Prevalence of Self-Reported Depressive Symptoms
In Young Adolescents
VICTOR J. SCHOENBACH, MSPH, PHD, BERTON H. KAPLAN, PHD, EDWARD H. WAGNER, MD, MPH,
ROGER C. GRIMSON, PHD, AND FRANCIS T. MILLER, PHD
Abstract: To investigate the significance and measurement of
depressive symptoms in young adolescents, 624 junior high school
students were asked to complete the Center for Epidemiologic
Studies Depression Scale (CES-D) during home interviews. In 384
usable symptom scales, item-scale correlations (most were above
.50), inter-item correlations, coefficient alpha (.85), and patterns of
reported symptoms were reasonable. Persistent symptoms were
reported more often by Blacks, especially Black males. Prevalence
of persistent symptoms in Whites was quite close to reported figures
for adults, ranging from I per cent to 15 per cent in adolescent males
and 2 per cent to 13 per cent in adolescent females. Adolescents
Introduction
Depressive symptoms, such as feeling sad, lonely, or
worthless, are common in young adolescents, but of uncer-
tain significance.' 2 Such symptoms may occur as part of a
depressive syndrome, but otherwise are not regarded as
clinically important.34 When severe, persistent, or recur-
rent, however, symptoms may be meaningful from a social
psychiatric perspective in which they and other "behaviors
of psychiatric interest"5 can be related to social and cultural
phenomena.
The interpretation of depressive symptoms in adoles-
cents is particularly problematic, because of: controversy
about the existence, prevalence, manifestations, and diagno-
sis of depression as a pediatric psychiatric disorderi4-8;
uncertainty about possible influences of biological and other
normal developmental processes on moods and behavior9'0:
and limitations in data about the prevalence, severity, dura-
tion, and distribution of depressive symptoms in this age
group.' 2 A basic obstacle to research has been the absence
of validated measurement instruments."I The present study
investigated the feasibility of measuring early adolescent
depressive symptoms by using an adult symptom scale.
Methods
Depressive symptom data were obtained in conjunction
with a pilot study of biosocial factors in adolescent sexual
behavior.'2 Requirements of that study dictated a study
population of between 500 and 1,000 students attending a
junior high school with grades seven through nine in one
building. For the first North Carolina school district to give
From the Department of Epidemiology, University of North Carolina
School of Public Health. Address reprint requests to Dr. Victor J. Schoen-
bach, Research Assistant Professor, Department of Epidemiology, Rosenau
Hall 201-H, University of North Carolina, School of Public Health, Chapel
Hill, NC 27514. This paper, submitted to the Journal April 19, 1982, was
revised and accepted for publication January 17, 1983.
© 1983 American Journal of Public Health 0090-0036/83 $1.50
AJPH November 1983, Vol. 73, No. 11
reported persistent vegetative symptoms less often and psychoso-
cial symptoms more often. Reports of symptoms without regard to
duration were much more frequent in the adolescents, ranging from
18 per cent to 76 per cent in White males, 34 per cent to 76 per cent
in White and Black females, and 41 per cent to 85 per cent in Black
males. The results support the feasibility of using a self-report
symptom scale to measure depressive symptoms in young adoles-
cents. Transient symptoms reported by adolescents probably reflect
their stage of development, but persistent symptoms are likely to
have social psychiatric importance. (Am J Puiblic Health 1983;
73:1281-1287.)
permission for the study, a school was chosen at random
from among several having the above characteristics.
Although the school roster listed 728 names, it devel-
oped that 119 of these students had moved from the area or
had been transferred to other schools. Fifteen students not
listed but attending the target school were discovered during
door-to-door canvassing, bringing the total number of eligi-
ble subjects to 624. Of these 624, 120 (19 per cent) were not
interviewed due to refusal of the parent to give informed
consent (N = 75) or to refusal of the subject to participate (N
= 45). Information in the roster indicated that refusals were
evenly distributed by grade level and by sex. However,
interviewers reported that refusal rates were higher in appar-
ently middle class households than in either well-to-do or
lower income households, perhaps due to the sensitive
nature of the questionnaire. This bias plus the exceptional
income and educational backgrounds of the students attend-
ing the target school (see "Results") indicate that the
prevalences to be presented must be interpreted with cir-
cumspection. For an additional 120 subjects (19 per cent),
depressive symptom scales were omitted by the subject or
were judged unsuitable for analysis. Sociodemographic data
were available for this latter group of nonrespondents and
are presented in the "Results" section.
Interviews, conducted in the subject's home, included
questions on sociodemographic data, physical development,
sexual feelings, attitudes, and experience, relationships with
parents and peers, and depressive symptoms during the past
week. The interviewer administered the initial section on
household composition, age, grade level, parental education,
and other routine factors, and remained available to answer
questions. Otherwise, the subject's responses were self-
recorded and were not known to interviewer or parents.
Measures taken to assure subjects that their responses
would be kept in confidence were judged to be adequate on
the basis of a pretest phase. Seventy per cent of the mothers
and 17 per cent of the fathers completed a shorter question-
naire at the same time.
The adolescent questionnaire, less the depressive symp-
tom component, was worded for sixth-grade reading level
1281
SCHOENBACH, ET AL.

TABLE 1-Per Cent of Usable Depression Scales by Characteristics of Subjects Interviewed

White Black White Black

Males Males females Females Total
(N = 153) (N = 104) (N = 155) (N = 92) (N = 504)
Characteristics

Age (years)
12 62t 61 77 81 73
13 77 44 90 83 73
14 86 74 96 76 86
15-16 67 63 83 74 70
Grade
7 70 44 81 78 69
8 80 61 94 86 81
9 78 72 90 70 80
Natural Father
In Home 79 60 86 76 79
Deceased 50t 70 1 00t 100 82
Divorced/other 74 50 91 75 70
Parental education*
Less than 12 years 64 53 94 75 65
12-15 years 74 61 88 82 77
16 years or more 86 50t 89 67t 86
TOTAL 76 57 88 78 76
*Number of years of school completed by natural parent living at home; if both natural parents present, then the higher value was
used.
tFewer than 10 subjects in this category.

and had been pretested several times with junior and senior
high school students. A validation study of the physical
development items, carried out in pediatricians' offices in
Lake County, Illinois, yielded Pearson correlations ranging
from .50 to .80 (all significant at the 5 per cent level) between
adolescent and pediatrician ratings of genital hair and global
rating of adolescent development.3 In the present data,
correlations of height with weight, pubertal development
with age, and sexual experience with physical maturity and
sexual attitudes allayed initial concerns that the adolescents
might not respond to the questionnaire in a serious manner.
Spearman correlations between the adolescent's report of
his/her parents' years of schooling and the parental report,
where available, ranged from .77 to .85 for four race-sex
groups.
Depressive symptom data were collected using a self-
report version of the Center for Epidemiologic Studies
Depression Scale (CES-D)'4 with slightly altered response
choices. Four additional items were appended but were not
used for most analyses. Development of the CES-D, its
validation with adults, and its use with community and
national samples have been described. 14.17.18 Use of the scale
with adolescents has not been reported by others.
The response choices used for this study ("Rarely or
None of the Time," "Some or a little of the time," "A Lot of
the time," or "Most or all of the time"), like those in the
standard form of the CES-D, refer to the duration or
frequency of recurrence of the symptom during the past
week. Although subject responses are usually combined into
a single score based on response values of zero through three
with three being the maximally depressive response, Craig
and Van Natta'9.20 have argued that overall scores disregard
the different clinical significance of persistent and transient
symptoms. Therefore the present study has examined the
item responses themselves and compared both "persistent"
and "present" symptoms across groups. For the 16 negative
items in the 20-item CES-D, "'persistent" refers to the
response "Most or all of the time." For the four positive
items, "persistent" refers to the response "Rarely or none
of the time." "Present" refers to any response other than
the non-depressive one ("Rarely or none of the time" for a
negative item, "Most or all of the time" for a positive item).
The response alternatives are hereinafter abbreviated as
"Rarely," "Some," "A lot," "Most."
Results
Subject Characteristics
Table I presents response rates (proportions of usable
depression scales among subjects interviewed) by selected
subject characteristics. Characteristics associated with high-
er response rates were female sex, White race (among
males), and parental education (except White females [WF]).
Many of the 84 subjects who skipped the depression scale
entirely also omitted other pages that contained a large
number of items, possibly due to poor reading ability. These
differential response rates should be kept in mind in inter-
preting the results. As discussed later, it is likely that the
respondents constituted a mentally healthier subset of the
student enrollment.
Among the 504 subjects interviewed, females tended to
be somewhat younger than males. White males (WM) were
underrepresented in seventh grade. Substantial proportions
of subjects were not living with both natural parents. Fully
half of White subjects but only 6 per cent of Black subjects
came from homes in which at least one natural parent had
four or more years of school beyond high school. Half of the
Blacks and only 12 per cent of the Whites came from homes
in which neither natural parent had completed high school.
Virtually the same statements could be made about the 384
subjects with satisfactory depressive symptom data (Table
2). These dramatic differences in the socioeconomic levels of
the White and Black students reflect the marked race-linked
socioeconomic differences in the school's enrollment, rein-
forced by the apparently higher refusal rate in middle class
households of both races.
Internal consistency of CES-D responses was generally
adequate. Item-scale Spearman correlations were above .50
for most items and above .40 for all but three. Values of

1 282 AJPH November 1983, Vol. 73, No. 11

 SELF-REPORTED DEPRESSIVE SYMPTOMS IN ADOLESCENTS

TABLE 2-Percentage Distribution of Sociodemographic Characteristics and .85 overall. This result is virtually identical to the value
of 384 Subjects with Satisfactory Depressive Symptom Data of .85 reported by Radlofft4 for White adults in community
surveys.
% Whites % Blacks Selected item-pairs* thought likely to be highly correlat-
Characteristics (N = 253) (N = 131)
ed were identified in advance of the data analysis. Spearman
Gender (N = 253) (N = 131) correlation coefficients for pairs of items of like polarity
Male 46 45 ranged from .21 to .57 (median = .39), and all were signifi-
Female 54 55 cant, with p < .001. Coefficients for pairs of opposite items
Age last birthday (years) (N = 253) (N = 131) ranged from -.09 to -.34 (median = .24), with a p < .05 for
12* 24 30
13 36 31 all pairs, p < .001 for all but two. Weighted Kappas22 with
14 34 30 linear disagreement weights ranged from .19 to .49 (median
15** 6 10 .30) for pairs of items of like polarity; Kappas for pairs of a
Grade in school (N = 253) (N = 131) negative item with the inverted response value of a positive
7 29 39
8 34 35 item were lower (.07 to .22, median .15).
9 37 26 Symptom Frequencies
Natural parents in subject's household (N = 250) (N = 129)
Both parents 67 31 Table 3 displays the response distributions for the entire
Father only 6 5 sample for each CES-D item. The general pattern is consist-
Mother only 27 64 ent with that expected for a healthy population. A majority
(Father deceased) 4 13 of respondents checked "Rarely" for the negative items.
Parental educationt (N = 241) (N = 106)
Less than 12 yrs 10 47 The distribution for most items descends rapidly toward the
12-15 yrs 37 47 more depressive responses, with "Most" chosen by fewer
16 yrs or more 53 6 than 12 per cent of respondents for any item. For the positive
Household incomett (N = 198) (N = 80)
Less than $6,000 5 55 items, the distributions are flatter, with somewhat larger
$6,000-$1 5,000 21 36 percentages reporting the most depressive response ("Rare-
$15,000-$25,000 26 9 ly"). Response distributions for WM and WF correspond
More than $25,000 48 0 closely to the aggregate distributions. Those for BM and BF
are similar but have higher percentages of depressive re-
*Includes 1 11-year-old female.
-Includes 4 16-year-old males. sponses.**
tNumber of years of school completed by natural parent living at home; if both natural "Persistent" symptoms were for the most part equally
parents lived with the subject, then the higher value was used. [Subject's report, unless only rare for WM and WF, but more frequent among Blacks,
a parent's report was available.]
ttlncome from all sources, including governmental assistance, as reported on the
parental questionnaire (1978 dollars). *AIl pairs from among the items "I felt that I could not shake off the blues
. ." (3), "I felt depressed" (6), "I thought that my life had been a failure"
(9), "I felt sad" (18), "I felt that I was just as good as other people" (4), "I
Cronbach's coefficient alpha2l calculated for the 270 subjects was happy" (12), "I enjoyed life" (16), "I felt great" (24). This was the only
who answered every item were above .80 in each race-sex analysis that included one of the items appended to the CES-D ("I felt great"
[24]).
group (.83, .83, .89, .86 for WM, WF, BM and BF, respec- **Race-sex specific response percentages for all items are available from
tively, based on 92, 109, 25, and 44 subjects, respectively) the first author on request.

TABLE 3-Distributions of Responses to Items In CES-Depression Scale, Entire Sample

Rarely or Some or A A Lot of

None of the Little of the the Most or All
Time Time Time of the Time
(0) (1) (2) (3)
-Item % % % % N

Bothered by things 42 44 11 4 379

Not feel like eating 56 30 10 4 377
Could not shake blues 50 34 12 4 373
Just as good as others* 12 28 32 29 375
Trouble keeping mind on things 31 40 22 8 376
Felt depressed 45 37 13 6 360
Everything was an effort 25 38 25 11 372
Hopeful about the future* 15 26 34 25 366
Life had been a failure 66 24 6 4 376
Felt fearful 54 31 10 5 376
Sleep was restless 53 26 15 6 370
Was happy* 9 19 36 36 377
Talked less than usual 40 35 16 9 372
Felt lonely 53 31 10 6 376
People were unfriendly 52 31 13 5 368
Enjoyed life* 11 17 32 40 370
Had crying spells 69 17 10 5 373
Felt sad 53 32 11 5 374
Felt people disliked me 53 31 10 6 375
Could not get going 48 33 11 7 374

Note: * indicates a positive item.

AJPH November 1983, Vol. 73, No. 11 1283

SCHOENBACH, ET AL.

TABLE 4-Symptom Persistence (Per Cent of Subjects Reporting Symptoms at Level Three)

White Black White Black All

Males Males Females Females Groups
Item % % % %

Bothered by things 3 0 2 11 4
Not feel like eating 2 2 4 9 4
Could not shake blues 3 7 4 4 4
Just as good as others* 11 10 10 17 12
Trouble keeping mind on things 5 7 8 14 8
Felt depressed 5 2 5 9 6
Everything was an effort 11 11 11 13 11
Hopeful about the future* 15 21 13 15 15
Life had been a failure 2 13 2 4 4
Felt fearful 4 11 2 6 5
Sleep was restless 6 14 3 7 6
Was happy* 7 18 5 11 9
Talked less than usual 6 20 4 17 9
Felt lonely 7 14 2 6 6
People were unfriendly 5 5 4 4 5
Enjoyed life* 5 17 8 19 11
Had crying spells 2 9 5 4 5
Felt sad 1 4 7 6 5
Felt people disliked me 4 9 6 6 6
Could not get going 3 14 5 13 7
Minimum number of subjects (110) (51) (131) (64) (360)

Note: *indicates that response values have been reversed.

especially BM (Table 4). The racial difference in percentages
reporting "persistent" symptoms is substantial. For symp-
toms reported as "present" (Table 5), the racial difference,
particularly among females, is attenuated, and a female
excess exists among Whites (among Blacks there is a slight
male excess). Moreover, there is a strong correspondence
across race-sex groups, in that symptoms reported relatively
more often by one group tend to be reported more often by
other groups.
Comparison of the CES-D responses of the young
adolescents with those of adults was carried out using
previously unpublished data collected during the first wave
of interviews from community surveys in Washington Coun-
ty, Maryland and Kansas City, Missouri. The surveys have
been reported by Comstock and Helsing'5 and Radloff. 14 For
reasons of space, the percentages for males and females
have been averaged with equal weighting. The summariza-
tion conceals sex differences in symptom prevalence among
adults and adolescents, particularly for some "present"
symptoms and sex-related behaviors, such as crying. In
Whites, persistent symptoms (Table 6) are neither consis-
tently more nor consistently less frequent in the young
adolescents than in the adults. The adolescents are less apt
to report persistent vegetative symptoms (loss of appetite,
restless sleep, could not get going) and more apt to report
symptoms with a developmental connotation (not just as

TABLE 5-Symptom Presence (Per Cent of Subjects Reporting Symptoms at Level One, Two, or Three)

White Black White Black All

Males Males Females Females Groups
Item % % % % %

Bothered by things 44 61 63 69 58
Not feel like eating 30 55 46 51 44
Could not shake blues 36 56 56 56 50
Just as good as others* 68 71 75 68 71
Trouble keeping mind on things 62 71 78 64 69
Felt depressed 45 58 59 63 55
Everything was an effort 76 80 72 72 75
Hopeful about the future* 68 85 76 74 75
Life had been a failure 19 51 34 46 34
Felt fearful 34 52 44 64 46
Sleep was restless 37 52 46 63 47
Was happy* 60 79 57 73 64
Talked less than usual 50 78 57 69 60
Felt lonely 38 66 47 46 47
People were unfriendly 49 51 48 46 48
Enjoyed life* 53 75 58 64 60
Had crying spells 18 41 37 35 31
Felt sad 39 59 51 42 47
Felt people disliked me 39 55 52 43 47
Could not get going 45 59 56 49 52
Minimum number of subjects (110) (51) (131) (64) (360)
Note: *indicates that response values have been reversed.

1 284 AJPH November 1983, Vol. 73, No. 11

 SELF-REPORTED DEPRESSIVE SYMPTOMS IN ADOLESCENTS

TABLE 6-Symptom Persistence in Young Adolescents and Community Adults (Per Cent of Subjects Reporting Symptoms at Level Three)

Whites Blacks

Community
Community Adults Adults
Young Young
Adolescents 18-24 25-64 65+ Adolescents 25-64 only

Bothered by things 2 4 5 6 6 10
Not feel like eating 3 9 6 9 5 13
Could not shake blues 4 5 4 5 6 5
Just as good as others* 11 5 9 9 14 6
Trouble keeping mind on things 7 14 7 6 11 5
Felt depressed 5 7 5 7 6 10
Everything was an effort 11 18 16 17 12 22
Hopeful about the future* 14 11 15 24 18 22
Life had been a failure 2 3 2 2 9 4
Felt fearful 3 3 2 2 8 3
Sleep was restless 5 12 11 14 11 9
Was happy* 6 3 6 8 15 10
Talked less than usual 5 7 4 4 18 6
Felt lonely 5 4 4 8 10 9
People were unfriendly 5 3 2 1 5 4
Enjoyed life* 7 4 6 8 18 6
Had crying spells 4 2 1 1 7 0
Felt sad 4 3 3 5 5 4
Felt people disliked me 5 2 1 1 7 4
Could not get going 4 8 7 9 14 6
Minimum number of subjects (110) (144) (707) (146) (51) (93)
Note: *indicates that response values have been reversed.
Source: Adult data from Washington County, Md., and Kansas City, Mo., community surveys ;kindly provided by Lenore Sawyer Radloff, Center for Epidemiologic Studies (NIMH). The
surveys are described in Radloff14.

good, fearful, others unfriendly, disliked, crying). Among
Blacks, most symptoms are reported as persistent more
often by the adolescents, with a few marked exceptions
(Table 6). For "present" symptoms (Table 7), the adolescent
percentages exceed those among adults, often dramatically,
in all race-sex groups and for nearly all items.
Discussion
Interpretation of the data presented above should take
into account the large numbers of nonparticipants (120 of
624) and of skipped or inadequately completed depressive
symptom scales (120 of 504), particularly for the Black
males. The likelihood that reading difficulties and lower
socioecc nomic status were associated with failure to com-
plete the symptom scale would probably lead to underreport-
ing of symptom prevalence, in view of Rutter, et al's report
of an association between reading difficulties and persistent
psychiatric disorder2 as well as the many reports'52324 of
associations between depressive symptoms and lower socio-
economic status. The data for White females are presumably
least affected.
It is more difficult to judge the impact of the refusals,
particularly since the study was principally concerned with
adolescent sexual attitudes and behavior rather than with
depression. Undersampling of middle class households prob-
ably overstates symptom prevalence. Of likely greater im-
portance, though, is the unusual nature of the school's
enrollment, with its high proportion of well-to-do families
and exaggerated socioeconomic disparity between Whites
and Blacks. The disparity could well have affected the
emotional experience of the students, especially of the
poorer ones. Therefore generalizability of the actual preva-
lences presented above is uncertain.
With regard to the measurement question, the analyses
of internal consistency support the reliability of the symptom
responses (although response tendencies could elevate such
consistency). The variation in response frequencies across
items, the correspondence across race-sex groups, and the
systematic differences between groups all suggest that the
adolescents were responding in a serious manner. "Persist-
ent" symptoms were relatively uncommon among the ado-
lescents, the prevalence in the Whites being no greater than
that reported for adults. Persistent symptoms were reported
equally often by White male and White female adolescents,
whereas the latter reported "present" symptoms more of-
ten. A similar pattern has been observed in adults.20 The
high rates of "present" symptoms among the adolescents, if
not a peculiarity of this study population, would extend
Craig and Van Natta's observation that the reporting of
"present" symptoms declines with age.20
Thus, in a number of regards, the adolescents' CES-D
responses are a plausible measure of depressive symptom
prevalence. The different relationship to age of reported
"persistent" and "present" symptoms may reflect the influ-
ence of developmental factors on emotional experience or on
reporting, possibly jeopardizing comparability of CES-D
scores across age groups. Future studies of depressive
symptoms should attempt to distinguish transient and non-
transient symptoms.
A possible implication of the present data is that if, as
suggested by Kovacs and Beck,4 depressive disorders in
children and adolescents have similar manifestations to their
adult expression, and if, as suggested by Craig and Van
Natta,'9 reported persistent depressive symptoms are more
indicative of clinical depression than are transient symp-
toms, then clinical depression is probably not more common
in early adolescence than in adults. Our attempt to interpret
CES-D responses in such a way as to identify a depressive
syndrome along the lines of the Research Diagnostic Crite-
ria25 for major depressive disorder yielded an estimated

AJPH November 1983, Vol. 73, No. 11 1 285

SCHOENBACH, ET AL.

TABLE 7-Symptom Presence in Young Adolescents and Community Adults (Per Cent of Subjects Reporting Symptoms at Level One, Two, or Three)

Whites Blacks

Community
Community Adults Adults
Young Young
Adolescents 18-24 25-64 65+ Adolescents 25-64 only

Bothered by things 54 45 38 23 65 33
Not feel like eating 38 35 22 23 53 31
Could not shake blues 46 36 20 14 56 26
Just as good as others* 72 30 19 13 70 14
Trouble keeping mind on things 70 60 39 26 68 35
Felt depressed 52 50 35 21 60 41
Everything was an effort 74 62 44 37 76 46
Hopeful about the future* 72 44 37 37 80 38
Life had been a failure 26 14 11 8 49 15
Felt fearful 39 29 15 10 58 18
Sleep was restless 42 52 42 42 57 39
Was happy* 59 35 29 26 76 37
Talked less than usual 54 42 24 16 74 36
Felt lonely 42 39 21 24 56 30
People were unfriendly 48 25 12 6 49 21
Enjoyed life* 55 29 20 20 69 25
Had crying spells 28 16 9 7 38 9
Felt sad 45 45 27 23 50 33
Felt people disliked me 46 25 10 6 49 18
Could not get going 50 53 39 28 54 30
Minimum number of subjects (110) (144) (707) (146) (51) (93)

Note: Indicates that response values have been reversed.

Source: Adult data from Washington County, Md., and Kansas City, Mo., community surveys kindly provided by Lenore Sawyer Radloff, Center for Epidemiologic Studies (NIMH). The
surveys are described in Radloff14.

prevalence of 2.9 per cent.26 The few other published
reports27,28 on population prevalence of depressive disorder
in young adolescents yield similar figures, although these
estimates are also derived from small numbers of affected
youngsters. A 1948 statewide sample of Minnesota ninth-
graders29 detected Minnesota Multiphasic Personality Inven-
tory profiles suggestive of clinical depression in 4 per cent of
boys and 1.5 per cent of girls. In contrast, reporting of
symptoms per se was much more frequent, both in our study
and in others. ' 2
The high rate of missing data and likely reading difficul-
ties among the Black males call for a conservative interpreta-
tion of their apparent high symptom prevalence. This high
prevalence is broadly consistent, however, with the findings
of higher depressive symptom prevalence in Black
adults,'52324'30although the relative ranking of males and
females has varied. In general, the racial difference in adults
has been attributed to socioeconomic differences, in that it
was eliminated by adjustment for income and/or education.
In one case,24 adjustment eliminated the significant racial
difference in mean CES-D scores but not in the prevalence of
high CES-D scores. Since Blacks below the poverty line had
the highest adjusted rate for any group in that study, not all
of the high depressive symptom prevalence could be ac-
counted for by reported income itself.24
Our data do not permit us to distinguish between the
relative importance of race and socioeconomic status varia-
bles, because of insufficient overlap between the socioeco-
nomic backgrounds of the two racial groups. Since the
majority of the Black students in the present study came
from households with 1978 income less than $6,000 and in
which no natural parent had 12 years of schooling, socioeco-
nomic depreviation is a prime suspect either alone or by
contrast to the relative affluence of the White students.
Other factors could also be involved. For example, the
junior high school studied is located in a White neighbor-
hood, with the Black students being bused in from else-
where. Rosenberg has reported lower self-esteem ratings for
adolescents in such "dissonant contexts."3' It is possible
that the Black males may have borne the brunt of racial
antagonism, which could account for their high symptom
rates.
In summary, we have presented evidence that persistent
depressive symptoms in young adolescents appear to be
measurable and meaningful. Future research is needed to
validate and refine measures of depressive symptoms and
their correlates, to assess duration and recurrence through
longitudinal observation, to ascertain behavioral conse-
quences, and to identify modifiable factors that contribute to
the experience of persistent or recurrent symptoms. A
particularly salient question is the relationship of reported
symptoms to subsequent suicide attempts or involvement in
other hazardous behaviors. Preventive and supportive inter-
ventions might include modification of school, home, and
community environments to reduce conflicting or untimely
demands, instruction in coping skills,32 and provision of
additional supports especially for high-risk groups.
REFERENCES
1. Albert N, Beck AT: Incidence of depression in early adolescence: a
preliminary study. J Youth Adol 1975; 4:301-306.
2. Rutter M, Graham P, Chadwick OFD, Yule W: Adolescent turmoil: fact
or fiction? J Child Psychol Psychiat 1976; 17:35-56.
3. Akiskal HS, McKinney WT Jr: Overview of recent research in depres-
sion. Arch Gen Psychiatry 1975; 32:285-305.
4. Kovacs M, Beck AT: An empirical-clinical approach toward a definition
of childhood depression. In: Schulterbrandt JG and Raskin A (eds):
Depression in Childhood. New York: Raven, 1977, pp 1-25.
5. Leighton AH: Psychiatric disorder and social environment. In: Kaplan
BH (ed): Psychiatric Disorder and the Urban Environment. New York:
Behavioral Publications, 1971, pp 20-67.

1 286 AJPH November 1983, Vol. 73, No. 11


6. Conners CK: Classification and treatment of childhood depression and
depressive equivalents. In: Gallant DM and Simpson GM (eds): Depres-
sion: Behavioral, Biochemical, Diagnostic and Treatment Concepts. New
York: Spectrum, 1976, pp 181-204.
7. Graham P: Depression in pre-pubertal children. Develop Med Child
Neurol 1974; 16:340-349.
8. Lefkowitz MM, Burton N: Childhood depression: a critique of the
concept. Psychological Bulletin 1978; 85:716-726.
9. Hamburg BA: Early adolescence: a specific and stressful stage of the life
cycle. In: Coelho GV, Hamburg DA, and Adams JE (eds): Coping and
Adaptation. New York: Basic Books, 1974, pp 101-124.
10. Hays SE: Strategies of psychoendocrine studies of puberty. Psychoneur-
oendocrinology 1978; 3:1-15.
11. Lipsitz J: Growing up Forgotten: A Review of Research and Programs
Concerning Early Adolescence. Lexington, MA: Lexington Books, 1977.
12. Bauman KE, Udry JR: Subjective expected utility and adolescent sexual
behavior. Adolescence 1981; 16:527-535.
13. Morris NM, Udry JR: Validation of a self-administered instrument to
assess stage of adolescent development. J Youth Adol 1980; 9:271-280.
14. Radloff LS: The CES-D scale: a self-repprt depression scale for research
in the general population. Applied Psychological Measurement 1977;
1:385-401.
15. Comstock GW, Helsing KJ: Symptoms of depression in two communi-
ties. Psychological Med 1976; 6:551-563.
16. Sayetta RB, Johnson DP: Basic data on depressive symptomatology,
United States, 1974-75. US Department of Health, Education, and
Welfare, Public Health Service, National Center for Health Statistics,
Series I 1, Number 216, (DHEW Pub. [PHS] 80-1666), 1980.
17. Weissman MM, Sholomskas D, Pottenger M, Prusoff BA, Locke BZ:
Assessing depressive symptoms in five psychiatric populations: a valida-
tion study. Am J Epidemiol 1977; 106:203-214.
18. Hankin JR, Locke BZ: The persistence of depressive symptomatology
among prepaid group practice enrollees: an exploratory study. Am J
Public Health 1982; 72:1000-1007.
19. Craig TJ, Van Natta PA: Presence and persistence of depressive symp-
toms in patient and community populations. Am J Psychiatry 1976;
133:1426-1429.
20. Craig TJ, Van Natta PA: Influence of demographic characteristics on two
measures of depressive symptoms. Arch Gen Psychiatry 1979; 36:149-
154.
21. Nunnally JC: Psychometric Thoery. New York: McGraw-Hill, 1967.
22. Cohen J: Weighted kappa: nominal scale agreement with provision for
scaled disagreement or partial credit. Psychological Bulletin 1968; 70:213-
220.
I ERRATUM
In: Landrigan PJ, Powell KE, James LM, Taylor PR: Paraquat and marijuana: epidemiologic risk
assessment. Am J Public Health 1983; 73:784-788.
In the footnote on page 784, column 1, the statement says "This paper, submitted to the Journal
September 30, 1983, was revised and accepted for publication February 1, 1983." The date of
submission was September 30, 1982-not 1983.
AJPH November 1983, Vol. 73, No. 11
SELF-REPORTED DEPRESSIVE SYMPTOMS IN ADOLESCENTS
23. Warheit GJ, Holzer CE, Schwab JJ: An analysis of social class and racial
differences in depressive symptomatology: a community study. J Health
& Social Behavior 1973; 14:291-299.
24. Eaton WE, Kessler LG: Rates of symptoms of depression in a national
sample. Am J Epidemiol 1981; 114:528-538.
25. Spitzer RL, Endicott J, Robins E: Research diagnostic criteria: rationale
and reliability. Arch Gen Psychiatry 1978; 35:773-782.
26. Schoenbach VJ, Kaplan BH, Grimson RC, Wagner EH: Use of a
symptom scale to study the prevalence of a depressive syndrome in young
adolescents. Am J Epidemiol 1982; 116:791-800.
27. Leslie SA: Psychiatric disorder in the young adolescents of an industrial
town. Brit J Psychiat 1974; 125:113-124.
28. Kashani J, Simonds JF: The incidence of depression in children. Am J
Psychiatry 1979; 136:1203-1204.
29. Hathaway SR, Monachesi ED: Adolescent personality and behavior.
Minneapolis: University of Minnesota, 1963.
30. Frerichs RR, Aneshensel CS, Clark VA: Prevalence of depression in Los
Angeles County. Am J Epidemiol 1981; 113:691-699.
31. Rosenberg M: Dissonant context and adolescent self-concept. In: Dragas-
tin SE and Elder GH (eds): Adolescence in the life cycle. New York:
Wiley, 1975.
32. Hamburg DA, Adams JE, Brodie HKH: Coping behavior in stressful
circumstances: some implications for social psychiatry. In: Kaplan BH,
Wilson RN, Leighton AH (eds): Further explorations in social psychiatry.
New York: Basic Books, 1976, pp 158-175.
ACKNOWLEDGMENTS
The authors wish to express their appreciation to J. Richard Udry, PhD,
and Karl E. Bauman, PhD, of the University of North Carolina at Chapel Hill
for collecting the data reported in this paper as part of their research on
adolescent sexual behavior. Lenore S. Radloff of the Center for Epidemiolog-
ic Studies, National Institute of Mental Health, provided helpful consultations
and kindly made available previously unpublished data from the Center for
Epidemiologic Studies' community survey in Washington County, Maryland
and Kansas City, Missouri. Ralph C. Patrick, PhD, Martha L. Smith, Marion
E. Schoenbach, MSPH, Sherman A. James, PhD, and Dr. Udry provided
consultation and assistance throughout the project. We are also grateful to
Michel A. Ibrahim, MD, PhD, for helpful editorial suggestions and to Patricia
G. Taylor for typing the manuscript through its various drafts. Dr. Schoen-
bach was supported by Public Health Service Traineeships from the National
Institute of Mental Health, under grant MH 13661. Data collection was
supported by grant No. ROI-HD12806-01 from the National Institute of Child
and Human Development to the Carolina Population Center.
I
1287