4855_e09_p69-76
2/4/04
1:18 PM
Page 69
JOURNAL OF WOMEN’S HEALTH
Volume 13, Number 1, 2004
© Mary Ann Liebert, Inc.
Higher Rates of Depression in Women: Role of Gender
Bias within the Family
JESSICA A. BROMMELHOFF, M.P.H.,1 KEVIN CONWAY, Ph.D.,2
KATHLEEN MERIKANGAS, Ph.D.,1,3 and BECCA R. LEVY, Ph.D.1
ABSTRACT
Objective: We sought to examine whether higher rates of depression in women than in men
can be explained partially by the artifact hypothesis, which suggests that when both sexes
have the same depressive symptoms, women are more likely than men to be diagnosed with
depression. We hypothesized that (1) this gender bias in identifying depression exists within
families and (2) family members will be more likely to attribute depressive symptoms to internal causes for women and external causes for men.
Methods: Our sample consisted of 205 adults, generated from the family members of 46
probands who participated in the Yale Family Study. To determine whether bias exists in the
family, we compared self-reports of depressive symptoms with family reports of depressive
symptoms for the same individual.
Results: As predicted, we found that compared with men, women were more likely to be
reported as depressed by a family member when they report themselves as not depressed.
We also found that family members were more likely to attribute depressive symptoms of females to internal causes. We did not, however, find any differences by gender in attribution
of depression to external causes.
Conclusions: Our findings suggest that gender bias within the family may contribute to the
higher recorded rates of depression in women.
INTRODUCTION
ences in rates of depression. Specifically, we seek
to determine whether there is a bias within families to recognize depression more often in female
rather than male family members. We also examine if there are differences in how family
members perceive the causes of depression in
male vs. female relatives. Family members frequently are selected to help individuals evaluate
their mental health, as it has been found that primary care physicians often do not discern mental disorders in their patients.8–11 Additionally,
clinicians and researchers frequently measure de-
B
EGINNING IN ADOLESCENCE AND CONTINUING
throughout the entire life span, women are
more likely than men to be diagnosed with depression.1,2 Studies have shown that the prevalence of depression among women is between
one and a half and three times more than the
prevalence among men.3–6 For nearly two centuries, scientists have tried to find the cause of
this difference.7 Our study examines whether reporting bias may contribute to the gender differ1Department
of Epidemiology and Public Health, Yale University, New Haven, Connecticut.
Institute on Drug Abuse, Bethesda, Maryland.
3National Institute of Mental Health, Bethesda, Maryland.
2National
69
4855_e09_p69-76
2/4/04
1:18 PM
Page 70
70
pression through proxy reports from family
members, especially in studies that include elderly people, children, or those who are cognitively impaired.12
Explanations for differential rates of depression
in women
Some contemporary theories attribute the difference in rates of depression between men and
women to social causes, specifically that women
may have a greater sensitivity to stressful life
events and traumas.13,14 According to the Vulnerability-Stress Model, women’s social role
makes them more susceptible to depression.15
This Model posits that because the role of caretaker, which women are expected to assume, is
not awarded much value in today’s society, women may feel lower self-worth than men, leading
to emotional vulnerability.15,16 Several studies,
however, have found results that contradict the
Vulnerability-Stress Model. For example, Kendler
et al.17 found that the difference in prevalence of
depression by gender is due neither to differences
in experiencing stressful life events nor to differential overall sensitivity to these events.
One frequently speculated reason for the differential rates of depression is the biological differences between men and women. Some researchers believe that women are particularly
prone to depression because of events associated
with the reproductive cycle. Examples of such
events include depression associated with oral
contraceptives, the luteal phase of the menstrual
cycle, the postpartum period, and menopause.18
The importance of these findings in explaining
gender differences in rates of depression has been
debated, however, as the scientific evidence is inconsistent.13,19 Furthermore, because these differences are seen across much of the life-span, it is
unlikely that biological differences fully explain
the differences in rates of depression.
An alternative explanation for the differential
rates of depression is the artifact hypothesis. This
suggests “that the rate of depression is equal
among men and women but that women express
and report more symptoms, seek help more frequently, and are subject to sex biases in diagnosis, thereby providing a false elevation in the
measurement of the rate of depression in women.”6(p91) Almost all studies examining the artifact hypothesis have concentrated on how women express and report symptoms. These studies
BROMMELHOFF ET AL.
found conflicting results about whether there is
any significant difference between how men and
women perceive depression.20,21
A few studies have examined the aspect of the
artifact hypothesis that concerns sex bias in diagnosis in a clinical setting.22–25 For two of these
studies exploring possible gender bias among primary care physicians,22,23 patients were asked to
self-report on the presence and severity of depressive symptoms. Subsequently, the researchers
reviewed each patient’s medical records to ascertain the presence or absence of a clinician’s diagnosis for depression. Both studies found that
being female increased the likelihood of the clinician’s diagnosing the patient with depression.
Furthermore, in the study conducted by Bertakis
et al.,22 when the results were stratified by depression scores, they found that women with
high scores were more likely to be diagnosed with
depression than men with high scores, and women with low scores were more likely to be diagnosed with depression than men with low
scores. Thus, women were more likely than men
to get a false positive diagnosis for depression.
Along similar lines, Borowsky et al.24 found that
physicians were more likely to be aware of depression in female patients than in male patients,
and Stoppe et al.25 found that this bias extends
into old age.
The current study builds on these findings in
a number of ways. First, we examine family member reports rather than physician reports, which
is advantageous because physicians may be less
familiar with the long-term history of patients. To
the best of our knowledge, no study has yet compared rates of agreement between family proxy
reports and self-reports by gender. Second, we
look at the types of causes to which family members attribute depression. Attribution theory describes how people make inferences about the
causes of negative events and behaviors in themselves and others.26
Some studies have found that the way depressed people make attributions is different
from how nondepressed people make attributions, in that an attributional style that explains
negative events by internal causes is likely to
characterize someone who is depressed.27,28 As it
pertains to gender, one study found that females
tend to believe that their depression is internally
caused, whereas men are more likely to attribute
their depression to external causes. In other
words, women are more likely to hold themselves
4855_e09_p69-76
2/4/04
1:18 PM
Page 71
71
GENDER BIAS AND DEPRESSION
responsible for their depressed mood, and men
tend to blame negative factors within their environment, such as unemployment or divorce, for
their depressed mood.29 We predict a similar dynamic might occur in families, such that individuals will be more likely to attribute depression in
female family members to internal causes and,
thus by implication, hold women more responsible for their mental illness than men. Furthermore, individuals will be more likely to attribute
depression in male family members to external
causes, therefore holding men less responsible for
their depression.
This study has two hypotheses. Our first hypothesis is that when an index self-reported no
depression, but a family member reported depression in the same individual, the index is significantly more likely to be female than male. Our
second hypothesis explores one of the reasons for
a potential gender bias in families. We predict
that depressive symptoms are more likely to be
attributed to external causes in men and internal
causes in women.
cluding 21 men and 25 women, were all between
the ages of 29 and 55, with an average age of 39
years. Twenty-two of these probands were initially recruited from the CMHC (14 males, 8 females), and 24 were initially recruited from the
general population (7 males, 17 females).
Every study participant fulfilled two roles: as
an index and as an informant. Each person selfreported about his or her own depression and
then reported about each adult family member
also in the study. The index self-reported on his
or her own depression through the Schedule for
Affective Disorders and Schizophrenia–Lifetime
Version (SADS-L),30 which is a semistructured interview. The informant reported on a family
member’s depression through an interview regarding family history, the Family History–Research Diagnostic Criteria interview (FH-RDC).31
These two instruments were used in the current
study because they were the measures of depression used for the YFS that our sample was
taken from.
Measures
MATERIALS AND METHODS
Sample population
The subjects in this study were all participants
in the Yale Family Study (YFS), conducted by the
Genetic Epidemiology Research Unit (GERU) at
Yale University in New Haven, CT. One of the
initial objectives of the YFS was to examine the
familial aggregation of substance abuse and psychiatric disorders, particularly anxiety and depression. The probands from the YFS were recruited from both outpatient treatment facilities
at the Connecticut Mental Health Clinic (CMHC)
in New Haven and from the general population
in the New Haven area via a random digit dialing procedure.
From the 260 probands in the YFS, we selected
all probands who had at least one family member age 60 or over who had been interviewed for
the study. We selected probands with older family members so that all ages of adults were represented in this study. This resulted in a sample
of 205 adults for this study (Table 1), who were
generated from the family members of the 46
probands who were recruited for the YFS. These
people included the parents, spouses, siblings,
and children over age 18. The 46 probands, in-
SADS-L: Depression section. In the YFS, a trained
clinician interviewed each subject, using the lifetime version of the semistructured SADS-L,
which was modified to obtain DSM-III-R criteria.32 To measure self-reported depression in this
study, we used the responses from three questions in the SADS-L about major depressive disorder (MDD). The depressive mood and ideation
section of the SADS-L has been shown to have a
joint interview (two raters observing the same interview) reliability coefficient of 0.95, a test-retest
reliability coefficient of 0.78, and internal consistency of 0.87.30 Across all diagnostic categories,
the interrater agreement is high, with a kappa of
0.83.33
FH-RDC: Depression section. For each index
case, at least one family member was interviewed
about his or her own assessment of the index
case’s history of depression and other psychiatric
disorders, using the FH-RDC. All the family
members used in this study were parents, sons,
daughters, spouses, or siblings of the index case.
In our sample, the number of completed family
histories of depression per index case ranged
from one to seven, with an average of three. This
interview was also modified to include DSM-IIIR criteria and questions to determine the degree
4855_e09_p69-76
2/4/04
1:18 PM
Page 72
72
of familiarity and amount of contact between the
informant and index, as described in detail later,
and ranged from extremely close and confiding
to having no current knowledge about the index
case.
Comparability between SADS-L and FH-RDC
The FH-RDC has been shown to have a general reliability of 73% when validated by the
SADS-L.31 The questions in the FH-RDC pertaining to MDD covered the same range of symptoms
related to depression as did the SADS-L. For example, in the SADS-L, the following questions
were asked to assess depression: (1) Has there
ever been a period of time when you were bothered most of the time, day and night, by feeling
depressed, sad, blue, hopeless, down in the
dumps, that you didn’t care anymore, or when
you were tearful or had crying spells? (2) Has
there ever been a period of time when you just
weren’t interested in or couldn’t enjoy the things
that you usually enjoy? Were you able to enjoy
anything? (3) What about a period when you felt
predominantly irritable or easily annoyed? The
parallel section in the FH-RDC was: Did he/she
have days, weeks, or months when he/she was
sad, hopeless, down in the dumps? What about
a period of time when he/she just wasn’t interested or couldn’t enjoy the things he/she usually
enjoyed? Or irritable, easily annoyed? There was
also a checklist, which included symptoms, such
as depressed mood, mood reactivity, and anhedonia.
Overreporting. Consistent with the literature,22,23 we used self-report as the standard to
which family reports were compared. Overreporting was defined as a case in which the index
did not self-report himself or herself as depressed
on the SADS-L, but an informant described this
index as depressed in the FH-RDC. In other
words, the FH-RDC overreports depression in
these individuals when compared with the corresponding SADS-L description. We expected
overreporting to be higher in women than in men.
Attribution. The section of the FH-RDC that
pertained to depression included an open-ended
portion in which the informant could elaborate
on the details of the index’s depression. In some
cases, the informant attributed the index’s depression to a cause. Each cause was then coded
BROMMELHOFF ET AL.
as either internal or external. Internal causes were
those where the depression was attributed to
something about the individual. These causes included such responses as “He/she is usually just
sad.” External causes were those where the depression was specifically attributed to something
beyond the control of the individual, including
alcohol/drugs, divorce, or loss of job. Two separate coders, who were blind to the index’s diagnosis and gender, coded the 59 cases, in which
the informant attributed the index’s depression
to a cause. Agreement between coders was 100%.
Covariates
Variables under consideration as potential confounders included age, marital status, socioeconomic status (SES), self-reported health status,
medication use, and levels of closeness between
the index and informant. SES was quantified using the Hollingshead two-factor index of social
position measure,34 which divided each participant into one of five SES levels based on the occupation and education of the head of household.
In our analyses, we collapsed this measure into
two levels: one that included the three highest
SES levels and another that included the lowest
two levels. Marital status was also collapsed into
two levels: married/remarried and not married
(including those who are single, divorced, or widowed/widowered).
Level of closeness between the index and informant was measured from the informant’s response to the following three-part question: How
close are you to [index]? Does [index] talk about
feelings or problems with you? Has someone else
in your family told you about [index]? For each
question, the informant chose one of six responses: (1) extremely close and confiding relationship, (2) quite close and confiding, would
know if [index] had problems, (3) not very close
or confiding, but has reliable second-hand information, (4) not very close, not confiding, but has
heard family gossip, (5) not at all close and has
little knowledge of [index], (6) has no knowledge
of [index].
Data analysis
To determine the effect of the gender of index
and informant on overreporting, bivariate analyses were run, using the chi-square procedure of
the Statistical Analytic Software (SAS) program,
version 8.1. Early analyses indicated that the co-
4855_e09_p69-76
2/4/04
1:18 PM
Page 73
73
GENDER BIAS AND DEPRESSION
TABLE 1.
Number
Mean age (years)
Range
Marital status
Married
Not married
Hollingshead SES
SES 1–3, high
SES 4–5, low
Contact with index
SAMPLE POPULATION DEMOGRAPHICS
All subjects
Women
Men
205
46.9
18–80
128
46.8
18–76
77
47.0
19–80
73.2% (150)
26.8% (55)0
69.5% (89)
30.5% (39)
79.2% (61)
20.8% (16)
66.9% (121)
33.1% (60)0
1.81
68.8% (75)
31.2% (34)
1.78
63.9% (46)
36.1% (26)
1.86
variates’ marital status, SES, age, and level of
closeness showed significant differences by gender. Therefore, they were used as covariates. After checking that the regression model assumptions were met, we selected multivariate logistic
regression analyses. We selected a multivariate
logistic regression model because both gender
and the overreporting of depression were dichotomous variables in this study.
RESULTS
Univariate analyses
The age of the participants ranged from 18 to
80 (M 46.9). Females made up 62.4% of this
sample. Of the 205 adults, 17.1% (n 35) self-reported depression. Of these 35 adults, 22 were
women and 13 were men. Men and women had
similar rates of marital status, age, SES, and contact with family members (Table 1). There was
also no difference between men and women in
their self-reported overall physical health or use
of psychiatric medication. All subjects in this
sample were Caucasian.
In all, there were 551 comparisons between an
index’s self-report and an informant’s family history report. In 78% of these comparisons, informants reported that they were either “extremely
close [and] confiding” with the index or that they
were “quite close and confiding [with the index
TABLE 2.
and] would know if [the index] had problems.”
Less than 1% of comparisons had the informant
indicate responses 5 or 6 for closeness level with
the index, 17.6% indicated a closeness of response
3, and 4% indicated number 4. In our analyses,
we excluded all comparisons in which the informant indicated 5 or 6 for level of closeness (n 5). We also excluded comparisons in which the
informant reported less than yearly contact with
the index (n 3), leaving a total of 543 index-informant comparisons. In 63.9% (n 347) of these
comparisons, the index was female.
Hypothesis 1: Overreporting of MDD will be
more common when index is female
For the 196 comparisons where the index was
male, 73.0% (n 143) showed index-informant
agreement, and only 3.6% (n 7) comparisons
overreported MDD. Among the 347 comparisons
with a female index, 64.3% (n 223) showed
agreement between index-informant reporting,
and 8.9% (n 31) of the comparisons overreported MDD (Table 2). In support of the first hypothesis, we found that overreporting was nearly
three times more likely when the index was female than when the index was male (OR 2.65,
CI 1.14, 6.13, p 0.025). Gender of the informant was not related to overreporting (OR 0.76, CI 0.39, 1.49, p 0.427).
In our multivariate model, we controlled for
SES, age, marital status, and closeness. The effect
INDEX-INFORMANT AGREEMENT
Total
Index-informant comparisons
Index-informant agree
Comparisons overreporting for MDD
543
61.9% (336)
7.0% (38)
BY
GENDER
Male
196
73.0% (143)
3.6% (7)
Female
347
64.3% (223)
8.9% (31)
4855_e09_p69-76
2/4/04
1:18 PM
Page 74
74
BROMMELHOFF ET AL.
TABLE 3.
Attribution
Internal
External
ATTRIBUTION TYPE
BY
GENDER
Total
Male index
Female index
59
27.1% (16)
72.9% (43)
17
11.8% (2)0
88.2% (15)
42
33.3% (14)
66.7% (28)
of gender remained significant (OR 2.62, CI 1.11, 6.18, p 0.028).
Hypothesis 2: Informants will attribute
depression in women to internal causes and
depression in men to external causes
Our secondary hypothesis posited that one of
the reasons for a potential gender bias in families
is that depressive symptoms are more likely to be
attributed to external causes in men and internal
causes in women. Because only 10.7% (n 59) of
the informant reports elaborated on the index’s
depression in the open-ended portion, this hypothesis could be examined only on an exploratory level (Table 3). There were 42 informants who attributed depression in a female
index. Of these 42 attributions, 14 were attributed
to internal causes, and 28 were attributed to external causes. There were 17 informant reports
that attributed depression in male indexes, 2 to
internal causes and 15 to external causes. We
found that depressive symptoms were not more
likely to be attributed to external symptoms in
men. However, informants were seven times
more likely to attribute a female relative’s depression than a male relative’s depression to internal causes (p 0.025).
DISCUSSION
As predicted, overreporting occurred more often when the index was female rather than male.
We also found that internal causes were much
more likely to be given as the reason for depression when the index was female.
The results from this study support one of the
aspects of the artifact hypothesis: that women are
more often subject to sex biases in diagnosis. Although we did not examine bias in clinical diagnosis, our results indicate that women are subject
to sex biases in the recognition of depression by
first-degree relatives. Such a bias could play a role
in seeking a diagnosis. If a family member is more
likely to discuss the possibility of depression with
a female relative than with a male relative, it is
more likely that a female will seek a diagnosis or
bring up the possibility of depression with a
physician. At the same time, if family members
are less likely to recognize depression in a male
relative, he may not consider the possibility that
he is suffering from depression. This implies that
the bias in reporting does not stem from women
being more in tune than men with their emotional
state but rather that other people are sensitizing
women to the possibility of depression in themselves.
The finding that women are likely to be the index in cases of overreporting can be applied to
the Vulnerability-Stress Model as well. The Vulnerability-Stress Model suggests that a woman’s
social role makes her vulnerable to depression.15
The assumption this model makes is that women
put others first, take “primary care and responsibility for children, home, spouse and dependent
relatives,”15(p61) and “often have only one source
of gratification in their lives—their families.”7(p78)
Today, however, many women are employed or
wait to have a family until they have a well-established career, yet the differences in rates of depression between men and women continue to
remain high. When our results are taken into consideration, it is possible that the informants are
the ones who assume a woman’s social role
makes her more vulnerable to depression. Even
though women are becoming more empowered,
the bias toward believing they are more susceptible to depression remains.
The results of this study have several implications. In our sample, it is not that women themselves exaggerate their depression. Rather, others
are more likely to report a woman as depressed,
even if she herself does not report herself as so.
Similar to our findings, Bertakis et al.22 found that
the clinicians in their study were more likely to
diagnose a woman as depressed even when the
level of severity of depression was the same for
a man. Our study found a similar dynamic in a
nonclinical setting. By not including the clinical
diagnosis for the participants in our study, we
were able to disentangle the self-reports and the
family history reports and examine agreement
between a person’s self-report and how others in
his or her family reported the absence or presence of depressive symptoms. The willingness of
both men and women to consider depression as
a female illness more readily is born of precon-
4855_e09_p69-76
2/4/04
1:18 PM
Page 75
75
GENDER BIAS AND DEPRESSION
ceptions that are far more widespread than clinical settings. It is, therefore, likely that these biases are formed in part by attitudes learned while
growing up.
More evidence of gender bias is found in the
patterns of attribution. Our results found that,
overall, informants attributed the depression to
external factors. For example, informants attributed depression to divorce and job stress in 31%
of all cases in which a reason for depression was
given. They also attributed depression in men to
alcohol, drugs, retirement, and unemployment,
which are all external factors. Depression in women, on the other hand, was frequently explained
by menopause, postpartum depression, premenstrual syndrome, or that the woman was “naturally” that way. All of these attributed causes of
depression in women are either biological or, as
in the case of the last example, an example of internal causality.
According to the DSM-III-R, a certain number
of mood and physical symptoms must be present
in order for a clinical diagnosis of MDD to be
given. However, by definition, MDD cannot be
caused by alcohol or drugs.32 Also, if the symptoms are due to an event, such as retirement or
unemployment, the individual would likely be
diagnosed with an adjustment disorder rather
than MDD. Of the men in this study who were
reported by an informant as having depressive
symptoms, most cases would not be classified as
MDD because of these exclusions. However, as
Calhoun et al.29 point out, men are likely to attribute their depressive mood to external causes
themselves. Furthermore, Johnson et al. found
that in one study, a clinical bias existed, “such
that severe mood problems [were] seen to result
from external, rather than internal causes.”35(p.603)
If depression in women, on the other hand, is seen
as internally caused, it is more likely that they
will receive medical attention. As a result, this
bias may cause women to get adequate, or possibly too much, medical attention, whereas men
may not get enough.
There were several limitations to this study.
First, all participants in this study were Caucasian, and, therefore, our results may not be generalizable to other ethnicities. Future studies
could use a similar approach to this study to examine if the same dynamic occurs across ethnicities. The participants in the YFS were not selected
for this particular study. However, this sample,
with reports of depression by an individual and
by a family member about the same individual,
offered us a chance to look at cases within families and, thus, extend the literature on a dynamic
that may contribute to gender differences in depression. Another limitation was that self-report
of depression was measured by the answers to
only three questions. We did not assess the severity of the depression or account for comorbidity
of other psychiatric disorders, such as anxiety disorders. Furthermore, both measures only asked
for lifetime incidence of depression, and we did
not determine any information regarding an individual’s involvement in counseling or the frequency and duration of any depressive episodes.
Additional research is needed to further understand the dynamic of gender bias within the family in regard to recognizing depression. Studies
examining the way different genders within a
family express their feelings and attitudes toward
mental health may further explain this dynamic.
CONCLUSIONS
To our knowledge, this is the first study comparing proxy reports with self-reports to examine gender bias in the family. Our findings suggest that gender bias within the family may
contribute to the higher rates of depression in women. This may occur by inflating rates of depression in women by sensitizing them to the
possibility of depression in themselves and
through processes accounted within families. Our
study also suggests that when family members
act as informants, mental health professionals
should take into account a tendency to perhaps
falsely report depression in women.
REFERENCES
1. Weissman MM, Klerman GL. Sex differences and the
epidemiology of depression. Arch Gen Psychiatry
1977;34:98.
2. Hankin BL, Abramson LY. Development of gender
differences in depression: An elaborated cognitive
vulnerability-transactional stress theory. Psychol Bull
2001;127:773.
3. Kessler RC. Gender differences in major depression.
In: Frank E, ed. Gender and its effects on psychopathology. Washington, DC: American Psychiatric Press, 2000:61.
4. Kohen D. Introduction. In: Kohen D, ed. Women and
mental health. Philadelphia: Routledge, 2000:1.
4855_e09_p69-76
2/4/04
1:18 PM
Page 76
76
5. Angst J. Epidemiology of depression. In: Honig A,
Van Praag HM, eds. Depression: Neurobiological,
psychopathological and therapeutic advances. New
York: John Wiley & Sons, 1997:17.
6. Shaw J, Kennedy SH, Joffe RT. Gender differences in
mood disorders: A clinical focus. In: Seeman MV, ed.
Gender and psychopathology. Washington, DC:
American Psychiatric Press, 1995:91.
7. Nolen-Hoeksema S. Sex differences in depression.
Stanford, CA: Stanford University Press, 1990.
8. Kessler LG, Cleary PD, Burke JD. Psychiatric disorders in primary care: Results of a follow-up study.
Arch Gen Psychiatry 1985;42:583.
9. Schulberg HC, Burns BJ. Mental disorders in primary
care: Epidemiologic, diagnostic, and treatment research directions. Gen Hosp Psychiatry 1988;10:79.
10. Borus JF, Howes MJ, Devins NP, Rosenberg R, Livingston WW. Primary health care providers’ recognition and diagnosis of mental disorders in their patients. Gen Hosp Psychiatry 1988;10:317.
11. Ormel J, Koeter MWJ, Van Den Brink W, Van De Villige G. Recognition, management, and course of anxiety and depression in general practice. Arch Gen Psychiatry 1991;48:700.
12. Neumann PJ, Araki SS, Getterman EM. The use of
proxy respondents in studies of older adults: Lessons,
challenges, and opportunities. J Am Geriatr Soc 2000;
48:1646.
13. Nolen-Hoeksema S. Gender differences in depression.
Curr Dir Psychol Sci 2001;10:173.
14. Hankin BL, Abramson LY. Development of gender
differences in depression: Description and possible
explanations. Ann Med 1999;31:372.
15. Holmshaw J, Hillier S. Gender and culture: A sociological perspective to mental health problems in women. In: Kohen D, ed. Women and mental health.
Philadelphia: Routledge, 2000:39.
16. Rosenfield S. Gender and dimensions of the self: Implications for internalizing and externalizing behavior. In: Frank E, ed. Gender and its effects on psychopathology. Washington, DC: American Psychiatric
Press, 2000:23.
17. Kendler KS, Thornton LM, Prescott CA. Gender differences in the rates of exposure to stressful life events
and sensitivity to their depressogenic effects. Am J
Psychiatry 2001;158:587.
18. Parry BL. Hormonal basis of mood disorders in women. In: Frank E, ed. Gender and its effects on psychopathology. Washington, DC: American Psychiatric Press, 2000:3.
19. Nolen-Hoeksema S. Epidemiology and theories of
gender differences in unipolar depression. In: Seeman
MV, ed. Gender and psychopathology. Washington,
DC: American Psychiatric Press, 1995:63.
20. Amenson CS, Lewinsohn PM. An investigation into
the observed sex difference in prevalence of unipolar
depression. J Abnorm Psychol 1981;90:1.
BROMMELHOFF ET AL.
21. Parker G. Sex differences in non-clinical depression.
Aust NZ J Psychiatry 1981;13:127.
22. Bertakis KD, Helms LJ, Callahan EJ, Azari R, Leigh P,
Robbins JA. Patient gender differences in the diagnosis of depression in primary care. J Wom Health Gender-Based Med 2001;10:689.
23. Williams JBW, Spitzer RL, Linzer M, et al. Gender differences in depression in primary care. Am J Obstet
Gynecol 1995;173:654.
24. Borowsky SJ, Rubenstein LV, Meredith LS, Camp P,
Jackson-Triche M, Wells KB. Who is at risk of nondetection of mental health problems in primary care?
J Gen Intern Med 2000;15:381.
25. Stoppe G, Sandholzer H, Huppertz C, Duwe H, Staedt
J. Gender differences in the recognition of depression
in old age. Maturitas 1999;32:205.
26. Abraham IL. Causal attributions of depression: Reliability of the “Causal Dimension Scale” in research
on clinical inference. Psychol Rep 1985;56:415.
27. Peterson C, Bettes BA, Seligman ME. Depressive
symptoms and unprompted causal attributions: Content analysis. Behav Res Ther 1985;23:379.
28. Seligman ME, Abramson LY, Semmel A, von Baeyer
C. Depressive attributional style. South Psychologist
1984;2:18.
29. Calhoun LG, Cheney T, Dawes AS. Locus of control,
self-reported depression, and perceived causes of depression. J Consult Clin Psychol 1974;42:736.
30. Endicott J, Spitzer RL. A diagnostic interview: The
Schedule for Affective Disorders and Schizophrenia.
Arch Gen Psychiatry 1978;35:837.
31. Andreasen NC, Endicott J, Spitzer RL, Winokur G.
The family history method using diagnostic criteria.
Arch Gen Psychiatry 1977;34:1229.
32. American Psychological Association. Diagnostic and
statistical manual of mental disorders, 3rd ed, Revised
(DSM-III-R). Washington, DC: American Psychiatric
Association, 1987.
33. Hesselbrock V, Stabenau J, Hesselbrock M, Mirkin P,
Meyer R. A comparison of two interview schedules.
Arch Gen Psychiatry 1982;39:674.
34. Hollingshead A. Two-factor index of social position.
New Haven, CT: Yale University Press, 1958.
35. Johnson RE, Calhoun LG, Boardman WK. The effect
of severity, consistency, and typicalness information
on clinicians’ causal attributions. J Clin Psychol 1975;
31:600.
Address reprint requests to:
Becca R. Levy, Ph.D.
Yale University School of Medicine
Department of Epidemiology and Public Health
PO Box 208034
New Haven, CT 06520-8034
E-mail: becca.levy@yale.edu