Journal of Personality and Social Ps~x~hology Copyright 1996 by the American Psychological Association, Inc.
1996, Vol. 70, No. I, 141-156 0022-3514/96/$3.00
Self-Concept Clarity:
Measurement, Personality Correlates, and Cultural Boundaries
Jennifer D. Campbell, Paul D. Trapnell, Steven J. Heine, Ilana M. Katz,
Loraine E Lavallee, and Darrin R. Lehman
University of British Columbia
Self-concept clarity (SCC)
references a structural aspect oftbe self-concept: the extent to which self-
beliefs are clearly and confidently defined, internally consistent, and stable. This article reports the
SCC Scale and examines (a) its correlations with self-esteem (SE), the Big Five dimensions, and
self-focused attention (Study l ); (b) its criterion validity (Study 2); and (c) its cultural boundaries
(Study 3 ). Low SCC was independently associated with high Neuroticism, low SE, low Conscien-
tiousness, low Agreeableness, chronic self-analysis, low internal state awareness, and a ruminative
form of self-focused attention. The SCC Scale predicted unique variance in 2 external criteria: the
stability and consistency of self-descriptions. Consistent with theory on Eastern and Western self-
construals, Japanese participants exhibited lower levels of SCC and lower correlations between SCC
and SE than did Canadian participants.
Within the last couple of decades, psychologists' view of the
self-concept has undergone a dramatic transformation (Markus
& Wurf, 1987). Early researchers treated the self-concept as a
unitary, monolithic entity--a stable, generalized view of the
self--and typically focused their research efforts on a single as-
pect of the self-concept, self-esteem. Contemporary research-
ers, in contrast, rely on a multifaceted, dynamic construal in
which the self-concept is defined as a cognitive schema--an or-
ganized knowledge structure that contains traits, values, epi-
sodic and semantic memories about the self and controls the
processing of self-relevant information (e.g., Greenwald & Prat-
kanis, 1984; Kihlstrom & Cantor, 1984; Kihlstrom et al., 1988;
Markus, 1977).
The current conceptualiTation allows a distinction between the
contents of the self-concept and its structure. The contents can
be usefully subdivided into knowledge components--Who/What
am I?--and evaluative components--How do I feel about myself?.
Examples of knowledge components include beliefs about one's
specific attributes (e.g, traits, physical characteristics), as well as
roles, values, and personal goals. Evaluative components include
the positivity of specific self-beliefs and self-esteem, a global self-
evaluation that is the product of viewing "the self" as an attitude
object. Structural characteristics of the self-concept refer to how
the knowledge components or specific self-beliefs are organized.
For example, Linville ( 1985, 1987) coined the term
self-complex-
Jennifer D. Campbell, Paul D. Trapnell, Steven J. Heine, Ilana M.
Katz, Loraine E Lavallee, and Darrin R. Lehman, Department of Psy-
chology, University of British Columbia, Vancouver, British Columbia,
Canada.
This research was funded by a grant from the Social Science and Hu-
manities Research Council.
Correspondence concerning this article should be addressed to Jen-
nifer D. Campbell, Department of Psychology, University of British
Columbia, 2136 West Mall, Vancouver, British Columbia, Canada
V6T l Z4.
ity
to represent the number of different or independent dimensions
that underlie the organization. Donahue and her associates
(Donahue, Robins, Roberts, & John, 1993) have focused on a
different aspect of complexity: the extent to which these dimen-
sions are integrated. Another structural variable can be found in
Showers's (1992) work on
compartmentalization, the
extent to
which positive and negative self-beliefs reside in different
dimensions.
The present article is concerned with another structural aspect
of the self-concept, namely,
self-concept clarity (SCC;
Campbell,
1990; Campbell & Lavallee, 1993). SCC is defined as the extent to
which the contents of an individual's self-concept (e.g., perceived
personal attributes) are clearly and confidently defined, internally
consistent, and temporally stable. A couple of points may serve to
bring the clarity construct into sharper focus.
First, clarity overlaps with a number of other, more tradi-
tional constructs. One construct with obvious overlap is that of
identity (achievement, status, integration, etc.). Identity, how-
ever, has a much richer and more complex set of elements than
clarity (e.g., Adler, 1959; Allport, 1961; Baumeister, 1986; Er-
ikson, 1959; Marcia, 1980), characteristics that render the
identity construct rather difficult to assess empirically. The lit-
erature also contains a plethora of overlapping constructs that
have a narrower focus than clarity. For example, Rosenberg's
( 1965 ) notion of self-concept stability focuses on the temporal
stability of self-beliefs, whereas the constructs of role variability
(Block, 1961a) and self-consistency (Gergen & Morse, 1967)
address the internal consistency of self-beliefs. In addition, the
Hogan Personality Inventory Identity scale (Hogan, 1986) and
the Identity Integration scale (O'Brien & Epstein, 1988) focus
on generalized certainty about the self, combined with manifes-
tations of certainty such as decisiveness and well-defined long-
term goals.
Second, clarity is a characteristic of people's beliefs about
themselves (i.e., their self-concepts). It is mute with respect to
the accuracy of those beliefs and therefore does not necessarily
141
142 CAMPBELL ET AL.
imply self-knowledge in the sense of insight or awareness of
one's behavioral potentials (Wicklund & Eckert, 1992). A per-
son could hold highly articulated self-beliefs that one might ar-
gue, on the basis of behavior, are inaccurate.
Clarity and the other aforementioned structural variables
(e.g., self-complexity) are theoretically independent of the
contents of the self-concept. That is, any particular set of self-
beliefs could, in principle, be organized with varying degrees of
complexity or be held with different levels of confidence and
stability. Campbell ( 1990; Campbell & Fehr, 1990; Campbell,
Chew, & Scratchley, 1991; see also Baumgardner, 1990), has
demonstrated, however, a connection between clarity and global
self-esteem, an evaluative component of the contents. Although
high-self-esteem people have positive, well-articulated beliefs
about the self, the prototypic low-self-esteem person does not,
in contrast, have a well-defined negative view of the self. The
self-concepts of low-self-esteem people are better described as
evaluatively neutral and, more important, are characterized by
relatively high levels of uncertainty, instability, and inconsis-
tency (i.e., low clarity).
The demonstration of the self-esteem-clarity relation has
proven to be theoretically useful in understanding many of the
contradictory, puzzling, and paradoxical findings in the self-es-
teem literature. Although lack of space precludes a review here,
a number of self-esteem phenomena (e.g., low self-esteem plas-
ticity, Brockner, 1984) that are not easily understood by refer-
encing differences in the positivity of self-beliefs are cogently
and parsimoniously explained by the fact that high- and low-
self-esteem people differ in the clarity or certainty of their self-
beliefs (e.g., Baumeister, 1993; Blaine & Crocker, 1993; Camp-
bell & Lavallee, 1993; Setterlund & Neidenthal,. 1993; Tice,
1993).
The validity of these explanations, however, hinges on the as-
sumption that clarity, like self-esteem, can be conceptualized as
a relatively stable trait. Although plausible, this assumption is
currently unsupported. The empirical research examining the
self-esteem-clarity relation has generally pretested participants
on standard self-report scales of self-esteem and subsequently
tested them on a number of unobtrusive measures of clarity
(e.g., the extremity, internal consistency, and temporal stability
of their self-descriptions); the evidence for the relation lies in
the convergent associations that have emerged across studies
and measures (Campbell & Lavallee, 1993 ).
Research Aims
One goal of the present article was to ascertain if self-concept
clarity is a relatively stable individual difference or trait and,
more specifically, if it is one that can be reliably captured in
self-reports. That is, would individuals whose self-perceptions
lacked ,~larity be sufficiently aware of their status to report it,
and would these self-reports remain stable over time? A further
aim, given the established empirical connection between clarity
and self-esteem, was to determine if trait measures of clarity
and self-esteem could be reliably differentiated from one an-
other in self-reports.
This first goal does not in any way deny the validity or utility
of conceptualizing Clarity, self-esteem, or any other component
of the self-concept as a state (e.g., the dynamic or working self-
concept, Markus & Wurf, 1987; identity images, Schlenker,
1985 ). In Conley's (1984) terms, self-concept clarity is a "self-
opinion"--a type of individual difference that is useful to assess
both as a state and as a trait because, although it is susceptible
to environmental influences, it also exhibits high levels of tem-
poral stability (Conley argued that, with time lags longer than
l 0 years, self-opinions are less stable than intelligence and per-
sonality traits). This dual nature of individual differences in
self-opinion is, in fact, highlighted in the postulated interplay
among clarity, temporary levels of self-esteem, and the working
self-concept in theoretical accounts of the self-esteem literature
(e.g., Campbell & Lavallee, 1993).
A second goal was to explore the nomological network of the
clarity construct by examining its relations with other person-
ality dispositions. The other traits we examined were based on
the Big Five model of personality, which is currently the model
of choice for measurement of normal-range trait dimensions,
owing to its demonstrated comprehensiveness (Briggs, 1992;
Goldberg, 1990, 1993; Wiggins & Trapnell, in press). We antic-
ipated that clarity would be related to Neuroticism, Conscien-
tiousness, and Extraversion (to avoid redundancy, we postpone
describing the theoretical rationale for these relations until the
General Discussion) and that empirically demonstrating these
relations might yield benefits comparable to those that have
emerged in the self-esteem literature.
A third, related goal was to examine the relation between clarity
and chronic attention to the self (i.e., self-consciousness). It has
been argued that higher levels of self-attention should result in a
more clearly articulated self-schema (e.g., Buss, 1980). A number
of studies have supported this proposition (e.g., Kernis & Gran-
nemann, 1988; Nasby, 1985, 1989; Scheier, Buss, & Buss, 1978)
and, consistent with this generally positive view of self-attention,
there are also studies showing that self-consciousness may buffer
the adverse effects of stressful life events (e.g., Mullen & Suls,
1982; Suls & Fletcher, 1985; cf. Frone & McFarlin, 1989). A more
negative view of self-attention, however, is suggested by the close
associations among self-focus, negative affective states, and psy-
chopathology. In his review, Ingram (1990) found elevated levels
of self-focus to be such a universal feature of psychopathological
states and disorders that he concluded '~he apparent ubiquitous-
ness of this process in disorder makes it difficult to find anything
dysfunctional that is not accompanied by increased self-focused
attention" ( p. 156 ). Other research suggests that negative mood is
associated with, and may in fact cause, increases in self-focused
attention (Wood, Saltzberg, & Goldsamt, 1990; cf. Salovey, 1992).
These two lines of evidence imply a "self-absorption paradox"
with respect to psychological adjustment: Despite their higher lev-
els of psychological distress and pathology, self-focused individuals
possess a clearer, better articulated self-structure than do less self-
focused individuals. Although this pattern is consistent with the
"sadder but wiser" phenomenon noted in the literature (Alloy &
Abramson, 1988), it stands in direct contradiction to the positive
clarity-self-esteem relation (Campbell, 1990), as well as to re-
search showing that people high in self-consciousness are more
susceptible to external cues and feedback (Hull, Van Treuren,
Ashford, Propsom, & Andrus, 1988).
In an effort to shed some light on these contradictory findings,
we examined clarity's relations with a number of scales that
measure dispositional self-focused attention. The most widely
SELF-CONCEPT CLARITY 143
used taxonomy ofdispositional self-consciousness (Buss, 1980)
is operationalized by the Public and Private Self-Consciousness
Scales (Fenigstein, Scheier, & Buss, 1975 ). The Public scale taps
attentiveness to or awareness of one's external self (the self as a
social object), whereas the Private scale taps attentiveness to or
awareness of one's inner thoughts, feelings, and attitudes. The
unidimensionality of the Private scale, however, has been ques-
tioned by studies (Burnkrant & Page, 1984; Gould, 1986; Len-
nox & Welch, 1987; Mittal & Balsubramanian, 1987; Piliavin
& Charng, 1988) showing two distinct factors (but see Britt,
1992; Bernstein, Teng, & Garbin, 1986, for arguments against
splitting the Private scale into two factors), factors usually re-
ferred to as Self-Reflectiveness (PRIV-SR) and Internal State
Awareness (PRIV-ISA). PRIV-SR consists of six items, includ-
ing "I'm always trying to figure myself out" and "I reflect about
myself a lot." PRIV-ISA consists of four items, including "I'm
aware of the way my mind works when I work through a prob-
lem" and "I'm alert to changes in my mood?' We examined
clarity's relation with the two factors separately (in addition to
the total scale) because, although low clarity could plausibly
lead individuals to engage in frequent self-analysis (to try to
"figure themselves out"), it could also be associated with a per-
ceived lack of awareness of one's internal states, moods, and
feelings.
We also used a second taxonomy of dispositional self-atten-
tiveness recently proposed by Trapnell and Campbell (1995).
This model distinguishes two motivational states presumed to
direct attention toward the self: curiosity and anxiety. The Re-
flection-Rumination Questionnaire (RRQ) was developed to
measure individual differences in these two motivationally dis-
tinct varieties of self-consciousness. The Reflection scale mea-
sures a voluntary, emotionally positive form of self-focus asso-
ciated with epistemic interest in the self(e.g., "I love exploring
my 'inner' self"). The Rumination scale measures an involun-
tary, emotionally negative form of self-focus associated with
threat or uncertainty ("Sometimes it is hard for me to shut off
thoughts about myself"). Scale development and validation is
reported elsewhere (Trapnell & Campbell, 1995), but we note
here that Reflection and Rumination (a) are relatively indepen-
dent forms of self-consciousness ( r < .25 ), (b) correlate equally
highly with the PRIV-SR subscale (rs > .50), (c) correlate in
opposite directions with the PRIV-ISA subscale, and (d) are as-
sociated with different Big Five dimensions--Reflection is re-
lated to Openness (r > .60 ), and Rumination is related to Neu-
roticism (r >.60).
A final research goal was to explore cultural boundaries of
the SCC construct. Markus and Kitayama ( 1991 ) argued that
people from Western and Eastern cultures have strikingly
different construals of the self. The Western (or independent)
construal views the self as an independent, self-contained, au-
tonomous entity. The self has a unique and stable configuration
of internal attributes that governs the individual's behavior
across situations. The Eastern (or interdependent) construal
views the self as an interconnected entity that is most meaning-
ful when cast within an interpersonal context. Internal attri-
butes are seen as situation specific and, to the extent that there
is a recognition of invariant attributes, they tend not to be
viewed as diagnostic of the self but as aspects to be controlled or
regulated to achieve harmony with others.
SCC is a construct that appears closely aligned with the West-
ern construal of the self--the independent, but not the interde-
pendent, self should possess a clearly defined, consistent set of
internal attributes that remain stable across situations. To in-
vestigate this possibility, we measured clarity and self-esteem in
Canadian and Japanese samples with the expectations that (a)
Canadians would exhibit higher average levels of clarity, and
(b) because clarity captures characteristics that are normatively
prescribed only in Western cultures, a more pronounced asso-
ciation between clarity and self-esteem would be found among
Canadians.
Overview of the Research
In Study 1 we report the development of the SCC Scale and
examine its associations with self-esteem, the Big Five person-
ality dimensions, and different forms of self-focused attention.
Assessing the convergent, divergent, and construct validity of a
scale consists of demonstrating that the pattern of correlations
with measures of other constructs conforms to (a) the definition
or conceptual specification of the construct and (b) theoretical
propositions regarding psychological antecedents and conse-
quences of the construct (Wiggins, 1973). Therefore, this study
expands the nomological network of the clarity construct but
also serves to establish the SCC Scale's validity to the extent
that the pattern of correlations adheres to theoretical expecta-
tions. A second aspect of validity is external or criterion validity,
the extent to which test scores relate to nontest manifestations
of the construct in a theoretically consistent manner (Wiggins,
1973). We examined the SCC Scale's external validity in Study
2, in which we used SCC scores to predict the internal consis-
tency and temporal stability of self-descriptions. In Study 3 we
explored cultural boundaries of the clarity construct by con-
trasting SCC scores and the SCC-self-esteem correlation in Ca-
nadian and Japanese samples.
Study 1
Method
Participants
Three samples of undergraduates enrolled in introductory psychol-
ogy classes at the University of British Columbia completed a battery of
personality instruments. In Study la we tested 471 participants whose
ages ranged from 17 to 44 (M = 19. I 1,
SD
= 2.51 ). Of those who indi-
cated their sex, 167 were male and 295 were female. Study lb included
608 participants, whose ages ranged from 17 to 42 (M = 19.37,
SD =
2.55 ); 262 were male and 328 were female. In Study l c there were 465
participants, whose ages ranged from 17 to 48 (M = 19.61,
SD = 3.15 );
138 were male and 312 were female. All participants received extra
course credit for participation.
Measures
Studies I a- lc were conducted in each of 3 consecutive years. Because
they differed in the content of the test batteries administered, we note
below which studies contained each measure. Unless otherwise noted,
participants responded to all scale items on 5-point Likert scales an-
chored by
strongly disagree
( l ) and
strongly agree ( 5 ).
SCC.
The Study la battery contained an initial pool of 40 items
designed to measure clarity. Some of the items assessed the perceived
144 CAMPBELL ET AL.
certainty, temporal stability, and internal consistency of self-beliefs,
whereas others tapped fairly direct ramifications of SCC, such as deci-
siveness and clearly articulated goals. The initial 40-item pool was gen-
erated both by culling (and sometimes revising) items from published
scales measuring related constructs (e.g., Rosenberg's [ 1965 ] Self-Con-
cept Stability Scale) and by constructing new items. The test batteries
administered in Studies lb and lc contained a 20-item subset of the
original item pool.
Self-esteem. We included two measures of self-esteem in Study I a:
the Rosenberg ( 1965 ) Self-Esteem Scale, which taps generalized, global
feelings of self-worth, and the Texas Social Behavior Inventory (TSBI;
Helmreich, Stapp, & Ervin, 1974), which focuses on social self-esteem
or perceived social competence. Studies lb and lc included only the
Rosenberg scale.
Big Five measures. In Studies l b and 1 c we measured the Big Five
personality factors with the NEO-FFI (Costa & McCrae, 1989), a 60-
item short form version of the NEO Personality Inventory (Costa &
McCrae, 1985 ). Reliability and validity evidence for the FFI is impres-
sive, with the five 12-item FFI scales accounting for approximately 75%
of the variance in convergent criteria of the full NEO-PI (Costa &
McCrae, 1989).
The Positive and Negative Affect Schedule (PANAS; Watson, Clark,
& Tellegen, 1988) consists of 20 adjectives, 10 that measure the disposi-
tional mood dimension of negative alfectivity (NA; e.g., nervous, hos-
tile, ashamed), and 10 that measure positive affectivity (PA; e.g., proud,
enthusiastic, inspired). Watson and Tellegen ( 1985 ) identified NA and
PA as the most superordinate dimensions of emotional experience.
Strong relations between NA and Neuroticism and between PA and Ex-
traversion (Meyer & Shack, 1989; Watson & Clark, 1992) have led some
theorists to propose that Neuroticism and Extraversion be relabeled
Negative Emotionality and Positive Emotionality (Tellegen, 1985; Tel-
legen & Wailer, in press).
Scales measuring selected aspects of neuroticism also were adminis-
tered. We included the Taylor Manifest Anxiety Scale (TMAS; Taylor,
1953 ) in Studies I a and I c, the Beck Depression Inveqtory ( BDI; Beck,
1967) in Studies 1 b and I c, and a short version of the Repression-Sen-
sitization Scale (R-S: Byrne, 1961 ) in Study lc. The TMAS is a true-
false questionnaire consisting of 50 items from the Minnesota
Multiphasic Personality Inventory (MMPI; Hathaway & McKinley,
1951) selected on the basis of face validity to measure manifest trait
anxiety. The scale is alternatively considered one of the best available
markers of NA or neuroticism (Watson & Clark, 1984). We used a 20-
item short form version of the TMAS developed by Bendig (1956) in
Study Ic instead of the original 50-item scale. The BDI is a 21-item
measure of depression that has been widely used with college popula-
tions (e.g., Alloy & Abramson, 1982). The R-S Scale consists of true-
false MMPI items selected on the basis of face validity to operationalize
the trait of repression-sensitization, the tendency, to cognitively avoid
versus approach threatening perceptions, thoughts, and feelings (Bell &
Byrne, 1978; Gordon, 1957). We used a 43-item short form version
constructed by Paulhus and Levitt (1983), who reported alpha reliabil-
ities and a factor structure based on the combined-sex sample closely
resembling that reported for the original R-S scale.
Finally, the Study lc battery included an 18-item short form version
of the Need for Cognition Scale (NCOG; Cacioppo, Petty, & Kao,
1984), which measures the tendency to engage in and enjoy thinking
(Cacioppo & Petty, 1982), a construct that overlaps with the Big Five
Openness dimension. The 18-item short form, constructed from factor
and psychometric analyses of the original scale, may be a less problem-
atic measure of the construct than the longer version, which has been
criticized for its multidimensionality.
Self-consciousness. We administered the Self-Consciousness Scale
(Fenigstein et al., 1975 ) in all three studies. Factor analyses in our own
samples consistently replicated the results reported by other investiga-
tors (e.g., Burnkrant & Page, 1984): The Public scale was unidimen-
sional, but the Private scale divided into the two expected factors: PRIV-
SR (6 items) and PRIV-ISA (4 items). We derived PRIV-SR and PRIV-
ISA subscores by summing responses to the 6 and 4 items, respectively.
In Studies lb and lc, participants also completed the RRQ (Trapnell &
Campbell, 1995 ).
Other measures. We administered the Marlowe-Crowne Social De-
sirability Scale (MC; Crowne & Marlowe, 1960), 25 true-false items,
in Studies la and lc to evaluate the extent to which responses to the
SCC Scale were associated with socially desirable response tendencies.
In Study la, participants also rated themselves on 16 adjectives, each
representing a label for 1 of the 16 sectors of Wiggins's (1979) circum-
plex model of interpersonal traits. We used these pretest ratings in Study
2, in which we examined the temporal stability of participants' self-
descriptions.
Results
We focus first on the data relevant to the development and
reliability of the SCC Scale. Then we examine the correlations
between the SCC Scale and measures of the other constructs.
Finally, we report regression analyses, which provide a portrait
of the independent personality concomitants of clarity.
SCC Scale
From the initial pool, we selected 20 items on the basis of
internal consistency and lack of item redundancy. This original
20-item scale (reported in Campbell, Katz, Lavallee, & Trap-
nell, 1991) contained three highly intercorrelated factors
(average r = .52) reflecting generalized clarity, goal-direct-
edness, and decisiveness. We subsequently decided that a
shorter, unidimensional scale focusing specifically on the epis-
temological status of the self-concept was preferable on both
theoretical and practical grounds. Our reasoning was that al-
though decisiveness and goal directedness are certainly logical
manifestations of clarity, both represent broad constructs of
their own and share many features with a large number of other
important trait constructs (e.g., dominance, conscientiousness,
vocational identity). Because inclusion of such items might un-
necessarily complicate interpretations of the SCC Scale and its
associations, we opted to keep the item set brief and as concep-
tually univocal as possible. We therefore retained only those 12
items that loaded on the first factor. The retained items tap per-
ceived internal consistency and temporal stability of self-beliefs,
along with more generic self-certainty items. ~ Table 1 shows the
12 items selected for the final scale, along with the average factor
loadings (principal-components analysis) and the average cor-
rected item-total correlations. Table 2 shows the scale statistics
for each of the three separate samples.
Internal consistency. The average alpha reliability coeffi-
cient was.86 (Table 2). The combined-sample corrected item-
total correlations ranged from .35 to .66, with an average item-
total correlation of.54 (Table 1 ). An examination of the com-
Item 12 contains both decisiveness and goal-directedness content,
but we retained it because it loaded most heavily on the first factor. For
investigators who have used the original 20-item scale, the correlations
reported in Studies I and 2 are highly similar to those obtained with the
earlier version of the scale.
SELF-CONCEPT CLARITY
Table 1
Structural Coefficients and Descriptive Statistics for Self-Concept Clarity Scale Items
145
Item PC1 rit M SD
1. My beliefs about myself often conflict with one another." .75 .66 3.57
2. On one day I might have one opinion of myself and on another day I might have a different opinion." .68 .57 3.07
3. I spend a lot of time wondering about what kind of person I really am." .65 .56 3.07
4. Sometimes I feel that I am not really the person that I appear to be. a .70 .61 3.16
5. When I think about the kind of person I have been in the past, I'm not sure what I was really like." .62 .52 3.55
6. I seldom experience conflict between the different aspects of my personality. .43 .35 3.08
7. Sometimes i think I know other people better than I know myself." .58 .48 3.39
8. My beliefs about myself seem to change very frequently." .76 .66 3.61
9. IfI were asked to describe my personality, my description might end up being different from one day
to another day." .68 .58 3.58
10. Even if I wanted to, I don't think I would tell someone what I'm really like." .49 .41 3.42
11. In general, I have a clear sense ofwho I am and what I am. .66 .57 3.61
12. It is often hard for me to make up my mind about things because I don't really know what I want." .56 .47 2.97
0.97
1.14
1.20
1.14
1.04
1.07
1.17
1.02
1.12
1.18
0.99
i.21
Note. N = 1,544. Scale ranges from ! (strongly disagree) to 5 (strongly agree). PC 1 = loading on first unrotated principal component; rlt ---
item-total correlation.
"Reverse-keyed item.
corrected
bined-sample interitem correlation matrix revealed that all of
the items were positively intercorrelated, with correlations
ranging from. 10 to.58, and a mean interitem correlation of.34
(Table 2). The alphas, item-total correlations, and interitem
correlations all indicate high levels of internal consistency, and
there was very little variability in these statistics across the three
samples.
Evidence for a general factor. In addition to the reliability
data, factor analyses of the 12 items yielded strong evidence of a
single, general factor. Principal-components and maximum-like-
lihood factor analyses in each sample typically yielded only one
factor with an eigenvalue greater than 1.0. Although some analy-
ses did yield a second factor with eigenvalues of approximately
1.0, the items that comprised this second factor replicated neither
across samples nor across solutions. In each sample, every item
loaded positively on the first unrotated factor, with a minimum
item loading of .34, and the average correlation between these
loadings in the three samples was .84. In the combined sample,
all items had loadings greater than .45 (Table l ). A scree test also
supported the unidimensionality of the scale. In the combined
sample, the first factor accounted for 41% of the total variance,
with the second factor accounting for only 8%.
Test-retest reliability. We readministered the scale to 155
Study la participants after a 4-month interval and to 61 Study
1 c participants after a 5-month interval. Despite the relatively
long intervals, the test-retest correlations of .79 and .70, respec-
tively (Table 2), revealed high levels of temporal stability.
Scale distributions. Table 2 shows the SCC Scale means,
standard deviations, and information relevant to sex and age
differences for each of the three samples. The overall scale mean
decreased across the three samples, indicating that in the 3-year
time period participants exhibited somewhat lower levels of
clarity. Although these differences could reflect a form of scale
unreliability (i.e., sensitivity to variations in the number or type
of other scale items included in the test battery), there are rea-
sons to doubt this interpretation. First, in Studies lb and lc we
constructed five different forms of each test battery in which we
systematically varied the location of the SCC items within the
test battery. Analyses of variance indicated that SCC scores
were invariant over the five forms of the batteries (both Fs < 1 ).
Second, scores on the other scales also indicated that the three
samples tended to exhibit generally higher levels of psychologi-
cal distress across time. For example, self-esteem scores de-
clined, and depression and neuroticism scores increased, over
the 3-year period. Taken together, these data suggest that the
decrease in SCC scores across studies may reflect an historical
trend in our samples rather than scale unreliability. 2
2 The decrease in SCC means over the 3-year period may reflect the
steadily increasing proportion of Asian students at the University of
Table 2
Scale Statistics: Study 1
N Total Males Females Interitem r
Study' Total Fem M SD M SD M SD x Range Test-retest r
la 464 295 42.12 8.19 42.93 8.29 41.67 8.11 .34 .15-.57 .79 .86
lb 595 328 39.68 8.16 40.18 7.71 39.27 8.49 .34 .10-.58 .86
lc 456 312 38.86 8.06 40.24 7.52 38.25 8.23 .32 .10-.56 .70 .85
Note. The test-retest correlation in Study la is based on 155 participants and a 4-month interval. The test-retest correlation in Study I c is based
on 61 participants and a 5-month interval. Fern = females.