Measuring global self-esteem: Construct validation of a single-item measure and the Rosenberg self-esteem scale

Full text

PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN
Robins et al. / MEASURING GLOBAL SELF-ESTEEM
Measuring Global Self-Esteem:
Construct Validation of a Single-Item Measure
and the Rosenberg Self-Esteem Scale
Richard W. Robins
Holly M. Hendin
Kali H. Trzesniewski
University of California, Davis
Four studies examined the construct validity of two global self-
esteem measures. In Studies 1 through 3, the Single-Item Self-
Esteem Scale (SISE) and the Rosenberg Self-Esteem Scale (RSE)
showed strong convergent validity for men and women, for differ-
ent ethnic groups, and for both college students and community
members. The SISE and the RSE had nearly identical correla-
tions with a wide range of criterion measures, including domain-
specific self-evaluations, self-evaluative biases, social desirabil-
ity, personality, psychological and physical health, peer ratings
of group behavior, academic outcomes, and demographic vari-
ables. Study 4 showed that the SISE had only moderate conver-
gent validity in a sample of children. Overall, the findings sup-
port the reliability and validity of the SISE and suggest it can
provide a practical alternative to the RSE in adult samples. More
generally, the findings contribute to the research literature by fur-
ther elaborating the nomological network of global self-esteem.
The measurement of self-esteem is a longstanding issue
in psychology, dating back to James’s (1890) seminal
writings on the self. Over the years, researchers have pro-
posed a wide range of self-esteem measures. In addition
to numerous self-report scales, there are “experience
sampling” measures (Savin-Williams & Jaquish, 1981),
pictorial measures for children (Harter, 1985), Q-sort
prototype measures based on observer judgments
(Waters, Noyes, Vaughn, & Ricks, 1985), peer ratings
(Demo, 1985), self-ideal discrepancy measures (Block &
Robins, 1993), measures based on letter preferences
(Kitayama & Karasawa, 1997), and reaction time mea-
sures (Greenwald, McGhee, & Schwartz, 1998). How-
ever, the vast majority of researchers rely on face valid
self-report scales. Of these, the Rosenberg Self-Esteem
Scale (RSE) (Rosenberg, 1965) is by far the most widely
used (Blascovich & Tomaka, 1991; Gray-Little, Williams, &
Hancock, 1997). The RSE also has received more
psychometric analysis and empirical validation than any
other self-esteem measure (Byrne, 1996; Gray-Little et al.,
1997; Wylie, 1989). In a recent review, Gray-Little et al.
(1997) concluded that the RSE is a reliable and valid
measure of global self-worth and “deserves its wide-
spread use and continued popularity” (p. 450). How-
ever, they also point out that the 10-item RSE “could be
shortened without compromising the measurement of
global self-esteem” (p. 450).
The present article proposes a new measure of self-
esteem based on a single item. We will present evidence
that this item has acceptable psychometric properties
and can serve as a useful proxy for the RSE in a variety of
research contexts. For example, a single-item measure
would be advantageous in large-scale surveys, pre-
screening packets (e.g., to select participants who are
high vs. low in self-esteem), longitudinal studies, experi-
ence sampling studies, and other research contexts in
which time constraints limit the number of items that
can be administered. These types of studies often
include shortened versions of the RSE, but unfortu-
151
Authors’ Note: This research was funded by a faculty research grant
(University of California, Davis), an Office of Educational Research
grant (University of California, Berkeley), a Provost’s Dissertation Re-
search Award (University of California, Berkeley), and a National Sci-
ence Foundation graduate fellowship. We thank Veronica Benet,
Jonathan Cheek, Chris Fraley, Samuel Gosling, Oliver John, Gerald
Mendelsohn, Delroy Paulhus, and Niels Waller for comments on an
earlier version of this article. We thank Brent Donnellan for help col-
lecting the data reported in Study 4. Correspondence should be ad-
dressed to Richard W. Robins, Department of Psychology, University of
California, Davis, CA 95616-8686; email: rwrobins@ucdavis.edu.
PSPB, Vol. 27 No. 2, February 2001 151-161
© 2001 by the Society for Personality and Social Psychology, Inc.

nately, researchers do not consistently select the same
items. For example, O’Malley and Bachman (1983) used
four of the RSE items, whereas Tafarodi and Swann
(1995) used six RSE items (only two items overlapped
between the two studies). Moreover, a single-item mea-
sure eliminates item redundancy and therefore reduces
the fatigue, frustration, and boredom associated with
answering highly similar questions repeatedly. In our
experience, even when the RSE items are interspersed
with items from other scales, participants frequently
write comments such as, “I have already answered this
question!” Such reactions may lead participants to skip
questions, respond randomly, and engage in other test-
taking behaviors that contribute to invalid protocols. For
these reasons, there is a need for a standard and brief
measure of self-esteem.
Despite the practical virtues of single-item measures,
there are a number of psychometric reasons to be skepti-
cal about them. First, multiple-item scales tend to be
more reliable. According to the Spearman-Brown for-
mula and classical reliability theory, item responses
reflect both random measurement error and true score
variance. By aggregating over multiple items, errors—if
they are random—cancel out. However, there are good
reasons to believe that measurement errors are not ran-
dom across items in the RSE. Due to the redundancy of
item content in the RSE, errors are unlikely to be aver-
aged out as the number of items is increased. Moreover,
asking more or less the same question 10 times may in
fact compound systematic errors (e.g., socially desirable
responding). As Burisch (1984) pointed out, “The
assumptions of the Spearman-Brown formula may be
unrealistic for most personality questionnaires” (p. 219).
A second reason to have multiple-item scales is to
ensure content validity for multifaceted constructs.
However, global self-worth refers to a person’s overall
positive versus negative feelings about the self; it is con-
ceptually unidimensional and explicitly not domain spe-
cific. Thus, the 10 items of the RSE do not tap into differ-
ent content domains (Gray-Little et al., 1997) but rather
ask about general feelings of self-regard using 10 differ-
ent phrasings of more or less the same item. Moreover,
global self-esteem is a subjective self-evaluation that has
no necessary link to specific behaviors. Consequently,
self-esteem may be easier to measure with a single item
than constructs such as conscientiousness, which refer to
an amalgamation of several domains of behaviors (e.g.,
punctuality, orderliness, responsibility). Thus, a consci-
entiousness scale with adequate content validity would
require at least one item to assess each domain, whereas
a global self-esteem scale may possess adequate content
validity with only a single item.
The benefits of using single-item measures, when
appropriate, have been noted by researchers (Burisch,
1984). For example, researchers studying life satisfac-
tion (Campbell, Converse, & Rodgers, 1976), subjective
well-being (Diener, 1984; Sandvik, Diener, & Seidlitz,
1993), affect (Russell, Weiss, & Mendelsohn, 1989), rela-
tionships (Aron, Aron, & Danny, 1992), and attachment
style (Hazan & Shaver, 1987) all have found single items
to be useful supplements to more extensive measure-
ment procedures. In short, single-item measures can
provide an acceptable balance between practical needs
and psychometric concerns.
We report findings from four studies to support the
construct validity of the Single-Item Self-Esteem Scale
(SISE). Study 1 is a longitudinal study in which the SISE
and the RSE were administered six times over the course
of 4 years of college; the SISE and the RSE were related to
each other and to a wide range of variables to examine
the construct validity of the two measures. In Study 2, we
examined correlations of the SISE and the RSE with
measures of socially desirable responding and examined
the effects of a 5-point versus a 7-point response scale. In
Study 3, we examined the correlation between the SISE
and the RSE in a community sample. In Study 4, we ex-
amined the convergent validity of the SISE in a sample of
fourth- to eighth-grade students by using the Global Self-
Esteem Scale from Harter’s (1985) Self-Perception Pro-
file for Children.
STUDY 1
Method
PARTICIPANTS AND PROCEDURE
This study used data from an ongoing longitudinal
study of 498 undergraduate students who attended the
University of California at Berkeley. The sample is
diverse in terms of ethnicity (42% Asian, 40% Caucasian,
11% Chicano/Latino, 6% African American, 1% Native
American), gender (56% women), socioeconomic sta-
tus (20% came from families with household incomes
less than $25,000 and 20% from families with household
incomes greater than $100,000), and academic ability
(combined Scholastic Aptitude Test [SAT] scores
ranged from 650 to 1540, M= 1183, SD = 180).
Participants were assessed annually throughout col-
lege. During the first week of college and at the end of
each academic year, participants were contacted by mail
and asked to complete an extensive questionnaire in
exchange for money or course credit. To validate our
single-item scale, measures of theoretical relevance to
self-esteem were selected from the larger set of measures
included in the longitudinal study. Table 1 shows the
assessment periods when participants completed each
measure.
152 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

MEASURES
Self-esteem. Participants completed three measures of
self-esteem: the SISE, the RSE, and the Texas Social
Behavior Inventory-Form A (TSBI) (Helmreich & Stapp,
1974). All three measures were rated on a 5-point scale
ranging from 1 (not very true of me)to5(verytrueofme)
and were administered along with several hundred items
from other scales. The SISE consists of the item, “I have
high self-esteem.” The RSE consists of 10 items assessing
global self-esteem (e.g., “On the whole, I am satisfied
with myself”). Previous studies have reported alpha
reliabilities for the RSE ranging from .72 to .88 (Gray-Lit-
tle et al., 1997). In the present sample, alpha reliabilities
ranged from .88 to .90 across the six assessments. The
TSBI consists of 16 items focused on self-esteem in social
settings (e.g., “I have no doubts about my social compe-
tence”). The alpha reliability of the TSBI was .82 in the
present sample.
Domain-specific self-evaluations. Participants rated their
competence in seven domains: general intelligence, quan-
titative ability, verbal ability, social skills, artistic ability,
athletic ability, and physical attractiveness (Pelham &
Swann, 1989). Ratings were made on a 10-point percentile
scale ranging from 1 (bottom 5% of college students)to10
(top 5%). Academic self-concept was assessed using a
standardized composite of seven items (αs ranged from
.79 to .88), tapping two facets of perceived academic abil-
ity: direct self-reports of ability (e.g., “Compared to the
average University of California [UC] Berkeley student,
how would you rate your academic ability?”) and
expected performance in college (e.g., “Realistically,
what overall grade point average [GPA] do you think you
will attain in college?”).
Self-evaluative biases. Self-enhancement bias was defined
as the degree to which participants overestimated their
academic ability relative to accepted measures of actual
academic ability. Specifically, we used a residual score
(computed via regression) reflecting the discrepancy
between the academic self-concept scale and a compos-
ite measure of academic ability based on standardized
test scores and high school grades (Robins & Beer, 1999;
see also Robins & John, 1997). Positive values indicate
self-enhancement (i.e., overestimation by the self) and
negative values indicate self-diminishment (i.e., under-
estimation by the self).
Self-serving attributional bias was defined as the ten-
dency to make internal attributions for academic success
Robins et al. / MEASURING GLOBAL SELF-ESTEEM 153
TABLE 1: Study 1: Assessment Periods for Self-Esteem Scales and Criterion Measures
Beginning End of End of End of End of End of
Measures of College First Semester First Year Second Year Third Year Fourth Year
Self-esteem
SISE X X X X X X
RSE X X X X X X
TSBI X
Domain-specific self-evaluations
Academic ability X X X X X X
Seven domains XXX
Self-evaluation bias
Self-enhancement X
Self-serving attribution X
Personality
NEO Big Five X X
Optimism XX
Shyness X
Psychological and physical well-being
Life satisfaction X X X X X X
Affect X
Depression X X X X
Perceived stress X X X X
Physical health X X X X
Group behavior
Peer ratings X
Academic outcomes
SAT scores X
High school GPA X
College GPA X X X X X X
Attrition X X X X X X
NOTE: SISE = Single-Item Self-Esteem Scale, RSE = Rosenberg Self-Esteem Scale, TSBI = Texas Social Behavior Inventory, and GPA = grade point
average.

and external attributions for academic failure. We used
the Multidimensional-Multiattributional Causality Scale
(MMCS) (Lefcourt, von Baeyer, Ware, & Cox, 1979) to
assess causal attributions to two internal factors (i.e., abil-
ity, effort) and two external factors (i.e., situation, luck)
separately for success and failure experiences. Self-
serving bias was computed as the sum of four scales:
internal attributions for success, external attributions
for failure, external attributions for success (reverse-
scored), and internal attributions for failure (reverse-
scored).
Personality. The NEO-Five Factor Inventory (NEO-
FFI) (Costa & McCrae, 1992) is a 60-item measure of the
Five-Factor Model completed by participants at the
beginning and end of college (αs = .83 and .82 for Extra-
version, .76 and .77 for Agreeableness, .81 and .83 for
Conscientiousness, .84 and .85 for Neuroticism, and .77
and .75 for Openness to Experience, respectively, for the
two assessments). Participants also completed Cheek
and Melchior’s (1990) three-component measure of shy-
ness, which assesses the physiological, cognitive, and
behavioral aspects of shyness. The three items assessing
the three components were summed to form a global
measure of shyness (α= .65). We assessed dispositional
optimism using the six-item Life Orientation Test—
Revised (Scheier, Carver, & Bridges, 1994) (α= .86).
Psychological and physical well-being. Participants com-
pleted four measures of psychological well-being: (a) the
20-item Center for Epidemiological Studies Depression
Scale (Radloff, 1977) (αs = .90 to .92), (b) the single-item
Overall Life Satisfaction Scale (Campbell et al., 1976),
(c) the six-item Perceived Stress Scale (Cohen, Kamarck, &
Mermelstein, 1983) (αs = .70 to .76), and (d) the Positive
and Negative Affect Scale (Watson, Clark, & Tellegen,
1988) (αs = .88 for positive affect and .88 for negative
affect). Two single-item measures of physical well-being
were used: (a) a global, subjective indicator (“How would
you rate your physical health right now?”) and (b) a more
objective indicator (“How often did you go to the stu-
dent health services or a private doctor this year?”).
Peer-rated group behavior. Participants interacted in
groups of four to five for a 20-minute decision-making
task. Following the interaction, participants rated them-
selves and the other group members on the following
dimensions: overall effectiveness in the group, talkative-
ness, cooperativeness, competitiveness, task orientation,
and creativity.
Academic outcomes. SAT scores, high school GPA,
cumulative college GPA, and attrition information were
obtained from university records.
Demographics. Gender, ethnicity, socioeconomic sta-
tus, and age were obtained from university records.
Results and Discussion
DESCRIPTIVE STATISTICS
Averaged across all six assessment periods, the SISE
had a mean of 3.5 (SD = 1.1), a median of 4, and a mode
of 4; the frequency distribution of the averaged scores
was as follows: 1 = 5.3%, 2 = 14%, 3 = 26.2%, 4 = 35.2%,
and 5 = 19.2%. Thus, the SISE is approximately normally
distributed with some negative skewness (–.47), that is,
most individuals reported having relatively high self-
esteem with a minority reporting lower self-esteem. Aver-
aged across assessment periods, the RSE had a mean of
3.8 (SD = .7) and a median of 3.9, with scores ranging from
1.3 to 5.0. Similar to the SISE, the RSE had an approxi-
mately normal distribution but with slightly more skew-
ness (–.63).
TEST-RETEST RELIABILITY AND STABILITY OVER TIME
Internal consistency reliability (e.g., Cronbach’s coef-
ficient alpha) cannot be computed for a single-item scale.
However, Heise (1969, Equation 9) developed a proce-
dure to estimate the reliability of a single-item scale
based on its pattern of autocorrelations over three points
in time. The Heise procedure provides an estimate of test-
retest reliability that separates true change from mea-
surement error. We computed the Heise estimate three
times using the across-time correlations of the SISE from
the beginning to the end of college.1The mean reliabil-
ity estimate for the SISE was .75. To demonstrate the
effectiveness of the Heise procedure, we also used this
method to estimate the reliability of the RSE. The mean
reliability estimate for the RSE was .88 (the same as its
alpha reliability).
To examine stability over time, correlations of the
SISE and RSE were computed across the six assessments.
The mean across-time correlation (averaged across all 15
possible time intervals) was similar for the SISE (mean r=
.61) and the RSE (mean r= .69). Moreover, the pattern of
across-time correlations was similar for the two mea-
sures, and both showed a simplex structure (i.e., higher
correlations between adjacent time periods and lower
correlations across longer time intervals).
CONSTRUCT VALIDITY
Concurrent correlations between the SISE and the
RSE ranged from .72 to .76 across the six assessments,
with a median of .75. (In comparison, the 10 RSE items
had median item-total correlations ranging from .46 to
.71.) These correlations were similar for men (Mdn =
.74) and women (Mdn = .73) and for Caucasians (Mdn =
.80), African Americans (Mdn = .71), Asians (Mdn = .70),
and Latinos (Mdn = .70). After correcting for attenua-
tion due to unreliability in the SISE and RSE, convergent
correlations in the total sample ranged from .89 to .94
with a median of .93. Thus, it appears that virtually all of
154 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

the reliable variance in the SISE overlaps with the RSE.
Correlations with the TSBI were somewhat lower; the
TSBI correlated .58 with the SISE and .62 with the RSE.
These lower convergent validity correlations most likely
reflect the fact that the TSBI is primarily a measure of
social self-esteem and not global self-esteem (Blascovich &
Tomaka, 1991).
Factor analyses of the 10 RSE items showed a single
general factor (Gray-Little et al., 1997): All items had
high loadings on the first unrotated factor, a scree test
showed an “elbow” after the first factor, and an analysis
using structural equation modeling showed relatively
good fit with a single-factor model (Comparative Fit
Index [CFI] = .90). Moreover, when two factors were
extracted and rotated using an oblique rotation, they
correlated .55, which also suggests a strong general fac-
tor. The SISE was strongly related to both factors; it cor-
related .81 with the first factor and .65 with the second
factor. These factors did not correspond to the positively
keyed and negatively keyed item factors found in some
previous studies of the RSE factor structure (Carmines &
Zeller, 1979; Hensley & Roberts, 1976).
Next, we compared the convergent and discriminant
validity of the two self-esteem measures. Table 2 shows
correlations of the SISE and the RSE with 37 different
criteria. The two measures showed remarkably similar
correlations with the domain-specific self-evaluations.
High-self-esteem individuals tended to see themselves as
academically competent, intelligent, socially skilled, ath-
letic, and physically attractive. These findings are consis-
tent with previous research linking global self-esteem to
specific self-evaluations (e.g., Marsh, 1986) as well as with
the recognition that global self-esteem is related but not
equivalent to a person’s domain-specific self-evaluations
(e.g., Harter, 1998; Pelham & Swann, 1989; Rosenberg,
1979).
Both measures also showed that high-self-esteem indi-
viduals are prone to self-evaluative biases. In particular,
high-self-esteem individuals’ beliefs about their aca-
demic ability were unrealistically positive (relative to
their test scores and achievement) and they made self-
serving attributions for their academic successes and fail-
ures (i.e., taking credit for success but externalizing
blame for failure). These findings are consistent with
previous research showing a link between self-esteem
and positive illusions about the self (e.g., Aspinwall &
Taylor, 1992; Brown, Collins, & Schmidt, 1988; Taylor &
Brown, 1988).
The SISE and RSE had highly similar Big Five profiles,
deviating more than slightly only in their correlations
with Agreeableness and Neuroticism. The Big Five corre-
lates are highly similar to those found in a very large sam-
ple (N= 127,838) obtained over the Internet between
the SISE and NEO-FFI measures of Extraversion (r=
.37), Agreeableness (r= .13), Conscientiousness (r=
.25), Neuroticism (r= –.50), and Openness (r= .18)
(Robins, Trzesniewski, Tracy, Gosling, & Potter, 1999).
The Big Five correlates are also similar to those reported
by Haig (1998), who found correlations between the
Robins et al. / MEASURING GLOBAL SELF-ESTEEM 155
TABLE 2: Study 1: Construct Validity of the Single-Item Self-Esteem
Scale (SISE) and the Rosenberg Self-Esteem Scale (RSE)
SISE RSE
Domain-specific self-evaluations
Academic ability .20* .24*
General intellectual ability .19* .20*
Quantitative ability .17* .15*
Verbal ability .12 .16*
Social skills .27* .27*
Artistic ability .13 .10
Athletic ability .19* .18*
Physical attractiveness .31* .31*
Self-evaluative biases
Self-enhancement bias .28* .32*
Self-serving attributions .30* .38*
Personality
NEO Extraversion .39* .41*
NEO Agreeableness .04 .23*
NEO Conscientiousness .23* .28*
NEO Neuroticism –.57* –.70*
NEO Openness to Experience .11 .16*
Optimism .44* .48*
Shyness –.26* –.28*
Psychological and physical well-being
Life satisfaction .45* .54*
Dispositional affect—positive .53* .56*
Dispositional affect—negative –.38* –.56*
Depression –.25* –.34*
Perceived stress –.36* –.39*
Physical well-being .23* .26*
Number of doctor visits –.01 –.07
Peer-rated group behavior
Effectiveness in group task .17* .15*
Talkative .15* .19*
Cooperative .08 .14*
Competitive .04 .04
Task-oriented .15* .19*
Creative .10 .17*
Academic outcomes
SAT scores –.08 –.07
High school GPA –.10 –.03
College GPA .05 .07
Attrition .00 –.01
Demographics
Sex (keyed toward male) .16* .12*
Socioeconomic status –.02 .02
Age .01 .00
NOTE: Ns ranged from 299 to 496. Values in the table are correlations
between self-esteem scores obtained in the first assessment, with crite-
rion variables aggregated across all available assessment periods. The
two columns in the table correlate at .98. For three of the criterion vari-
ables (NEO Agreeableness, NEO Neuroticism, and Dispositional Af-
fect—Negative), the correlation with the SISE differed significantly (p
< .01) from the correlation with the RSE, based on a t-test for the differ-
ence between dependent correlations. GPA = grade point average.
*p< .01.

RSE and NEO-PI-R measures of Extraversion (r= .45),
Agreeableness (r= .13), Conscientiousness (r= .43),
Neuroticism (r= –.69), and Openness (r= .10) in a sam-
ple of 345 college students. Both self-esteem measures
were strongly positively correlated with optimism and neg-
atively correlated with shyness, which is consistent with
research showing that high-self-esteem individuals tend
to be optimistic (Scheier et al., 1994) and low in shyness
(Cheek & Melchior, 1990).
The SISE and RSE had similar relations with measures
of psychological and physical health. The most notable
difference between the two measures was the stronger rela-
tion between the RSE and measures of depressive and neg-
ative affect, perhaps reflecting the fact that the RSE
includes items tapping negative feelings toward the self
(e.g., “All in all, I am inclined to feel that I’m a failure.”).
Supporting this possibility, the negatively keyed RSE items
had stronger relations with the maladjustment measures
than did the positively keyed items.2Whether this is an
advantage or disadvantage of the RSE relative to the SISE
depends on one’s theory about how strongly global
self-esteem should be related to psychological distress.3
The findings for the peer ratings of behavior were
consistent with the self-report measures of personality
and adjustment. High-self-esteem individuals on both
the SISE and the RSE were described by their peers as
effective, talkative, cooperative, task-oriented, and cre-
ative in the group interaction. Together, the self- and
peer criterion measures paint a portrait of the high-self-
esteem individual as functioning effectively in both the
intrapersonal and interpersonal domains.
Neither self-esteem measure was strongly associated
with any of the academic outcomes. The correlations are
consistent with, although somewhat weaker than, those
reported in previous studies; for example, Hansford and
Hattie (1982) conducted a meta-analysis of the relation
between self-esteem and academic performance and
found an average effect size of .14 for college students.
Thus, our findings replicate previous research showing
that self-esteem and objective academic outcomes are
only weakly related and further confirm the realization
that self-esteem may not be the panacea for academic
problems that many hoped it would be (California Task
Force on Self-Esteem, 1990).
Consistent with previous research, men reported
slightly higher self-esteem than did women at every assess-
ment period (e.g., Major, Barr, Zubek, & Babey, 1999;
Robins, Trzesniewski, et al., 1999). The magnitude of
this gender difference was approximately the same for
the SISE (Cohen’s d= .26, p< .05) and the RSE (Cohen’s
d= .21, p< . 05). There were significant ethnic differ-
ences for the SISE, F(3, 474) = 4.3, p< .05, and the RSE,
F(3, 474) = 3.8, p< .05; the pattern of differences was the
same for both measures: Asian Americans reported sig-
nificantly lower levels of self-esteem than Caucasians
(ds = .36 and .32, ps < .05, respectively) and marginally
significantly lower self-esteem than African Americans
(ds = .31 and .44, ps < .15, respectively). Neither the SISE
nor the RSE had significant correlations with socioeco-
nomic status or age. The similarity between the SISE and
the RSE in their demographic correlates provides addi-
tional support for the psychological equivalence of the
two scales.
To further explore the similarity between the SISE
and the RSE, we recomputed all of the correlations in
Table 2 separately for men and women. The two columns
correlated .97 for men and .97 for women, and for both
genders the correlations differed only minimally in abso-
lute level. In addition, to address the possibility that aggre-
gating over time reduces error variance in the SISE more
than in the RSE, we also computed all possible concur-
rent correlations between the self-esteem measures and
each criterion variable and then averaged the resulting
correlations. We found a nearly identical pattern for the
SISE and the RSE (correlation between columns = .98)
and a minimal difference in absolute level. Finally, to
address the possibility that repeated assessments of the
SISE increase its validity, we computed the correlations
in Table 2 using only the first wave of self-esteem data.
Again, the pattern of correlations for the SISE and the
RSE was nearly identical (correlation between columns =
.99), and there were minimal differences in absolute
level.
In summary, the findings from Study 1 provide sub-
stantial support for the convergence between the SISE
and the RSE. The two measures were highly correlated
with each other and had similar relations with a broad
range of criterion measures selected because of their
theoretical or empirical link with self-esteem. In virtually
all cases, the correlations were consistent with previous
research on global self-esteem. Thus, the findings sup-
port the construct validity of both the SISE and the RSE
and contribute to the general research literature by fur-
ther elaborating the nomological network of global self-
esteem.
STUDY 2
Study 2 had three goals. First, we examined the con-
vergent correlation between the SISE and the RSE in an
independent sample. Second, we tested whether a 7-
point rating scale would increase the convergent validity
of the SISE relative to the 5-point scale used in Study 1.
The length of the response scale might affect the validity
of the measure because (a) it might increase the amount
of reliable variance (if individuals can reliably discrimi-
nate among seven scale points when rating their self-
esteem then the 5-point scale would lose valid informa-
tion) and (b) it might affect the distributional properties
156 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

of the measure, including the degree of skewness and
kurtosis. Third, we examined the degree to which the
SISE and the RSE are related to two different facets of
socially desirable responding (Paulhus, 1994, 1998): self-
deceptive enhancement (the tendency to give honestly
held but inflated self-descriptions) and impression man-
agement (the tendency to consciously provide inflated
self-descriptions as a form of self-presentation to an
audience).
Method
SAMPLE
The study included 208 undergraduate students (61%
women) who participated in this study as part of a course
requirement.
MEASURES
Participants completed the SISE, the RSE, and the 40-
item Balanced Inventory of Desirable Responding
(BIDR) (Paulhus, 1994). The BIDR measures two forms
of socially desirable responding: Self-Deceptive Enhance-
ment (SDE) and Impression Management (IM). One
hundred thirty-nine participants completed the ques-
tionnaire using a 5-point rating scale, and 69 participants
used a 7-point scale. Alpha reliabilities for the 5-point
and 7-point versions, respectively, were .89 and .93 for the
RSE, .72 and .79 for the SDE, and .78 and .78 for the IM.
Results and Discussion
The SISE and RSE correlated .74 for the 5-point scale
and .73 for the 7-point scale (see Table 3). These values
are similar to the median correlation of .75 found in
Study 1 and provide further support for the convergent
validity of the SISE. However, the length of the response
scale (5-point scale vs. 7-point scale) does not appear to
affect the convergent validity of the SISE.
In terms of socially desirable responding, the SISE
and SDE correlated .21 and .47 and the RSE and SDE
correlated .22 and .49, respectively, for the 5-point and
7-point scales. The SISE and IM correlated .04 and .05
and the RSE and IM correlated .13 and .08, respectively,
for the 5-point and 7-point scales. Thus, both the SISE
and the RSE were associated with self-deceptive
enhancement but neither was associated with impression
management. This is consistent with previous research
on self-esteem and socially desirable responding
(Paulhus, 1994, 1998) and provides further support for
the relation found in Study 1 between self-esteem and
self-enhancement bias.
Men reported higher self-esteem than women on
both the SISE (Cohen’s d= .24 and .70, for the 5-point
and 7-point scales) and the RSE (Cohen’s d= .02 and .34,
for the 5-point and 7-point scales). This sex difference is
consistent with the findings from Study 1.
In summary, Study 2 replicated the correlation between
the SISE and the RSE in a second sample of college stu-
dents and showed that this correlation is equally strong
for a 5-point and 7-point rating scale. Study 2 also showed
that high scorers on both the SISE and the RSE show a
tendency toward self-deceptive enhancement but they
do not seem to be consciously managing their impressions.
STUDY 3
The goal of Study 3 was to replicate the convergent
correlation between the SISE and the RSE in a noncollege
student sample. It is possible that the single-item mea-
sure of self-esteem is only valid in college student samples,
where the participants are highly educated and gener-
ally focused on their competencies, goals, and aspira-
tions. If so, the high convergence between the SISE and
the RSE found in Studies 1 and 2 may not generalize to a
community sample.
Method
SAMPLE
Participants were 66 individuals recruited from sev-
eral community locations, including shopping malls, bus
stations, day care centers, and movie theaters. The sam-
ple was diverse in terms of gender (64% women), age
(range = 21 to 61, Mdn = 44), and ethnicity (67% Cauca-
sian, 17% Chicano/Latino, 12% Native American, 3%
African American, 1% Asian). Participants were employed
in a wide range of occupations, including manual labor
(e.g., sheet metal worker), clerical (e.g., administrative
assistant), managerial (e.g., store manager), and profes-
sional (e.g., architect).
MEASURES
Participants completed the SISE and the RSE (α= .87),
which were embedded in a longer questionnaire.
Robins et al. / MEASURING GLOBAL SELF-ESTEEM 157
TABLE 3: Correlation Between the Single-Item Self-Esteem Scale
(SISE) and the Rosenberg Self-Esteem Scale (RSE) Across
Three Studies
Total Sample Men Women
Study 1 .75* .74* .75*
(N= 496) (.93) (.91) (.92)
Study 2a.74*/.73* .76*/.71* .74*/.70*
(N= 208) (.91)/(.87) (.93)/(.87) (.90)/(.84)
Study 3
(N= 66) .80* (.99) .79* (.95) .79* (> .99)
NOTE: Values in parentheses are corrected correlations, disattenuated
for unreliability in the RSE and the SISE (the SISE reliability estimate
from Study 1 was used for all corrections).
a. Correlations for 5-point/7-point response scales, respectively.
*p< .01.

Results and Discussion
The correlation between the SISE and the RSE was
.80, even higher than in the two college student samples
(see Table 3). The higher correlation was not due to
greater variance in the self-esteem scores; the standard
deviations of the SISE and the RSE were slightly lower in
the community sample than in the college student sam-
ples. The correlation between the SISE and the RSE was
not moderated by the occupational status of the partici-
pant; the correlation was .81 for individuals employed in
jobs requiring a college degree (n= 29) and .79 for indi-
viduals employed in jobs that did not require a college
degree (n= 37). This suggests that the SISE may be a
valid measure of global self-esteem in noncollege stu-
dent samples, although we recommend that additional
validation evidence be obtained before using the SISE in
such contexts.
Men reported slightly higher self-esteem than did wo-
men, which is consistent with Studies 1 and 2. The mag-
nitude of this gender difference was approximately the
same for the SISE (Cohen’s d= .40, p< .05) and the RSE
(Cohen’s d= .51, p< .05). Age was not significantly corre-
lated with self-esteem for either the SISE (r= .14) or the
RSE (r= .09).
STUDY 4
Study 4 had three goals. The first goal was to deter-
mine whether the SISE shows convergent validity in a
sample of children. To test this, we collected self-esteem
data on a sample of children in Grades 4 through 8 and
correlated the SISE with the Global Self-Esteem (GSE)
Scale from the Self-Perception Profile for Children
(Harter, 1985), the most widely used self-esteem scale for
children. The second goal was to determine whether
observer ratings using the SISE show convergent validity
with observer ratings using the GSE. To test this, we cor-
related parent ratings of their child’s self-esteem using
the SISE and the GSE. The third goal was to determine
whether agreement between child and parent ratings of
self-esteem is comparable for the two measures of
self-esteem.
Method
PARTICIPANTS
The study included 212 children (58% girls) and
their parents or primary caregivers. The sample con-
sisted of fourth- to eighth-grade students recruited from
a kindergarten through eighth-grade school in a pre-
dominantly middle-class suburban community. The sam-
ple was 65% Caucasian, 15% Chicano/Latino, 10% Afri-
can American, 5% Asian, and 5% Native American.
PROCEDURE
To control for differences in reading ability, the chil-
dren were interviewed individually and the questions
were presented visually on a computer screen and read
aloud to the children through headphones. The com-
puter recorded response latencies for each item. A
trained interviewer was present throughout the inter-
view to assist the child and to answer any questions the
child might have about the meaning or content of the
questions. Parents completed the self-esteem measures
at home and returned the questionnaire by mail. For
both children and parents, the self-esteem items were
embedded in a larger questionnaire.
SELF-ESTEEM MEASURES
SISE. Children rated their self-esteem using the SISE.
The item was worded exactly as in Studies 1 through 3 (“I
have high self-esteem”), and responses were made on a
5-point scale ranging from 1 (strongly disagree)to5(strongly
agree). A parent or primary caregiver also rated the child’s
self-esteem using the SISE (“My son/daughter has high
self-esteem”). For a subset of participants, two parents or
caregivers provided SISE data, in which case we averaged
their ratings.
Self-Perception Profile for Children. Children completed
the six-item GSE Scale of the Self-Perception Profile for
Children (Harter, 1985). A parent or primary caregiver
described his or her child’s self-esteem using the same
set of items worded in the third person. For a subset of
participants, two parents or caregivers completed the
scale, in which case we averaged their scale scores. The
alpha reliability was .77 for child-reported self-esteem
and .84 for parent reports of the child’s self-esteem.
RESULTS AND DISCUSSION
The correlation between the SISE and the GSE Scale
was .52 in the total sample (rs = .55 for boys and .50 for
girls). There were no systematic trends across Grades 4
to 8 (convergent rs = .59, .37, .66, .50, and .43 for Grades
4 to 8, respectively). Overall, these convergent correla-
tions are moderate in magnitude but substantially lower
than we found in Studies 1 through 3. This suggests that
the SISE may not be as valid for children as it is for adults.
Nonetheless, the convergence between the SISE and the
GSE is sufficiently high to justify further research into its
construct validity for children (e.g., correlations with
external criteria).
Our second question concerned the convergence
between parental reports of their child’s self-esteem
using the SISE and the GSE. The correlation between
the SISE and the GSE was .74, which is near the reliability
of the two measures. This suggests that parental ratings
of self-esteem are highly similar for the two measures.
158 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

Our third question concerned the convergence
between child and parent reports of self-esteem, that is,
to what extent do the children’s self-reports of self-esteem
correspond with parent reports of the child’s self-
esteem? The child SISE correlated .24 with the parent
SISE and .30 with the parent GSE; the child GSE corre-
lated .34 with the parent GSE and .35 with the parent
SISE. Thus, child-parent convergence was comparable
in magnitude for the two self-esteem measures but some-
what higher for the multiple-item GSE Scale.
How can we explain the fact that the convergent valid-
ity of the SISE was lower in the child sample than in the
college student samples (Studies 1 and 2) or the commu-
nity sample (Study 3)? The lower validity may simply
reflect the limited language proficiency of fourth- to
eighth-grade students (e.g., they do not fully understand
the meaning of self-esteem). However, this seems unlikely
given that the convergent correlations do not increase
with age. Another possibility is that the SISE is simply less
strongly associated with the item content on the GSE
than the item content on the RSE. However, this also
seems unlikely because the GSE items (“I like the kind of
person I am”) are similar in content to the RSE items
(e.g., “I take a positive attitude toward myself”) and,
moreover, the convergent correlations were very high
for the parent ratings of the SISE and GSE.4Part of the
difference may be due to the lower reliability of the GSE
Scale (α= .77) compared to the RSE. However, even the
disattenuated convergent correlation is only .68 (cor-
recting for unreliability in the GSE and the SISE), which
is considerably lower than the disattenuated correlations
reported in Studies 1 through 3 (see Table 3).
Finally, a developmental interpretation is plausible.
The findings from all four studies suggest an interesting
developmental trend: The validity of the SISE may in-
crease from childhood to late adolescence to adulthood.
We found the lowest convergence in the child sample (r=
.52), much higher convergence in the two college stu-
dent samples (rs = .73 to .75), and even higher conver-
gence in the community sample (r= .80), which had a
median age of 44. We might speculate that this trend
reflects the development of a consistent and coherent
self-schema; from this perspective, children may show
lower convergence because they do not have fully devel-
oped global self-esteem schemas. Consistent with this
interpretation, we found that response latency on the
SISE moderated the relation between the SISE and the
GSE (beta of the interaction term in a moderated multi-
ple regression analysis = –.21); specifically, the conver-
gent validity of the SISE was .72 for individuals 1 SD
below the mean in response latency (schematics), .51 for
individuals at the mean, and .30 for individuals 1 SD
above the mean (aschematics) (Aiken & West, 1991).5
Interestingly, the estimated convergent correlation for
schematic children is comparable to what we found for
college students in Studies 1 and 2.
GENERAL DISCUSSION
Together, the findings from Studies 1 through 4 fur-
ther elaborate the nomological network of global
self-esteem and show that global self-esteem has impor-
tant and wide-ranging implications for interpersonal
and intrapsychic functioning. In general, the findings
replicate previous research on the correlates of self-
esteem. The findings also provide information about
self-esteem correlates that have received relatively little
attention in the self-esteem literature, including how
self-esteem relates to peer ratings of behavior in a group
interaction task, how self-esteem relates to the Big Five
personality traits, and the degree to which parent and
child ratings of self-esteem converge. The replication of
our findings using two measures of self-esteem strength-
ens the generalizability of the research.
In addition to contributing to an understanding of
the correlates of global self-esteem, the present research
provides further support for the construct validity of the
most widely used measure of global self-esteem—the
RSE—as well as supporting the validity of a new measure
of global self-esteem based on a single item—the SISE.
The SISE had very high convergent validity with the RSE
in all three studies using adult samples (see Table 3) and
lower, but still moderate, convergent validity with Har-
ter’s GSE scale in the child sample. The strong conver-
gence between the SISE and the RSE held (a) for males
and females, (b) for different ethnic groups, (c) for both
college students and community members, (d) for dif-
ferent occupational statuses, (e) across 4 years of col-
lege, and (f) for a 5-point and 7-point rating scale. More-
over, the SISE and the RSE had a nearly identical pattern
of correlates with a wide range of criterion variables,
including domain-specific self-evaluations, self-evaluative
biases, two forms of social desirability, personality traits,
psychological and physical health, peer ratings of group
behavior, demographic characteristics, and several
objective measures of academic outcomes. Thus, the
findings suggest that in most cases, researchers using the
SISE will find virtually the same relations as they would
have had they used the RSE. Given its greater practicality,
therefore, the SISE may be a useful alternative to the RSE
in some research contexts.
Clearly, the SISE has limitations. First, the SISE is not
as reliable as the RSE. Yet, despite its lower reliability, the
SISE exhibited essentially the same pattern of correla-
tions as the RSE. Second, the SISE may be more suscepti-
ble to extremity and acquiescence response styles
because it has only one positively keyed item. However,
this concern is ameliorated by the fact that the SISE cor-
relates almost as highly with a composite of the five nega-
Robins et al. / MEASURING GLOBAL SELF-ESTEEM 159

tively keyed RSE items (rs = .66 to .70 across Studies 1 to 3)
as it does with a composite of the five positively keyed
RSE items (rs = .71 to .75 across Studies 1 to 3). Third, the
SISE is a blatantly face-valid measure and shows some
susceptibility to socially desirable responding, in particu-
lar to self-deceptive enhancement. This is undoubtedly a
problem but one that is equally applicable to the RSE (as
Study 2 shows) and the GSE. If one wants a subtle mea-
sure of deep-seated feelings of self-worth, neither the
SISE nor the RSE nor the GSE is the measure of choice.
Fourth, scale validation is an ongoing process (one never
validates a scale but rather provides progressively more
evidence for a particular interpretation of the scale) and
researchers should be cautious about using the SISE
without validating it in other contexts and in a wider
range of samples. For example, Study 4 suggests that the
SISE may not be as valid in childhood and further evi-
dence is needed for its construct validity in that age group.
The finding that a single item can provide a valid mea-
sure of global self-esteem is itself of theoretical impor-
tance, which may have implications for the development
of self-esteem and the way in which evaluative knowledge
about the self is organized and retrieved. We have specu-
lated that a single item may provide an adequate mea-
sure of self-esteem because most adult individuals are
schematic for self-esteem. Self-esteem is an indicator of
social status in group and interpersonal contexts and
helps to motivate and regulate behavior in achievement
contexts, and thus serves important adaptive functions
(Leary, 1999; Robins, Norem, & Cheek, 1999). Conse-
quently, global self-esteem is likely to be repeatedly acti-
vated in a wide range of situations and is therefore likely
to be chronically accessible by adulthood. The fact that
that most adults are schematic for global self-esteem may
reduce random error in self-reports of self-esteem (but
not systematic errors that are correlated across items).
Thus, in terms of measuring global feelings of self-worth,
it may simply be necessary to activate the self-esteem
schema (which can be done with a single item) so redun-
dant items provide limited additional information. This
interpretation is supported by the lower convergent
validity found for the SISE in the child sample, as well as
the moderating effect of response latency.
This raises an important caveat: We are not arguing
that single-item self-report scales are generally useful in
psychological assessment or that they can be used to
measure all psychological constructs. For example, it is
unlikely that a single-item measure would be valid for a
multifaceted and psychodynamically complex construct
such as narcissism or for a construct on which a nontrivial
proportion of adults are not schematic, such as inde-
pendence (Markus, 1977). The present findings are infor-
mative about the circumstances under which a single
item can provide a suitable alternative to multiple-item
scales. Specifically, a single self-report item may be ade-
quate when the construct is (a) highly schematized for
most individuals, (b) unidimensional in content, and
(c) primarily reflects subjective experience. In the pres-
ent case, we are convinced by the accumulated evidence
that the SISE sacrifices little in the way of reliability and
validity while at the same time providing a practical alter-
native to the RSE. In short, if the goal is to discover
whether individuals consciously experience positive self-
regard, it may be possible, as Allport (1942) suggested, to
simply ask them.
NOTES
1. The Heise procedure assumes an equivalent interval between
assessments so we excluded the data from the end of the first semester
assessment; this left five assessment periods, each separated by 1 year.
2. Overall, however, the pattern of correlations was highly similar
for the negatively keyed Rosenberg Self-Esteem Scale (RSE) items and
the positively keyed RSE items. When we recomputed all of the correla-
tions in Table2 separately for the negatively and positively keyed items,
the correlation between the two sets of correlations (computed across
the 37 criteria) was .99, and they differed minimally in absolute level.
3. Psychological theory is somewhat ambiguous about what the
nomological net of self-esteem should look like with regard to mea-
sures of neuroticism and depression. Self-esteem and neuroticism are
generally considered to be distinct constructs, and very high conver-
gent correlations between a measure of self-esteem and a measure of
neuroticism could suggest a lack of discriminant validity, perhaps due
to shared method variance, item content overlap, and/or response-
style effects (e.g., socially desirable responding). On the other hand, in
some psychological theories (e.g., Beck, 1972), negative beliefs about
the self are considered central to depression and the two constructs are
not entirely distinct. From this perspective, one would expect very
strong relations between measures of self-esteem and measures of psy-
chological distress.
4. In addition, in a pilot study of 304 fifth- and eighth-grade stu-
dents, we administered the Single-Item Self-Esteem Scale (SISE) and
the RSE and found a convergence correlation of .51, which is almost
exactly the same as we found for the SISE and the Global Self-Esteem
(GSE) Scale.
5. To ensure that this effect was not due to general processing
speed, we conducted another analysis in which we subtracted mean
response latency for a set of control items (ratings of teachers) from
response latency for the SISE. This corrected measure of response
latency also significantly moderated the relation between the SISE and
the GSE. To ensure that the effect was not due to a few outliers, we con-
ducted two additional moderated multiple regression analyses exclud-
ing individuals who were 3 or more and 2 or more standard deviations
above the mean; in both analyses, response latency continued to signif-
icantly moderate the relation between the SISE and the GSE.
REFERENCES
Aiken, L. S., & West, S. G. (1991). Multiple regression:Testing and interpret-
ing interactions. Newbury Park, CA: Sage.
Allport, G. W. (1942). The use of personal documents in psychological science
(Bulletin 49). New York: Social Science Research Council.
Aron, A., Aron, E. N., & Danny, S. (1992). Inclusion of Other in Self
Scale and the structure of interpersonal closeness. Journal of Person-
ality and Social Psychology,63, 596-612.
Aspinwall, L. G., & Taylor, S. E. (1992). Modeling cognitive adaptation:
A longitudinal investigation of the impact of individual differences
and coping on college adjustment and performance. Journal of Per-
sonality and Social Psychology,63, 989-1003.
Beck, A. T. (1972). Depression: Causes and treatments. Philadelphia: Uni-
versity of Pennsylvania Press.
Blascovich, J., & Tomaka, J. (1991). The Self-Esteem Scale. In J. P. Rob-
inson, P. R. Shaver, & L. S. Wrightsman(Eds.), Measures of personality
160 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

and social psychological attitudes (pp. 115-160). New York: Academic
Press.
Block, J., & Robins, R. W. (1993). A longitudinal study of consistency
and change in self-esteem from early adolescence to early adult-
hood. Child Development,64, 909-923.
Brown, J. D., Collins, R. L., & Schmidt, G. W. (1988). Self-esteem and
direct versus indirect forms of self-enhancement. Journal of Personal-
ity and Social Psychology,55, 445-453.
Burisch, M. (1984). Approaches to personality inventory construction:
A comparison of merits. American Psychologist,39, 214-227.
Byrne, B. M. (1996). Measuring self-concept across the life span: Issues and
instrumentation. Washington, DC: American Psychological
Association.
California TaskForce on Self-Esteem. (1990). Toward a state of self-esteem:
The final report of the California task force to promote self-esteem and per-
sonal and social responsibility. Sacramento: California Department of
Education.
Campbell, A., Converse, P. E., & Rodgers, W. L. (1976). The quality of
American life: Perceptions, evaluations, and satisfactions. New York: Rus-
sell Sage.
Carmines, E. G., & Zeller, R. A. (1979). Reliability and validity assessment.
Beverly Hills, CA: Sage.
Cheek, J. M., & Melchior, L. A. (1990). Shyness, self-esteem, and
self-consciousness. In H. Leitenberg (Ed.), Handbook of social and
evaluation anxiety (pp. 47-82). New York: Plenum.
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. (1992). Revised NEO Personality Inventory
(NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI) professional man-
ual. Odessa, FL: Psychological Assessment Resources.
Demo, D. H. (1985). The measurement of self-esteem: Refining our
methods. Journal of Personality and Social Psychology,48, 1490-1502.
Diener, E. (1984). Subjective well-being. Psychological Bulletin,95,
542-575.
Gray-Little, B., Williams, V.S.L., & Hancock, T. D. (1997). An item
response theory analysis of the Rosenberg Self-Esteem Scale. Person-
ality and Social Psychology Bulletin,23, 443-451.
Greenwald, A. G., McGhee, D. E., & Schwartz, J.L.K. (1998). Measuring
individual differences in implicit cognition: The implicit associa-
tion test. Journal of Personality and Social Psychology,74, 1464-1480.
Haig, J. R. (1998). Self-esteem and depression: An exploration of a facet level
relationship of the five-factor personality model. Poster presented at the
American Psychological Society, Washington, DC.
Hansford, B. C., & Hattie, J. A. (1982). The relationship between self
and achievement/performance measures. Review of Educational
Research,52, 123-142.
Harter,S. (1985). Manual for the Self-Perception Profile for Children (revision
of the Perceived Competence Scale for Children). Denver, CO: University
of Denver.
Harter, S. (1998). The development of self-representations. In N.
Eisenberg (Ed.), Handbook of child psychology (Vol. 3, pp. 553-617).
New York: John Wiley.
Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an
attachment process. Journal of Personality and Social Psychology,52,
511-524.
Heise, D. R. (1969). Separating reliability and stability in test-retest cor-
relation. American Sociological Review,34, 93-101.
Helmreich, R., & Stapp, J. (1974). Short forms of the Texas Social
Behavior Inventory (TSBI): An objective measure of self-esteem.
Bulletin of the Psychonomic Society,4, 473-475.
Hensley, W. E., & Roberts, M. K. (1976). Dimensions of Rosenberg’s
Scale of Self-Esteem. Psychological Reports,38, 583-584.
James, W. (1890). The principles of psychology (Vols. 1 & 2). New York:
Holt.
Kitayama, S., & Karasawa, M. (1997). Implicit self-esteem in Japan:
Name letters and birthday numbers. Personality and Social Psychology
Bulletin,23, 736-742.
Leary, M. R. (1999). Making sense of self-esteem. Current Directions in
Psychological Science,8, 32-35.
Lefcourt, H. M., von Baeyer, C. L., Ware, E. E., & Cox, D. J. (1979). The
multidimensional-multiattributional causality scale: The develop-
ment of a goal specific local of control scale. Canadian Journal of
Behavioural Science,11, 286-304.
Major, B., Barr, L., Zubek, J., & Babey, S. H. (1999). Gender and self-
esteem: A meta-analysis. In W. B. Swann, J. H. Langlois, & L. A. Gilbert
(Eds.), Sexism and stereotypes in modern society: The gender science of Janet
Taylor Spence (pp. 223-253). Washington, DC: American Psychologi-
cal Association.
Markus, H. (1977). Self-schemata and processing of information about
the self. Journal of Personality and Social Psychology,35, 63-78.
Marsh, H. W. (1986). Global self-esteem: Its relation to specific facets of
self-concept and their importance. Journal of Personality and Social
Psychology,51, 1224-1236.
O’Malley, P. M., & Bachman, J. G. (1983). Self-esteem: Change and sta-
bility between ages 13 and 23. Developmental Psychology,19, 257-268.
Paulhus, D. L. (1994). Reference manual for the Balanced Inventory of Desir-
able Responding, Version 6. British Columbia, Canada: University of
British Columbia, Department of Psychology.
Paulhus, D. L. (1998). Interpersonal and intrapsychic adaptiveness of
trait self-enhancement: A mixed blessing? Journal of Personality and
Social Psychology,74, 1197-1208.
Pelham, B. W., & Swann, W. B. (1989). From self-conceptions to self-
worth: On the sources and structure of global self-esteem. Journal of
Personality and Social Psychology,57, 672-680.
Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale
for research in the general population. Applied Psychological Mea-
surement,1, 385-401.
Robins, R. W., & Beer, J. S. (1999). A longitudinal study of the adaptive and
maladaptive consequences of positive illusions about the self. Davis: Uni-
versity of California, Department of Psychology.
Robins, R. W., & John, O. P. (1997). Effects of visual perspective and
narcissism on self-perception: Is seeing believing? Psychological Sci-
ence,8, 37-42.
Robins, R. W., Norem, J. K., & Cheek, J. M. (1999). Naturalizing the
self. In L. Pervin & O. P. John (Eds.), Handbook of personality (pp.
443-477). New York: Guilford.
Robins, R. W., Trzesniewski, K. H., Tracy, J., Gosling, S. D., & Potter, J.
(1999). Self-esteem across the life span: Age and sex differences. Davis: Uni-
versity of California, Department of Psychology.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ:
Princeton University Press.
Rosenberg, M. (1979). Conceiving the self. New York: Basic Books.
Russell, J. A., Weiss, A., & Mendelsohn, G. A. (1989). Affect grid: A sin-
gle-item scale of pleasure and arousal. Journal of Personality and
Social Psychology,57, 493-502.
Sandvik, E., Diener,E., & Seidlitz, L. (1993). Subjective well-being: The
convergence and stability of self-report and non-self-report mea-
sures. Journal of Personality,61, 317-342.
Savin-Williams, R. C., & Jaquish, G. A. (1981). The assessment of ado-
lescent self-esteem: A comparison of methods. Journal of Personality,
49, 324-336.
Scheier, M. F., Carver, C. S., & Bridges, M. W. (1994). Distinguishing
optimism from neuroticism (and trait anxiety, self-mastery, and
self-esteem): A reevaluation of the Life Orientation Test. Journal of
Personality and Social Psychology,67, 1063-1078.
Tafarodi, R. W., & Swann, W. B., Jr. (1995). Self-liking and self-compe-
tence as dimensions of global self-esteem: Initial validation of a
measure. Journal of Personality Assessment,65, 322-342.
Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social psy-
chological perspective on mental health. Psychological Bulletin,103,
193-210.
Waters, E., Noyes, D. M., Vaughn, B. E., & Ricks, M. (1985). Q-sort defi-
nition of social competence and self-esteem: Discriminant validity
of related constructs in theory and data. Developmental Psychology,
21, 508-522.
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and vali-
dation of brief measures of positive and negative affect: The PANAS
scales. Journal of Personality and Social Psychology,54, 1063-1070.
Wylie, R. C. (1989). Measures of self-concept. Lincoln: University of
Nebraska Press.
Received March 29, 1999
Revision accepted December 8, 1999
Robins et al. / MEASURING GLOBAL SELF-ESTEEM 161