derived components is highly warranted. At present, one
study has conducted such a comparison. Clark et al. (2006) com-
pared their enhanced cognitive therapy package against a basic
program consisting of exposure and applied relaxation. Results
indicated significantly stronger effects for the enhanced package
across all measures. However, this study did not compare the
enhanced package against a more traditional treatment package
consisting of exposure plus cognitive restructuring. In addition, it
did not assess changes in any putative mechanisms of treatment
efficacy, and hence it is not clear to what extent the enhanced
cognitive therapy modified these theoretically important processes.
The purpose of the current study was to determine whether a
treatment package for social phobia that incorporated the “newer
generation” of techniques would result in a greater reduction in
social anxiety than a traditional package that focused entirely on
verbal cognitive restructuring and in-vivo exposure. Because the
comparison “standard” CBT condition used by Clark et al. (2006)
did not include cognitive restructuring, which has long been con-
sidered key in the management of social phobia (Butler, 1985), and
because they utilized applied relaxation, which has previously
shown considerably smaller effects than other standard treatments
(Fedoroff & Taylor, 2001), we included a comparison CBT con-
dition that utilized a comprehensive combination of cognitive
restructuring and in vivo exposure. Aside from early work by
Heimberg et al. (1990), there have also been few comparisons of
exposure-based treatments against nonspecific skills-based treat-
ments. Hence, the current trial also included a comparison treat-
ment that focused on relaxation and general stress management but
did not include specific exposure and cognitive components aimed
at modifying social threat appraisals.
Participants for the study were 195 individuals meeting Diag-
nostic and Statistical Manual of Mental Disorders (4th ed.; DSM–
IV; American Psychiatric Association, 1994) criteria for social
phobia, randomly allocated to one of three treatment conditions:
(1) stress management, (2) cognitive restructuring and in vivo
exposure (standard), and (3) cognitive restructuring and exposure
“enhanced” with the additional strategies (enhanced). A planned
sample size (N ? 160) was determined to allow a minimum
difference of .50 standard deviation units to be expected to be
detected at an alpha of .05 with 80% power (Faul & Erdfelder,
1992) for main effects.
Participants were included if they were over 18 years and met
criteria for social phobia as their main (or most interfering) disor-
der. To maximize external validity, we minimized exclusions. The
only planned exclusions were problems requiring immediate at-
tention—such as clear suicidal intent, severe substance abuse or
dependence, or florid psychosis—which were assessed during the
structured interviews. Concurrent pharmacotherapy or psychother-
apy was allowed as long as dosages had been consistent for 3
months and there were no plans to change. No participants were in
concurrent psychotherapy. However, 26.7% were taking medica-
tions for their anxiety.
Diagnoses for Axis I disorders were made by graduate students
in clinical psychology using a structured clinical interview: the
Anxiety Disorders Interview Schedule for DSM–IV (ADIS-IV; Di
Nardo, Brown, & Barlow, 1994). All pretreatment interviews were
recorded, and 30 were randomly selected for coding by a second
rater for reliability. A principal diagnosis of social phobia was
rated with high reliability (? ? .86, p ? .001). Clinical severity of
each disorder was rated on a 0–8 scale reflecting severity relative
to other people with the disorder and life impact. The intraclass
correlation for the clinical severity rating of the principal diagnosis
among the current reliability sample was r ? .85 (95% confidence
interval ? .68–.93). Following treatment, all participants were
reinterviewed with the ADIS-IV addressing all anxiety disorders
plus any additional disorders that they met criteria for at pretreat-
ment. Reliability for posttreatment interviews was not assessed.
Among the current sample, 91.4% met criteria for the general-
ized subtype of social phobia. As would be expected, Axis I
comorbidity was also high: 37% met criteria for an additional
anxiety disorder (most commonly generalized anxiety disorder,
30%; specific phobia, 8%; obsessive–compulsive disorder, 5%);
29% met criteria for an additional mood disorder (major depres-
sion, 17%; dysthymia, 11%); and 9% met criteria for another
additional disorder (alcohol abuse, 3%). The mean age of the
sample was 33.7 years (SD ? 11.1), and 51.8% were female.
Measures were included for two main purposes: first to assess
clinical outcome and second to assess changes in relevant pro-
cesses. All measures were administered at pretreatment and post-
In addition to the diagnostic measures described above, ques-
tionnaire measures were designed to tap two constructs: symptoms
of social anxiety and life interference.
Social Interaction Anxiety Scale (SIAS) and Social Phobia Scale
(SPS; Mattick & Clarke, 1998).
the main fears and avoidance of social phobia, respectively focus-
ing on interaction fears and more specific performance-based
fears. They have excellent psychometric properties (Peters, 2000).
Albany Panic and Phobia Scale—Social Phobia subscale
(APPQ-S; Rapee, Craske, & Barlow, 1994).
items designed to tap social fears that are relatively distinct from
overlap with agoraphobic fears. Later examination has shown
consistent factor structure, solid reliability, and clear concurrent
validity (Brown, White, & Barlow, 2005).
Life Interference Scale (LIS).
impact of individuals’ social fears, we asked respondents to indi-
cate the impact of their fears on various components of their
life—including work, family life, and leisure activities—using six
Likert scales ranging from 0 to 8. The scales were summed to
provide a total interference rating from 0 (no interference) to 48
(maximum interference). Previous analysis in our center has shown
that the six items show excellent internal consistency (? ? .90),
and the total correlates significantly with the 12-item Short Form
Health Survey (Ware, Kosinski, & Keller, 1996)—Mental Com-
ponent subscale (r ? .56).
These companion scales assess
This is a brief set of
To provide a measure of the life
RAPEE, GASTON, AND ABBOTT
To assess the role of several putative psychological processes,
we also had participants complete measures of the following
processes at pretreatment and posttreatment.
We assessed this construct using the
seven public self consciousness items from the Self Consciousness
Scale (Fenigstein, Scheier, & Buss, 1975) plus an additional six
items designed for the current study to extend the scope of this
measure. The additional items were based on face validity and
aimed to assess the tendency to take an external, observer’s per-
spective during interactions—a construct that has been argued as
central to the maintenance of social phobia (Hackmann, Surawy, &
Clark, 1998). Items included the following: “I try to see myself
through other people’s eyes,” “I am aware of people’s reactions to
me,” and “I wonder how I look to other people.” For the current
sample, internal consistency of the 13-item scale was ? ? .93, and
it correlated moderately with the SIAS (r ? .56).
Safety behaviors and subtle avoidance.
gage in subtle avoidance or safety behaviors during social inter-
actions was assessed with the 32-item Subtle Avoidance Fre-
quency Examination (Cuming et al., 2008). Items include “speak
softly,” “wear cool items so as not to sweat,” and “make excuses
about your appearance.” Psychometric data indicate that the mea-
sure has strong internal consistency (? ? .91) and clear convergent
and discriminant validity (Cuming et al., 2008). For the current
sample, internal consistency of the scale was ? ? .90, and it
correlated moderately with the SIAS (r ? .59).
Negative mental representation of appearance and perfor-
We assessed the individual’s overall perception of their
abilities and appearance using a 15-item measure that was used in
previous experimental research that tapped two related factors of
perception of personal abilities and perception of personal appear-
ance (Rapee & Abbott, 2006). Items included “I look attractive,”
“I am a good dancer,” and “others perceive me as boring.” The
measure has previously been shown to discriminate people with
social phobia from nonclinical controls (Rapee & Abbott, 2006).
For the current sample, internal consistency of the scale was ? ?
.78, and it correlated moderately with the SIAS (r ? ?.53).
Probability and cost of negative evaluation.
sure of the probability and cost of negative evaluation was devel-
oped from previous research (Foa, Franklin, Perry, & Herbert,
1996; J. K. Wilson & Rapee, 2005). Participants read hypothetical
negative events and then rated (a) the likelihood that the event
would occur to them (probability) and (b) how bad it would be if
it did occur (cost). Items included “you will blush while being
introduced to a new person” and “you will be ignored by someone
you know.” Internal consistencies for the two sets of ratings were
?s ? .83 and .89, respectively, and they both correlated moder-
ately with the SIAS (rs ? .53, .58, respectively).
A 20-item list of core beliefs relevant to social
anxiety was developed, and participants were asked to rate the
extent to which they generally believed each one. Items included
“I am unlovable,” “I am inept,” and “I am dumb/stupid.” Internal
consistency for the current sample was ? ? .95, and it correlated
moderately with the SIAS (r ? .62).
We assessed participants’ perceptions
of the credibility of the treatment they were allocated to using three
items based on the measure by Borkovec and Nau (1972). The
The tendency to en-
A 13-item mea-
measure was administered at the end of Session 1 and Session 4 of
The 36-item Working Alliance Inventory
(Horvath & Greenberg, 1989) was given to all participants in
Session 4 to assess the relationship, support, and collaboration that
participants perceived from their therapists.
The 20-item Group Attitude Scale (Evans &
Jarvis, 1986) was also given to all participants in Session 4 to
assess each participant’s sense of belonging, positivity, and com-
mitment to the group.
Behavioral Observation Assessment
To provide an additional measure of treatment outcome, we had
participants attend a laboratory session before and following treat-
ment where they were asked to perform an impromptu speech.
This assessment was conducted by a research assistant who was
blind to group allocation and study hypotheses. Participants
were instructed to deliver a speech about a topic of their own
choosing to the experimenter and a video camera. Participants
were allowed to talk on any topic aside from their participation
in the research. They were given 2 min of preparation time and
were instructed to talk for 3 min. Any participant who stopped
in less than 3 min was briefly encouraged by the experimenter
to continue, but if he or she refused, the postspeech measures were
administered at that point. At pretreatment, a total of 8 participants
refused to give the speech, and an additional 18 participants failed
to complete the entire 3 min. At posttreatment, only 1 participant
refused to give the speech, and an additional 14 participants failed
to complete the entire 3 min.
Participants completed several measures of relevance to other
research (e.g., Rapee & Abbott, 2007). The main measure of
relevance to the current study was an assessment of state anxiety
during the speech. The measure comprised 10 items that were
conceptually related to a sense of current anxiousness in relation to
the speech (e.g., “I felt nervous” and “I felt like stopping the
speech”). Items were scored on a scale ranging from 0 to 4, and
participants were instructed to complete the items indicating how
they felt during delivery of their speech. Internal consistency of the
10 items in the current sample was ? ? .95.
General Treatment Factors
All three treatments were conducted in groups of approximately
6 participants. Therapists were clinical psychologists with specific
expertise in the treatment of social phobia. In most cases, a
graduate psychology student acted as cotherapist. All primary
therapists conducted groups in each of the treatment conditions.
Therapists received regular weekly supervision from Jonathan E.
Gaston. Supervision was primarily focused on maintaining the
treatment protocols and ensuring that relevant differences were
maintained between conditions. Groups extended for 12, 2-hr
sessions across 12 weeks. All three treatments followed manuals
for the therapists and were supported by printed materials and
handouts for participants.
Standard treatment comprised what would probably be referred
to as a “basic” cognitive–behavioral package. Treatment was
ENHANCED TREATMENT OF SOCIAL PHOBIA
based loosely on previous cognitive–behavioral programs (e.g.,
Heimberg et al., 1990; Mattick & Peters, 1988) but was restricted
to cover purely cognitive restructuring and in vivo exposure.
Cognitive restructuring followed principles described by Beck
(e.g., Beck, Emery, & Greenberg, 1985) and was based on refu-
tation of unrealistic beliefs through consideration of evidence.
Extension to broader overarching constructs (such as schemas) or
to cognitive processes (such as attentional bias or use of hypothesis
testing) was not done. In vivo exposure exercises were conducted
in a graduated and systematic fashion and were individually tai-
lored to target the specific social fears of each participant. The
rationale was based on the principles of habituation. Participants
were directed to develop individualized hierarchies and to conduct
systematic exposure between sessions. Assertiveness skills and
some general social skills practice were also included toward the
end of the program.
The enhanced treatment was developed by members of our unit
on the basis of a similar earlier program (Rapee & Sanderson,
1998). Sessions covered cognitive restructuring and in vivo expo-
sure in a similar manner to the standard treatment, although the
rationale for exposure followed more of a cognitive, evidence-
gathering concept, and the overlap between cognitive restructuring
and in vivo exposure was emphasized through use of hypothesis
testing. Data gathering was extended to incorporate identification
and refutation of underlying beliefs and overarching life scripts by
identification of these broader beliefs followed by examination of
evidence from across the individual’s life. In vivo exercises also
included elimination of safety behaviors and subtle avoidance as
well as realistic appraisal and feedback of performance. Partici-
pants also practiced exercises to train controlled attentional re-
sources away from negative evaluation and the self and toward the
task at hand through meditation-type exercises and in vivo practice
of focusing of attention to different features of the environment.
The purpose of the treatment was to provide credible procedures
that were aimed at management of stress and general anxiety but
did not specifically address social threat concerns or fears of
negative evaluation. The provided rationale was based on teaching
management of the fight or flight response. Specific components
were mostly focused on training in relaxation skills but also
included problem solving, time management, and healthy lifestyle
habits. Participants were encouraged to apply relaxation and other
skills when feeling stressed, but they were not instructed to sys-
tematically approach feared objects while relaxing.1
All treatment sessions were audiotaped, and a random sample
was coded for content. Of a total of 360 possible treatment ses-
sions, 60 (16.7%) were coded. Coders were therapists who had
been trained in and ran the treatments (although naturally they did
not rate their own groups), and they were blind to the specific
condition that they were rating. Coding was based on a measure
that listed 34 techniques or types of information that were included
in one or more of the treatment conditions. Coders listened to the
entire session and checked a given item when they heard that
technique or information covered by the therapist. A research
assistant later “unblinded” the allocated condition and determined
whether the checked items were “correct” with respect to the
allocated condition (i.e., were part of the allocated treatment con-
dition) or “incorrect” (i.e., should not have been included in that
condition). For example, the item “discouraged use of safety
behaviors” would have been rated correct in a session from the
enhanced condition but incorrect if it appeared in a session from
the other two conditions, whereas the item “look for evidence for
or against clients’ thoughts” would have been marked correct in
either the enhanced or standard conditions but incorrect if it
appeared in the stress management condition. Each treatment
condition had a different number of possible correct and incorrect
items from the checklist. Hence, to provide a simple indication of
treatment integrity, we calculated the number of correct compo-
nents identified by each rater as a percentage of the total number
of components identified. In other words, if only correct compo-
nents were identified, the percentage would have been 100%; if
equal correct and incorrect components had been identified, the
percentage would have been 50%. Raters also “guessed” which of
the treatment conditions the tape they listened to was from.
Eighteen (30%) of the integrity tapes were coded by a second
rater who was blind to treatment condition. There was complete
agreement between pairs of raters on the proportion of correct
components for 17 of the 18 sessions (there was no variance for
one cell, and hence statistics could not be calculated). Pairs of
raters agreed on the guessed condition represented by each tape in
17 of the 18 cases (? ? .91, p ? .01).
The procedures were approved by the Human Research Ethics
Committee, and all participants gave informed consent. Potential
participants contacted the Centre for Emotional Health at Macqua-
rie University (Sydney, Australia) between September 2002 and
February 2005 through usual referral sources, including general
practitioners, mental health professionals, occasional media cov-
erage, and word of mouth. They were screened via telephone, and
those who appeared to have anxiety-related difficulties were in-
vited in for a structured interview and behavioral assessment. A
total of 494 participants were assessed with structured interview,
of whom 265 did not meet study criteria, and an additional 34 were
randomly allocated to another study. The remaining 195 who met
inclusion criteria were randomly allocated to one of the three
conditions. Randomization was done on a group basis by the
project manager via a preassigned random schedule. Participants
were allocated in blocks of six to eight to allow for group delivery
and group start times. An individual participant was not allocated
to a condition until after he or she was accepted into the study—
that is, after initial diagnostic and questionnaire assessment. Fol-
lowing the structured diagnostic interview, participants completed
the questionnaire measures and were scheduled for the behavioral
assessment, usually within 1 week. Treatment generally began
1Copies of the manuals can be obtained from Ronald M. Rapee or
Jonathan E. Gaston.
RAPEE, GASTON, AND ABBOTT
within the following few weeks. At the conclusion of treatment (12
weeks), questionnaire, diagnostic interview, and laboratory mea-
sures were repeated and were conducted by clinicians who were
blind to participants’ group allocations. Because of ethical require-
ments, participants who requested additional treatment were then
offered the enhanced program, and the theoretically relevant com-
parison ended. In principle, this offer was open to all participants;
however, in reality, therapists did not generally extend the offer to
participants in the enhanced condition because they had already
received this treatment.
Primary outcomes for this trial were a reduction in clinical
diagnoses and severity of social phobia as assessed by the ADIS-
IV, reduction in social phobia symptoms, and reduction in self-
rated life interference. In addition, analyses were conducted to
examine differences between treatments on reductions in theoret-
ically meaningful cognitive variables, and mediation of changes in
social anxiety by these cognitive variables was also examined.
To reduce the number of repetitive statistical tests, we calculated
a social phobia symptom composite score by totaling standardized
scores on the SPS, SIAS, and APPQ-S. This variable is hereafter
referred to as social anxiety.
Analyses were based on intention-to-treat, including all partic-
ipants who entered the trial. Every attempt was made to collect
12-week (posttreatment) data on all participants, even those who
did not complete treatment. Analysis of data was also conducted
on the basis of participants who completed at least three sessions
of treatment and returned posttreatment data (N ? 160; 82.1%).
The pattern of means and effect sizes was identical to the
intention-to-treat analyses and, hence, is not described further. The
flow of participants through the study is presented in Figure 1.
We examined differences between treatments in the reductions
on continuous data using hierarchical mixed models containing
random intercept and random slope terms as well as fixed effects
for treatment received (Gibbons et al., 1993). Because randomiza-
tion was based on treatment groups, the individual treatment group
that each participant was allocated to was entered into each anal-
ysis as a first level effect followed by the interaction between
group and individual. Finally, to examine the role of specific
cognitive factors in mediating treatment effects, any process vari-
ables that showed a significantly greater change in the enhanced
condition were subjected to mediation analyses to assess their role
as mediators of the change in diagnostic severity associated with
the enhanced treatment. All analyses were conducted with SPSS
Consolidated Standards of Reporting Trials (CONSORT) style flow chart of study participants.
ENHANCED TREATMENT OF SOCIAL PHOBIA
Demographic data on the sample broken down across the allo-
cated groups are presented in Table 1. As can be seen from the
table, the groups did not differ significantly in demographic fea-
tures or on their main diagnostic and clinical characteristics.
Treatment Integrity and Attitudes to Therapy
The percentage of correct components relative to total identified
by the raters was compared between conditions. There was no
significant difference across the three conditions: enhanced (M ?
99.7%, SD ? 1.5), stress management (M ? 95.5%, SD ? 11.1),
and standard (M ? 98.9%, SD ? 4.7), F(2, 59) ? 2.39, p ? .10.
As a better indication, the mode and median of percentage of
correct components in each condition was 100%, and incorrect
components (i.e., less than 100%) were only identified in five tapes
(8.3%). Hence, there was very little contamination between con-
There was also significant agreement between the condition that
the raters guessed the selected tape was from and the actual
condition it had been selected from: enhanced, 96.6% correct
guesses; standard, 94.4% correct guesses; and stress management,
84.6% correct guesses (? ? .89, p ? .001).
We compared treatment credibility across groups at the two
assessment times (Sessions 1 and 4) using a repeated measures
analysis of variance. As can be seen in Table 1, the groups did not
differ significantly on credibility of treatment, F(2, 126) ? 0.00,
p ? .97, partial ?2? .01, and there was not a significant Group ?
Time interaction, F(2, 126) ? 0.25, p ? .78, partial ?2? .01.
Working alliance measured in Session 4 was compared between
the groups and showed no significant differences, F(2, 150) ?
0.71, p ? .49, ?2? .01. Similarly, there was no significant
difference between the groups on the attitudes toward the group,
F(2, 150) ? 0.79, p ? .46, ?2? .01.
We compared the number of participants who no longer met
criteria for a diagnosis of social phobia at the end of treatment
across conditions using chi-square. When data at posttreatment
were missing, the pretreatment diagnosis was substituted. There
was a significant overall difference between groups, ?2(2, N ?
195) ? 7.28, p ? .026. Follow-up pairwise chi-squares indicated
that the enhanced group (n ? 29; 40.8%) and the standard group
(n ? 23; 34.8%) did not differ significantly, ?2(1, N ? 137) ?
0.52, p ? .470, but both showed significantly greater proportion of
diagnosis free participants than the stress management condition
(n ? 11; 19.0%), ?2(1, N ? 129) ? 7.14, p ? .008, ?2(1, N ?
124) ? 3.91, p ? .048, respectively.
To determine the clinical significance of effects, we determined
the percentage of participants in each group who showed a reliable
change that placed them closer to the mean of a community control
group than to the mean of the current socially phobic sample at
posttreatment (Jacobson & Truax, 1991). These clinically signifi-
cant cutoffs were determined for the SPS and SIAS (Mattick &
Clarke, 1998) because these are two of the most widely and
consistently used measures of social anxiety. This required a
two-step process as follows. First, cutoff scores were calculated on
the basis of Formula “c” reported by Jacobson and Truax (1991).
The mean scores for the SPS and SIAS reported by Mattick and
Clarke (1998) for their community sample were used and com-
pared to the mean scores for each measure reported in the current
sample. This resulted in clinical significance based on the SPS at
a score of 23 or below and on the SIAS at 35 and below. Next,
reliable change indices were calculated for each participant on the
basis of the 12-week retest reliability reported by Mattick and
Clarke. Thus, clinical significance for a given participant was
defined as showing a reliable change from pre- to posttreatment
plus reaching a posttreatment absolute score that was below the
cutoff for clinical significance. According to these criteria, the
following proportions of participants demonstrated clinically sig-
nificant effects at posttreatment in each condition: SPS—enhanced
(42.4%), standard (22.0%), stress management (10.2%), ?2(2, N ?
158) ? 15.01, p ? .001; SIAS—enhanced (31.1%), standard
(18.0%), stress management (12.2%), ?2(2, N ? 160) ? 6.26, p ?
Descriptive Data Across the Three Groups
Mean age (SD)
Never married (%)
Avoidant PD (%)
Additional Axis I (%)
Treatment credibility, M (SD)
Working alliance, M (SD)
Group attitudes, M (SD)
F(2, 192) ? 1.42, p ? .24, ?2? .02
?2(2, N ? 195) ? 1.61, p ? .49
?2(4, N ? 187) ? 2.59, p ? .63
?2(4, N ? 185) ? 7.66, p ? .11
?2(2, N ? 187) ? 0.65, p ? .73
?2(2, N ? 195) ? 3.07, p ? .22
?2(2, N ? 195) ? 1.14, p ? .57
?2(2, N ? 195) ? 1.50, p ? .48
F(2, 153) ? 0.10, p ? .90, ?2? .01
F(2, 135) ? 0.74, p ? .48, ?2? .01
F(2, 150) ? 0.71, p ? .46, ?2? .01
F(2, 192) ? 0.79, p ? .46, ?2? .01
PD ? personality disorder.
RAPEE, GASTON, AND ABBOTT
.044. Follow-up pairwise chi-squares indicated that the enhanced
group showed significantly more clinically significant change than
the standard group on both the SPS, ?2(1, N ? 137) ? 9.32, p ?
.002, and the SIAS, ?2(1, N ? 137) ? 5.56, p ? .018. The
enhanced group also showed more clinically significant change
than the stress management group on both the SPS, ?2(1, N ?
129) ? 18.26, p ? .001, and the SIAS, ?2(1, N ? 129) ? 7.39, p ?
.007. The standard CBT and stress management groups did not
differ significantly on either the SPS, ?2(1, N ? 124) ? 2.39, p ?
.122, or the SIAS, ?2(1, N ? 124) ? 0.25, p ? .619.
Continuous Clinical Measures
Mixed model analysis comparing the three treatments across
time on the clinician rating of social phobia severity showed a
significant main effect of condition, F(2, 23.9) ? 3.60, p ? .043;
a significant main effect of time, F(1, 184.8) ? 170.0, p ? .001;
and a significant Condition ? Time interaction, F(2, 184.7) ?
8.01, p ? .001. Follow-up tests demonstrated significant Condi-
tion ? Time interactions comparing enhanced treatment with
standard treatment, t(186.3) ? 2.84, p ? .005; and enhanced
treatment with stress management, t(181.5) ? 3.80, p ? .001; but
not between stress management and standard treatment, t(186.4) ?
1.00, p ? .317. Estimated marginal means, standard errors, and
effect sizes are presented in Table 2.
The three treatment conditions were also compared across time
on the composite measure of social anxiety. There was a signifi-
cant main effect of condition, F(2, 185.6) ? 3.70, p ? .027; a
significant main effect of time, F(1, 168.7) ? 187.56, p ? .001; and
a significant Condition ? Time interaction, F(2, 168.7) ? 5.73, p ?
.004 (see Table 2). Follow-up tests did not demonstrate a significant
Condition ? Time interaction comparing enhanced treatment with
standard treatment, t(169.2) ? 0.45, p ? .651; however, significant
effects were demonstrated between enhanced treatment and stress
management, t(167.5) ? 3.20, p ? .002, and between standard
treatment and stress management, t(169.3) ? 2.68, p ? .008.
Examination of effects on the LIS indicated no significant main
effect of treatment condition, F(2, 184.5) ? 1.13, p ? .327; however,
examination of effects on the LIS indicated a significant main effect
of time, F(1, 169.3) ? 155.39, p ? .001, which was qualified by a
significant Condition ? Time interaction, F(2, 169.3) ? 4.07, p ?
.019 (see Table 2). Follow-up tests demonstrated no significant Con-
dition ? Time interaction comparing enhanced treatment with stan-
dard treatment, t(169.9) ? 0.80, p ? .425; however, a significant
effect was demonstrated comparing enhanced treatment with stress
management, t(168.0) ? 2.80, p ? .006, and a trend approaching
traditional levels of significance comparing standard treatment and
stress management, t(170.0) ? 1.97, p ? .051.
We compared state anxiety in response to the brief impromptu
speech pre- and posttreatment between the three treatment condi-
tions using a mixed model analysis. There was no significant main
effect of treatment condition, F(2, 167.3) ? 0.56, p ? .575;
however, there was a significant main effect of time, F(1, 134.5) ?
69.91, p ? .001, and a significant Condition ? Time interaction,
F(2, 134.5) ? 3.17, p ? .045 (see Table 2). Follow-up tests
demonstrated a trend approaching traditional levels of significance
comparing enhanced treatment with standard treatment, t(136.7) ?
1.67, p ? .098, and a significant interaction between enhanced
treatment and stress management, t(132.7) ? 2.44, p ? .016. The
interaction between standard treatment and stress management was
not significant, t(134.0) ? 0.70, p ? .488.
Mixed model analysis comparing the three treatments across
time on the measure of probability of negative evaluation showed
a main effect of condition, F(2, 25.7) ? 4.31, p ? .024; and a
significant main effect of time, F(1, 165.6) ? 39.67, p ? .001; but
failed to show a significant Condition ? Time interaction, F(2,
165.6) ? 0.35, p ? .702. Marginal means, standard errors, and
effect sizes are presented in Table 3.
Comparison between the three treatments across time on the
measure of the cost of negative evaluation showed a significant
main effect of condition, F(2, 185.5) ? 9.25, p ? .001; and a
significant main effect of time, F(1, 173.1) ? 141.90, p ? .001; as
well as a significant Condition ? Time interaction, F(2, 173.1) ?
6.68, p ? .002 (see Table 3). Follow-up tests demonstrated a trend
approaching significance for the Condition ? Time interaction
comparing enhanced treatment with standard treatment, t(173.8) ?
1.89, p ? .060; a significant interaction comparing enhanced
treatment with stress management, t(171.5) ? 3.65, p ? .001; and
a trend approaching traditional levels of significance for standard
treatment versus stress management, t(173.9) ? 1.76, p ? .080.
Estimated Marginal Means and Standard Errors for Continuous Outcome Measures Over Time
Clinician-rated diagnostic severityPretreatment
Social anxiety composite1.211.15 0.64
Life Interference Scale1.121.060.66
State anxiety to speech0.900.63 0.41
Effect size expressed as Cohen’s d, on the basis of pre–post treatment change within conditions.
ENHANCED TREATMENT OF SOCIAL PHOBIA
Comparison between the three treatments across time on the
measure of self-focused attention showed no significant main
effect of condition, F(2, 183.0) ? 2.28, p ? .106, but a significant
main effect of time, F(1, 164.5) ? 46.85, p ? .001. However, the
Condition ? Time interaction did not reach significance, F(2,
164.4) ? 0.91, p ? .403 (see Table 3).
Comparison between the three treatments across time on the
measure of the negative mental representation showed no signifi-
cant main effect of condition, F(2, 23.5) ? 1.70, p ? .205.
However, there was a significant main effect of time, F(1,
162.6) ? 85.41, p ? .001, as well as a significant Condition ?
Time interaction, F(2, 162.6) ? 8.56, p ? .001 (see Table 3).
Follow-up tests demonstrated significant Condition ? Time interac-
tions comparing enhanced treatment with standard treatment,
t(163.0) ? 2.37, p ? .019, and enhanced treatment with stress
management, t(161.8) ? 4.10, p ? .001. However, the interaction
between standard treatment and stress management only approached
traditional levels of significance, t(163.0) ? 1.75, p ? .083.
Comparison between the three treatments across time on the
measure of the safety behaviors failed to show a significant main
effect of condition, F(2, 181.0) ? 1.83, p ? .163, or a significant
Condition ? Time interaction, F(2, 163.1) ? 2.06, p ? .131.
However, there was a significant main effect of time, F(1,
163.1) ? 104.0, p ? .001 (see Table 3).
Finally, comparison between the three treatments across time on
the measure of core beliefs showed no significant main effect of
condition, F(2, 182.2) ? 1.32, p ? .269. However, there was a
significant main effect of time, F(1, 154.6) ? 47.50, p ? .001, but
the Condition ? Time interaction was not significant, F(2,
154.6) ? 1.00, p ? .372 (see Table 3).
Mediation of Treatment Effects
Given the significant Condition ? Time interactions on two of
the process measures—negative mental representation and cost of
negative evaluation—an analysis was run to examine whether
changes in these process variables might mediate between the
differences between treatments and change in outcome. The
unique effect of the enhanced treatment program was represented
by a dummy code on allocated treatment condition that compared
the enhanced condition against the other two conditions combined.
Change in treatment outcome was represented by a difference
score from pretreatment to posttreatment on the clinician’s rating
of diagnostic severity. Change scores were also used to represent
change in the two process variables. Test of mediation for multiple
mediators was based on the macro and principles described by
Preacher and Hayes (2004, 2007).
The difference between treatments accounted for significant
variance in the change in diagnostic severity, B ? 1.02, SE ? 0.27,
p ? .001 (Path c); however, after the two process variables were
included in the regression, this path was markedly reduced, B ?
0.52, SE ? 0.25, p ? .041 (Path c?). The total indirect path from
the difference between treatments to change in diagnostic severity
via change in mental representation and change in cost of negative
evaluation was significant, B ? 0.50, SE ? 0.15, p ? .001. Hence,
significant, partial mediation was demonstrated. The overall re-
gression accounted for 29% of the variance in change in diagnostic
severity, F(3, 149) ? 20.46, p ? .001.
The overall pattern of results indicated that the enhanced treatment
was generally superior to a traditional skills-based treatment, although
the traditional treatment was efficacious and generally superior to
nonspecific stress management. The general pattern of results and the
effect sizes shown in Tables 1 and 2 indicated that both exposure-
based treatments showed good efficacy and that the standard CBT
was significantly better than general stress management on two main
outcome measures. In fact, the traditional cognitive restructuring plus
exposure treatment appeared to show slightly larger effects than has
been demonstrated in some previous research. Early studies that used
pure exposure or exposure plus cognitive restructuring have resulted
in effect size changes of around 0.9 on self-reported symptoms of
social phobia (Fedoroff & Taylor, 2001; Feske & Chambless, 1995;
Gould et al., 1997). In the current study, the standard treatment
produced changes of over 1 standard deviation on the main outcome
measures, which compares favorably with these early studies. Nev-
the enhanced program showed slightly larger effects across measures,
and these differences reached significance on several key variables.
Estimated Marginal Means and Standard Errors for Process Measures Over Time
Probability of negative evaluationPretreatment
Cost of negative evaluation1.46 1.010.56
Self-focused attention 0.600.500.33
Negative mental representation0.97 0.500.28
Use of safety behaviors 0.970.710.52
Core beliefs0.48 0.42 0.35
Effect size expressed as Cohen’s d, on the basis of pre–post treatment change within conditions.
RAPEE, GASTON, AND ABBOTT
Specifically, the enhanced treatment showed significantly greater
changes than the standard treatment on the clinician’s rating of
diagnostic severity and the measures of clinical significance, and it
also showed a trend toward statistical significance on the obser-
vational measure of anxiety in response to a speech. An inability
to demonstrate statistical significance across some of the other
measures may have been due to the group format, brevity of the
program, or sample size.
The delivery of all treatments in group format may have limited
variance between treatments. There is some evidence that group
treatment of social phobia may not produce as large effects as
individual delivery (Mortberg, Clark, Sundin, & Wistedt, 2007;
Stangier et al., 2003). This might be particularly the case for the
enhanced treatment that depends quite heavily on individualization
of treatment components, especially performance feedback, explo-
ration of schemas, and dropping of safety behaviors. For maximum
efficacy, these components depend on tailoring of their application
to an individual’s specific circumstances, and this is difficult to do
in the context of group delivery. The current program was also
somewhat briefer than similar programs conducted by Clark and
his colleagues (Clark et al., 2003; Mortberg et al., 2007), and
selection of participants was slightly less specific. Consistent with
these factors, the effect sizes found in the current study were
somewhat smaller than reported in studies by Clark’s group (Clark
et al., 2003, 2006). Delivery of treatment in a group format may
have also provided some in vivo exposure for all participants,
further reducing variance between conditions. In some ways, then,
the demonstration of larger effects in the enhanced program than
the other treatments on several measures is even more impressive.
Another factor that may have influenced outcomes was the
competence of therapists. It is unlikely that therapists were not
competent because they were clinical psychologists with experi-
ence in management of social phobia and received supervision
from an expert in anxiety and social phobia. They also demon-
strated excellent adherence to the treatment protocols. However,
adherence is not the same as competence, and it is a limitation of
the study that competence was not assessed. Demonstrating com-
petence is extremely difficult because there is no “gold standard”
for delivery of these techniques. In addition, given that the tradi-
tional program essentially formed a subset of the enhanced pro-
gram, it is less likely that lack of competence influenced differ-
ences between these key conditions. However, variable therapist
competence in any of the treatment conditions may have influ-
enced the results in a variety of directions, and future studies
would benefit from development of competence measures.
The current study represents one of the larger clinical trials for
the management of social phobia. Nevertheless, even this trial only
had sufficient power to detect a moderate effect size difference
between groups on main effects and even larger effect sizes for the
relevant interactions. As a result, we did not control the
experiment-wise Type 1 error rate, and this provides a limitation to
the study. Strict control of the Type 1 error rate may have pre-
vented some differences between the active treatments to reach
significance, yet we believe that this would not be a reasonable
conclusion on the basis of examination of the overall pattern of
results. Hence, we chose to maintain a balance between Type 1 and
Type 2 errors. Ideally, a study of this kind would include multiple
sites and a sample size of several hundred. Only with such samples
will researchers be able to truly gauge the effects of relatively
subtle improvements to already active treatments.
Perhaps the greatest interest from the current study was in the
effects of treatment on process measures. Given that the enhanced
treatment was designed to modify theoretically important maintaining
factors, the key question of interest is whether it did. Several factors
included in maintenance models of social phobia (Clark & Wells,
1995; Rapee & Heimberg, 1997) were assessed. Unfortunately, given
of the measures were newly developed and, hence, had little estab-
lished reliability or validity. However, such a limitation would be
conservative with respect to the hypotheses because limited reliability
in particular would reduce the power of the study to demonstrate
relationships. As a result, a lack of demonstrated relationships might
simply reflect limitations of the measures. Similarly, the lack of
reliability data on the posttreatment ADIS-IV means that diagnostic
criteria may have been assessed less reliably at posttest, and this may
have contributed to noise in the data.
Significant changes across treatment were demonstrated on all the
assessed process variables. It is possible that these changes were due
simply to the effects of time because no waiting list control condition
was included. However, social phobia is one of the more chronic
anxiety disorders (Massion et al., 2002), and waiting list conditions in
other studies rarely demonstrate marked changes (e.g., Rapee, Abbott,
Baillie, & Gaston, 2007). Therefore, it appears that even broad-based
generic treatments are associated with some reduction in factors that
maintain social phobia; as a result, the changes in social phobia
severity and symptoms associated with all three treatments are not
surprising. Importantly, engagement in the enhanced treatment was
associated with greater changes on two of the variables—the cost of
negative evaluation and the perception of personal abilities and at-
tractiveness (negative mental representation)—than engagement in
the stress treatment and was associated with greater changes on the
negative mental representation than engagement in the standard CBT
treatment while the difference between these conditions on the cost of
negative evaluation approached significance. Of greater interest, the
extent to which the enhanced treatment was associated with reduc-
tions in the severity of social phobia was mediated by changes in the
cost of negative evaluation and the negative mental representation. It
is important to point out that data from the mediators and outcome
variables were assessed at the same time. Hence, the results cannot
indicate a causal relationship, and it remains for future research to
determine whether changes in these process variables cause changes
in outcome. The findings are consistent with some previous data
showing that changes in the perceived cost of negative social events
mediate change in social anxiety (Foa et al., 1996; Hofmann, 2004;
J. K. Wilson & Rapee, 2005). Although some research has pointed to
the importance of a negative mental representation in maintaining
social anxiety (Hirsch, Clark, Mathews, & Williams, 2003; Hirsch,
Mathews, Clark, Williams, & Morrison, 2006; Rapee & Abbott,
2007) and has demonstrated reductions in this factor following suc-
cessful treatment (Abbott & Rapee, 2004), mediation of changes in
social anxiety through reductions in the negative mental representa-
tion has not previously been demonstrated. Importantly, the current
results show that the effect of the enhanced program on reducing the
perceived cost of negative evaluation and the negative mental repre-
sentation were associated with key differences between the enhanced
treatment and the other treatment packages. Future more tightly con-
ENHANCED TREATMENT OF SOCIAL PHOBIA
trolled studies utilizing more highly developed measures may be able
to demonstrate a similar role for other theoretically important factors.
A limitation of the current study, and one that plagues most treat-
ment outcome research, is the fact that the study was conducted in a
specialized unit that was involved in developing the enhanced treat-
ment program. The therapists were independent employees and
it is possible that the investigators and clinical supervisor may have
communicated preexisting expectations about the likely outcome.
This type of bias is very difficult to overcome, although as above,
multicenter trials may have the best chance of reducing the influence
of such biases. As one method of checking for differential therapist
enthusiasm and expectation, we assessed working alliance and group
attitudes. There were no significant differences between conditions
indicating that participants did not appear to detect differences in the
way they were being treated and did not differ in their enthusiasm for
the group. Similarly, the conditions did not differ in credibility of
treatments indicating that the therapists did not “undersell” any of the
any therapist bias was sufficient to be detected by participants; how-
enthusiastic about any particular treatment.
Another limitation of the study was the lack of follow-up data.
Hence, the results speak only to effects in the short term and do not
allow any conclusions to be drawn about the maintenance of
differences between treatments. To the extent that the enhanced
treatment produced a greater effect on putative maintenance mecha-
nisms, it may be expected that the relatively minor differences be-
tween the conditions after 12 weeks would have increased over time.
On the other hand, it is also possible that following some degree of
change after 12 weeks, participants in the standard CBT program may
have begun to shift some of these same maintaining mechanisms
during their increasing interactions with the world, and any initial
treatment differences may have reduced. Future studies need to in-
clude longer term comparisons so that these possibilities can be
Although the current mediational results are cross-sectional and
cannot be used to demonstrate causality, they are consistent with
theoretical suggestions that the extent to which treatment packages
for social phobia address certain key mediating variables—such as
use of safety cues, perceived consequences of negative evaluation,
self-focused attention, and a negative mental representation—the
greater will be their efficacy. Hence, continued clinical develop-
ments aimed at developing novel methods of changing these
factors should continue to improve the efficacy of treatment pack-
ages for social phobia. Several possibilities—including imagery
rescripting (Wild, Hackmann, & Clark, 2008), attentional retrain-
ing (Amir, 2007; Schmidt, 2007), and retraining of interpretation
biases (Murphy, Hirsch, Mathews, Smith, & Clark, 2007; E. J.
Wilson, MacLeod, Mathews, & Rutherford, 2006)—are beginning
to be explored. These innovative laboratory procedures that are
aimed at the modification of theoretical maintaining factors in
social phobia may be one way to further improve future clinical
packages for the treatment of this complex disorder.
Abbott, M. J., & Rapee, R. M. (2004). Post-event rumination and negative
self-appraisal in social phobia before and after treatment. Journal of
Abnormal Psychology, 113(1), 136–144.
American Psychiatric Association. (1994). Diagnostic and statistical man-
ual of mental disorders (4th ed.). Washington, DC: Author.
Amir, N. (2007). Attention and interpretation training as treatments for
social anxiety. Paper presented at the V World Congress of Behavioural
and Cognitive Therapies, Barcelona, Spain.
Beck, A. T., Emery, G., & Greenberg, R. L. (1985). Anxiety disorders and
phobias: A cognitive perspective. New York: Basic Books.
Bo ¨gels, S. M. (2006). Task concentration training versus applied relax-
ation, in combination with cognitive therapy, for social phobia patients
with fear of blushing, trembling, and sweating. Behaviour Research and
Therapy, 44(8), 1199–1210.
Borkovec, T. D., & Nau, S. D. (1972). Credibility of analogue therapy
rationales. Journal of Behavior Therapy and Experimental Psychiatry, 3,
Brown, T. A., White, K. S., & Barlow, D. H. (2005). A psychometric
reanalysis of the Albany Panic and Phobia Questionnaire. Behaviour
Research and Therapy, 43, 337–355.
Butler, G. (1985). Exposure as a treatment for social phobia: Some in-
structive difficulties. Behaviour Research and Therapy, 23, 651–657.
Clark, D. M., Ehlers, A., Hackmann, A., McManus, F., Fennell, M., Grey,
N., et al. (2006). Cognitive therapy versus exposure and applied relax-
ation in social phobia: A randomized controlled trial. Journal of Con-
sulting and Clinical Psychology, 74(3), 568–578.
Clark, D. M., Ehlers, A., McManus, F., Hackman, A., Fennell, M., Camp-
bell, H., et al. (2003). Cognitive therapy versus fluoxetine in generalized
social phobia: A randomized placebo-controlled trial. Journal of Con-
sulting and Clinical Psychology, 71(6), 1058–1067.
Clark, D. M., & Wells, A. (1995). A cognitive model of social phobia. In
R. G. Heimberg, M. R. Liebowitz, D. A. Hope, & F. R. Schneier (Eds.),
Social phobia: Diagnosis, assessment, and treatment (pp. 69–93). New
York: Guilford Press.
Cuming, S., Rapee, R. M., Kemp, N., Abbott, M. J., Peters, L., & Gaston,
J. E. (2008). A self report measure of subtle avoidance and safety
behaviours relevant to social anxiety: Development and psychometric
properties. Manuscript submitted for publication.
Di Nardo, P. A., Brown, T. A., & Barlow, D. H. (1994). Anxiety Disorders
Interview Schedule for DSM–IV— Lifetime version. Albany, NY: Gray-
Evans, N. J., & Jarvis, P. A. (1986). The group attitude scale: A measure
of attraction to group. Small Group Behavior, 17(2), 203–216.
Faul, F., & Erdfelder, E. (1992). GPOWER: A priori, post-hoc, and
compromise power analyses for MS-DOS [Computer program]. Bonn,
Germany: Bonn University.
Fedoroff, I. C., & Taylor, S. (2001). Psychological and pharmacological
treatments of social phobia: A meta-analysis. Journal of Clinical Psy-
chopharmacology, 21(3), 311–324.
Fenigstein, A., Scheier, M. F., & Buss, A. H. (1975). Public and private
self-consciousness: Assessment and theory. Journal of Consulting and
Clinical Psychology, 43, 522–527.
Feske, U., & Chambless, D. L. (1995). Cognitive behavioral versus expo-
sure only treatment for social phobia: A meta-analysis. Behavior Ther-
apy, 26, 695–720.
Foa, E. B., Franklin, M. E., Perry, K. J., & Herbert, J. D. (1996). Cognitive
biases in social phobia. Journal of Abnormal Psychology, 105, 433–439.
Garcia-Palacios, A., & Botella, C. (2003). The effects of dropping in-
situation safety behaviors in the treatment of social phobia. Behavioral
Interventions, 18(1), 23–33.
Gibbons, R. D., Hedeker, D., Elkin, I., Waternaux, C., Greenhouse, J. B.,
Shea, M. T., et al. (1993). Some conceptual and statistical issues in
analysis of longitudinal psychiatric data: Application to the NIMH
Treatment of Depression Collaborative Research Program data set. Ar-
chives of General Psychiatry, 50, 739–750.
Gould, R. A., Buckminster, S., Pollack, M. H., Otto, M. W., & Yap, L.
(1997). Cognitive-behavioral and pharmacological treatment for social
RAPEE, GASTON, AND ABBOTT
phobia: A meta-analysis. Clinical Psychology-Science and Practice, Download full-text
Hackmann, A., Surawy, C., & Clark, D. M. (1998). Seeing yourself though
others’ eyes: A study of spontaneously occurring images in social
phobia. Behavioural and Cognitive Psychotherapy, 26, 3–12.
Harvey, A. G., Clark, D. M., Ehlers, A., & Rapee, R. M. (2000). Social
anxiety and self-impression: Cognitive preparation enhances the bene-
ficial effects of video feedback following a stressful social task. Behav-
iour Research and Therapy, 38(12), 1183–1192.
Heimberg, R. G., Dodge, C. S., Hope, D. A., Kennedy, C. R., Zollo, L. J.,
& Becker, R. E. (1990). Cognitive behavioral group treatment for social
phobia: Comparison with a credible placebo control. Cognitive Therapy
and Research, 14, 1–23.
Heinrichs, N., & Hofmann, S. G. (2001). Information processing in social
phobia: A critical review. Clinical Psychology Review, 21(5), 751–770.
Hirsch, C. R., Clark, D. M., Mathews, A., & Williams, R. (2003). Self-
images play a causal role in social phobia. Behaviour Research and
Therapy, 41(8), 909–921.
Hirsch, C. R., Mathews, A., Clark, D. M., Williams, R., & Morrison, J. A.
(2006). The causal role of negative imagery in social anxiety: A test in
confident public speakers. Journal of Behavior Therapy and Experimen-
tal Psychiatry, 37(2), 159–170.
Hofmann, S. G. (2004). Cognitive mediation of treatment change in social
phobia. Journal of Consulting and Clinical Psychology, 72(3), 392–399.
Hofmann, S. G. (2007). Cognitive factors that maintain social anxiety
disorder: A comprehensive model and its treatment implications. Cog-
nitive Behaviour Therapy, 36(4), 193–209.
Horvath, A. O., & Greenberg, L. S. (1989). Development and validation of
the Working Alliance Inventory. Journal of Counseling Psychology,
Jacobson, N. S., & Truax, P. (1991). Clinical significance: A statistical
approach to defining meaningful change in psychotherapy research.
Journal of Consulting and Clinical Psychology, 59(1), 12–19.
Kim, E.-J. (2005). The effect of the decreased safety behaviors on anxiety
and negative thoughts in social phobics. Journal of Anxiety Disorders,
Massion, A. O., Dyck, I. R., Shea, M. T., Phillips, K. A., Warshaw, M. G.,
& Keller, M. B. (2002). Personality disorders and time to remission in
generalized anxiety disorder, social phobia, and panic disorder. Archives
of General Psychiatry, 59(5), 434–440.
Mattick, R. P., & Clarke, J. C. (1998). Development and validation of
measures of social phobia scrutiny fear and social interaction anxiety.
Behaviour Research and Therapy, 36, 455–470.
Mattick, R. P., & Peters, L. (1988). Treatment of severe social phobia:
Effects of guided exposure with and without cognitive restructuring.
Journal of Consulting and Clinical Psychology, 56, 251–260.
Mortberg, E., Clark, D. M., Sundin, O., & Wistedt, A. A. (2007). Intensive
group cognitive treatment and individual cognitive therapy vs. treatment
as usual in social phobia: A randomized controlled trial. Acta Psychiat-
rica Scandinavica, 115(2), 142–154.
Murphy, R., Hirsch, C. R., Mathews, A., Smith, K., & Clark, D. M. (2007).
Facilitating a benign interpretation bias in a high socially anxious
population. Behaviour Research and Therapy, 45(7), 1517–1529.
Musa, C. Z., & Lepine, J. P. (2000). Cognitive aspects of social phobia: A
review of theories and experimental research. European Psychiatry, 15,
Norton, P. J., & Price, E. C. (2007). A meta-analytic review of adult
cognitive-behavioral treatment outcome across the anxiety disorders.
Journal of Nervous and Mental Disease, 195, 521–531.
Peters, L. (2000). Discriminant validity of the Social Phobia and Anxiety
Inventory (SPAI), the Social Phobia Scale (SPS) and the Social Inter-
action Anxiety Scale (SIAS). Behaviour Research and Therapy, 38(9),
Pinto-Gouveia, J., Castilho, P., Galhardo, A., & Cunha, M. (2006). Early
maladaptive schemas and social phobia. Cognitive Therapy and Re-
search, 30, 571–584.
Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for
estimating indirect effects in simple mediation models. Behavior Re-
search Methods, Instruments, & Computers, 36(4), 717–731.
Preacher, K. J., & Hayes, A. F. (2007). SPSS and SAS macros for estimat-
ing and comparing indirect effects in multiple mediator models. Re-
trieved November 29, 2007, from http://www.comm.ohio-state.edu/
Rapee, R. M., & Abbott, M. J. (2006). Mental representation of observable
attributes in people with social phobia. Journal of Behavior Therapy and
Experimental Psychiatry, 37, 113–126.
Rapee, R. M., & Abbott, M. J. (2007). Modelling relationships between
cognitive variables during and following public speaking in participants
with social phobia. Behaviour Research and Therapy, 45(12), 2977–
Rapee, R. M., Abbott, M. J., Baillie, A. J., & Gaston, J. E. (2007).
Treatment of social phobia through pure self help and therapist-
augmented self help. British Journal of Psychiatry, 191, 246–252.
Rapee, R. M., Craske, M. G., & Barlow, D. H. (1994). Assessment
instrument for panic disorder that includes fear of sensation-producing
activities: The Albany Panic and Phobia Questionnaire. Anxiety, 1,
Rapee, R. M., & Heimberg, R. G. (1997). A cognitive-behavioral model of
anxiety in social phobia. Behaviour Research and Therapy, 35, 741–756.
Rapee, R. M., & Sanderson, W. C. (1998). Social phobia: Clinical appli-
cation of evidence-based psychotherapy. Northvale, NJ: Aronson.
Rapee, R. M., & Spence, S. H. (2004). The etiology of social phobia:
Empirical evidence and an initial model. Clinical Psychology Review,
Rodebaugh, T. L., & Chambless, D. L. (2002). The effects of video
feedback on self-perception of performance: A replication and exten-
sion. Cognitive Therapy and Research, 26(5), 629–644.
Rodebaugh, T. L., Holaway, R. M., & Heimberg, R. G. (2004). The
treatment of social anxiety disorder. Clinical Psychology Review, 24(7),
Schmidt, N. B. (2007). Attention retraining as a treatment for social
anxiety disorder. Paper presented at the V World Congress of Behav-
ioural and Cognitive Therapies, Barcelona, Spain.
Smits, J. A. J., Powers, M. B., Buxkamper, R., & Telch, M. J. (2006). The
efficacy of videotape feedback for enhancing the effects of exposure-
based treatment for social anxiety disorder: A controlled investigation.
Behaviour Research and Therapy, 44(12), 1773–1785.
Stangier, U., Heidenreich, T., Peitz, M., Lauterbach, W., & Clark, D. M.
(2003). Cognitive therapy for social phobia: Individual versus group
treatment. Behaviour Research and Therapy, 41(9), 991–1007.
Ware, J. E., Kosinski, M., & Keller, S. D. (1996). A 12-item short-form
health survey: Construction of scales and preliminary tests of reliability
and validity. Medical Care, 34(3), 220–233.
Wild, J., Hackmann, A., & Clark, D. M. (2008). Rescripting early mem-
ories linked to negative images in social phobia: A pilot study. Behavior
Therapy, 39, 47–56.
Wilson, E. J., MacLeod, C., Mathews, A., & Rutherford, E. (2006). The
causal role of interpretive bias in anxiety reactivity. Journal of Abnormal
Psychology, 115(1), 103–111.
Wilson, J. K., & Rapee, R. M. (2005). The interpretation of negative social
events in social phobia: Changes during treatment and relationship to
outcome. Behaviour Research & Therapy, 43(3), 373–389.
Received February 22, 2008
Revision received November 24, 2008
Accepted November 26, 2008 ?
ENHANCED TREATMENT OF SOCIAL PHOBIA