The illusion of transparency and the alleviation of speech anxiety

Journal of
Experimental
Social Psychology
Journal of Experimental Social Psychology 39 (2003) 618–625
www.elsevier.com/locate/jesp
The illusion of transparency and the alleviation of speech anxietyq
Kenneth Savitskya,* and Thomas Gilovichb
a Department of Psychology, Bronfman Science Center, Williams College, Williamstown, MA 01267, USA
b Department of Psychology, Uris Hall, Cornell University, Ithaca, NY 14853, USA
Received 1 July 2002; revised 25 March 2003
Abstract
Individuals often believe their internal states are more apparent to others than is actually the case, a phenomenon known as the
illusion of transparency. In the domain of public speaking, for example, individuals who are nervous about delivering a public speech
believe their nervousness is more apparent to their audience than it actually is, a ?nding we document in Study 1. We contend that
the illusion of transparency can play a role in the self-exacerbating nature of speech anxiety, and show in Study 2 that an awareness
of the illusion can improve the quality of a speakerÕs performance, from both the speakerÕs own perspective and in the eyes of
observers. Discussion focuses on the application of these ?ndings to the treatment of speech anxiety and other forms of social
anxiety.
Ó 2003 Elsevier Science (USA). All rights reserved.
I turn pale at the outset of a speech and quake in every limb and
McFarland, 1987, 1991; Vorauer & Ross, 1999). This
in all my soul.
illusion derives from the di?culty people have in getting
Cicero, De Oratore
beyond their own phenomenological experience when
Individuals often wish to conceal their internal states.
attempting to determine how they appear to others. The
Anxiety over approaching a potential romantic partner,
adjustment one makes from the ‘‘anchor’’ of oneÕs own
feelings of disgust over a disagreeable entreee served at a
phenomenology, like adjustments to anchors generally,
dinner party, or nervousness over delivering a public
tends to be insu?cient (Chapman & Johnson, 2002;
speech—all are internal states one may wish, for a va-
Epley & Gilovich, 2001, 2003; Jacowitz & Kahneman,
riety of reasons, to keep private.
1995; Savitsky & MacIntosh, 2003). As a result, people
How well can people conceal their internal states,
exaggerate the extent to which their internal states
and how well do they believe they can do so? Research
‘‘leak out’’ and overestimate the extent to which others
suggests that individuals are typically better at dis-
can detect their private feelings.
guising their internal states than they believe—i.e.,
The illusion of transparency has been documented in
people are prone to an illusion of transparency, or a
a variety of domains and across a number of procedural
belief that their thoughts, feelings, and emotions are
variations (Gilovich & Savitsky, 1999). In one study,
more apparent to others than is actually the case
participants who were asked to tell lies believed they had
(Gilovich,
Savitsky,
&
Medvec,
1998;
Miller
&
leaked more cues to their deception than they actually
had and overestimated the extent to which others could
detect their falsehoods. In another study, participants
q
who sampled foul-tasting drinks in view of an observer
Portions of this research were presented at the annual meeting of
the Eastern Psychological Association, Boston, 1998. This research
believed that their disgust was more apparent than was
was supported by Research Grant MH45531 from the National
actually the case (Gilovich et al., 1998). Other research
Institute of Mental Health and SBR9319558 from the National Science
has demonstrated that individuals overestimate the ex-
Foundation. We thank Adi Bar-Lev and Lucy Berk for serving as
tent to which their internal states are apparent to others
experimenters in Study 2.
*
in contexts as varied as negotiations (Van Boven,
Corresponding author. Fax: 1-413-597-2085.
E-mail address: ksavitsk@williams.edu (K. Savitsky).
Gilovich, & Medvec, in press; Vorauer & Claude, 1998),
0022-1031/$ - see front matter Ó 2003 Elsevier Science (USA). All rights reserved.
doi:10.1016/S0022-1031(03)00056-8

---

K. Savitsky, T. Gilovich / Journal of Experimental Social Psychology 39 (2003) 618–625
619
police interrogations (Kassin & Fong, 1999), and po-
illusion of transparency, their anxiety can become self-
tential emergencies (Gilovich et al., 1998).
perpetuating. An individual who experiences some
One purpose of the present investigation is to expand
anxiety while giving a speech may believe it is more
further the range of situations to which the illusion of
apparent to the audience than is actually the case. This
transparency applies by investigating a domain in which
thought—that the audience is aware of just how nervous
such a misconception can be both profound and con-
he or she feels—may ironically serve to make the speaker
sequential: public speaking. Many individuals report
all the more nervous. The speaker may then believe that
experiencing signi?cant anxiety when called upon to
this newfound nervousness is itself apparent to others,
speak in public (Leary & Kowalski, 1995; McCroskey,
leading to still more nervousness, concerns about leak-
1970, 1977)—anxiety that can interfere with oneÕs social
age, and so on.1
life and education, lead to job absenteeism, and even
Speech anxiety may thus be an example of what
force one to change careers (Hohenstein, 1986; Monroe,
Storms and McCaul (1976) have termed a ‘‘self-exac-
Borzi, & Burrell, 1992; Stein, Walker, & Forde, 1996).
erbating syndrome.’’ They note that the experience of
These individuals may anchor on their internal sensa-
certain conditions, such as stuttering, insomnia, or
tions of anxiety, correct insu?ciently for the fact that
writerÕs block, can lead to an ironic increase in those
others are less privy to those sensations than they are
very conditions (cf., Ascher & Schotte, 1999; Wegner,
themselves, and consequently overestimate the extent to
1997; Wegner, Broome, & Blumberg, 1997). We believe
which their anxiety is apparent to onlookers. As Miller
the illusion of transparency may lend a similar self-
and McFarland (1991) note, ‘‘in anxiety-provoking sit-
exacerbating quality to speech anxiety. At the same
uations, it is often very di?cult for people to believe
time, however, the realization that oneÕs nervousness is
that, despite feeling highly nervous, they do not appear
less apparent than one thinks may be useful in allevi-
highly nervous’’ (p. 310).
ating speech anxiety: If individuals can be convinced
Indeed, the literature on speech anxiety provides
that their internal sensations are not manifested in
some evidence consistent with this possibility. In par-
their external appearance, one source of their anxiety
ticular, several studies have shown discrepancies be-
can be attenuated, allowing them to relax and even
tween speakersÕ self-ratings and ratings provided by
improving the quality of their performance. Thus,
others. Behnke, Sawyer, and King (1987) found that
speakers who know about the illusion of transparency
public speakers felt more nervous than audience mem-
may tend to give better speeches than speakers who do
bers believed they did, and Rapee and Lim (1992) found
not.
that participants evaluated their own speeches more
We put this speculation to empirical test in Study 2,
negatively than they were evaluated by their audience.
examining whether knowledge of the illusion of trans-
In the only existing study that reports the key compar-
parency can alleviate speech anxiety. But ?rst we seek, in
ison between how speakers thought they would be
Study 1, to document the illusion of transparency in the
evaluated by others and how others actually evaluated
domain of public speaking.
them, Mansell and Clark (1999) found that socially
anxious individuals did indeed overestimate how anx-
ious they appeared during an extemporaneous speech
Study 1
performance, relative to the judgments of observers.
Unfortunately, however, their report does not permit an
Participants delivered extemporaneous speeches in
assessment of whether this e?ect held true for their
pairs. Each speaker then rated the degree to which he or
sample as a whole.
she appeared nervous during his or her speech, as well
Given the uncertain nature of the existing evidence on
as the extent to which his or her counterpart appeared
this issue, our ?rst goal in the present investigation was
nervous. We expected speakers to overestimate the
to collect unambiguous evidence of an illusion of
transparency in the domain of public speaking. Our
second goal was to explore a signi?cant implication of
1 In fact, an individual need not even experience any bona ?de
the illusion of transparency—how it can exacerbate the
nervousness to set this cycle in motion. Normal preparatory arousal,
intensity of speech anxiety, and, more important, how
stemming from the simple need to ‘‘get with it’’ and mobilize whatever
cognitive and motor programs are required to perform the task at
an awareness of the illusion (that is, an appreciation that
hand, may be misattributed to pangs of anxiety and self-doubt. As
oneÕs anxiety is typically not as apparent to others as one
Olson (1988) notes, ‘‘the distinction between arousal caused by
suspects) can help alleviate that anxiety. Many people
Ôgearing up for a speechÕ and arousal caused by Ôspeech anxietyÕ is
report that one component of their fear of public
subtle’’ (p. 766). Thus, the individual may reason heuristically and
speaking is the concern that they will ‘‘tremble, shake, or
conclude, ‘‘If I am feeling these feelings of arousal, I must be nervous
about giving this speech’’ (Savitsky, Medvec, Charlton, & Gilovich,
show some other signs of anxiety’’ (Stein et al., 1996, p.
1998). The concern that such feelings are apparent to others, then, may
172). In other words, people are often nervous about
give rise to a more genuine experience of anxiety, which the individual
looking nervous. If public speakers are prone to an
may think is also leaking out, and so on.

---

620
K. Savitsky, T. Gilovich / Journal of Experimental Social Psychology 39 (2003) 618–625
extent to which their nervousness was apparent, and
order: ?rst vs. second) Â 2 (target: self vs. other) re-
thus to rate themselves as having appeared more ner-
peated-measures analysis of variance (ANOVA), which
vous, on average, than they were rated as having ap-
revealed only the predicted main e?ect of target,
peared.
F ð1;19Þ ¼ 16:38; p < :001. Speakers overestimated how
nervous they appeared to observers.
Method
This result was echoed in participantsÕ responses to
the ?nal item. As anticipated, a majority of the partici-
Participants
pants (27 of 40; 67.5%) indicated that they had appeared
Forty Cornell University students participated in
more nervous than their counterpart, p < :05 by the
pairs.
binomial test. Indeed, in eight pairs, both participants
felt they appeared more nervous than the other speaker;
Procedure
in only one pair did both feel the other speaker had
The experimenter explained that the study was con-
appeared more nervous than they had.
cerned with extemporaneous public speaking. In turns
(determined by coin ?ip), each participant was asked to
Discussion
stand at a podium at the front of the room and was
handed an index card containing a speech topic. All
Together with previous ?ndings, these results estab-
participants were asked to begin immediately and speak
lish an illusion of transparency in the domain of public
for 3 min on one of two topics—The Best and Worst
speaking. Individuals asked to deliver extemporaneous
Things About Life Today and The Most Important Moral
speeches overestimated the extent to which their ner-
Lessons to Teach One’s Children. The participants within
vousness was apparent, rating themselves as having
each pair always spoke on di?erent topics with topic
appeared more nervous, on average, than they actually
order counterbalanced across pairs.
appeared to another participant in the same experi-
After both participants spoke, they were escorted to
mental session. Indeed, in many cases, both participants
separate cubicles to complete a questionnaire containing
believed they had appeared more nervous than their
the dependent measures. Participants rated the degree to
counterpart (cf., Miller & Prentice, 1994).
which they felt anxious or nervous while delivering their
Readers may take issue with one element of the
speech, the degree to which they believed they had ap-
procedure used in Study 1, however. Rather than rely on
peared anxious or nervous, and the degree to which their
passive observers, we had speakers evaluate their ap-
counterpart had appeared anxious or nervous. All
pearance vis-
a
a-vis the impressions they made on another
judgments were made on scales ranging from 0 (not at
speaker. Thus, it could be argued that the speakers in
all) to 10 (very). Finally, participants were asked to
Study 1 made poor observers because they were unduly
check one of the following two statements: (a) I ap-
focused on their own upcoming (or recently delivered)
peared more nervous than the other participant or (b) The
speech (Gilbert, Pelham, & Krull, 1988).
other participant appeared more nervous than I did.
To address this possibility, we replicated Study 1,
using observers who did not themselves serve as speak-
Results
ers, and who were therefore free to devote their full at-
tention to speakersÕ presentations. We recruited 42
Participants indicated that they did indeed feel
Williams College students, 17 of whom served as
somewhat nervous as they delivered their speech
speakers and the rest as observers. Speakers reported to
(M ¼ 5:95). As expected, however, participants rated
the laboratory individually, were videotaped as they
themselves
as
having
appeared
more
nervous
delivered an extemporaneous speech on The Best and
(M ¼ 6:55)2 than they were rated by their counterparts
Worst Things About Life Today, and rated the degree to
as having appeared (M ¼ 5:25). To examine the statis-
which they believed they had appeared nervous on a
tical signi?cance of this e?ect, we performed a 2 (speaker
scale from 0 (not at all) to 10 (very). Groups of ?ve
observers then viewed the speakersÕ videotapes and rated
the degree to which the speakers appeared nervous.
2 That speakers indicated they appeared somewhat more nervous
Each group of observers viewed the speeches of three or
than they felt runs counter to the anchoring-and-adjustment mecha-
four speakers. We collapsed observersÕ ratings of each
nism and the phenomenology of ‘‘leakage’’ that we have argued lie at
speaker into a single measure and compared the result-
the heart of the illusion of transparency—and diverges from previous
?ndings, in which participants typically report experiencing phenom-
ing index to speakersÕ own ratings. As expected, speak-
enological states more intensely than they believe they are manifested
ers believed they appeared more nervous (M
externally (Gilovich et al., 1998). We suspect that the anomalous
¼ 5:71)
than they actually appeared to the observers (M
?nding here stems from the fact that nervousness is a widely
¼ 4:24),
t
acknowledged component of the public-speaking ‘‘script’’—a script
ð16Þ ¼ 2:94, p < :01.
Once again, some readers may ?nd fault with this
speakers may have expected observers to be over-zealous in applying
to their performance.
procedure. Speci?cally, the fact that observers viewed

---

K. Savitsky, T. Gilovich / Journal of Experimental Social Psychology 39 (2003) 618–625
621
several speeches may have encouraged them to use the
pect. Psychologists have documented what is called an ‘‘illusion
response scale in a comparative fashion that di?ered
of transparency.’’ Those speaking feel that their nervousness is
from the way speakers used it. Nevertheless, it is note-
transparent, but in reality their feelings are not so apparent to
observers. This happens because our own emotional experience
worthy that the results from two divergent procedures,
can be so strong, we are sure our emotions ‘‘leak out.’’ In fact,
each with its own advantages and disadvantages, con-
observers arenÕt as good at picking up on a speakerÕs emotional
verge on the same ?nding: Individuals who deliver
state as we tend to expect. So, while you might be so nervous
public speeches are subject to an illusion of transparency
youÕre convinced that everyone can tell how nervous you are,
and overestimate the extent to which their nervousness
in reality thatÕs very rarely the case. WhatÕs inside of you typi-
cally manifests itself too subtly to be detected by others. With
is apparent to others.
this in mind, you should just relax and try to do your best.
Know that if you become nervous, youÕll probably be the only
one to know.3
Study 2
For speakers in the reassured condition (n ¼ 18), the
experimenter provided verbal reassurances but did not
Our goal in Study 2 was to explore the role of the il-
inform them about the illusion of transparency:
lusion of transparency in the exacerbation of speech
anxiety. Since a portion of some individualsÕ nervousness
I think it might help you to know that you shouldnÕt worry
over public speaking stems from the fear that their ner-
much about what other people think. Psychologists have found
that you donÕt need to be concerned about other peopleÕs im-
vousness will be detected by others (Stein et al., 1996),
pressions. This is hard to do because our own emotional expe-
the conviction that they are transparent (even if they are
rience of anxiety can be so strong, but past research has shown
not) can cause them considerable anxiety—anxiety they
that we shouldnÕt be worried about this. With this in mind, you
may believe is itself apparent to their audience, and so
should just relax and try to do your best. Know that if you be-
on. This implies that disabusing individuals of the notion
come nervous, you probably shouldnÕt worry about it.
that their nervousness is ‘‘written all over their face’’ may
This condition was included to control for the pos-
forestall the spiral of anxiety described above. This, in
sibility that any di?erences in the speeches of those in
turn, may improve the quality of individualsÕ speeches—
the informed and control conditions might stem not
both in their own eyes and in the eyes of others.
from whether or not speakers possessed an appreciation
of the illusion of transparency, but from the implicit
Method
suggestion that one should not be concerned with ap-
pearing nervous that the instructions in the informed
Participants
condition provided.
One hundred seventeen Cornell University students
Speakers then delivered their speeches in front of a
served as participants, 77 as speakers and 40 as ob-
videocamera in the presence of a second experimenter
servers. A portion of the data from 9 of the speakers was
who was unaware of their condition. In addition,
inadvertently misplaced; some analyses were therefore
speakers were told that other members of the research
conducted on data from all 77 speakers and some on
team were stationed behind a one-way mirror; in reality,
data from 68 speakers.
there were no observers behind the mirror. When ?n-
ished, speakers completed a questionnaire containing
Procedure
the dependent measures. Unless otherwise speci?ed, all
Speakers reported to the laboratory individually,
ratings were made on scales from 1 (not at all) to 7
where they were informed that they would deliver a
(very). First, speakers rated the overall quality of their
3-min speech on race relations at Cornell. Speakers were
speech on a scale from 1 (very poor quality) to 7 (very
randomly assigned to one of three conditions and were
high quality), along with their e?ectiveness as a speaker
given 5 min to prepare. Those in the control condition
and how expressive they had been. Next, speakers rated
(n ¼ 20) were given no additional instructions. To those
the extent to which they were relaxed before giving their
in the other two conditions, the experimenter remarked:
speech, and the extent to which they were nervous
during their speech. Speakers were then asked to antic-
I realize you might be anxious. ItÕs perfectly natural to be anx-
ipate how an observer who viewed their speech would
ious when confronted with a public speaking task. Many people
rate its overall quality (on the same 1–7 scale), how
become anxious not only because theyÕre concerned about
whether or not theyÕll do well, but also because they believe they
expressive an observer would rate them as having been,
will appear nervous to those who are watching. TheyÕre nervous
about looking nervous.
3 Approximately half of the participants assigned to the informed
For participants in the informed condition (n ¼ 39),
condition were given these instructions before they prepared their
the experimenter continued by saying:
speech; the others were given these instructions after the 5-min
preparation period but before they delivered their speech. Because
I think it might help you to know that research has found that
there were no signi?cant di?erences between these two conditions on
audiences canÕt pick up on your anxiety as well as you might ex-
any of the dependent measures, we collapsed across them.

---

622
K. Savitsky, T. Gilovich / Journal of Experimental Social Psychology 39 (2003) 618–625
and how anxious they thought they appeared to ob-
Table 1
servers. Finally, speakers indicated the extent to which
Mean ratings of speeches by speakers and observers, Study 2
they had discussed the assigned topic prior to the ex-
Type of rating
Condition
periment, and how comfortable they usually were in
Informed
Control
Reassured
public speaking situations.
Speakers’ self-ratings
In the second phase of the experiment, speakersÕ
Speech quality
3.50Ã
3.04
2.83
videotaped speeches were viewed by observers. First, to
How relaxed
3.94
3.47
3.25
familiarize observers with the task and give them an idea
Anticipated quality
3.54y
3.00
3.15
of the range they could expect to see in the speeches,
Relaxed appearance
4.20ÃÃÃÃ
3.35
2.69
each observer was shown a pair of speeches that varied
Observers’ ratings
widely in overall quality. Observers then viewed 8–10
Composed appearance
4.65ÃÃÃ
3.90
3.94
speeches and rated how relaxed the speakers appeared
Speech quality
4.23ÃÃ
3.50
3.62
before giving their speech, how composed they appeared
Note. Scores could vary from 1 to 7. Some items have been reverse-
during their speech, how expressive they were, and how
scored such that higher numbers indicate more favorable ratings on all
e?ective they were as a speaker. These ratings were made
measures. Signi?cance tests refer to planned contrasts that assign a
weight of +2 to the informed condition and weights of )1 to both the
on scales from 1 (not at all) to 7 (very). In addition,
control and reassured conditions.
observers rated the quality of each speakerÕs speech,
* p < :05.
both stylistically and substantively, on scales from 1
** p < :01.
***
(very low quality) to 7 (very high quality).
p < :005.
**** p < :
Observers made their ratings individually or in
001.
y p < :06.
groups of two or three. When more than one observer
was present, dividers were placed between them so that
they were unable to view one anotherÕs responses. Each
lapsed the ?rst three measures into an index of self-rated
speech was rated by ?ve observers and the experimenter
speech quality and the latter two (after appropriate re-
who oversaw these sessions was unaware of each
verse-scoring) into an index of the degree to which
speakerÕs condition.
speakers were relaxed. Speakers informed about the il-
lusion of transparency evaluated their speeches more
Results
positively than did speakers in the control or reassured
conditions, F ð1; 65Þ ¼ 4:47; p < :05. On the other hand,
First, it is reassuring that neither the extent to which
although speakers in the informed condition rated
speakers had previously discussed the assigned topic nor
themselves as somewhat more relaxed than speakers in
their typical comfort with public speaking di?ered
the other two conditions, this di?erence fell short of
across conditions, Fsð2; 65Þ ¼ 0:95 and 1.61, respec-
signi?cance, F ð1; 65Þ ¼ 2:43, ns.
tively, ns. But did our manipulation have an e?ect on
How did speakers believe they appeared in the eyes of
participantsÕ speech performance during the experiment
observers? SpeakersÕ predictions of how observers would
itself? To ?nd out, we examined observersÕ evaluations
rate the quality of their speech were correlated with their
of speakers, as well as speakersÕ evaluations of their own
predictions of how expressive observers would rate them
performance (see Table 1). Because our hypotheses
(r ¼ :61), so we collapsed the two variables into an index
centered around a comparison of speakers informed
of anticipated ratings of speech quality. Analysis of this
about the illusion of transparency and those in the
index revealed that speakers in the informed condition
other two conditions, and because preliminary analyses
expected their speech to be rated more positively than
revealed no signi?cant di?erences between the control
did speakers in the control or reassured conditions,
and reassured conditions on any of the dependent
F ð1;65Þ ¼ 3:70; p < :06. In addition, analysis of
measures, we used planned contrasts with the error
speakersÕ predictions of how anxious they thought they
terms from one-way ANOVAs to analyze the data. In
appeared to observers (reverse-scored so that higher
each case, the contrasts assigned weights of +2 to the
numbers indicate less anxiety) revealed, as expected, that
informed condition and )1 to the control and reassured
speakers in the informed condition thought they ap-
conditions.
peared signi?cantly more relaxed than did speakers in
the other two conditions, F ð1; 65Þ ¼ 12:30; p < :001.
Speakers’ self-ratings
SpeakersÕ ratings of the quality of their speech, their
Observers’ ratings
e?ectiveness as a speaker, and the extent to which their
Recall that observers evaluated speakersÕ perfor-
speech was expressive were highly correlated (mean
mance on six dimensions. We grouped two of these—
r ¼ :71), as were speakersÕ ratings of how relaxed they
how relaxed speakers appeared before their speech and
were before giving their speech and how nervous they
how composed they appeared during their speech—into
were during their speech (r ¼ À:70). We therefore col-
an index of how composed speakers appeared (r ¼ :89).

---

K. Savitsky, T. Gilovich / Journal of Experimental Social Psychology 39 (2003) 618–625
623
The remaining items—how expressive speakers were,
can plague individuals experiencing speech anxiety.
how e?ective they were, the quality of their speech on a
Knowing about the illusion of transparency, in other
stylistic basis, and the quality of their speech on a sub-
words, allows speakers to be better speakers.
stantive basis—were grouped into an index of overall
speech quality (mean r ¼ :86). Analyses of these indexes
revealed strong support for our hypothesis. First,
General discussion
speakers informed about the illusion of transparency,
who were freed from worry that their nervousness would
When individuals are called upon to speak in public,
be apparent to observers, appeared signi?cantly more
they do not appear as nervous as they think they do. In
composed than did speakers in the control or reas-
Study 1, participants who gave extemporaneous spee-
sured conditions, F ð1; 74Þ ¼ 9:49; p < :005. Moreover,
ches overestimated the extent to which their nervousness
speakers in the informed condition were rated as having
was apparent to others. This result extends the illusion
delivered better speeches than their counterparts in the
of transparency to the domain of public speaking.
other two conditions, F ð1; 74Þ ¼ 7:94; p < :01.4
The results of Study 2 suggest that the illusion of
transparency can play a role in the exacerbation of
Discussion
speech anxiety, and point to one way that such anxiety
can be circumvented. Public speakers are often nervous
These results demonstrate that speakers informed
over the (largely illusory) prospect that their nervous-
about the illusion of transparency evaluate their spee-
ness is apparent to their audience—a concern that serves,
ches more positively, and expect observers to evaluate
ironically, to increase their nervousness. Speakers who
their speeches more positively (and them as more re-
were informed that their nervousness was not as ap-
laxed), than speakers not so informed. Moreover, the
parent as they thought (but not participants who were
bene?ts of being informed about the illusion of trans-
merely reassured by the experimenter) were able to es-
parency carry over to the actual evaluations of their
cape this spiral of nervousness and concern over leak-
speeches by impartial observers. Informed speakers were
age. As a result, they delivered speeches that were rated
seen as more composed by observers, and their speeches
more positively than the speeches of those not so in-
were evaluated more positively, than those not informed
formed. Our results thus lend credence to the notion
of the illusion. The same bene?ts were not obtained
that ‘‘the truth can set you free’’: Knowing the truth
from a simple reassurance and instructions not to worry
about the illusion of transparency set participants free
about what others think, suggesting that the instructions
from the cycle of anxiety that can plague those who
in the informed condition were e?ective because they
engage in public speaking—and helped them deliver
allowed participants to escape the spiral of nervousness
better speeches.
and concern over the leakage of that nervousness that
Because speech anxiety can be so consequential for
some people, there is interest in developing e?ective
treatment techniques. Might the technique used in the
4 At ?rst glance, one might expect participants informed about the
present research—informing individuals about the illu-
illusion of transparency to display it less than those not informed. If
sion of transparency—be useful in clinical settings for
informed participants take their instruction to heart, they are likely to
treating speech anxiety and other forms of social anxi-
lower their estimates of how apparent their nervousness is to others.
ety? Our ?ndings suggest that it might be. Simply in-
But note that because informing participants about the illusion
forming participants that their nervousness was less
succeeded in reducing their anxiety, the judgesÕ ratings and the
speakersÕ estimates of the judgesÕ ratings are likely to be lowered in
apparent to others than they thought improved the
lock-step.
quality of their speeches not only in their own eyes, but
Accordingly, we did not design this study with the goal of tracking
in the eyes of observers. The literature on the use of
the magnitude of the illusion of transparency across conditions. In-
‘‘cognitive modi?cation’’ reinforces the view that alter-
deed, the dependent measures completed by speakers and observers
ing individualsÕ beliefs in this way can be an e?ective
were not parallel, making comparison dubious (speakers indicated how
anxious they thought they appeared whereas observers rated how re-
treatment for speech anxiety (Allen, Hunter, & Don-
laxed speakers appeared). Nevertheless, because speakersÕ ratings were
ohue, 1989). Moreover, it is particularly tempting to
reverse-scored in Table 1, readers may be tempted to compare them
consider our approach since, unlike the use of misat-
with observersÕ ratings. To formalize this comparison, we conducted a
tribution techniques to alleviate speech anxiety (i.e.,
2 Â3 mixed-model ANOVA, comparing how anxious speakers be-
encouraging individuals to attribute their anxiety about
lieved they appeared to observers (reverse-scored) with observersÕ
ratings of how composed speakers appeared during their speeches,
public speaking to ‘‘subliminal noise’’ or other illusory
across the three conditions. Although the discrepancy between pre-
sources; Olson, 1988; Olson & Ross, 1988), our manip-
dicted and actual ratings was smaller in the informed condition
ulation does not require deception—indeed, it involves
(M ¼ 0:36) than the control or reassured conditions (Ms ¼ 0:78 and
the disclosure of actual research ?ndings—and so does
1.22), the interaction was not signi?cant, F ð2;65Þ ¼ 1:74, ns. As ex-
not raise ethical concerns when applied to a patient
pected, informing participants about the illusion of transparency did
not reduce its magnitude signi?cantly.
population.

---

624
K. Savitsky, T. Gilovich / Journal of Experimental Social Psychology 39 (2003) 618–625
But there are (as always) reasons to be cautious in
psychology of intuitive judgment (pp. 120–138). Cambridge, En-
applying the present laboratory ?ndings to a clinical
gland: University Press Cambridge.
Epley, N., & Gilovich, T. (2001). Putting adjustment back in the
setting. For one thing, informing individuals about the
anchoring and adjustment heuristic: Di?erential processing of self-
illusion of transparency would be expected to decrease
generated and experimenter-provided anchors. Psychological Sci-
anxiety only in individuals who are made nervous by the
ence, 12, 391–396.
belief that their nervousness is apparent to others. Al-
Epley, N., & Gilovich, T., (2003). Are adjustments insu?cient?
though this fear is su?ciently common for our manip-
Harvard University, manuscript submitted for publication.
Gilbert, D. T., Pelham, B. W., & Krull, D. S. (1988). On cognitive
ulation to have been e?ective, there are undoubtedly
busyness: When person perceivers meet persons perceived. Journal
some individuals who su?er from speech anxiety for
of Personality and Social Psychology, 54, 733–740.
whom this is not a part of their concern (Stein et al.,
Gilovich, T., & Savitsky, K. (1999). The spotlight e?ect and the illusion
1996). Moreover, note that the speeches delivered by
of transparency: Egocentric assessments of how we are seen by
participants in Study 2 followed closely on the heels of
others. Current Directions in Psychological Science, 8, 165–168.
Gilovich, T., Savitsky, K., & Medvec, V. H. (1998). The illusion of
the manipulation. We have no way of knowing whether
transparency: Biased assessments of othersÕ ability to read our
the e?ects of our manipulation would have endured
emotional states. Journal of Personality and Social Psychology, 75,
beyond the con?nes of the experimental session, as one
332–346.
would desire in a therapeutic intervention. It thus re-
Hohenstein, R. D. (1986). Communication apprehension and its
mains for future research to determine if the e?ects of
e?ect on job absenteeism, turnover, satisfaction, and promotion.
Unpublished doctoral dissertation, University of Oregon.
informing individuals about the illusion of transparency
Jacowitz, K. E., & Kahneman, D. (1995). Measures of anchoring in
can be of use to a clinical population.
estimation tasks. Personality and Social Psychology Bulletin, 21,
Do the present ?ndings apply beyond the domain of
1161–1166.
speech anxiety, to other varieties of social anxiety? We
Kassin, S. M., & Fong, C. T. (1999). ‘‘IÕm innocent!’’: E?ects of
chose to focus on speech anxiety in part because so
training on judgments of truth and deception in the interrogation
room.
many otherwise well-adjusted people report signi?cant
Law and Human Behavior, 23, 499–516.
Leary, M. R., & Kowalski, R. M. (1995). Social anxiety. New York:
anxiety about public speaking (Leary & Kowalski,
Guilford.
1995). Nevertheless, what distinguishes speech anxiety
Mansell, W., & Clark, D. M. (1999). How do I appear to others? Social
from other forms of social anxiety is little more than the
anxiety and processing of the observable self. Behaviour Research
context in which it occurs. Indeed, one can readily
and Therapy, 37, 419–434.
McCroskey, J. C. (1970). Measures of communication-bound anxiety.
imagine the same processes described in this article
Speech Monographs, 37, 269–277.
plaguing, say, an individual who is chronically nervous
McCroskey, J. C. (1977). Oral communication apprehension: A
about approaching potential romantic partners. Here
summary of recent history and research. Human Communication
again, concerns that oneÕs nervousness is visible may
Research, 4, 78–96.
give rise to more nervousness, heightened fears of
Miller, D. T., & McFarland, C. (1987). Pluralistic ignorance: When
similarity is interpreted as dissimilarity.
leakage, and so on. And again, the realization that oneÕs
Journal of Personality and
Social Psychology, 53, 298–305.
nervousness is largely concealed may go a long way
Miller, D. T., & McFarland, C. (1991). When social comparison goes
towards quelling oneÕs anxiety and improving oneÕs
awry: The case of pluralistic ignorance. In J. Suls, & T. Wills (Eds.),
‘‘performance.’’
Social comparison: Contemporary theory and research (pp. 287–
In general, then, we suggest that any time people are
313). Hillsdale, NJ: Erlbaum.
Miller, D. T., & Prentice, D. A. (1994). Collective errors and errors
nervous over the very prospect of appearing nervous,
about the collective. Personality and Social Psychology Bulletin, 20,
the illusion that they are more transparent than they
541–550.
actually are can add fuel to the ?res of their anxiety. An
Monroe, C., Borzi, M. G., & Burrell, R. D. (1992). Communication
appreciation of the illusion of transparency may help to
apprehension among high school dropouts. School Counselor, 39,
dampen the ?ames.
273–280.
Olson, J. M. (1988). Misattribution, preparatory information, and
speech anxiety. Journal of Personality and Social Psychology, 54,
758–767.
References
Olson, J. M., & Ross, M. (1988). False feedback about placebo
e?ectiveness: Consequences for the misattribution of speech anx-
Allen, M., Hunter, J. E., & Donohue, W. A. (1989). Meta-analysis of
iety. Journal of Experimental Social Psychology, 24, 275–291.
self-report data on the e?ectiveness of public speaking anxiety
Rapee, R. M., & Lim, L. (1992). Discrepancy between self- and
treatment techniques. Communication Education, 38, 54–76.
observer-ratings of performance in social phobics. Journal of
Ascher, L. M., & Schotte, D. E. (1999). Paradoxical intention and
Abnormal Psychology, 101, 728–731.
recursive anxiety. Journal of Behavior Therapy and Experimental
Savitsky, K., & MacIntosh, A. E. (2003). If youÕre happy and you
Psychiatry, 30, 71–79.
know it, will your face surely show it?: The illusion of transparency
Behnke, R. R., Sawyer, C. P., & King, P. E. (1987). The communi-
in facial expressions of emotion. Williams College, manuscript
cation of public speaking anxiety. Communication Education, 36,
submitted for publication.
138–141.
Savitsky, K., Medvec, V. H., Charlton, A. E., & Gilovich, T. (1998).
Chapman, G. B., & Johnson, E. J. (2002). Incorporating the irrelevant:
‘‘What, me worry?’’: Arousal, misattribution, and the e?ect of
Anchors in judgments of belief and value. In T. Gilovich, D.
temporal distance on con?dence. Personality and Social Psychology
Gri?n, & D. Kahneman (Eds.), Heuristics and biases: The
Bulletin, 24, 529–536.

---

K. Savitsky, T. Gilovich / Journal of Experimental Social Psychology 39 (2003) 618–625
625
Stein, M. B., Walker, J. R., & Forde, D. R. (1996). Public-speaking
Vorauer, J. D., & Claude, S.-D. (1998). Perceived versus actual
fears in a community sample: Prevalence, impact on functioning,
transparency of goals in negotiation. Personality and Social
and diagnostic classi?cation. Archives of General Psychiatry, 53,
Psychology Bulletin, 24, 371–385.
169–174.
Vorauer, J. D., & Ross, M. (1999). Self-awareness and feeling
Storms, M. D., & McCaul, K. D. (1976). Attribution processes and
transparent: Failing to suppress oneÕs self. Journal of Experimental
emotional exacerbation of dysfunctional behavior. In J. H.
Social Psychology, 35, 415–440.
Harvey, W. J. Ickes, & R. F. Kidd (Eds.), New directions in
Wegner, D. M. (1997). When the antidote is the poison: Ironic mental
attribution
research (Vol.
1,
pp. 143–164). Hillsdale, NJ:
control processes. Psychological Science, 8, 148–150.
Erlbaum.
Wegner, D. M., Broome, A., & Blumberg, S. J. (1997). Ironic e?ects of
Van Boven, L., Gilovich, T., & Medvec, V. H. (in press). The illusion
trying to relax under stress. Behaviour Research and Therapy, 35,
of transparency in negotiations. Negotiation Journal.
11–21.

---

Document Outline

• The illusion of transparency and the alleviation of speech anxiety
  • Study 1
    • Method
      • Participants
      • Procedure
    • Results
    • Discussion
  • Study 2
    • Method
      • Participants
      • Procedure
    • Results
      • Speakers’ self-ratings
      • Observers’ ratings
    • Discussion
  • General discussion
  • References

---