Shyness, public and private, and its relationship to other measures of social behavior1

Full text

shyness, public and private, and
its
relationship
to
other measures
of
social behavior'
Paul A. Pilkonis,* Stanford
University
Complaints of social anxiety and interpersonal difficulties are
common among adolescents and young adults (Borkovec, Stone,
O'Brien,
&
Kaloupek, 1974; Bryant
&
Trower, 1974; Martinson &
Zerface, 1970). Zimbardo, Pilkonis, and Norwood (1974), for
example, foimd that 42 percent of a sample of high school and col-
lege age persons {N = 817) called themselves disposidonaUy
"shy
persons."
Of the respondents who had ever considered them-
selves shy persons (past, present, or both), 86 percent did not Hke
being shy, and
63
percent called shyness a real "problem." Clearly,
shyness (defined as a tendency to avoid social interacdon and to,
fail to pardcipate appropriately in social situadons) deserves
addidonal attention because of both its prevalence and its clinical
importance.
The present research was conceived as part of an effort to
provide such attention. First, the research was designed to ad-
dress some preliminary "fact-finding" quesdons: How rehable
are the earlier prevalence results? What are the convergent and
discriminant reladonships between self-reports of shyness ,.and
offiCT'measafes of isocial Mxfety7seff-consciousnesr,~seK^
ing, and neurodcism? Second, the work involved an attempt to
identify different types of shy people in order to give more preci-
1.
This article is based on a dissertation submitted to the Department of
Psychology, Stanford University. The research was generously supported by
NIMH training grant MH12283 in social psychology, the Boys Town Genter for
Youth Development at Stanford University, and NIMH grant MH27543 to Philip
Zimbardo. While the research was conducted, the author held first a predoctoral
fellowship from NSF and then a Research Service Award from NIMH. The
author is especially grateful to Philip Zimbardo for his help and encouragement
during every stage of the project, and would also like to thank Albert Hastorf and
Daryl Bem for their advice and suggestions.
2.
Requests for reprints should be addressed to the author, who is now in
the Department of Psychiatry, University of Pittsbur^ Medical School, 3811
O'Hara Street, Pittsburgh, Pa.
15261.

586 Pilkonis
sion to a commonly used, but global term. Earlier inventories of
social anxiety (e.g., Dixon, de Monchaux,
&
Sandier, 1957; Watson
&
Friend, 1969) have been composed primarily of items from five
major categories—items designed to tap (a) intemal discomfort
in social situations (e.g., emotional upset and physiological
arousal); (b) fear of negadve evaluadon; (c) avoidance of social
situations; (d) failures to respond appropriately in social situa-
tions (e.g., a reluctance to talk, an avoidance of eye contact); and
(e) awkward behaviors arising from attempts to respond (e.g., an
inability to be fluent or articulate, physical clumsiness). The
first two categories focus on private events, while the last three
are concemed with public behaviors (or the lack of behavior).
Therefore, an effort was made to distinguish among various types
of shyness by assessing the relative importance of its private and
public aspects for different individuals. Shy respondents in the
present sample were asked to rank order the importance of these
aspects of the experience, and a cluster analysis was performed
on the rankings in order to create a typology of shy individuals.
A survey, then, was conducted to discover whether the earlier
questionnaire results were replicable, to examine the relationships
between self-reports of shyness and other measures of social
competence and anxiety, and to provide the data necessary for
the cluster analysis.
METHOD
')
Students {N = 263) enrolled in an introductory psychology class at
Stanford University were administered a short form of the Stanford
Shyness Survey (Pilkonis, 1976); the Self-Consciousness Scale, (Fenig-
stein, Scheier,
&
Buss, 1975); the Self-Monitoring Scale (Snyder, 1974);
and Form
A
of thaEysenck Personality Inventory (Eysenck
&
Eysenck,
1968).
The sample had a majority of males (57.6 percent) and was
composed primarily of students in their first two undergraduate years
(63.8 percent freshmen, 24.6 percent sophomores, 8.1 percent juniors,
and 3.5 percent seniors).
T^e Ej^senjck Personality Inventory (EPI) was scored for neuroti-
cism,
j^xtraversion,
and sociability, one of the two subscales of the extra-
version ficfor (Eaves & Eysenck, 1975). The Self-Monitoring Scale,
which measures sensitivity to cues of social appropriateness and the
ability to control one's social behavior, yields a single score. In Snyder's
use of the term, self-monitoring is a social skill (characteristic of actors,
for example) that shy people should be less adept at exercising.

Shyness and social behavior
587
The Self-Consciousness Scale (SCS) contains three subscales: (a)
public self-consciousness, wbich measures concern about oneself as a
social object; (b) private self-consciousness, which measures sensitivity
to inner thoughts and feelings; and (c) social anxiety, which measures
distress m the presence of others. A score for total self-corisciousness
was computed i5y Summing ttercores for public and private self-con-
sciousness.
'" The Stanford Shyness Survey included a number of different items
designed to elicit a thorough self-report about a respondent's shyness.
Among them were: (a) a yes-no, dispositional choice about one's shy-
ness ("Do you presendy consider yourself to be a shy person? In
general, is shyness an important part of your personality?"); (b) a 7-
point rating of the degree of one's shyness; (c) a 7-point rating of the
variability of one's shyness across situations, regardless of its general
level; and (d) a 5-point comparison to one's peers.
Also included was a 7-point rating of introversion-extroversion,
where these terms were defined on the basis of one's usual focus of
attention, regardless of the anxiety or behavior associated with tbis
perspective. An introvert was defined as "one whose thoughts and
interests are primarily directed inward," while an extrovert was speci-
fied as "one primarily interested in others or in the environment." In
addidon, all individuals who had labeled themselves dispositionally
^y were asked to assess how much of a problem their shyness was (on
a 7-point scale), to indicate their wlL^gness to seek help to overcome
their social anxiety, and to rank order the importance of the five as-
pects described earlier.
A hierarchical cluster analysis (Johnson, 1967,1968) was performed
on tbe ranks which individuals assigned to the five aspects of shyness
included on the questionnaire. The product-moment correlation be-
tween the rankings of any two individuals provided a measure of simi-
larity between those individuals, and the similarity coefficients
consdtuted the data for the analysis. Johnson's "diameter" or "maxi-
mum" method was employed, a technique designed to produce maxi-
mally compact clusters.* One hundred shy respondents (from a total of
107) provided complete data on the rankings, and these were the
subjects included in the analysis.
RESULTS
AND DISCUSSION :
Two basic analyses were performed. Firsti two-way, un-
weighted means analyses of variance were carried out on each of
3,
Special thanks are due Daryl Bem, whose modified version of the HIGLUS
program was used to perform the present analysis.

588 Pilkonis
the measures, with shyness (yes-no) and sex of subject as the
independent variables. Second, in order to examine the reladon-
ships among the measures, intercorrelations were computed (a)
for the entire sample, and (b) for males and females separately.
Two findings from the overall sample are especially important.
First, the percentage of self-reported shy persons was impres-
sively large and closely replicated earher results;
41
percent of the
undergraduates surveyed called themselves "shy persons." Second,
shyness, in large part, was judged to be situadon specific. In-
formants were asked to rate how much they varied in their shy-
ness from situation to situation, regardless of the absolute level
of their shyness. The mean rating of variability (M = 4.39, SD
= 1.51, on a 7-point scale) fell between "vary moderately" and
"vary to a large degree," and this latter category was the inedal
response. The two-way analysis of variance revealed thaf sliy
respondents were more variable in behavior than their nopSiy
counterparts, P(^I, 255)=36.74, p <
.001,
highlighting the situa-'
tional nature
6F
social anxiety, even in those vi^ho are willing to
apply a
farait-Bke
label to themselvesfi In order to justify diat
labelpsliy persons may l^Tthat their shyness still occurs in more
situadons than is true for others, or that it occurs more intensely
in those contexts where it is a problem. It is easier, after all, to
imagine being outgoing and relatively relaxed across a wide
range of situadons than it is to imagine being shy and chronically
anxious in all of one's encounters. Anxiety, by its nature, would
seem to be a less consistent occurrence.
Analyses of Variance
The Sex X Shyness analyses of variance made it clear that
simple, yes-no self-reports possess a high degree of validity. All
of the expected shy versus not shy differences were apparent.
Individuals who labeled themselves as "shy persons" rated them-
selves as more shy in general and more shy in comparison to their
peers.
They were less extroverted (on both the 7-point self-report
and the Eysenck scale), less sociable, less capable of monitoring
their social behavior, and more socially anxioush In addition, shy
respondents were significantly more neurodc as measured by the
EPI,
reinforcing the contention that self-ascribed shyness is a
valid indicator of poorer personal adjustment, at least among a

Shyness and social behavior 589
late adolescent sample. All of these differences were significant
beyond the .001 level.
Although an earlier survey had found no sex difference in the
prevalence of shyness, a somewhat unexpected difference did
exist in the present sample. A higher percentage of males re-
ported themselves to be shy (46.4 percent vs. 33.0jpe«?fent for fe-
males;
z = 2.23, p < .03). It may be the case"fiiat women who
successfully compete for places at a srfecdve, private university
are relatively more asserdve than their male peers. Other sex dif-
ferences appeared on (a) die 7-point, introversion-extroversion
rating, where females described themselves as more outwardly
oriented dian males, F(l, 254) = 7.02, p <
.01,
and (b) the
Self-
Monitoring Scale, where males scored significandy higher, indi-
cating that they were more adept at controlling their social be-
havior, F(l, 233) = 15.27, p < .001.
No differences of any kind were found for private jelf-con-
sciousness, but the public self-consciousness scores ppaduced both
a Sex X Shyness interaction, F(l, 259) =
4.93-,
p < .03, and a
main effect for shyness, shy > not shy, F^lfi^9) = 4.08, p < .05.
The interaction occurred because tljeliiale groups lay at the ex-
tremes (shy males were most j>^-conscious and not shy males
were least self-conscious^, while the two female groups were
idendcal and fell in tfe middle. The pattem suggests that public
self-consciousijesS'does not play as large a role in the development
of social aa^ty among women. Perhaps women are more uni-
forinly^ociahzed to be publicly self-conscious, and for that reason,
tifieir public self-awareness is a less hkely predictor of shyness.
Results from Dispositionally Shy Respondents
Participants who called theinselyes "jshy pfirsons" were asked^
seriesiSr quesdons about the difficulties that their shyness pre-
sented and about its most important aspects. The consensus (see
Table 1) was that shyness posed a moderate problem, and as a
result, a quarter (24 percent) of the respondents said that they
would be willing to seek help to overcome their social anxiety.
Table 1 also includes the mean rankings of the importance of
various aspects of shyness. Comparisons between means (on a
within-subjects basis) were performed for all possible pairs of
means. This analysis revealed that (a) among the first four items,
"failures to respond" was ranked significantly more highly than

3.57*
3.34"'
3.54"
3.06'
1.89°
1.20
1,24
1.44
1,28
1.26
590 Pilkonis
Table 1. Results from disposidonaUy shy persons.
Extent of Problem
"How much of a problem is your shyness?" M = 3.08*
SD = 1,11
Need for intervention
"Would you be willing to go to a shyness clinic if one were set up
to counsel shy students?"
Yes = 24% No = 18% Maybe = 58%
Overall rankings of aspects of shyness
M" SD
Failures to respond
Internal discomfort
Fear of negative evaluation
Awkward behavior
Avoidance
No>e.—^Total N = 107. There were no sex differences on any of these measures. Mean
rankings not bearing a common superscript differ at the .05 level.
' 3 := "somewhat of a problem," on a 7-polnt scale with a possible range of 1 "not at
all a problem" to 7 "extreme problem."
" Range = 1 "least important" to 5 "most important."
"fear of negadve evaluation" and "awkward behavior," and (b)
each of the first four items was seen as more important than the
fifth, "avoidance of social situadons." The latter result probably
occurred for two reasons. First, a college campus, with its variety
of group settings (dormitories, dining halls, classrooms) makes it
virtually impossible to avoid interacting with others, and second,
avoidance may well be the most extreme reacdon to social anxiety
of the five aspects sampled. Since the present respondents were
not, in general, "pathologically" shy, they may not have reached
the point of actively avoiding others.
Intercorrelations
The pattem of intercorreladons (see Table 2) was consistent
with the shy-not shy differences reported earher, but the relation-
ships tended to be stronger with males than females. Across both
sexes,
die 7-point sh)Tiess rating was posidvely correlated with the
independent scale of social anxiety (r = .67), while s|gificant
negadve correladons were found between the shyness s^lf-report
and the mehswes gi p^jgwxsXSi (intrpv^rsion-extroversion, r
•= -.38; Eysenckjextoaversion, f
^^.43;
sociability, r = -.39).
However, only in males did several of the other expected reladon-
ships obtain. In men, the shyness scale was cQi;rielat£id positwely

591
§??
SS
• • •
^^
I"
XX
I'XX
'^"^
''XX
I'J-X
582
ta=
2::2
SS!^
§8
•>-i
I'XX I'XX
•~='^
I'X
|
.H.2
SSS
*sk|
8
58
vv
a a
SI
2
I
S Efes
^ ^
>.^
•
fes
ss
i
g
gg
••§1
Is
o o
uu
J
o
I
II
.2
1-
IO
I
CL
2
0
II
l/>
I

592 Pilkonis
Table
3.
Mean ratings
of
five aspects
of
shyness within
clusters.
Aspects of
shyness
Internal
discomfort
Fear of
negative
evaluation
Avoidance of
social
situations
Failures to
respond
Awkward
behavior
1
Avoidance and
failures
(N = 7)
(4)"
(5)
0)
(2)
(3)
2.14"
1.43
4,57
4.43
2.43
2
Performance
deficits
(N = 37)
(3)
(4)
(5)
(2)
(1)
3.30
2.16
1.30
4,03
4.22
Cluster
3
Subjective
discomfort
(N = 27)
(1)
(2)
(5)
(3)
(4)
4.30
4,07
1.52
3.26
1.85
4
Fear
behaviora
(N =
(4)
(1)
(5)
(2)
(3)
and
1 deficits
22)
2.36
4.82
1.55
3.45
2.82
* Rank orderings of the means are given in parentheses.
" Range = 1 "least important" to 5 "most important."
with public self-consciousness (r = .27) and neurodcism (r =
.39),
and negatively with self-monitoring (r = -.25). Similarly,
only in men was public self-consciousness associated with social
anxiety (r = .26).
Cluster Analysis
Four clusters, capturing
93
of the 100 subjects, were iaterpret-
able within the presetot soludon. The Ns for each cluster and the
mean ratings of each of the five aspects of shyness within the
clusters are included in Table 3.
The first cluster was the smallest {N = 7), but it was quite
homogeneous (mean intracluster similarity correladon = .77)
and consisted of those people who avoided social situadons and
failed to respond appropriately while in them. As mentioned
earlier, avoidance is probably the most extreme of the aspects
ranked, and the size of the cluster suggests that only a small per-
centage of the present respondents were attempting to eliminate
all interaction with others.
The second cluster was the largest {N = 37, mean intra-
cluster correladon = .62) and was composed of people who
emphasized performance deficits (both awkward behavior and
failures to respond) in describing their shyness. Members of this
cluster focused on pubhc behavior in labeling themselves shy.

Shyness and social behavior 593
At the other extreme, subjecdve discomfort (intemal arousal
and fear of negative evaluadon) was the major complaint of indi-
viduals in the third cluster {N =
27,
mean intracluster correlation
= .64). These persons focused on the private aspects of shyness.
Members of the fourth cluster (N = 22, mean intracluster
correladon = .61) were a pardcular mix of the latter two types,
pointing; to both fear of nejcative evaluadon and behavioral
deficits in calling themselves shy.
The third and fourth clusters were, however, more similar
than the second and fourth, and when combined, they formed a
larger group that was still reasonably homogeneous (mean intra-
cluster correlation = .50). By disregarding the first cluster
(which seems least relevant for the present populadon), the solu-
tion can therefore be reduced to two primary types: persons who
are publicly shy (cluster 2) and focus on behavioral deficits, and
persons who are privately shy (clusters 3 and 4) and focus on
intemal arousal and anxiety. There were no sex differences in
frequency of membership in specific clusters, or in the larger
pubHc-private clusters.
This contrast between public and private, between behavior
and phenomenology, is a disdncdon that seems justified by the
present cluster analysis and makes good intuidve sense. But is
there any addidonal evidence for its validity? Yes, there is; as
one would expect, members of the private cluster were more
self-
conscious than their outwardly oriented peers, *(84) = 3.07, p <
.003,
on public self-consciousness, and *(84) = 2.40, p < .02, on
total self-consciousness.
A
second difference between the public and private clusters is
also important. Although publicly and privately shy people rated
the extent of their shyness as similar, individuals in the public
cluster tended to say that their shyness was more of a "problem,"
t{83) = 1.73, p < .09. This is somewhat surprising, since the
phenomenology of subjects in the private cluster seems to be more
aversive. They complained of greater self-consciousness, and their
shyness would appear to be more affectively laden as a result.
However, it was pubhcly shy persons who reported that they had
more difficulty coping with social anxiety. Behavioral deficits,
rather than intemal experience, carried more weight in decisions
about how much of a problem shyness poses.

594 Pilkonis
CONCLUSIONS AND IMPLICATIONS
Results from the present sample both confirmed and extended
earlier findings. Once again, the prevalence of self-reported shy-
ness was substandal, and the vahdity of these simple, direct
self-
reports was strengthened by their reladonships with other inde-
pendendy validated and more complex measures of extroversion,
neurodcism, self-monitoring, self-consciousness, and social anxi-
ety. For some assessment purposes, then, yes-no self-reports and
single, 7-point ratings may provide an adequate and economical
method of identifying populadons of interest.
Addidonal research should supplement such self-reports by
examining behavioral differences between shy and not shy indi-
viduals. The reported variability of shyness across situations, and
differences between males and females on some of the self-report
measures suggest that even if rehable behavioral differences do
exist between shy and not shy persons, one should not expect
them to appear equally in all settings and idendcaUy in both sexes.
It was assumed that all shy people experience both pubhc and
private components of shyness. However, it seemed likely that
different types of shy people would vary in the significance they
attached to privately experienced anxiety or to publicly mani-
fested behavior, and this a priori distincdon did receive substan-
dal support from the cluster analysis. This dichotomy should be
of potendal use in (a) explaining some of the behavioral and
affecdve differences among various sorts of shy people, (b)
designing treatment techniques sensidve to the need to change
both the anxiety and sldlls deficits components of shyness, and
(c) interpreting some of the common discrepancies which occur
between self-perceptions and the perceptions of others. Previous
quesdonnaire and interview work (Zimbardo et al., 1974) demon-
strated that sizable groups of respondents described themselves as
"shy, but extroverted," or said, "I'm shy, but my acquaintances
don't recognize it." These respondents would seem to be people
who are outgoing, but whose public demeanor does not express
something they feel privately is important about themselves, per-
haps increasing their sense of isoladon as a result. Efforts to
change shyness should be sensitive to both phenomenology and
behavior, attempting not just to enhance social skills, but attend-
ing also to the role of affect and arousal and the evaluation and
labeling of intemal experience.

Shyness
and
social
behavior
595
SUMMARY
A survey of collage undergraduates revealed that, consistent
with past results, a substandal propordon (41 percent) of such
students called themselves "shy persons." The vahdity of these
self-reports was supported by dieir reladonships with other mea-
sures of extroversion, neurodcism, self-monitoring, self-conscious-
ness,
and social anxiety. Shy respondents reported their shyness
to be a moderate problem, and 24 percent of them said they
would be willing to seek some help to overcome their social
anxiety. A cluster analysis performed on rank orderings of the
importance of various aspects of shyness suggested diat shy
individuals can be usefully distinguished by the degree to which
they emphasize publicly manifested behavior or privately ex-
perienced anxiety in describing their shyness.
BEFEBENCES
Borkovec,
T. D., Stone, N. M., O'Brien, G. T., & Kaloupek, D. G. Identification
and measurement of a clinically relevant target behavior for analogue outcome
research. Behavior Therapy, 1974, 5, 503-513.
Bryant, B. M., & Trower, P. E. Social difficulty in a student sample. British
Joumal of Educational Psychology, 1974, 44,
13-21.
Dixon, J. J., de Monchaux, C., & Sandier, J. Pattems of anxiety: An analysis of
social anxieties. British Joumal of Medical Psychohgy, 1957, 30, 107-112.
Eaves,
L., & Eysenck, H. The nature of extraversion: A genetical analysis.
Joumal of
Persoruility
and Social Psychohgy, 1975, 32, 102-112.
Eysenck, H. J.,
&
Eysenck, S. B. G. Manual for the Eysenck personality inventory.
- Stm Diego: Educational and Industrial Testing Service, 1968.
Fenigstein, A., Scheier, M. F., & Buss, A, H. Public and private self-conscious-
ness:
Assessment and theory. Joumal of Consulting and Clinical Psychohgy,
1975,
43, 522-527.
Johnson, S. C. Hierarchical clustering schemes. Psychometrika, 1967, 32, 241-
254.
Johnson, S. C. How to use HICLUS: A hierarchical cluster aruilysis program.
Unpublished manuscript. Bell Laboratories, 1968. (Revised by and available
from Stanford Computation Center, Stanford, California.)
Martinson, W. D., & Zerface, J. P. Comparison of individual counseling and a
social program with non-daters. Joumal of Counseling Psychohgy, 1970, 17,
36-40.
fjJJgpij^P.
A. Shyness: Public behavior and private experience. Unpublished
doctoral dissertation, Stanford University, 1976.
Snyder, M. Self-monitoring of expressive behavior. Journal of Persoruility and
Social Psychohgy, 1974, 30, 526-537.
Watson, D., & Friend, R. Measurement of social-evaluative anxiety. Jourruil of
Consulting and Clinical Psychohgy, 1969, 33, 448-457.
Zimbardo,
P. G., Pilkonis, P. A., & Norwood, R. M. The silent prison of shyness
(ONR Tech. Rep. Z-17). Stanford: Stanford University, November 1974.
Manuscript received January 4, 1977.