Literature Review

PublicSpeakingAvoidanceasaTreatmentModeratorforSocial.pdf

Home >Psychology homework help >Literature Review

Public Speaking Avoidance as a Treatment Moderator for Social Anxiety Disorder

Bita Mesria,1, Andrea N. Nilesa,2, Andre Pittigb,3, Richard T. LeBeaua,4, Ethan Haika,5, and Michelle G. Craskea,6

aUniversity of California, Los Angeles, Department of Psychology, 405 Hilgard Avenue, Los Angeles, California, United States of America, 90095-1563

bInstitute of Clinical Psychology and Psychotherapy, Technische Universität Dresden, Chemnitzer Str. 46, Dresden, Germany, D-01187

Abstract

Background and Objectives—Cognitive behavioral therapy (CBT) and acceptance and commitment therapy (ACT) have both garnered empirical support for the effective treatment of

social anxiety disorder. However, not every patient benefits equally from either treatment.

Identifying moderators of treatment outcome can help to better understand which treatment is best

suited for a particular patient.

Methods—Forty-nine individuals who met criteria for social anxiety disorder were assessed as part of a randomized controlled trial comparing 12 weeks of CBT and ACT. Pre-treatment

avoidance of social situations (measured via a public speaking task and clinician rating) was

investigated as a moderator of post-treatment, 6-month follow-up, and 12-month follow-up social

anxiety symptoms, stress reactivity, and quality of life.

Results—Public speaking avoidance was found to be a robust moderator of outcome measures, with more avoidant individuals generally benefitting more from CBT than ACT by 12-month

follow-up. In contrast, clinician-rated social avoidance was not found to be a significant moderator

of any outcome measure.

Limitations—Results were found only at 12-month follow-up. More comprehensive measures of avoidance would be useful for the field moving forward.

Please address correspondence to Michelle Craske, Ph.D., Department of Psychology, UCLA, 1193 Franz Hall, Box 951563, Los Angeles, CA 90095-1563. Telephone: 1-310-825-8403; fax: 1-310-825-9048; [email protected]. [email protected] [email protected] [email protected] [email protected] [email protected] [email protected]

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

HHS Public Access Author manuscript J Behav Ther Exp Psychiatry. Author manuscript; available in PMC 2018 June 01.

Published in final edited form as: J Behav Ther Exp Psychiatry. 2017 June ; 55: 66–72. doi:10.1016/j.jbtep.2016.11.010.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

Conclusions—Findings inform personalized medicine, suggesting that social avoidance measured behaviorally via a public speaking task may be a more robust factor in treatment

prescription compared to clinician-rated social avoidance.

Keywords

Social anxiety; Moderator; Treatment outcome; Cognitive behavioral therapy; Acceptance and commitment therapy

1. Introduction

Cognitive behavioral therapy (CBT) is a well-established treatment for social anxiety

disorder (Butler, Chapman, Forman, & Beck, 2006; Hofmann & Smits, 2008). Recently,

acceptance and commitment therapy (ACT), a third-wave behavioral therapy, has garnered

support as another effective treatment for social anxiety (Swain, Hancock, Hainsworth, &

Bowman, 2013; Bluett, Homan, Morrison, Levin, & Twohig, 2014) with comparable

treatment outcomes to CBT (Craske et al., 2014). Clinically significant response rates of

individual patients following these interventions are around 50–55%, ranging from 43% to

70% (for a review see Loerinc et al., 2015; Craske et al., 2014; Leichsenring et al., 2014;

Lincoln et al., 2005). Identifying treatment moderators may be a key to improving response

rates, as they clarify for whom and under which circumstances treatments have different

effects. Knowledge of such moderators can help clinicians better match patients with

existing treatments from which they are likely to glean the greatest benefit (Kraemer,

Wilson, Fairburn, & Agras, 2002).

Unfortunately, though several predictors of treatment outcome have been identified, little

research exists on treatment moderators. This is likely due to the fact that the majority of

prior studies on social anxiety disorder do not compare two active treatments, which is

required for assessing treatment moderators. To our knowledge, only a few papers have

reported moderators of psychological treatments for individuals with social anxiety disorder.

The findings are detailed below.

In a previously published article on the current sample, individuals with social anxiety

disorder who were rated as high in experiential avoidance (i.e., self-reported unwillingness

to accept negative emotions) measured by the Acceptance and Action Questionnaire

reported greater symptom reduction at 12-month follow-up in CBT than ACT (Craske et al.,

2014). The same pattern of moderation was found in a separate study with a mixed anxiety

sample (Wolitzky-Taylor, Arch, Rosenfield, & Craske, 2012). We speculated that individuals

with high experiential avoidance benefit more from CBT in the long-term because they are

motivated to practice skills (e.g., exposures) designed to decrease avoidance of anxious

thoughts, feelings, and sensations. Compared to CBT, ACT emphasizes acceptance rather

than reducing uncomfortable internal experiences. Conversely, in the same mixed anxiety

sample, individuals with high behavioral avoidance of negative physical sensations (i.e.,

unwillingness to continue a hyperventilation task) were more likely to benefit from ACT

than CBT (Davies, Niles, Pittig, Arch, & Craske, 2015). However, this study did not

examine moderators separately by diagnosis and thus it is possible that this finding was

Mesri et al. Page 2

J Behav Ther Exp Psychiatry. Author manuscript; available in PMC 2018 June 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

driven by patients with anxiety primarily related to bodily sensations (e.g., those with panic

disorder and health anxiety), which is a common but not essential or primary component of

social anxiety disorder.

A measure of avoidance that is more specific to social anxiety disorder would be avoidance

of social situations. Behavioral measures of social avoidance including public speaking tasks

are ecologically valid and easily implemented in research, but rarely used in clinical

assessments (Beidel, Turner, Jacob, & Cooley, 1989; Hofmann, Newman, Ehlers, & Roth,

1995; Levin et al., 1993; Moscovitch, Suvak, & Hofmann, 2010). Instead, clinicians

typically make judgments of behavioral avoidance based on patient self-report. However,

anxious patients’ estimates of their avoidance can be at odds with their actual behavior

(Rachman & Lopatka, 1986; Taylor & Rachman, 1994). To our knowledge there is no

previous study evaluating behavioral measures of social avoidance as moderators of

treatment outcome for social anxiety disorder.

Theoretically, experiential and behavioral avoidance are two separate parts of anxiety.

Whereas experiential avoidance is centered on avoidance of internal experiences such as

thoughts, feelings, and physical sensation, behavioral avoidance is centered on avoidance of

external experiences such as social events, public speaking, and meetings. It would seem

likely that individuals who are avoidant of feared internal experiences would also be

avoidant of feared external experiences. Moreover, both experiential avoidance and

behavioral avoidance are indicators of poor emotion regulation (Craske, Street, & Barlow,

1989; Hayes, Wilson, Gifford, Follette, & Strosahl, 1996). From a deficit correction model,

it is likely that those who show deficits in emotion regulation would benefit from a treatment

that is targeting said deficit (e.g., CBT) compared to a treatment that is not targeting emotion

regulation (e.g., ACT).

Given prior evidence that individuals who report high levels of experiential avoidance

(indicator of poor emotion regulation) respond more positively to CBT than ACT, we

hypothesized that those with the most overt social avoidance (another indicator of poor

emotion regulation), would similarly respond more positively to CBT than ACT. To evaluate

the effects of in vivo versus clinician-rated social avoidance, we analyzed avoidance via a

public speaking task and clinician rating prior to treatment. To isolate the effect of social

avoidance above social fear, we analyzed public speaking avoidance, clinician-rated social

avoidance, public speaking fear, and clinician-rated social fear as moderators of all

outcomes.

2. Materials and methods

2.1. Participants

Forty-nine individuals who met diagnostic criteria for principal or co-principal generalized

social anxiety disorder as diagnosed using the Anxiety Disorders Interview Schedule IV

(Brown, Di Nardo, & Barlow, 1994, see Craske et al., 2014, for more details) were included

in the current analyses. Fifty-two participants completed treatment but follow-up behavioral

and self-report data were missing for 3 individuals. A clinician severity rating of 4 or higher

on the ADIS-IV indicated clinical severity and served as the cutoff for study eligibility.

Mesri et al. Page 3

J Behav Ther Exp Psychiatry. Author manuscript; available in PMC 2018 June 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

Individuals were a subset of a larger sample that included randomization to a waitlist

condition (Craske et al., 2014). Because moderator analyses examine differential response to

two active treatments and not differential response to active treatment versus control, we did

not include participants assigned to the waitlist in these analyses. Demographics for the

current subsample are in Table 1. There were no significant group differences on any

demographic or diagnostic variable at baseline.

Exclusion criteria included active suicidal ideation, pregnancy, substance abuse or

dependence within the last 6 months, bipolar disorder, psychosis, or certain medical

diseases. Additional exclusion criteria (i.e., left handedness, metal implants, claustrophobia)

were included due to a neuroimaging component. Individuals were permitted to receive

concurrent psychotherapy or psychotropic medication if they were stabilized on

benozodiazepines and beta blockers for a minimum of 1 month; on SSRIs, SNRIs,

heterocylics, and MAO inhibitors for a minimum of 3 months; and on non-anxiety related

psychotherapy for a minimum of 6 months prior to study entrance. Individuals were

recruited through online and newspaper advertisements as well as community flyers and

referrals from the greater Los Angeles area. The study took place at the Anxiety Disorders

Research Center in the University of California, Los Angeles (UCLA).

2.2. Design

Individuals were assessed prior to treatment (i.e., pre-treatment), within 6 weeks after the

end of treatment (i.e., post-treatment), 6 months after pre-treatment (i.e., 6-month follow-

up), and 12 months after pre-treatment (i.e., 12-month follow-up)1.

2.3. Treatments

Individuals in CBT and ACT groups received 12 weekly, 1-hr individual therapy sessions

based on standard manuals2. ACT and CBT were matched on number of exposure sessions

but differed in framing of the intent of exposure. CBT and ACT were administered by

advanced clinical psychology students at UCLA (see Craske et al., 2014). Therapists

received a two-day training session in CBT and ACT by Drs. Craske and Hayes,

respectively. They received weekly group supervision by Dr. Craske and members of Dr.

Craske’s and Hayes’s teams.

CBT—The 12-session CBT protocol has been effective for social anxiety disorder (Craske et al., 2014; Arch et al., 2012). Session 1 included assessment, psychoeducation, and self-

monitoring. Sessions 2–4 covered cognitive restructuring, hypothesis testing, and breathing

retraining. Session 5–11 included exposures to social stimuli. Session 12 focused on relapse

prevention.

ACT—Session 1 included psychoeducation and experiential exercises. Sessions 2–3 covered creative hopelessness. Sessions 4–5 covered mindfulness, acceptance, and cognitive

16-month follow-up was approximately 3 months after treatment completion and 12-month follow-up was approximately 9 months after treatment completion. 2See authors for a copy of the CBT treatment manual (CBT manual modified from Hope, Heimberg, Juster, & Turk, 2000); the ACT manual is published (Eifert & Forsyth, 2005).

Mesri et al. Page 4

J Behav Ther Exp Psychiatry. Author manuscript; available in PMC 2018 June 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

defusion. Sessions 6–11 honed previous skills and introduced value exploration. Exposures

were used throughout to observe and accept anxiety as well as to engage in valued activities

despite anxiety. Session 12 created a plan for future use of skills.

2.4. Moderator Variables

2.4.1. Public Speaking Avoidance and Fear—At pre-treatment, individuals were asked to give a 3-minute speech in front of a video camera and two confederates. Speech

topics included global warming and corporeal punishment. These topics were selected to be

moderate in terms of difficulty and controversy. Individuals were given 5 minutes to prepare

the speech on one or both topics. They were instructed to rate their fear level using a 0–100

Subjective Units of Distress Scale (SUDS; Wolpe, 1990) with 0 being no fear and 100 being maximum fear at the start of the speech, at each 1-minute interval, and at the end of the speech. After 3 minutes, individuals were given the opportunity to continue speaking for up

to 3 more minutes. Mean SUDS ratings were calculated for each individual and analyzed as

a measure of fear on the public speaking task. Number of minutes spoken was used as a

measure of avoidance. Individuals who refused the public speaking task altogether were

given a score of 0 minutes and SUDS rating of 100. See appendix A for the brief protocol

used to assess public speaking avoidance.

2.4.2. Clinician-Rated Social Avoidance and Fear—As part of the pre-treatment ADIS-IV, clinicians rated individuals’ avoidance and fear (0 = none, 8 = extreme anxiety or avoidance) of 13 social situations (e.g., dating, public speaking, speaking with unfamiliar people). Avoidance scores for all 13 situations were averaged to create a clinician-rated

social avoidance score (α = .74). Fear scores for all 13 social situations were also averaged to create a clinician-rated social fear score (α = .77).

2.5. Outcome Variables

2.5.1. Symptom Composite Score—The self-report version of the Liebowitz Social Anxiety Scale (LSAS-SR; Fresco et al., 2001) is a 24-item measure of fear and avoidance of

social and performance situations. Total ratings demonstrate good test-retest reliability (r = . 83), internal consistency (α = .95), convergent validity and the scale is sensitive to change following treatment (Baker, Heinrichs, Kim, & Hofmann, 2002). The Social Interaction

Anxiety Scale (SIAS; Mattick & Clarke, 1998) is a 20-item measure of thoughts, feelings,

and behaviors in social situations. The SIAS correlates highly with other measures of social

phobia and has good internal consistency (α = .90) (Osman, Gutierrez, Barrios, Kopper, & Chiros, 1998). The Social Phobia Scale (SPS; Mattick & Clarke, 1998) is a 20-item measure

of being observed by others during routine activities (e.g., eating, writing). The SPS

correlates highly with other measures of social phobia and has good internal consistency (α = .91) (Osman et al., 1998). Alphas for the LSAS-SR, SIAS, and SPS were all at or above .

90 in this sample across all time points (Niles, Mesri, Burklund, Lieberman, & Craske,

2013). To improve construct validity for the measurement of social anxiety severity, a

composite was created from the three scales. Z-scores for each measure were combined to

create a standardized measure with mean 0 and standard deviation 1. The composite score

includes averages of all three measures at pre, post, and 12-month follow-up. The LSAS-SR

was not administered at 6-month follow-up, which includes only the SPS and SIAS.

Mesri et al. Page 5

J Behav Ther Exp Psychiatry. Author manuscript; available in PMC 2018 June 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

2.5.2. State-Trait Anxiety Inventory—The State-Trait Anxiety Inventory – A State (STAI AState; Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983) is a 20-item measure

of temporary anxiety in response to a stressor. Example items include “I feel nervous” and “I

feel tense.” Each item is rated on a scale from 1 to 4, with 1 being not at all and 4 being to a great extent. The STAI A-State demonstrates good internal consistency (α = .83 – .92) (Spielberger et al., 1983). The STAI was administered at the start of the laboratory

assessment (which included a hyperventilation task, a public speaking task, and computer

tasks) in order to assess stress reactivity. Because the laboratory assessment was not

conducted at 6-month follow-up, STAI data were analyzed only at pre, post, and 12-month

follow-up.

2.5.3. Quality of Life Inventory—The Quality of Life Inventory (QOLI; Frisch, 1994a, 1994b) is a measure of satisfaction with regard to 16 broad life domains. Each domain is

first rated for importance on a scale from 0 to 2, with 0 being not important and 2 being extremely important. Then, individuals rate their life satisfaction with that domain on a −3 to +3 scale, with −3 being very dissatisfied and +3 being very satisfied. The QOLI demonstrates good test-retest reliability (r = .80 – .91), internal consistency (α = .77 – .89) and is sensitive to treatment change (Frisch et al., 2005).

2.6. Statistical analyses

A multi-level model with repeated measures design was used. Pre-treatment scores were

modeled as a covariate rather than a repeated measure to minimize the variance in the

outcome measures (Tabachnick & Fidell, 2006). This model has been previously used in

examining moderators of treatment outcome (Craske et al., 2014; Niles et al., 2013;

Wolitzky-Taylor et al., 2012).

Analyses were run in Stata 13 using the xtmixed command. A two level growth curve model

was used. Time (post-treatment, 6-month follow-up, 12-month follow-up) was modeled on

level 1 as a continuous linear predictor. On level 2, we included baseline levels of the

outcome measures (as a covariate), Group (CBT or ACT), status (0 = completed 12-month

measures, 1 = not completed 12-month measures) and the moderators. To test specificity of

public speaking avoidance as a moderator above fear, we included fear during the public

speaking task as a covariate. When testing public speaking fear, we included public speaking

avoidance as a covariate. Pairwise correlations between public speaking avoidance and

public speaking fear revealed only a moderate correlation, r = −.39, p < .001. However, pairwise correlations between clinician-rated social avoidance and clinician-rated social fear

revealed a strong correlation, r = .81, p < .001. Hence, we did not include clinician-rated social fear in the model when analyzing clinician-rated social avoidance and vice versa.

Models were fitted using maximum likelihood. Random effects of intercept and time were

included in all models.

Because moderators may interact with Group (CBT or ACT) or Time, both of these

interactions, and the three-way interaction between moderator, Group, and Time were

included in each analysis. Quadratic relationships between moderator, Group, and Time

were assessed. If there was no quadratic relationship, Time was dropped and a moderation of

Mesri et al. Page 6

J Behav Ther Exp Psychiatry. Author manuscript; available in PMC 2018 June 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

Group without Time was assessed. Tests of simple effects were used to explain moderation

effects. More specifically, 1 SD above and below the mean was used to categorize high

avoidant/ fear or low avoidant/ fear individuals. 1 SD was used in order to capture

representative avoidance or fear behavior in a social anxiety group and is typical in previous

moderation studies (Niles et al., 2013).

3. Results

As reported in Craske et al. (2014), CBT and ACT were each more effective than a waitlist

comparison control for symptoms of social anxiety, with no differences between them.

3.1. Moderator of Symptom Composite

Public speaking avoidance significantly interacted with Group and Time to moderate

symptom composite, z = −2.25, p = .045 (see Fig. 1). Tests of simple effects revealed that at 12-month follow-up, more avoidant individuals (operationally defined as 1 SD above the

mean) reported .87 SD fewer symptoms following CBT than ACT, 95% confidence interval

(CI) = .05 to 1.70, z = 2.07, p = .038. No group differences were found for low avoidant individuals (1 SD below the mean), p > .05. Public speaking avoidance did not moderate post-treatment or 6-month follow-up, ps > .05. Neither fear on the public speaking task nor clinician-rated social avoidance or social fear were significant moderators of symptom

composite at any time point, ps > .05.

3.2. Moderator of Stress Reactivity

Public speaking avoidance significantly interacted with Group and Time to moderate stress

reactivity (measured by STAI A-State prior to a stressful laboratory assessment), z = −3.87, p < .001 (see Fig. 2). Tests of simple effects revealed that at 12-month follow-up, more avoidant individuals reported 15.77 fewer points in stress reactivity following CBT than

ACT, CI = 8.38 to 23.17, z = 4.18, p < .001. No group differences were found for low avoidant individuals, p > .05. Public speaking avoidance did not moderate at post-treatment or 6-month follow-up, ps > .05. Neither fear on the public speaking task nor clinician-rated social avoidance or social fear were significant moderators of stress reactivity at any time

point, ps > .05.

3.3. Moderator of Quality of Life

Clinician-rated social fear significantly moderated quality of life, z = −2.12, p = .006 (see Fig. 3). Tests of simple effects revealed that at 6-month follow-up, less fearful individuals

reported 1.32 fewer points in quality of life following CBT than ACT, CI = −2.33 to −.31, z = −2.56, p = .010 and more fearful individuals reported 1.26 more points in quality of life following CBT than ACT, CI = .003 to 2.52, z = 1.96, p = .049. There were no significant differences between high and low clinician-rated fearful individuals in CBT and ACT at

post-treatment and 12-month follow-up, ps > .05. Therefore, this finding is no longer discussed in this paper. Public speaking fear, public speaking avoidance, and clinician-rated

social avoidance were not significant moderators of quality of life at any time point, ps > . 05.

Mesri et al. Page 7

J Behav Ther Exp Psychiatry. Author manuscript; available in PMC 2018 June 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

4. Discussion

The current study tested social avoidance as a moderator of treatment outcome for social

anxiety disorder. Understanding moderators of treatment outcome allow us to better match

patients to a particular treatment, which has important implications for improving treatment

outcome. Our findings suggest that individuals who are more avoidant during a public

speaking task benefit more, in terms of long-term symptoms and stress reactivity, from CBT

than ACT.

Conversely, fear during the public speaking task did not moderate the treatment effects,

suggesting that the results were specific to public speaking avoidance versus fear. Moreover,

clinician-rated social avoidance did not moderate treatment effects, which could imply that

the results were specific to avoidance of public speaking in particular rather than social

avoidance in general. Alternatively, these results may suggest that clinicians may not be

particularly accurate judges of a patient’s degree of social avoidance in their daily life. Such

judgments are likely to be heavily reliant on a patient’s self-report, particularly at an initial

assessment when the clinician has limited information about the patient, and self-report of

avoidance behavior may not be an exact indicator of actual avoidance behavior in laboratory

paradigms (Gamez, Kotov, & Watson, 2010; McNeil, Ries, & Turk, 1995).

We found that more avoidance on the public speaking task predicts better long-term outcome

in CBT than ACT. One possible explanation is that CBT targets avoidance in a structured

way through creation of an exposure hierarchy followed by in-session and homework

exposure assignments. Avoidant individuals may benefit from this structure. A similar

finding has been reported in a panic disorder sample that was randomly assigned to exposure

therapy with an active therapist who guided patients through exposures or a less active

therapist who was not present during assigned exposures (Hamm, et al., 2016). Overall,

panic disorder patients benefitted from exposure therapy; however, patients with greater

public speaking avoidance benefitted even more from therapist-directed exposures than self-

directed exposures. This finding may highlight the added benefit of structure during

exposures (which may be more present in CBT than ACT) for patients with high public

speaking avoidance. Although ACT includes exposure, these exposures are less structured

and their focus is not on fear reduction. Rather, in ACT, individuals conduct exposures in

order to be present, open, mindful, and accepting of their anxious feelings with the eventual

goal of taking committed action toward their values. Thus, in contrast to CBT in which

exposures are a critical strategy for alleviating symptoms, the connection between exposures

and treatment goals is more removed in ACT and possibly simply one of many approaches

toward valued living. Indeed, there was greater adherence to behavioral exposures in CBT

than ACT in the present sample (Craske et al., 2014).

Moderation was found only at the 12-month follow-up, which replicated our prior studies in

the same and different samples (Craske et al., 2014; Niles et al., 2013; Wolitzky-Taylor et

al., 2012). In prior studies, we proposed that experiential avoidance motivated continued

exposure practice over the months following treatment, in turn leading to improved long-

term outcomes (Wolitzky-Taylor et al., 2012). Perhaps those who were most avoidant of

public speaking similarly perceived the benefits of continued exposure practice following the

Mesri et al. Page 8

J Behav Ther Exp Psychiatry. Author manuscript; available in PMC 2018 June 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

end of treatment resulting in better long-term outcome in CBT than ACT. It is also important

to note that CBT was supervised directly by Dr. Craske and her team, whereas ACT was

only supervised by Dr. Hayes’s team and not himself. It is possible that if Dr. Hayes had

supervised the therapists, outcomes from ACT may have differed. Moreover, more

comprehensive measures of avoidance would be useful for the field moving forward.

Despite limitations, this is one of few studies that investigated moderators of ACT and CBT

for social anxiety disorder. Asking patients to give a speech and identifying how long they

are willing to speak may be a simple way of assessing behavioral avoidance. It may provide

useful long-term prognostic information not gleaned by traditional methods such as rating

levels of social avoidance based largely on patient self-report. Furthermore, should these

results be replicated, they suggest that those who are more behaviorally avoidant may benefit