Journal of Social and Clinical Vol. 9, No. 4, 1990, pp.

439-455
Psychology,

SELF-CONCEALMENT:
CONCEPTUALIZATION, MEASUREMENT,
AND HEALTH IMPLICATIONS
DALE C. LARSON
Santa Clara University
ROBERT L. CHASTAIN
Stanford University

This article introduces the construct of self-concealment, the active concealment

from others of personal information that one perceives as negative or distressing. A
Self-Concealment Scale (SCS) was developed and was included in a questionnaire
battery completed by 306 subjects. The SCS had excellent psychometric properties.
Self-concealment was conceptually and empirically distinguished from self-disclosure.
Self-concealment significantly correlated with self-report measures of anxiety,
depression, and bodily symptoms and accounted for a significant incremental per
centage of the variance in physical and psychological symptoms even after controlling
for occurrence of trauma, trauma distress, disclosure of the trauma, social support,
social network, and self-disclosure. The implications of these findings are discussed
and directions for further research are briefly outlined.

Self-concealment is a familiar human

experience. Most people have un
comfortable feelings, thoughts, and information about themselves that
they avoid telling others. These secrets can range from mildly embarrassing
to highly distressing. Sometimes these secrets have been told to only
one or two persons and sometimes to no one at all. Clinical practice and

research, as well as casual observation, indicate that some individuals

tend to self-conceal more than others do and that the most painful or
traumatic experiences are often concealed; examples are sexual abuse as
a child (Russell, 1986; Stark, 1984), rape (Binder, 1981; Burgess & Holm-
strom, 1974), grief (Evans, 1976), family secrets (Karpel, 1980; Saffer,

This work was supported by a Presidential Research Grant from Santa Clara University.
The authors would like to thank Allen Kanner, Edward McQuarrie, and Karen Trocki for
their comments on earlier versions of the manuscript and suggestions for data analyses,
for her assistance with the research. Correspondence and requests for
and Jody Kaplan
reprints should be sent Larson, Division of Counseling Psychology and Education,
to Dr.

Santa Clara University, 220 Bannan Hall, Santa Clara, CA 95053.

439
440 LARSON AND CHASTAIN

Sansone, & Gentry, 1979), strong negative thoughts about oneself or

unhappiness in one's relationships, and serious medical conditions, in

cluding acquired immune deficiency syndrome (AIDS).
The aims of this article are threefold: first, to present a definition
of self-concealment and to describe the psychometric properties of a self-
concealment scale; second, to determine empirically whether self-con
cealment and self-disclosure are two separate and distinct constructs;
and third, to examine the predictive validity of self-concealment in relation
to psychological and physical health.

CONCEPTUALIZATION OF SELF-CONCEALMENT

Self-concealment is defined here predisposition to actively conceal

as a

from others personal information that

one
perceives as distressing or
of seen here as a
negative. The process self-concealment, special instance
of boundary regulation in the maintenance of privacy (Derlega & Chaikin,
1977), can also be viewed within a model of self-presentation or image
management in which social interactions present opportunities to disclose
distressing or negative personal information (Schlenker, 1980). This self-
concealed personal information is (a) a subset of private personal infor
mation, (b) consciously accessible to the individual, and (c) actively kept
from the awareness of others. If disclosed at all, it is usually confided
to only a small number of persons. Thus, self-concealment involves the
conscious concealment of personal information (thoughts, feelings, actions,
or events) that is
highly intimate and negative in valence.

BACKGROUND

Work in the self-disclosure area and

investigations of the link between
confiding and health status
provide important background for the current
study and suggest that the systematic study of self-concealment can
extend these lines of inquiry in valuable directions.
The relation between self-disclosure, the act of revealing personal
information to others (Archer, 1980), and self-concealment, the act of
information from others, is an important conceptual
concealing personal
and research issue. One possible relation is that these two constructs
are
simply the reverse of each other: the self-concealing individual is
not disclosing, and the low-disclosure individual is
self-concealing.
However, the argument made here is that self-concealment and self-
disclosure are two separate and distinct,
though related, constructs.
SELF-CONCEALMENT 441

SELF-DISCLOSURE RESEARCH AND THEORY

A review of the self-disclosure literature reveals that, although self-

concealment has not conceptually differentiated from self-disclosure,
been
several authors have
pointed to the significance of what may be identified
as self-concealment
phenomena. For example, Jourard (1959, 1971a, 1971b)
emphasized the negative health consequences of hiding or actively con
cealing significant aspects of the self as well as the positive health con
sequences of disclosing important self-information (Goodstein & Reinecker,
1974). However, subsequent research has not directly studied the former
dimension. In his review of the self-disclosure literature, Cozby (1973)
pointed out that self-disclosure research has focused almost exclusively
on the factors
promoting disclosure and encouraged study of the factors
inhibiting disclosure.
Other investigators have proposed a refocusing of self-disclosure
research toward the study of how one manages one's most private self-
information (Fisher, 1984; Goodstein & Reinecker, 1974). Fisher (1984)
argued that the expanded class of behaviors included within the general
rubric of self-disclosure has prevented an adequate testing of Jourard' s
disclosure/health hypothesis and has contributed to the equivocal nature
of the findings relating self-disclosure to various indices of health status
(Blotcky, Carscaddon, & Grandmaison, 1983; Chelune, 1979).
The study of self-concealment is consonant with the foregoing ob
servations and recommendations it includes an explicit focus on the
forces inhibiting disclosure and a concern with negative self-information
of the most intimate and private nature. In these ways, the study of
self-concealment builds upon conceptual developments within the self-
disclosure research tradition.

SELF-CONCEALMENT AND HEALTH STATUS

The implications of self-concealment for health status have long been

noted by clinical practitioners. Ellenberger (1970) has traced the historical
importance of the concept of the "burdening" or "pathogenic" secret in
the development of dynamic psychotherapy. For the psychoanalytically
oriented therapist, the client's conscious secrets from the therapist rep
resent a form of "deliberate, overt resistance" (Ekstein & Caruth, 1975,

p. 202) that must be overcome. A focus on clients' most intimate and

disturbing experiences and their difficulties in revealing them is not
to analytically oriented clinicians. Indeed, the therapist's role as
unique
confidant someone to whom clients can disclose their most private
442 LARSON AND CHASTAIN

thoughts, feelings, and behaviors is a hallmark of the therapeutic re

lationship (Towbin, 1978).

Work in the social support area has demonstrated that having a
confidant, someone to whom one disclose has
can
fully, a salutary impact.
In several studies the availability of a confidant predicted health status
and coping outcomes following stressful life events (Brown, Bhrolchain,
& Harris, 1975; Lowenthal & Haven, 1968; Miller & Ingham, 1976).
The health significance of self-concealment has been even more

explicitly underlined by recent evidence that not confiding traumatic

events has undesirable long-term health consequences. In a series of
studies Pennebaker and his colleagues have examined what they call
the confiding- illness relation or the inhibition- disease link (Pennebaker
& Chew, 1985) and have found that not expressing thoughts and feelings
about traumatic events (divorce of parents, death of spouse, death of
parent, and sexual traumas) leads to long-term health effects, even when
social support levels are controlled (Pennebaker & O'Heeron, 1984). In
a recent study, subjects writing about traumatic experiences showed
improved cellular immune system functioning (Pennebaker, Kiecolt-Glaser,
& Glaser, 1988).
Pennebaker concludes that "the act of not discussing or
confiding
the event with another may be more damaging than having experienced
the event per se" (Pennebaker, 1985, p. 82) and theorizes that the mech
anism accounting for these effects is increased physiological work resulting
from the behavioral inhibition accompanying not confiding the traumatic
event (Pennebaker & O'Heeron, 1984). The inabilityor
unwillingness to
discuss major upheavals with others is attributed to either circumstances
or individual differences & O'Heeron, 1984). One purpose
(Pennebaker
of the present study is to introduce a measure of the general
tendency
to self-conceal that would permit more direct assessment of this issue.
Another purpose is to broaden the scope of self-concealed
personal
information to include not only traumatic events but any events or ex

periences the revelation of which is perceived to be

potentially threatening
or harmful.
The above presentation leads to three main research ad
questions
dressed in this study.
1. Are there individual differences in the
tendency to conceal personal
information, and they be reliably and validly measured using
can

a
self-report instrument? The psychometric properties of a scale
designed to measure this new construct will be tested.
2. Is the self-concealment construct
separate and distinct from self-
disclosure? When examined in relation to other variables, how
SELF-CONCEALMENT 443

similar or dissimilar are the correlational

profiles of self-concealment
and self-disclosure?
3. Does the self-concealment variable
predict psychological and
physical health status? To be a useful addition to theory, self-
concealment should contribute uniquely and significantly to the
prediction of psychological and physical health.

METHOD

SUBJECTS

Three methods were used to gather subjects. First, questionnaires were

mailed to 248 persons on a mailing list maintained for the Health Psy
chology Program at the first author's university. This list consisted primarily
of human services workers (e.g. , nurses,physical therapists, social work
ers, clergy, and volunteers in social service agencies). Second, ques
tionnaires were distributed to 366 persons attending professional training
conferences at which the first author delivered an address. Conference
attendees were asked whetherthey would like to volunteer to participate
in a study of health and self-disclosure. One group of subjects at these
conferences was immediately given copies of the research questionnaire
with a return envelope; other subjects were asked for their names and
addresses and were later mailed questionnaires. Finally, 225 questionnaires
were distributed to
graduate counseling psychology students (not currently
in any of the author's courses) volunteering to participate in the study.
Because of the highly intimate nature of many of the questions asked,
extreme care was taken to protect the confidentiality of respondents. All

subjects were told not to put their names anywhere on the questionnaire
and were given a postcard to send separately for a report of the findings
of the study.
A total of 306 questionnaires was returned. Of the respondents, 277
were female and 29 were male, reflecting the small percentage of males

in the populations sampled. Response rates for the three samples were,

respectively, 29% (n 73), 41% (n 151), and 36% (n 82), giving an

= = =

overall response rate of 36%. The response rates were not significantly
(p < .05) different among the three samples. These response rates could
indicate a self-selection problem due to nonrespondent bias, but they
the highly personal and time-demanding nature
appear acceptable given
of the questionnaire. Although nonrespondent bias could not be estimated
in these samples, the respondents represent a subsample of primary
444 LARSON AND CHASTAIN

human services workers to this kind ot

interest willing to respond
questionnaire.
to
analysis of variance (MANOVA) was performed
A multivariate
test the comparability of the three samples and males and females
on

the nine major variables of the study. These variables represented health,
A
social support, social networks, self-concealment, and self-disclosure.
two-factor MANOVA (sex by group) revealed the following: (a) sex, F(9,
223) 1.60, p > .10; (b) group, F(18, 446)
=
0.70, p > .80; (c) sex-by-
=

0.67, p > .80. Thus, there is no evidence

group interaction, F(18, 446)
=

of group differences in the

sex or total sample (N 306). =

characteristics of the final sample included an average

Demographic
age of 42 years (range, 21 to 79;
SD 10.9); 82% with college education
=

and 60% with education beyond college; 61% married, 16% divorced,
15% single, 4% separated, and 4% widowed; 43% Protestant, 28% Catholic,
7% other Christian 10% no preference, and 12% other. Eighty-
religion,
seven percent of the subjects indicated that they were currently working
in the human services.

MEASURES

Self-Concealment making up the Self-Concealment

Scale. The 10 items
Scale used in the data
analyses reported below are presented in
(SCS)
Table 1 Scale items refer to (a) a self-reported tendency to keep things
.

to oneself (e.g., "There are lots of things about me that I keep to

myself"); (b) possession of a personally distressing secret or negative

thoughts about oneself that have been shared with few or no other
about myself that I never share
persons (e.g., "I have negative thoughts
with anyone"); and (c) apprehension about the disclosure of concealed

personal information (e.g., "If I shared all my secrets with my friends,

they'd like me less").

The internal consistency estimate of Cronbach's alpha showed a =

.83 (N 306). Test-retest reliability was assessed in an independent

=

sample female graduate counseling psychology students (n 43) with

of =

a 4-week interval between testing and revealed r .81. =

Self-Disclosure Index. Subjects also completed the Self-Disclosure Index

(SDI), an 11-item scale developed by Miller and her associates (Miller,

Berg, & Archer, 1983). The version used here, which specifies willingness
to disclose to a same-sex
stranger in the future, can be considered an

"expectation" measure and thus better

predictor
a of future disclosing
behavior than a
"history" Subjects indicated extent of disclosure
measure.

on a scale from 0 (discuss not at all) to 4 (discuss

fully and completely).
The SDI had an internal consistency of a .91 (N 306).
= =
SELF-CONCEALMENT 445

TABLE 1
Means and Standard Deviations of Individual SCS Items

SCS Items M SD

1. I have that I haven't

an
important secret
shared with anyone. 2.73 1.42
2. If I shared all
my secrets with my friends,
they'd like me less. 2.44 1.10
3. There lots of about that I
are
things me
keep
to myself. 3.00 1.15
4. Some of
my secrets have really tormented me. 2.85 1.31
5. When something bad happens to me, I tend to
keep it tomyself. 2.49 1.09
6. I'm often afraid I'll reveal I don't
something
want to. 2.33 1.01
7. Telling a secret often backfires and I wish I
hadn't told it. 2.59 .91
8. I have a secret that is so
private I would lie if
anybody asked me about it. 2.40 1.32
9. My secrets too to share with
are
embarrassing
others. 2.42 1.03
10. I have negative thoughts about myself that I
never share with anyone. 2.71 1.21
Total scale 25.92 7.30

Note. The scale uses a

5-point Likert scale ranging from 1 to 5
(strongly disagree)
(strongly agree).

Social Support and Social Network Measures. Measures of self-reported

social support and social network strength were included. Subjects were
asked to think of the several kinds of social support they might receive
from different people in their lives (spouse or partner, friends, closest
relatives, neighbors, and work associates) and were asked to rate each
of these persons in terms of how much support they received from them.
The exact instructions were as follows: "The support we get from others
can take many forms. It includes people giving us information and
guidance, being there when we need them, boosting our spirits, making
us feel that they for us, and helping out with small favors like
care

errand
doing household chores. Using these examples of
or
running an

as a guideline, rate each of the following person(s) in terms of

support
how much support you feel you receive from them." Two variables
Social Support from Spouse/Partner and Social Support from Others
(friends, closest relatives, neighbors, and work associates combined)
were generated from these ratings. To obtain measures of social network

strength for friends and relatives, subjects were asked to indicate separately
446 LARSON AND CHASTAIN

number of close ("feel at ease with, can talk to about matters,

private
and can call help")
on for friends and close relatives.1
Physical Symptom Checklist. A 39-item physical symptom checklist,
which shows adequate reliability and validity (Cohen & Hoberman, 1983;
Cohen, Kamarck, & Mermelstein, 1983), asked subjects to indicate the
extent to which they had been bothered by 39 commonly occurring
symptoms (e.g., back pain, headache, blurred vision, muscle soreness)
during the immediately preceding 4-week period. This checklist uses a
5-point scale ranging from "Not at all" to "Extremely" and had an internal
consistency of a .86 (N 306).
= =

Anxiety and Depression Scales. Two measures of psychological distress

taken from the Typology of Psychic Distress instrument (PSYDIS; Mel-

linger, Baiter, Manheimer, Cisin, & Parry, 1978) were included: a Mood
Depression scale (4 items) and a Mood Anxiety scale (6 items). The items
were
originally taken, for the most part, from an early version of the
Hopkins Symptom Checklist (HSCL) and have been used extensively in
research, demonstrating excellent psychometric properties (Mellinger et
al., 1983). This instrument was chosen because it asks about symptoms
experienced during the past year, consistent with the research goal of
measuring more long-standing health status. The two scales had good
internal reliability for scales with so few items (Mood Anxiety, alpha =

.78; Mood Depression, alpha .79). The measures used the following
=

response format: 1 Not Bothered at All, 2 Bothered Some but Not

= =

Much, and 3 Bothered

=
a Lot.

Life Events or Experiences. Another section of the questionnaire asked

about the occurrence of traumatic or
distressing life events
or
experi
ences death of parent(s) before age 17, sexual molestation as a child,
physical abuse as a child, rape experience as an adult, and 20 other
events or
experiences, including death of a child, abortion, divorce of
parents before age 17, affair/infidelity, alcoholic parent(s), alcohol or

drug use, and unhappiness or difficulties in one's marriage. For each of

the events or
experiences that applied to them, subjects were asked to
indicate (1) whether they had kept its occurrence secret from anyone,

(2) how many persons they had told about it ("No one," "1-3," or "4
or more"), and (3) how much it had distressed them during the past
year. A "Secret Total" variable
was
computed as the total number of life
events or
experiences
subject reported having kept secret. The total
a

number of life events or experiences a subject indicated not

having
disclosed to anyone was also computed.

1. The social network items were from research conducted

by the Human Population
Laboratory, California
Department of Health Service Research, and the Centers for Disease
Control, U.S. Department of Health and Human Services (Kaplan & Camacho, 1983). The
social support items were adapted from the Dimensions of Social
Support Scale developed
by Frances Cohen, University of California, San Francisco.
SELF-CONCEALMENT 447

RESULTS

PSYCHOMETRIC PROPERTIES OF THE SCS

An
exploratory maximum-likelihood factor analysis was performed on
the 10 SCS items to determine dimensionality. Although two factors
were extracted with two eigenvalues greater than 1 (6.57 and 1.38, re
spectively), the SCS was essentially unidimensional because (a) the first
factor accounted for over 65% of the common variance, (b) the second
factor was uninterpretable even after orthogonal and oblique rotations,
and (c) item loadings on the first factor ranged from .46 to .71. The
internal consistency of the SCS (a .83) also supported unidimensionality.
=

The moderate corrected item- total correlations indicated that the SCS
items were not redundant. In addition, the mean for the interitem cor
relations, which provides a relatively pure index of scale homogeneity,
was .34. Thus, the SCS
appears to be a reliable and essentially unidi
mensional instrument.

SELF-CONCEALMENT VERSUS SELF-DISCLOSURE

Is self-concealment well
conceptually distinct from self-
empirically as as

disclosure? If these two constructs cannot be clearly differentiated, then

self-concealment may merely be the reciprocal or "mirror image" of self-
disclosure. This important question was answered by a series of analyses.
The first step was to subject the 11 SDI items and the 10 SCS items to
a maximum-likelihood factor
analysis. The null hypothesis that one factor
was sufficient to account for the within-set variance was rejected in favor

of the alternate hypothesis that more factors were needed, x2 (189) =

1062.6, p < .0001. The two-factor solution showed that the 11 SDI items
had on the first factor (.50 to .78) and low
relatively high loadings loadings
on the second factor (.02 to -.22). The 10 SCS items had high loadings
on the second factor (.44 to .70) and low loadings on the first factor (.00
to .15). These results were essentially duplicated using a combined
cross-validation sample of 110 undergraduates and 60 graduate students
(126 females and 44 males).2 Although related to self-disclosure, the SCS
represents a new construct that is separate and distinct from self-disclosure.

2. Further evidence for the two-factor solution was provided by results from a series of

confirmatory factor analyses using LISREL. The two-factor model was superior to the one-
factor model (GFI .848 versus .630, adjusted GFI
=
.814, coefficient of determination
= =

.915, and RMS .092). The 10 SCS item loadings ranged from .41 to .95, and the 11 SDI
=

items ranged from .52 to .95 on their respective factors (all other loadings were set to 0).
These results were duplicated in two independent cross-validation samples. In the above
self-concealment latent construct and the self-
analyses, the correlations between the
latent construct ranged from .017 to -

.317 in the different sets of confirmatory

-

disclosure
analyses (which are close to the -.27 correlation in the total sample; see Table 2).
448 LARSON AND CHASTAIN

PREDICTIVE VALIDITY

The next question addressed was the "so what" begging often left
question
in much theoretical work. Even if this new
reliably we can measure

construct, the crucial test is whether self-concealment uniquely contributes

to predicting important health outcomes. What is the predictive validity
of the SCS in the presence of other competing explanatory variables,

especially self-disclosure?
The current research design permits assessment of the contribution
of self-concealment to health outcomes physical symptoms (as measured
by the 39-item physical symptom checklist) and psychological distress
(as measured by the anxiety and depression scales from the PSYDIS
measure).
The analyses of these relations included correlations, analysis of
variance (ANOVA), and hierarchical multiple regression. Results showing
the relations between self-concealment and the health measures are
pre
sented below.
Table 2 shows the correlational profiles for both the SCS and the
SDI. None of the correlations between the SDI and health outcomes was

TABLE 2
Correlational Profiles for Self-Concealment (SCS) and Self-Disclosure (SDI)
in Total Sample and in Cross- Validation Sample

TOTAL SAMPLE CROSS-VALIDATION

(N- 306)
=
SAMPLE (N 110) =

VARIABLE SCS SDI SCS SDI

Self-concealment (SCS) 1.00 nn****

1.00 -.23*
7Q*=t-*X-
Physical symptoms .00 .30** -.17

Depression 4j**** -.08 .32*** -.14

OJ**** A-l

Anxiety .02 ****

-.14
Social support, Others -.27*"* .01 -.17 .13
Social support, Spouse -.10 -.12* -.14 .00
Social network,
Friends -.33**** .22*** -.16 .22*
Social network,
Relatives .12* .06 -.05 .12
Secret total .14* .15* .20* -.00
Told no one 41 ****
-.07 .29** -.26**
*

p < .05.
"

p < .01.

*"p < .001.

***>< .0001.
SELF-CONCEALMENT 449

significantly different from zero, whereas the SCS

significantly relatedwas

to all three health outcomes. These results upheld in the cross-

were

validation sample of 110 undergraduates and are also reported in Table

2. Thus, self-concealment was more highly and significantly related to
all three health outcomes than was self-disclosure.
A different look at the relations between the SCS and the health
outcomes is provided by forming high- and low-SC groups using different
grouping definitions
(e.g., mean, median, and quartile splits). Results
from one-way ANOVAs also showed that the high-SC group had sig
nificantly more bodily symptoms, depression, and anxiety than the low-
SC group. A 2 x 2 ANOVA (high/low on SCS with high/low on SDI)

performed on physical symptoms showed only a significant main effect

for SCS grouping (F 9.59, p=
.002). Neither SDI grouping nor the
=

interaction between SCS and SDI groups was significant (F 0.93, p = =

.337, and F 2.21, p

=
.138, respectively).
=

A hierarchical regression procedure (Cohen & Cohen, 1975) was

used to test the incremental contribution of self-concealment to health
outcomes. In this analysis five alternative sets of predictor variables were
entered into the regression equation prior to the addition of self-con
cealment. For instance, it is reasonable to suppose that the mere fact of
trauma (e.g., molestation as a child) may lead to negative health outcomes
and that disclosure of this trauma and general levels of social support
could buffer these health effects. Only if self-concealment
uniquely predicts
health outcomes, after these alternative explanations, is
controlling for
it appropriate to conclude that self-concealment itself has a negative

impact on health. The five competing sets of predictor variables, in order

of entry, are (1) trauma incidence4 four variables (physically abused
as a child, sexually molested as a child, raped as an adult, and death of

parent(s) before age 17); (2) trauma distress during past year; (3) trauma
disclosure; (4) social support and social network four variables (social
support spouse/partner, social support others, social network friends,
social network relatives); and (5) SDI score (see Table 3).
With bodily symptoms as the dependent measure, trauma incidence
accounted for a significant proportion of the variance (R2 .060, p < =

.01). Trauma disclosure did not make a significant contribution, but

trauma distress (R2 .042, p < .05) added significantly to the prediction
=

of outcome. Self-disclosure, however, did not make a significant con

tribution. Thus, the five sets of alternative predictor variables were able

3. Further were not cross-validated because of insufficient sample sizes.

analyses
4. Number of traumas were as follows: sexually molested as a child,
subjects reporting
55; physically abused as a child, 25; rape experience as an adult, 26; death of parent(s)
before age 17, 22.
 OJ

OO O LD
Pi o o o
u
2

^JH t^ \Q r^
cn cn oo cn
o

H
w
-J
2
uj
CQ
<
w
2
U
2
H
2
w
D
2
w
Cl, ^ cj * oo ^ ^
W O
Q
CD

H
2
UJ

UJ c< mom
o o o
OS
U
2

vO OO ^ C7>
* ^ 00 fN

CD

O QJ
X
UJ
O OJ
TJ ra TJ
C "C c
re tn
~ > C1J

o ^ 3 E -a
.
a s -a -a L-* O
3
3 u
O
S
5
U-h uin "^
c o ffl in
'oj at O
t/i ai h

450
SELF-CONCEALMENT 451

to account for about 15% of the variance in

bodily symptoms (see Table
3), and the addition of self-concealment increased prediction still further
(R2 .036, p < .01).
=

The patterns of results for the depression and anxiety measures

were
quite similar. For each, only the social support/social network variables
significantly explained variance in the outcome variable (p =s .001), with
self-concealment making a significant additional contribution (p < .001).
For depression, the addition of self-concealment enabled a further im

provement to 26.5% of the variance (R2 increment =

.074, p < .001). For
anxiety, the addition of self-concealment improved prediction to 18.6%
of the variance (R2 increment
.052, p < .001).
=

To summarize, after controlling for trauma incidence, trauma distress,

trauma disclosure, social support and social network, and self-disclosure

levels, self-concealment accounted for significant increments in variance

explained for all three outcomes. Although the total variance accounted
for in thedependent variables may seem lower than desirable, the pre
dictivevalidity coefficients (total multiple Rs) ranged from .429 to .515.
In addition, these coefficients were quite acceptable given that the de

pendent measures were not corrected for attenuation due to unreliability.

Thus, although the existence of trauma has a negative impact on health
and although social support can ameliorate this impact, at least with
psychological symptoms, it remains the case that self-concealment has
a
uniquely negative impact on mental and physical health.

DISCUSSION

The SCS is a reliable measure of self-concealment. Self-concealment and

self-disclosure are distinct and separate constructs. Self-concealment is

significantly related to physical and psychological symptoms. The ex

pectation that self-concealment would contribute uniquely and significantly
and even after
to the prediction of psychological physical status accounting
for trauma level, trauma distress, trauma disclosure, social support and
social network variables, and general self-disclosure levels was confirmed.
Self-concealment contributes significantly and uniquely to increased
and physical symptoms, after controlling
depression, higher anxiety, more

for the other variables. The social support and social network variables
contributed significantly depression and anxiety, but self-concealment
to

made an additional and significant contribution. Thus, the effect of the

self-concealment variable on health outcomes is not a function of differing
levels of social support for the high- and low-self-concealment groups.
It is important to note, however, that the relation between self-
concealment and physical symptoms, when controlling for anxiety and
452 LARSON AND CHASTAIN

depression, is demonstrated in an interaction, rather than in a direct

effect.5 Subjects with high levels of trauma distress and high self-con
cealment reported significantly (p < .0001) more physical symptoms even
after controlling for anxiety and depression. The obtained relation between
self-concealment and physical symptoms must be interpreted quite cau
tiously for another reason. Recent work has shown that neuroticism or
negative affectivity is related to somatic complaints but not to disease
(Costa & McCrae, 1987; Smith, Pope, Rhodewalt, & Poulton, 1989); Watson
& Pennebaker, 1989), and further, that correlations between various
personality variables and health reports may actually reflect shared variance
with neuroticism. The present study employed a self -report physical
symptom measure and did not establish the discriminant validity of the
SCS relative to neuroticism. Thus, the obtained relation between self-
concealment and physical symptoms could reflect an association between
chronic distress and somatic complaints rather than a relation between
self-concealment and actual physical illness.
Although generalizability of these findings may be limited because
of the small percentage of males and possible self-selection, three points
should be noted: (a) cross-validation samples showed similar measurement
distributions and relations, (b) males and females did not differ on the
measures of interest in any of the samples, and (c) if respondents were

not representative because they are more willing to disclose personal

information or were more psychologically astute, the relations between
health outcomes and the psychological constructs would be more likely
to be attenuated rather than exaggerated (i.e., respondents will be less
variable with respect to these measures).
What are the mechanisms accounting for the relations obtained be
tween self-concealment and the health measures? One possible pathway
for the effects of self-concealment on health status is physiological work
(Pennebaker & O'Heeron, 1984) resulting from behavioral inhibition as

sociated with self-concealment. As discussed above, many writers have

concluded that self-concealment processes whether labeled as non
disclosure (Jourard), a failure to confide (Pennebaker), or readjustment
of the privacy regulation system (Margulis, 1977) exact a price and
function as internal stressors.
Self-concealment as construed and measured here includes a be
havioral inhibition component. The SCS items ask about the propensity
to keep things to oneself, about fear of revealing something one doesn't
want to, and about the readiness to lie if asked about concealed self-

information; these items tap inhibitory processes like those described

by Pennebaker. The SCS item "When something bad happens to me, I

5. Partial r =
.06, p =
.16
SELF-CONCEALMENT 453

tend to keep it to myself" directly assesses the tendency not to confide

traumatic experiences. Thus, one hypothesis is that health effects are
mediated by behavioral inhibition accompanying self-concealment.
A second pathway could be through the role of self-concealment as
a
personality variable mediating coping responses to distressing external
events or troubling inner experiences. In this view, inhibition of the
disclosure of troubling experiences restricts the range of available coping
responses, preventing more active, problem-focused coping responses
and leaving primarily emotion-focused responses (Lazarus & Folkman,
1984) as possible alternatives. In their study of coping and health, Folkman,
Lazarus, Dunkel-Schetter, DeLongis, and Gruen (1986) identified self-
control as a form of emotion-focused coping assessed by items such as
"I tried to keep my feelings to myself" and "Kept others from knowing
how bad things were." Subjects tended to use more self-control coping
when threat to self-esteem was high, consistent with self-concealment
theory. In a separate study, Folkman and Lazarus (1986) reported that
subjects with high levels of depressive symptoms used more self-control
coping strategies than did subjects low in symptoms.
A third and related hypothesis is that self-concealment affects health
status by limiting the range and frequency of helping behaviors offered

by significant others. By avoiding discussions of problems, particularly

those threatening to self-esteem, high self-concealers are deprived of
important kinds of social support. A possible resulting scenario is that
social comparison data are not received, difficulties are not normalized,
the stakes involved in confiding increase, and additional help is not
sought (Snyder & Ingram, 1983).
Further research is planned to study behavioral inhibition, coping,
and help seeking as pathways for the direct and indirect effects of self-
concealment on health over time in different populations. Using objective
measures of health status will address the limitations of self-report outcome

measures like those used in the present study. An additional goal of

this research is to clarify the relations between self-concealment, negative

affectivity (Watson & Pennebaker, 1989), emotional control (Watson &
Greer, 1983), self-control (Folkman et al., 1987), social desirability (Crowne
& Marlowe, 1960), and other relevant variables.
The investigation of self-concealment could encompass a wide range
of research endeavors. For example, research exploring the origins of
self-concealment could provide important insights for clinical theory and
treatment approaches. The role of self-concealment in the development
and dissolution of intimate relationships is another arena for subsequent
work.
construct that clinicians have
The present study operationalizes a

in the etiology and treatment of psychological

long viewed as significant
454 LARSON AND CHASTAIN

psychosomatic disorders. Although further work is needed for

a
and
fuller of the psychology of self -concealment, initial evidence
understanding
suggests that this new construct is an important addition to research on

personality and health.

REFERENCES

Archer, R. L. (1980). Self-disclosure. In D. M. Wegner & R. R. Vallacher (Eds.), The self in

New York: Oxford University Press.
social psychology (pp. 183-205).
Binder, R. (1981). Why women don't report sexual assault. Journal of Clinical Psychiatry,
42, 437-438.
and
Blotcky, A. D., Carscaddon, D. M., & Grandmaison, S. L. (1983). Self-disclosure
physical health: In support of curvilinearity. Psychological Reports, 53, 903-906.
Brown, G. W., Bhrolchain, M. H., & Harris, T. (1975). Social class and psychiatric disturbance
225-254.
among women in an urban population. Sociology, 9,
Burgess, A. W., & Holmstrom, L. L. (1974). Rape trauma syndrome. American Journal of
Psychiatry, 131, 981-986.
Chelune, G. J. (Ed.). (1979). Self-disclosure: Origins, patterns, and implications of openness in
interpersonal relationships. San Francisco: Jossey-Bass.
Cohen, J., & Cohen, P. (1975). Applied multiple regression/correlation analysis for the behavioral
sciences. Hillsdale, NJ: Erlbaum.
Cohen, S., & Hoberman, H. M. (1983). Positive events and social support as buffers of

lifechange Journal
stress. of Applied Social Psychology, 13, 99-125.
Cohen, S., Kamarck, T, & Mermelstein, R. (1983). A global measure of perceived stress.
Journal of Health and Social Behavior, 24, 385-396.
Costa, P. T, Jr., & McCrae, R. R. (1987). Neuroticism, somatic complaints, and disease:
Is the bark worse than the bite? Journal of Personality, 55, 299-316.

Cozby, P. (1973). Self-disclosure: A literature review. Psychological Bulletin, 79, 73-91.

Crowne, D. P., & Marlowe, D. (1960). A new scale of social desirability independent of
psychopathology. Journal of Consulting Psychology, 24, 349-354.
Derlega, V. J., & Chaikin, A. L. (1977). Privacy and self-disclosure in social relationships.
journal of Social Issues, 33, 102-115.
Ekstein, R., & Caruth, E. (1975). Keeping secrets. In P. Giovacchini (Ed.), Tactics and
techniques in psychoanalytic therapy (pp. 200-215). New York: Aronson.
Ellenberger, H. F. (1970). The discovery of the unconscious: The history and evolution of dynamic
psychiatry. New York: Basic Books.
Evans, N. (1976). Mourning as a
family secret, journal of the American Academy of Child

Psychiatry, 15, 502-509.

Fisher, D. V. (1984). A conceptual analysis of self-disclosure, journal for the Theory of Social
Behavior, 14, 277-296.
Folkman, S., & Lazarus, R. S. (1986). Stress processes and depressive symptomatology.
journal of Abnormal Psychology, 95, 107-113.
Folkman, S., Lazarus, R. S., Dunkel-Schetter, C, DeLongis, A., & Gruen, R. J. (1986).
Dynamics of a stressful encounter:
Cognitive appraisal, coping, and encounter out
comes. Journal of Personality and Social
Psychology, 50, 992-1003.
Goodstein, L. D., & Reinecker, V. M. (1974). Factors affecting self-disclosure: A review of
the literature. In B. A. Maher (Ed.), Progress in experimental personality research (pp.
49-77). New York: Academic Press.
Jourard, S. M. (1959). Health personality and self-disclosure. Mental Hygiene, 43, 499-507.
SELF-CONCEALMENT 455

Jourard, S. M. (1971a). Self-disclosure: An experimental analysis of the transparent self. New

York: Wiley-Interscience.
Jourard, S. M. (1971b). The
transparent self (rev. ed.). New York: Van Nostrand Reinhold.
Kaplan, G. A. & Camacho, T. (1983). Perceived health and
mortality: A nine-year follow-
up of the human population laboratory cohort. American Journal 117 of Epidemiology
292-304.

Karpel, M. A. (1980). Family secrets.

Family Process, 19, 295-306.
Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York:
Springer.
Lowenthal, M. R, & Haven, C. (1968). Interaction and adaptation: Intimacy as a critical
variable. American Sociological Review, 33, 20-30.
Margulis, S. T. (1977). Conceptions of privacy: Current
status and next steps. Journal of
Social Issues, 33, 5-21.

Mellinger, G. D., Baiter, M. B., Manheimer, D. I., Cisin, I. H., & Parry, H. J. (1978).
Psychic distress, life crisis, and use of psychotherapeutic medications. Archives of
General Psychiatry, 35, 1045-1052.

Mellinger, G. D., Baiter, M. B., Uhlenhuth, E. H., Cisin, I. H., Manheimer, D. I., &
Rickels, K. (1983). Evaluating a household survey measure of psychic distress. Psy-
chological Medicine, 13, 607-621.
Miller, L. C, Berg, J. H., & Archer, R. L. (1983). Openers: Individuals who elicit intimate
self-disclosure. Journal of Personality and Social Psychology, 44, 1234-1244.
Miller, P., & Ingham, J. G. (1976). Friends, confidants, and symptoms. Social Psychiatry,
11, 51-68.
Pennebaker, J. W. (1985). Traumatic experience and psychosomatic disease: Exploring the
roles of behavioral inhibition, obsession, and confiding. Canadian Psychology, 26, 82-
95.
Pennebaker, J. W., & Chew, C. (1985). Behavioral inhibition and electrodermal activity
during deception, journal of Personality and Social Psychology, 49, 1-7.
Pennebaker, J. W., Kiecolt-Glaser, J. K., & Glaser, R. (1988). Disclosure of traumas and
immune function: Health implications for psychotherapy. Journal of Consulting and
Clinical Psychology, 56, 239-245.
Pennebaker, J. W., & O'Heeron, R. C. (1984). Confiding in others and illness rate among

spouses of suicide and accidental death victims. Journal ofAbnormal Psychology, 93,
473-476.
Russell, D. E. (1986). The secret trauma: Incest in the lives of girls and women. New York: Basic
Books.
Saffer, J. B., Sansone, P., & Gentry, J. (1979). The awesome burden upon the child who
must keep family secret. Child Psychiatry and Human Development, 10, 35-40.
a

Schlenker, B. R. (1980). Impression management: The self-concept, social identity, and interpersonal
relations. Monterey, CA: Brooks/Cole.
Smith, T. W., Pope, M. K., Rhodewalt, R, & Poulton, J. L. (1989). Optimism, neuroticism,
coping, and symptom reports: An alternative interpretation of the Life Orientation
Test. Journal of Personality and Social Psychology, 56, 640-648.

Snyder, C. R., & Ingram, R. E. (1983). "Company motivates the miserable": The impact
of consensus information on help-seeking for psychological problems. Journal of
Personality and Social Psychology, 45, 1118-1126.
Stark, E. (1984). The unspeakable family secret. Psychology Today, 18, 38-46.
Towbin, A. P. (1978). The confiding relationship: A new paradigm. Psychotherapy Theory,
Research and Practice, 15, 333-343.
Watson, D., & Pennebaker, J. W. (1989). Health complaints, stress, and distress: Exploring
negative affectivity. Psychological Revieiv, 96, 234-254.
the central role of
Watson, M., & Greer, S. (1983). Development of a questionnaire measure of emotional
control. Journal of Psychosomatic Research, 27, 299-305.