Content uploaded by John G. Orme
Author content
All content in this area was uploaded by John G. Orme on Dec 27, 2018
Content may be subject to copyright.
FACTORIAL AND DISCRIMINANT VALIDITY OF THE CENTER
FOR EPIDEMIOLOGICAL STUDIES DEPRESSION (CES-D) SCALE
JOHN
G.
ORME
University
of
Chicago
JANET
REIS
AND ELICIA
J.
HERZ
Center
for
Health Services and Policy Research
North western University
The factorial and discriminant validity of the Center for Epidemiological
Studies Depression (CES-D) scale was examined for a sample of
116
parents
who were praticipating in family support programs designed to prevent child
abuse and neglect. Participants’ self-reports of depressive symptoms as
measured by the CES-D were analyzed in relation to their self-esteem
(measured with the Rosenberg Self-Esteem scale) and state and trait anxiety
(measured with Spielberger’s State-Trait Anxiety Inventory). Factorial
validity was adequate, and results indicated a moderate correlation between
the CES-D and self-esteem and state anxiety. However, a high correlation
was obtained between the CES-D and trait anxiety, which suggests that the
CES-D measures in large part the related conceptual psychological domain
of
predisposition for anxiousness.
The Center for Epidemiological Studies Depression scale (CES-D) (Radloff, 1977)
is a widely used self-report measure designed to provide an index
of
the number and
frequency of depressive symptoms. Since its publication in 1977, at least
50
published
articles have reported use
of
the CES-D as a measure
of
depression, and these include
its use in epidemiological as well as in clinical studies (e.g.,
Brown
&
Lewinsohn, 1984;
Eaton
&
Kessler, 1981; Fava, 1983; Husaini, Neff, Newbrough,
&
Moore, 1982).
The extensive application of the CES-D suggests that it is perceived by many clini-
cians and researchers as a reliable and valid instrument, and, indeed, various aspects
of the reliability and validity of the CES-D have been examined extensively. Results
indicate that the CES-D exhibits good internal consistency reliability across diverse
population subgroups and relatively good short-term test-retest reliability (Aneshensel,
Clark,
&
Frerichs, 1983; Fava, 1983; Radloff, 1977; Roberts, 1980; Ross
&
Mirowsky,
1984). Moreover, the CES-D differentiates fairly well between those in treatment and
those not in treatment for psychological and psychiatric problems in general, and depres-
sion in particular, and CES-D scores have shown improvement for treated populations
(e.g., Boyd, Weissman, Thompson,
&
Myers, 1982; Husaini, Neff, Harrington, Hughes,
&
Stone, 1980; Radloff, 1977; Weissman, Sholomskas, Pottenger, Prusoff,
&
Locke,
1977). Finally, the CES-D correlates well with other scales designed to measure depres-
sion (e.g., Radloff, 1977; Weissman et al., 1977).
Although various aspects of the reliability and criterion and convergent validity
of the CES-D have been examined, and these results appear very promising, to date
the discriminant validity of the CES-D has been less well documented and supported.
Preparation
of
this article was supported in part by National Istitute
of
Mental Health Grant
IT32-MH17053
to John G. Orme and by grants
from
the Illinois Department
of
Children and Family Services and the Pitt-
way Charitable Trust, Chicago, Illinois, to the Center
for
Health Services and Policy Research
of
Northwestern
University.
Correspondence should be addressed to Janet Reis, Center
for
Health Services and Policy Research, North-
western University,
629
Noyes St., Evanston, Illinois
60201.
28
CES-D
Factorial and Discriminant Validity
29
Previous assessments
of
this issue suggest that the
CES-D
is less successful in measur-
ing the single construct
of
depression than originally thought. Vernon and Roberts (1981),
for example, report that the correlation of the CES-D with the total score of the
Demoralization scale (a broad-band measure of psychological distress that measures anx-
iety, depression, self-esteem, and feelings of helplessness and hopelessness) was as high
as possible given the reliabilities of the two measures. These and other similar results
(e.g., Weissman et al., 1977) suggest that the CES-D might be viewed best as a measure
of nonspecific psychological distress and not interpreted solely as a measure of depression.
The growing popularity of the CES-D and the questions that have been raised as
to its discriminant validity underscore the importance of further exploration of this issue.
In the present study, the relationships among the CES-D and three important, related
psychological constructs
-
self-esteem, trait anxiety, and state anxiety
-
will be examined
in order to assess the CES-D’s discriminant validity.
METHOD
Subjects
The CES-D, along with measures of self-esteem, trait anxiety, and state anxiety,
was completed by
116
individual parents who were participating in family support pro-
grams designed to prevent child abuse and neglect. Data collection took place in five
Illinois service delivery sites. All sites are situated in communities with high rates of
unemployment, poverty, and negative indices of health status. The battery of tests were
administered to the majority of parents by their third program visit. A total of 89%
of the sample were women, 46% were Black, and
50%
were Caucasian. All subjects
were at least 16 years old, and the mean age was 21.30
(SD
=
7.16).
Measures
The CES-D is a 20-item self-report measure that includes questions that pertain
to a wide range of depressive symptoms. Respondents are asked to rate the frequency
of occurrence of each symptom in the past week on a 4-point scale, which ranges from
rarely or none of the time (less than 1 day) to most or all of the time (5-7 days). The
potential range of scores is from
0
to 60; higher scores indicate greater depression. A
score of 16 or greater typically is employed as a cut-off that indicates clinical depression
(e.g., Boyd et al., 1982).
The Rosenberg Self-Esteem scale (Rosenberg, 1969, a well-validated (cf. Eppinger
&
Lambert, 1983) 10-item self-report scale, was used to measure self-esteem. Each item
is a declarative statement for which respondents are asked whether they strongly agree,
agree, disagree, or strongly disagree. The potential range of scores is from
0
to 30, and
higher scores indicate
less
self-esteem.
Spielberger’s State-Trait Anxiety Inventory (STAI) (Spielberger, Gorsuch,
&
Lushene, 1970) was employed to measure state and trait anxiety because it is a well-
validated measure
of
these constructs (cf. Beutler
&
Crago, 1983; Roberts, Aronoff,
Jensen,
&
Lambert, 1983). The STAI is a self-report measure comprised of a 20-item
trait anxiety scale and a 20-item state anxiety scale. For the trait anxiety scale, respondents
are required to rate the frequency with which they usually feel particular anxiety symp-
toms, and for the state anxiety scale respondents are required to rate the intensity of
their anxiety at the time
of
testing. On both scales respondents are required to provide
their ratings on a 4-point scale, and each scale has a potential range of scores from
0
to
60.
Higher scores indicate greater anxiety.
30
Journal
of
Clinical Psychology, January
1986,
Vol.
42,
No.
I
RESULTS
Reliability
As
shown in Table
1,
the internal consistency reliability of each of the four measures
is adequate and consistent with previous results (e.g., Eppinger
&
Lambert, 1983; Radloff,
1977; Spielberger et
al.,
1970). Coefficients alpha ranges from
-80
for self-esteem to
.88
for the CES-D.
Table
1
Scale Reliabilities, Means, and Standard Deviations
(N
=
116)
CES-D Self-esteem Trait anxiety State anxiety
~
cy
M
SD
.88 .so .81 .87
14.31 6.47 12.85 10.30
10.26 4.98 5.95 5.17
Factorial Validity
If the CES-D has good factorial validity, its items should correlate relatively highly
with the CES-D total score and should correlate more highly with its total score than
with self-esteem, trait anxiety, or state anxiety total scores. Such an analysis of item-
total correlations is equivalent to
a
confirmatory multiple-group factor analysis with
units on the main diagonal of the inter-item correlation matrix; the resultant item-total
correlations are equal to factor loadings obtained from such
a
factor analysis (Hudson,
1982; Nunnally, 1978, p. 394).
As
shown in Table 2, the correlated item-total correlations between the CES-D items
and the CES-D total score are all positive. Moreover, with the exception of correlations
for three items (items
2,
4,
and
7),
all of the corrected item-total correlations for the
CES-D exceed .30, and the mean correlation, computed using Fisher’s
Z
transforma-
tion, is .52.
Table
2
Correlations
of
CES-D
Items
with Total Scale Scores
Measures
CES-D items CES-D Self-esteem Trait anxiety State anxiety
__
1
2
3
4
5
6
7
8
9
10
11
~
.5
I
.28
.68
.25
.59
.61
.13
.31
.62
.46
.33
.29
.20
.40
.I7
.38
.38
.06
.I2
.43
.27
.31
.42
.28
.56
.13
.51
.46
.I1
.I6
.43
.43
.51
.29
.10
.34
.16
.46
.39
.
.02
.06
.39
.20
.I8
CES-D
Factorial and Discriminant Validity
31
Table
2
Correlations
of
CES-D Items with Total Scale Scores
Measures
CES-D items CES-D Self-esteem Trait anxiety State anxiety
12 .57 .36 .41 .42
13
.51 .36 .36 .17
14 .68 .44 .52 .26
15
16
17
18
19
.37
.62
.63
.70
.57
.29
.38
.31
.50
.52
.3
1
.39
.49
.55
.52
.12
.38
.23
.36
.2f
20
.49 .43 .44 .I7
Also as shown in Table
2,
the correlations between the CES-D items and the total
scores for self-esteem, trait, and state anxiety are less than the correlations between these
items and the CES-D total score, with only two exceptions (items
2
and
11).
Moreover,
the mean item-total correlation of the CES-D items, as computed using Fisher’s
Z
transformation, is greater for the CES-D
(52)
than for self-esteem
(.34),
trait anxiety
(.41),
or state anxiety
(.25).
However, the CES-D items do load on the self-esteem, trait,
and state anxiety scales, and for some items these correlations are relatively high, espe-
cially with trait anxiety.
Discriminant Validity
The intercorrelations among the scales are displayed in Table
3.
These data indicate
that the CES-D is correlated moderately with self-esteem and state anxiety and correlated
highly with trait anxiety. Correcting for attenuation in reliability, the disattenuated cor-
relation of
-84
between the CES-D and trait anxiety indicates further the especially high
correlations between these two measures.
Table
3
Scale Intercorrelations
Self-esteem Trait anxiety State anxiety
CES-D
.58 .71 .44
Self-esteem
-
.46 .44
Trait anxiety
-
.43
Note.-For each correlation
p
<
.001.
A
stepwise multiple regression analysis was conducted that used the CES-D as the
dependent variable and self-esteem, trait, and state anxiety as independent variables in
order to determine the amount of variance in the CES-D that could be predicted by
self-esteem, trait, and state anxiety. Trait anxiety entered into the equation first (ad-
justed
R2
=
.494,
F(1,
114)
=
113.03,
p
<
.001),
followed by self-esteem (adjusted
R2
change
=
.082,
F(2,
113)
=
22.86,
p
<
.001);
state anxiety did not add a statistically
32
Journal
of
Clinical Psychology, January
1986,
Vol.
42,
No.
1
significant amount of variance beyond that accounted for by trait anxiety and self-esteem.
The total adjusted
R2
=
.582
(F(2,
113)
=
78.79,
p
c
.001).
DISCUSSION
In general, the CES-D scale achieved
a
good level of factorial validity, as evidenced
by the fact that the CES-D items are correlated more highly with the CES-D total score
than with the total scores for self-esteem, state anxiety, or trait anxiety. However, many
of the CES-D items have substantial correlations with these three related but putatively
distinct constructs, and, in particular, strong relationships were observed between the
CES-D items and trait anxiety. Furthermore, bivariate and multiple correlations between
the total score
of
the CES-D and total score measures of self-esteem, state anxiety, and
trait anxiety further underscore the fact that the CES-D does not measure solely the
delimited conceptual domain of depression, as originally intended in the development
of the measure (Radloff, 1977). Therefore, the CES-D appears to measure aspects of
self-esteem, state anxiety, and, especially, trait anxiety in addition to depression.
A
determination of whether correlations of the CES-D with other measures of
depression are higher than correlations with measures of other constructs, such as anx-
iety and self-esteem, would be desirable for a more complete examination of discrimi-
nant validity, but data from the present study do not allow
a
direct answer to this ques-
tion. Results from previous studies (cf. Radloff, 1977; Weissman et al., 1977) do sug-
gest that correlations between the CES-D and other measures of depression generally
are higher than either of the correlations between the CES-D and self-esteem or state
anxiety obtained
in
the present study. However, relative to correlations between the
CES-D and other measures of depression as obtained in these previous studies, as well
as in absolute terms, the correlation obtained here between the CES-D and trait anxiety
suggests that the CES-D measures trait anxiety as much as depression.
Further analysis of the convergent and discriminant validity of the CES-D should
be undertaken. For example, examination of the discriminant and convergent validity
of the CES-D within the context
of
a
carefully designed multi-trait, multi-method analysis
(Campbell
&
Fiske, 1959) might be especially fruitful. In the interim, the present results
recommend caution in using the CES-D as
a
measure of depression in clinical assess-
ment. Further, caution should be exercised in the use of the CES-D in the evaluation
of interventions designed to alleviate depression (e.g., Brown
&
Lewinsohn, 1984) because
it might pose
a
threat to the construct validity
of
the evaluation (Cook
&
Campbell, 1979).
REFERENCES
ANESHENSEL, C.
S.,
CLARK,
V.
A.,
&
FRERICHS, R.
R.
(1983).
BEUTLER,
L.
E.,
&
CRAGO,
M.
(1983).
Race, ethnicity, and depression:
A
confirmatory
analysis.
Journal
of
Personality and Social Psychology,
44,
385-398.
Self-report measures
of
psychotherapy outcome. In
M.
J.
Lambert,
E.
R.
Christensen,
&
S.
S.
DeJulio (Eds.),
The assessment
of
psychotherapy outcome
(pp.
453-497).
New
York:
John Wiley.
Screening
for
depression in
a
com-
munity sample: Understanding the discrepancies between depression symptom and diagnostic scales.
Ar-
chives
of
General Psychiatry,
39,
1195-1200.
A
psychoeducational approach to the treatment
of
depression:
Comparison
of
group, individual, and minimal contact procedures.
Journal
of
Consulting and Clinical
Convergent and discriminant validation by the multi-trait multi-
Quasi-experimentation: Design and analysis issues forfield set-
BOYD,
J.
H.,
WEISSMAN,
M. M.,
THOMPSON,
D.,
&
MYERS,
J.
K.
(1982).
BROWN,
R.
A.,
&
LEWINSOHN,
P.
M.
(1984).
Psychology, 52,
774-783.
CAMPBELL,
D.
T.,
FISKE, D. W.
(1959).
method matrix.
Psychological Bulletin,
56,
81-105.
COOK,
T.
D.,
&
CAMPBELL, D.
T.
(1979).
tings.
Chicago: Rand McNally.
CES-D
Factorial and Discriminant Validity
33
EATON, W. W.,
&
KESSLER, L.
G.
(1981).
EPPINGER, M.
G.,
&
LAMBERT, M. J.
(1983).
Rates of symptoms
of
depression in a national sample.
American
Journal
of
Epidemiology,
114,
528-538.
Assessment
of
habit disorders: A tripartite perspective in measur-
ing change. In M.
J.
Lambert, E. R. Christensen,
&
S.
S.
DeJulio (Eds.),
The assessment
of
psychotherapy
outcome
(pp.
356-405).
New
York:
John Wiley.
Assessing depressive symptoms across cultures: Italian validation
of
the CES-D self-
rating scale.
Journal
of
Clinical Psychology,
39,
249-251.
FAVA,
G.
A.
(1983).
HUDSON, W. W.
(1982).
HUSAINI,
B.
A., NEFF,
J.
A,, HARRINGTON,
J.
B.,
HUGHES, M. D.,
&
STONE, R. H.
(1980).
HUSAINI,
B.
A., NEFF, J. A., NEWBROUGH, J. R., &MOORE, M. C.
(1982).
The clinical measurement package.
Homewood,
IL:
Dorsey Press.
Depression in
rural communities: Validating the CES-D scale.
Journal
of
Community Psychology,
8,
20-27.
The stress-buffering
role
of
social
support and personal competence among the rural married.
Journal
of
Community Psychology,
10,
Psychometric theory
(2nd ed.). New
York:
McGraw-Hill.
409-426.
NUNNALLY,
J.
C.
(1978).
RADLOFF, L.
S.
(1977).
The CES-D scale: A self-report depression scale
for
research in the general popula-
tion.
Applied Psychological Measurement,
I,
385-401.
ROBERTS, R. E.
(1980).
Reliability
of
the CES-D scale in different ethnic contexts.
Psychiatry Research,
2,
125-134.
ROBERTS,
S.,
ARONOFF,
J.,
JENSEN,
J.,
&
LAMEXERT,
M.
J.
(1983).
Measurement
of
outcome in anxiety disorders.
In M.
J.
Lambert,
E.
R.
Christensen,
&
S.
S.
DeJulio (Eds.),
The assessment ofpsychotherapy outcome
(pp.
304-355).
New
York:
John Wiley.
ROSENBERG, M.
(1965).
Ross, C.
E.,
Components of depressed mood in married men and women: The
Center for Epidemiologic Studies depression scale.
American Journal
of
Epidemiology,
119,
997-1004.
SPIELBERGER, C. D., GORSUCH, R.
L.,
~r
LUSHENE, R. E.
(1970).
Test Manual
for
the State-Trait Anxiety
Inventory.
Palo Alto, CA: Consulting Psychologists Press.
VERNON,
S.
W.,
~r
ROBERTS, R. E.
(1981).
Measuring nonspecific psychological distress and other dimen-
sions of psychopathology.
Archives
of
General Psychiatry,
38,
1239-1247.
WEISSMAN, M. M., SHOLOMSKAS, D., POTTENGER, M., PRUSOFF,
B.
A,,
&
LOCKE,
B.
Z.
(1977).
Assessing
depressive symptoms in five psychiatric populations: A validation study.
American Journal
of
Epidemiology,
106,
203-2 14.
Society and the adolescent self-image.
Princeton: Princeton University Press.
MIROWSKY,
J.
(1984).