Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Sex Differences in Vulnerability to Undesirable Life Events

Author(s): Ronald C. Kessler and Jane D. McLeod


Source: American Sociological Review, Vol. 49, No. 5 (Oct., 1984), pp. 620-631
Published by: American Sociological Association
Stable URL: http://www.jstor.org/stable/2095420
Accessed: 01-09-2016 14:59 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Sage Publications, Inc., American Sociological Association are collaborating with JSTOR to digitize,
preserve and extend access to American Sociological Review

This content downloaded from 92.241.95.11 on Thu, 01 Sep 2016 14:59:48 UTC
All use subject to http://about.jstor.org/terms
SEX DIFFERENCES IN VULNERABILITY TO
UNDESIRABLE LIFE EVENTS*

RONALD C. KESSLER JANE D. MCLEOD


University of Michigan

Past research has shown that the emotional impact of undesirable life events is
significantly greater among women than men. This finding has led to speculation
that women possess a deficit in coping capacity or in access to social support that
renders them pervasively disadvantaged in responding emotionally to problematic
situations. We present a different argument in this paper. We hypothesize and then
document that women are not pervasively more vulnerable to the effects of
undesirable events. A disaggregated analysis of life-event effects shows, further, that
female vulnerability is largely confined to "network" events: life events that do not
occur to the focal respondent but to someone in his or her social network who is
considered important. Further results are presented to argue that this greater
vulnerability is due to the greater emotional involvement of women in the lives of
those around them. It is demonstrated that this emotional cost of caring is
responsible for a substantial part of the overall relationship between sex and distress.

It is well documented that women in Western ble for the higher levels of distress among
society have significantly higher rates of psy- women than men.
chological distress than men (Al-Issa, 1982). In Although provocative, this indirect line of
sociology, most discussions of this fact have analysis cannot take us far in understanding the
revolved around the idea that women's social processes that create distress among men and
roles are more stress provoking and less ful- women. Only gross comparisons across con-
filling than those occupied by men (Gove, stellations of status profiles are possible, and
1978). This social-role explanation has fostered these lend themselves to multiple interpreta-
a substantial body of research on sex dif- tions including the possibility that prior psy-
ferences in chronic role-related stress. This chological status or predispositions lead to
work has typically been based on indirect mea- selection into or out of certain statuses (Kess-
sures of stress. Marital status, numbers and ler and McRae, 1984). There have been a few
ages of children, and employment status have attempts to move beyond this indirect assess-
been used to make inferences about chronic ment, by measuring role stresses and resources
stress (Gove, 1972; Radloff, 1975; Gove and directly (Pearlin, 1975; Kessler and McRae,
Geerken, 1977; Aneshensel et al., 1981). For 1982), but this has proven to be quite difficult.
example, the sex-distress relationship has The role-stress measures developed so far are
been shown to be more pronounced among the highly subjective in content, and this has ham-
married than the previously married or never pered attempts to make causal inferences. (For
married (Gove, 1972; Fox, 1980). This specif- a critique, see Kessler [1983].)
ication has been interpreted as evidence that In the face of this difficulty, some re-
role-related stress and resources are responsi- searchers have turned to an examination of sex
differences in exposure and response to unde-
sirable life events (Dohrenwend, 1973; Kess-
*Direct all correspondence to: Ronald C. Kessler,
ler, 1979; Radloff and Rae, 1981). This ap-
Institute for Social Research, University of Michi-
proach offers methodological advantages over
gan, Box 1248, Ann Arbor, MI 48106.
This research was supported by NIMH Research the analysis of chronic stress in that exposure
Scientist Development Award IK02 MH00507 to the can be unambiguously measured. Further-
senior author, by NIMH research grants IROI more, the consideration of life events is neces-
MH30479 and T32 MH16806, and by a National In- sary for a comprehensive interpretation of sex
stitutes of Health Biomedical Research Support differences in stress exposure.
Grant to the Vice President for Research at the Uni- This research has documented that although
versity of Michigan. We would like to thank
men and women do not differ greatly in the
Elizabeth Douvan, Richard Kulka, Ben Locke,
number of undesirable life events they experi-
Jerome Myers, Lenore Radloff and Joseph Veroff for
ence, women are significantly more affected
making their data available to us. We would also like
to thank Duane Alwin, Elaine Wethington and emotionally than men. Indeed, this greater fe-
Camille Wortman for helpful comments on earlier male vulnerability to life-event effects ac-
drafts. counts for a substantial part of the overall re-

620 American Sociological Review, 1984, Vol. 49 (October:620-631)

This content downloaded from 92.241.95.11 on Thu, 01 Sep 2016 14:59:48 UTC
All use subject to http://about.jstor.org/terms
UNDESIRABLE LIFE EVENTS 621

lationship between sex and psychological dis- ered stressful. In addition, women reported a
tress. number of events that men failed to mention.
A variety of interpretations have been of- The latter consisted of life crises that had oc-
fered to account for this vulnerability. For curred to members of the respondents' social
example, Belle (1982) has suggested that networks: family, friends, and neighbors.
women have access to fewer supportive re- Women reported that such network events
lationships that ameliorate the distressing ef- were very distressing to them, while men sel-
fects of life crises. Radloff and Monroe (1978) dom mentioned events of this type.
have argued that women are socialized in such Research on male and female roles provides
a way that they are susceptible to reactive de- a basis for understanding this vulnerability of
pression. Pearlin and Schooler (1978) have women to network events. From early child-
suggested that the coping strategies used by hood, women develop a moral sense that em-
women are less effective than those used by phasizes caring for others, just as men con-
men. struct a morality that emphasizes individual
Arguments of this sort imply that women are rights and responsibilities (Gilligan, 1982).
disadvantaged by a pervasive deficit in coping Women are more likely than men to be attuned
resources. However, this implication remains to the needs of others and to feel responsible
untested. All of the research on life-event ef- for meeting those needs (Miller, 1976). This
fects to date has worked with aggregated in- sense of responsibility for the life events of
dices that combine many different events into aloved ones could lead women to report more of
single measure of exposure. We do not know these events and to experience these events as
from this research whether the vulnerability of more distressing.
women is pervasive or limited to particular On the basis of the above considerations, we
types of events. Clearly, the development of propose two hypotheses for evaluation in this
theory in this area would be advanced by such paper. First, we hypothesize that women are
a specification. not pervasively more vulnerable than men to
Some evidence that women are not perva- the emotional effects of undesirable life events.
sively more vulnerable than men can be found The confirmation of this hypothesis would call
in case studies of crisis coping. For example, into
it doubt the relevance of recent research on
seems that women adjust somewhat better than sex differences in overarching stress-buffering
men to the loss of a spouse (Stroebe and resources, such as social support and coping
Stroebe, 1983). Women also seem to be less styles, for an understanding of vulnerability to
emotionally affected than men by financial dif- acute life stress. Second, we hypothesize that
ficulties (Kessler, 1982). There is evidence that network life events-events that occur to
women cope more adequately than men with people who are important to the respondent-
marital separation and divorce after an initial play a central part in the aggregate vulnerabil-
period (Wallerstein and Kelly, 1980). These ity of women to undesirable events. The con-
examples point to the idea that the different firmation of this hypothesis would significantly
roles occupied by men and women, and redirect research on sex differences in vul-
perhaps the salience of these role domains, are nerability to life events. Furthermore, it would
implicated in vulnerability to particular events. provide an opportunity for research on the
If this suggestive evidence can be rigor- stress of female roles to be integrated with
ously supported in general population surveys, research on life events, by demonstrating that
a fundamental reorientation will be required in the demands for nurturance that are at the
current thinking about the vulnerability of center of female role obligations are also cen-
women. Rather than searching for pervasive trally involved in the events to which women
coping deficits, research on sex differences in are more emotionally vulnerable than men.
vulnerability to life events would be directed to
the role domains in which female vulnerability DATA AND METHODS
is isolated. But if female vulnerability is con-
Data
centrated in this way, what types of events are
involved'? Although no research has explicitly We present analyses based on five epi-
estimated disaggregated models to pinpoint the demiologic surveys of' the general popula-
effects of particular events, Dohrenwend tion. Table 1 describes the sample sizes, sam-
(1977) has presented suggestive evidence about pling frames, years of administration, and mea-
what these events are. She interviewed a sam- sures of psychological distress included in each
ple of men and women to determine the eventsof the surveys. The smallest and oldest of the
that they had experienced in the recent past five is a panel survey conducted in New
which, by their own accounts, were most dis- Haven, Connecticut, by Jerome Myers and his
tressing. This inductive search uncovered associates (NH). We pooled the two waves of
many events that both men and women consid- this panel into a synthetic cross-section in the

This content downloaded from 92.241.95.11 on Thu, 01 Sep 2016 14:59:48 UTC
All use subject to http://about.jstor.org/terms
622 AMERICAN SOCIOLOGICAL REVIEW

Table 1. Data Sets and Screening Scales Included in the Analysis

Screening Scales

Survey Sample Psycho-


Symbol Year Population Size Depression physiological Source
NH 1967, 1969 New Haven, CT 720 Ss GURIN Myers et al.
21 or older (1974)
CMHA I 1971-73 Washington County, 1673 CESD, PHYSL Comstock and
MD, 18 or older PSYCL Helsing (1976)
CMHA II 1971-73 Kansas City, MO 1173 CESD, PHYSL Comstock and
18 or older PSYCL Helsing (1976)
CMHA III 1971-73 Washington County, 1089 CESD Comstock and
MD, 18 or older Helsing (1976)
SML 1976 U.S., 21 or older 2264 ZUNG GURIN Veroff et al.
(1981)

CESD: Center for Epidemiologic Studies Depression Scale (Radloff, 1977).


GURIN: Gurin Scale (Myers et al., 1974).
PHYSL: Psychophysiological/Physical Subscales of the Langner Scale (Langner, 1962; Dohrenwend and
Crandell, 1970).
PSYCL: Psychological Subscale of the Langner Scale (Langner, 1962; Dohrenwend and Crandell, 1970).
SS: Study-Specific Scale (Weissman and Myers, 1978).
ZUNG: Modified Zung Scale (Zung, 1965; Veroff et al., 1981).

analysis. (See Kessler and Greenberg operationalized in perceptions of bodily


[1981:157-62] for a description of this tech- feelings known to be associated with anxiety
nique.) The largest and most recent survey is and depression. Psychological distress is mea-
the Survey of Modern Living (SML) adminis- sured seven times with a depressed-mood scale
tered by Joseph Veroff and associates. The and four times with a scale of psycho-
three other surveys are part of the Community physiological symptoms in the five surveys.
Mental Health Assessment (CMHA) program These 11 replications form the basis of the
administered through the Center for analysis. 1
Epidemiologic Studies at the National Institute Life events. Life events were assessed with
of Mental Health. Two of the five surveys were life-event lists in four of the five surveys (the
conducted in urban areas (NH and CMHA II), exception is SML). The lists differ in length
two in rural areas (CMHA I, CMHA III), and and in the precision with which they dif-
one in the total U.S. population (SML). ferentiate events (for example, asking about
Our decision to use a number of surveys in minor illnesses, serious illnesses, and serious
the analysis was based on the fact that the rate injuries separately versus combining them into
of occurrence of specific life events is very low one overall illness/injury category) and in the
in surveys of this type. Therefore, analyses of inclusion of some less serious events (for
particular life-event effects have low power example, death of a pet). All of the lists include
within surveys. Combining the results from the most serious life crises, such as death of a
many surveys helps resolve this problem by loved one and divorce.
increasing effective sample size. Methods have In these four surveys, respondents were
been developed to combine tests of sig-
nificance across surveys to arrive at overall
levels of statistical significance. These tests of
I Although there is a significant relationship be-
pooled significance are discussed in more de-
tween scores on the two types of screening scales,
tail below.
depressed mood and psychophysiological symp-
Our analysis employs three central vari-
tomatology can be discriminated in factor analysis.
ables: sex, psychological distress, and life Furthermore, scores on both types of scales dis-
events. A description of the latter two vari- criminate people with either depression or anxiety
ables and their operationalization is necessary. disorders from others in the normal population. It is
Psychological distress. The outcome vari- not clear why some people manifest these disorders
ables of interest are several different scales of in high scores on one type of screening scale rather
subjectively experienced psychological dis- than the other, although there has been some
speculation that cultural differences play a part
tress, described in the sixth and seventh col-
(Dohrenwend and Crandell, 1970; Wheaton, 1982).
umns of Table 1. The scales used vary from
For our purposes, though, this uncertainty is of less
one survey to the next, but they can collec-
importance than the fact that each type of scale pro-
tively be divided into two groups: those that vides some information about psychopathology in-
measure depressed mood and those that mea- dependent of the other, a fact which argues for the
sure psychophysiological distress (PPS) as use of both scales in a comprehensive analysis.

This content downloaded from 92.241.95.11 on Thu, 01 Sep 2016 14:59:48 UTC
All use subject to http://about.jstor.org/terms
UNDESIRABLE LIFE EVENTS 623

asked to report which, if any, of the events on Death of a loved one is the most frequently
the list happened to them in the past year. This reported serious network event in standard
time frame is consistent with evidence from life-event inventories, with about 15 percent of
retrospective case-control studies that events adults in the normal population affirming that
have a significant impact on acute onset of someone important to them died in the past
clinical disorders for no more than one year year. Less dramatic events obviously occur in
(Brown and Harris, 1978; Paykel, 1978; Surtes social networks with greater frequency, but
and Ingham, 1980). The list was then repeated they are not reported by most respondents un-
and respondents were asked whether any of less they occurred to members of their im-
the events happened to someone important to mediate family. For this reason, we aggregated
them. all other undesirable network events into a
No life-event list was included in the SML single scale in the first part of our analysis.
survey. Instead, respondents were asked a We recognize that the combination of these
single open-ended question: "Now think about diverse events into substantive groups masks
the last time something really bad happened to important variability among crises, both in
you. What was it about?" Each respondent their meanings and their emotional effects.
reported one event and the date of its occur- However, this partial disaggregation is more
rence. To approximate the one-year measures true to reality than the conventional approach
available in the other surveys, only those of combining all undesirable events into one
events experienced in the last year were in- summary measure. It is sufficiently detailed to
cluded for respondents in the SML survey. test the two central hypotheses and, as we
Investigations of life-event effects typically document below, it allows us to offer a rela-
use weighted sums of the life events on the list tively concise, yet theoretically rich, analysis
as the measure of exposure, with weights being of life-event effects.
assigned on the basis of some objectively de-
termined criterion for the seriousness of the
Methods of Analysis
event. We present a different approach. We
disaggregate the list into six theoretically cohe- A model of the form
sive categories. This allows a more precise
determination of which events are implicated D = bo + b1S + b2LE
in the pattern of differential response observed + b3(S)x(LE) + b4C (1)
between men and women. We consider only
undesirable events. We do not weight the indi- is typically estimated in research on the re-
vidual events or the categories for seriousness lationship between sex and response to life
but, instead, use simple counts of the number events. D is a measure of distress, S is a
of reported events within each category.2 dummy variable coded 1 for men and 0 for
The six categories of events are income loss, women, LE is an index of life events, (S) x (LE)
separation and divorce, other love loss, ill is a multiplicative interaction of S and LE, and
health, death of a loved one, and other network C is a series of control variables. Estimates of
events. Most of these are diverse. Income loss, b3 are usually negative and statistically signifi-
for example, includes such things as being cant, meaning that the impact of life events on
fired, business failure, and foreclosure of a distress is more pronounced among women
mortgage. Other love loss consists in broken than men.3
engagements, breaks with a steady girlfriend or Our analysis strategy modified this basic
boyfriend, and the termination of other impor- prediction equation in two ways. First, we di-
tant friendships. vided the sample of women into two groups:
women in the labor force (WLF) and
homemakers (HM).i This division is justified
2 Previous research has shown that weighting will
by the fact that the life situations of employed
not normally affect the results of life-event analysis
because the variance of the item weights is small and unemployed women are very different.
relative to the between-person variance in event oc-
currence (Shrout, 1981). This is also true in our
analysis, in which events are divided into six 3An alternative interpretation is that the putative
categories. We considered using individual life interaction is masking a nonlinearity in the effect of
LE on D (Southwood, 1978). This is possible,
events in the analysis instead of event categories, but
decided against this option for two reasons. First, though, only if LE and S are strongly related to each
even with the five surveys considered here, there are other, which is not true in these data since, as we will
too few reports of some life events to allow a pow- note, mean levels of LE do not differ substantially by
erful analysis. Second, an analysis of the many dif- sex.
ferent undesirable life events included in the life- 4 The analysis was limited to whites age 65 and
event list would prove difficult to interpret and, younger, excluding students, the retired and the dis-
thereby, theoretically wasteful. abled.

This content downloaded from 92.241.95.11 on Thu, 01 Sep 2016 14:59:48 UTC
All use subject to http://about.jstor.org/terms
624 AMERICAN SOCIOLOGICAL REVIEW

Furthermore, previous evidence shows that three subsamples and follow with between-
labor force participation is the most important group comparisons.
role-related descriptive characteristic among
women which pertains to their psychological
Exposure to Life Events
well-being (Kessler and McRae, 1982). Our es-
timation equation included this elaboration by As noted above, past research has found that
introducing separate dummy variables for each the greater importance of undesirable events
of the two groups of women in place of the for the well-being of women than men can be
single dummy variable for sex in equation (1). largely accounted for by differences in emo-
We included only men in the labor force (MLF) tional response to events. Differences in expo-
in the omitted contrast group, as there are few sure have been comparatively unimportant.5
unemployed, retired, or disabled men in these Nevertheless, it is useful to begin by examining
surveys. the frequency with which events of different
The second change in the prediction equa- types are reported to get a sense of the entire
tion involved the life-events term. We included process through which life events affect the
six life-event categories instead of one. This emotional functioning of men and women.
means that there were six interaction terms Table 2 presents the average number of life
between sex and life events for each subgroup events reported in the five surveys, within sub-
comparison and twelve interaction terms in all. groups, by type of event.6 It also lists z-
This model was estimated for each of the out- statistics for the test of sex differences in aver-
come variables in the five surveys. Control age number of events reported, pooled across
variables for marital status (dummy variables the five surveys.
for previously married and never married) and There is a wide range in the number of
age were included because many of the event events reported. The most common are net-
types are either related to one of these vari- work events and the least common are marital
ables (as the presence of health problems is to separation/divorce. There are so few of the
age) or confined to persons in one category of latter that slope estimates for the impact of
these variables (as marital separation is to the marital disruption on emotional functioning
married). should be interpreted with caution. All other
With these modifications to equation (1), it event types are much more common.
was possible to evalute the differential impact Men report a considerably higher rate of in-
of life events on men- and women. In addition, come loss than either women in the labor force
we estimated within-group models to evaluate or homemakers. They also report significantly
the significance of life-event effects separately more physical health problems, although the
among men in the labor force, women in the substantive difference is small (1 i percent more
labor force, and homemakers. Each of these physical health problems among men than
equations was estimated 11 times, once for women). Overall, men and women are very
each of the outcome variables enumerated in much alike in the frequency with which
Table 1. they report marital disruptions and other love
The results of these replicated analyses were losses, although women in the labor force have
summarized by calculating pooled significance higher rates than homemakers.7
tests. A number of methods are available for
doing this (Rosenthal, 1978). Although we used 5 Makosky (1980) and Belle (1982) have noted that
several of these methods, they yielded similar life-event inventories represent the typical life events
results and so we present only one here: the experienced by men better than those experienced
method of adding t-values. In this approach, by women. None of the surveys considered here, for
example, includes abortion or rape. All of them, in
the t-values for a parameter are summed across
contrast, include entering or leaving the armed
replications and divided by
forces. This slanted representation must be borne in
mind when interpreting evidence about sex dif-
{j [df/(df- 2)]11/2.
ferences in exposure.
This ratio is a standard normal deviate that can 6 Respondents in WCII were not asked about
be used to evaluate the overall significance of Other Love Loss or Network events. Other Love
the parameter across the replications. Loss events were also not included in the life-event
list in NH. Therefore, the sample sizes for these two
types of events are different from those listed in
RESULTS Table 2. The correct sample sizes for Other Love
Loss and Network events, for MLF, WLF, and HM
In this section we consider both differences in
respectively are 1429, 1214, 845, and 1862, 1406,
exposure to life events across the three groups 1201.
and differences in the slopes of distress on life 7 In the population as a whole, of course, the num-
events. We begin by describing the absolute bers of men and women who experience a marital
levels of exposure and response within the disruption are identical since it takes one of each sex

This content downloaded from 92.241.95.11 on Thu, 01 Sep 2016 14:59:48 UTC
All use subject to http://about.jstor.org/terms
UNDESIRABLE LIFE EVENTS 625

Table 2. Average Number of Reported Life Events, by Type of Event, for Men in the Labor Force (MLF),
Women in the Labor Force (WLF), and Homemakers (HM) and z-statistic for Sex Differences
in Average

Average Number of
Life Events z-statistic

Type of Event MLF WLF HM MLF vs. WLF MLF vs. HM

Income .102 .056 .042 4.001 5.296


Separation/Divorce .028 .036 .018 - 1.174 1.617
Other Love Loss .053 .082 .039 -2.818 1.318
Death of a Loved One .118 .146 .170 -2.075 -4.231
IlII Health .159 .138 .139 1.773 2.301
Network .719 .892 .958 -4.074 -6.315
(N)II (2221) (1616) (1472)
a Sample sizes apply to all categories of eve
discussion of the differing sample sizes.

Women report significantly greater expo- Response to Life Events


sure to death than men. This is part of a
Table 3 presents partially standardized slopes
broader pattern of reporting more network
of distress on each event type averaged over
events of many different kinds. Homemakers
the 11 replications of the regression analysis.8
report the highest level of exposure, followed
These slopes describe the expected increases
by women in the labor force and then men in
in distress, in standard deviation units, associ-
the labor force. Approximately 12 percent of
MLF report having known someone who diedated with the occurrence of each event type,
controlling for age and marital status. The
in the past year, as compared to 15 percent of
standard deviation is defined on the basis of the
WLF and 17 percent of homemakers. The pat-
total sample, rather than separately within the
tern is the same for other network events. Men
sex-specific subsamples, so that comparisons
in the labor force report an average of .719
across the subsamples can be made. The num-
events, WLF an average of .892, and HM an
bers in parentheses under each average slope
average of .958 events. The sex differences for
are the z-statistics of these slopes, pooled
both event types are highly significant.
across the 1 1 replications. Comparable statis-
Most of these results make good sense. It is
tics for the pooled significance of sex dif-
obvious, for example, that more events occur
ferences in the slopes are presented in the table
in networks than to their individual members.
as well.
It is also reasonable that employed people ex-
Fourteen of the 18 pooled subgroup regres-
perience more income events than homemak-
sion coefficients are significant at the . 10 level.
ers. It is much more difficult, though, to under-
One of the remaining four coefficients is sub-
stand the significant difference in the frequen-
stantively significant but based on such a small
cies with which men and women report net-
number of events that it is not significant in
work events. Evidence suggests that men
statistical terms (separation/divorce among
know more people than do women (Veroff et
HM). The remaining three coefficients-for the
al., 1981), yet women are aware of more of the
influence of deaths and network events on men
crises that occur to the people around them.
and the influence of income losses on
This could occur as a result of women being
homemakers-are substantively as well as
more attuned to their interpersonal envi-
statistically insignificant.
ronments or because the members of these en-
The sex comparisons show that several
vironments look to women more than men for
event types affect men and women equally.
support and comfort in times of trouble. We
Income losses have similar effects on the
explore these possibilities more fully in our
well-being of both men and women in the labor
interpretation of the results.

8 The 11 replications are not independent, as they


are taken from only five separate surveys. There-
to make (and break) a marriage. The differences we fore, the significance tests here are not exact. We
observe between women in the labor force and experimented with two other ways of combining the
homemakers reflect two facts: marital disruptions results: (1) selecting only one outcome variable from
are more likely to occur among women employed each of the surveys; and (2) averaging the results
outside the home than among homemakers; and within surveys before combining the five averages in
marital disruption is often the occasion for women a final analysis. These different approaches yielded
who were previously homemakers to seek a job out- very similar results to those reported here, which
side the home. documents that the results are robust.

This content downloaded from 92.241.95.11 on Thu, 01 Sep 2016 14:59:48 UTC
All use subject to http://about.jstor.org/terms
626 AMERICAN SOCIOLOGICAL REVIEW

Table 3. Average Slopes of Distress on Life Events for Men in the Labor Force (MLF), Women in the Labor
Force (WLF), and Homemakers (HM) and z-statistics for Sex Differences in Slopesa

Average b z-statistic

Type of Event MLF WLF HM MLF vs. WLF MLF vs. HM

Income .342 .330 .043 -.114 -2.759


(9.455) (4.744) (.500)
Separation/Divorce .143 .263 .236 .927 1.088
(2.101) (2.980) (1.533)
Other Love Loss .374 .354 .320 -.679 -.430
(5.100) (4.595) (2.514)
Death of a Loved One -.003 .193 .222 1.529 3.902
(-.002) (2.360) (4.119)
Ill Health .176 .245 .538 1.235 5.444
(5.133) (5.456) (8.307)
Network -.028 .129 .045 4.297 1.991
(-.115) (5.071) (2.789)

a The numbers in parentheses are pooled z-statistics for the significance of the slope estimates. Slopes are
pooled across all 11 dependent variables except for Other Love Loss (available for 8 dependent variables) and
Network events (available for 10 dependent variables).

force. The effects of marital disruptions on The most dramatic of these sex differences is
men and women are statistically indistinguish- the extreme vulnerability of homemakers to
able, as are the effects of other love losses. On physical health problems. We prefer not to
the basis of these results, it is perfectly clear interpret this coefficient, as it is difficult to
that women do not suffer from a pervasive know whether selection of the most seriously
emotional vulnerability to stressful experiences ill women out of the labor force is responsible.
when compared to men. Therefore, the first of The slope difference between men and women
our two hypotheses is supported. in the labor force for ill health events is in the
At the same time, there are sex differences in same direction, but not significant."'
the response to some kinds of life events. Much more important for our purposes are
These overwhelmingly show men to be less the findings that women are considerably more
responsive than women. In the 12 male-female affected than men by the death of a loved one
comparisons only one shows women to be sig- and by other network events. The latter is par-
nificantly less affected than men while six ticularly striking, given that men do not report
show women to be significantly more affected as many network events as women. A closer
than men. The single instance of greater male examination of the results in NH indicate that
vulnerability is in the case of income losses, this underreporting takes a rather special form:
where homemakers have a markedly low men report fewer ininor events and events at
slope. The six cases of greater female vulnera- the edges of their caring networks. The correc-
bility are associated with ill health, death of a tion of this sex-related bias in reporting would,
loved one, and network events.9 in all likelihood, magnify the differential impact
of network events.
9 The greater impact of death on women than men
is not due to widowhood. Indeed, as noted above,
the available evidence suggests that women adjust contains either one or two.) The average slope of
somewhat better than men to the death of a spouse distress on widowhood in this subsample is .64. For
(Stroebe and Stroebe, 1983). Although only a small homemakers, finally, there are small numbers of re-
number of widowhoods are reported in our surveys cently widowed people in each survey. (The mini-
compared to other types of deaths (in CMHA-I, for mum is three in CMHA-II; the maximum, eight in
example, there were 18 widowhoods and 344 deaths CMHA-I and SML.) The average slope here is .46,
of other important people in the year prior to the with six of I I slopes significantly greater than zero.
interview), our results are consistent with this result. None of the possible tests of slope differences be-
There is only one survey, SML, where more than a tween males and females is significant nor, for that
single male respondent was recently widowed. Here matter, close to significance.
there are five such men. The average slope of dis- "' These sex differences are equally clear when we
tress on widowhood in this subsample (comparing confine ourselves to the depression outcomes. We
these five men to those who are still married, con- took this extra step in specification because Thoits
trolling age) is .88 across the two distress outcomes (1981) has suggested that illness events might be
in that survey. Despite the small number of men confounded with psychophysiological indicators of
involved, both slopes are significant. Among women psychological distress, and because this might be
in the labor force the only survey with more than a more true of women than men. That the differential
trivial number of recent widowhoods is CMHA-1, vulnerability appears when depressed mood is taken
where there are nine. (Each of the other surveys as the outcome argues against this interpretation.

This content downloaded from 92.241.95.11 on Thu, 01 Sep 2016 14:59:48 UTC
All use subject to http://about.jstor.org/terms
UNDESIRABLE LIFE EVENTS 627

Table 4. Decomposition of Sex Differences in Distress from Pooled Sex-Specific Regression Equations

MLF vs. WLF MLF vs. HM

Type of Event Response Exposure Total Response Exposure Total

Income .000 - .016 - .016 - .021 - .012 - .033


Separation/Divorce .004 .001 .005 .003 - .002 .001
Other Love Loss -.001 .010 .009 -.003 -.004 -.007
Death of a Loved One .026 .002 .028 .033 .006 .039
Ill Health .011 -.005 .006 .055 -.008 .047
Network .126 .009 .135 .061 .001 .062
Total .166 .001 .167 .128 -.019 .109
Residual Mean Difference .041 .220
Observed Mean Difference .208 .329

Decomposition of the Overall ing the observed and residual mean differences
Sex-Distress Relationship in the last two rows of the table. Women in the
labor force have an average distress score .208
We noted earlier that previous research has
standard deviations higher than men, while
found sex differences in vulnerability to unde-
homemakers have an average distress score
sirable events more important than sex dif-
.329 standard deviations higher than men.
ferences in exposure to these events in ac-
Were we to remove the observed sex dif-
counting for the higher levels of distress found
ferences in exposure and response to stressful
among women than men. This is also true for
life events, these differences would be reduced
the disaggregated data presented here, a result
to .041 and .220, respectively. This means that
that can be most clearly seen by using a
we can explain 80 percent of the MLF-WLF
demographic-mean decomposition procedure
difference and 33 percent of the MLF-HM
(Jams and Thornton, 1975). Our use of this
difference in distress by taking life events into
procedure is heuristic, as we created synthetic
account.
regression equations from the average means
The row and column totals pinpoint the crit-
and slopes in Tables 2 and 3 and performed the
ical ways in which life events are implicated.
decomposition on these.1" Nonetheless, even
From the row totals we see that differential
in this rough sort of manipulation the over-
response accounts entirely for the influence of
whelming importance of differential vulnera-
life events on the excess distress observed
bility is clear.
among women. Differential exposure is of no
The results of the decomposition are pre-
importance in the MLF-WLF contrast and
sented in Table 4. There are two decomposi-
works to the advantage of women in the
tions reported: one for the comparison of
MLF-HM contrast.
MLF versus WLF and one for the comparison
Response to network events accounts for
of MLF versus HM. For each life event, the
over 75 percent (.126/.166) of the response
decomposition derived two components: (1)
component in the MLF-WLF contrast and 48
(bw-bm) (Xm+Xw)/2, the differential-response
percent (.061/.128) in the MLF-HM contrast.
component, which represents the expected
Deaths affect women more than men as well,
change in the distress of men if the slope
accounting for 16 percent (.026/.166) and 26
among men were the same as it is among
percent (.033/.128) of the differential response
women and the mean level of exposure were
components in the two contrasts.
equal to the observed mean, averaged over
Women also report the occurrence of more
both men and women; (2) [(bm+bw)/2]
network events, which means that the dif-
(Xw-Xm), the differential-exposure compo-
ferential exposure components in the decom-
nent, which represents the expected change in
positions are meaningful for deaths and other
distress among men if mean exposure were the
network events. When we combine these ef-
same among men as it is among women and the
fects with the influences of differential re-
slopes were equal to the average slope among
sponse it is possible to explain 78 percent
men and women.
(.163/.208) of the observed MLF-WLF dif-
The first thing to note in this table is that a
ference in distress and 30 percent (.101/.329) of
substantial part of the observed sex difference
the MLF-HM difference in distress.
in distress can be explained by the life events
considered here. This can be seen by compar-
DISCUSSION
I An exact decomposition would require us to
work separately with each of the 11 replications. We Our analysis has documented that previous
hesitated to do this as the parameter estimates in evidence of female vulnerability to undesirable
each replication are unstable. life events is confined to network life crises

This content downloaded from 92.241.95.11 on Thu, 01 Sep 2016 14:59:48 UTC
All use subject to http://about.jstor.org/terms
628 AMERICAN SOCIOLOGICAL REVIEW

including the death of a loved one. This means serious problem. Women are between 30 per-
that pervasive deficits in coping resources- cent and 50 percent more likely than men to be
like limited access to social support, personal- mentioned as helpers in such surveys (Kessler
ity predisposition or coping styles-need not et al., forthcoming).
be called on to understand the ways in which There is no direct evidence that support pro-
life events help create distress among women. vision is distressing, but indirect evidence sug-
Instead, we need to learn more about the ways gests that it might be in some situations. Case
in which network life crises impinge on the studies of helping transactions show that help-
emotional lives of women and more about how ers often do things that reduce their own dis-
men manage to avoid these damaging effects. tress, like minimizing the victim's problem or
A consideration of these issues should begin attempting to be inappropriately cheerful
by recognizing that women are doubly disad- (Dunkel-Shetter and Wortman, 1982; Wortman
vantaged by network events. They report more and Lehman, forthcoing), even though these
of them than men and they are more emotion- things make the victim feel more inadequate,
ally responsive to them than men. We consider isolated, or misunderstood.
these two issues in turn. Despite these findings, the available evi-
There are two main possibilities for under- dence is inconsistent with the view that sup-
standing the greater exposure of women to port provision by itself is capable of explaining
network events: they might actually know of the vulnerability of women to network events.
more network crises than men; and they might The contrary evidence comes from the previ-
define a wider range of people as significant ously mentioned surveys of help seeking,
others than men. We believe that both of these which show that men provide a substantial
possibilities have merit. Our belief that women amount of support to crisis victims. It is true,
know about more network events is based on as noted above, that women far outnumber
evidence in the help-seeking literature that men among informal helpers. Nonetheless,
people in need of emotional support are much help-seekers describe receiving considerable
more likely to seek out women than men as help from men, and they report that the types
confidants (see Kessler et al. [forthcoming] forof help received include listening, comforting,
a review). In many cases, these confidants are and talking things out-the very sorts of emo-
given exclusive information about the prob- tional support that are presumably most dis-
lems of those who seek their support. This tressing for providers. If support provision
means that women actually know about more were centrally involved in the distress caused
network events than men. by network events, we would expect to find
At the same time, it is difficult to examine some evidence of this among men. That we
the available evidence and conclude that dif- find no such evidence argues against this in-
ferences in knowledge about the occurrence of terpretation.
network events account entirely for sex dif- There are other possibilities. One is that
ferences in self-reported exposure. The fact providing support leads to distress only when
that women report deaths of loved ones more the provider is already overloaded with so
frequently than men, for example, is unlikely many demands for nurturance that his or her
to be due to special access to the information. capacities for keeping emotional distance
A more plausible interpretation is that women break down. Evidence exists that women have
consider more people important than men do. more demands of this sort made on them by
Our conception of sex differences in vul- their networks than men (Fischer, 1982; Gove
nerability to network events is a good deal and Hughes, 1984). In fact, some social net-
.more complex. One possibility is that the pro- works are so demanding that their female
vision of social support is centrally implicated: members are in better mental health if they are
women are more involved as supporters and isolated from other members of the network
this involvement is distressing. Unfortunately, (Cohler and Lieberman, 1980).
our life-event surveys did not probe respon- Another possibility is that support provision
dents about their involvement with the victims is of only secondary importance and that the
of network events that were reported. We must impact of network events on women stems
consequently use evidence -outside these sur- from emotional involvement in the lives of
veys to examine this interpretation. their loved ones. Evidence for this ability to
One central aspect of this interpretation has become involved is widespread. We know that
been documented: women are more likely to be women are more oriented to the needs of those
sought out as supporters during times of crises. around them (Gilligan, 1982) and that women
The evidence for this comes from surveys of have a more emphatic orientation than men in a
help seeking in which respondents are asked to wide range of experimental and natural situa-
describe who helped them during the last pe- tions (Eisenberg and Lennon, 1983). General
riod in their life when they needed help with a population surveys, furthermore, show that

This content downloaded from 92.241.95.11 on Thu, 01 Sep 2016 14:59:48 UTC
All use subject to http://about.jstor.org/terms
UNDESIRABLE LIFE EVENTS 629

women are more likely than men to cite the closely at life events. Women are exposed to
well-being of their spouses (Campbell et al., more network events than men and less
1976), parents (Brody, 1981), and children shielded from their emotional effects, just as
(Menaghan, 1978) as important sources of con- qualitative research and theoretical arguments
cern. have suggested that women are more burdened
This interpretation fits nicely with an elab- than men by daily hassles and provided with
oration of the analysis that was not reported fewer resources to cope with them within their
above. We reasoned that if women have a role situations.
wider field of concern than men, vulnerability Like the ongoing stresses linked to their
to network events should be smallest for roles, the network events that damage female
events that occur to close family members and mental health are more important for their con-
larger for events that occur to more distant stancy and ability collectively to overwhelm
others. This is exactly what we found in than for their individual emotional effects.
analysis of the New Haven data, the only Network events have quite small effects on
survey that allowed us to focus on the person distress compared to personal life crises like
to whom the network event occurred. In this marital disruption or income loss. However,
survey, men were as distressed as women by network events are much more common than
serious life crises that occurred to their the latter. The small accumulation of effects
spouses and children. Yet, as soon as we across a number of network crises creates the
moved beyond this small field of concern the distress found among women, a result that
impact of network events on men vanished, once again ties in with thinking about the stres-
while it persisted among women. ses associated with traditional female roles.
The overload and emotional-involvement Our analysis has concentrated on the un-
interpretations are probably both implicated in usual stress experience rather than the ordi-
the vulnerability of women to network events. nary. In doing this we have not accounted en-
The time and energy demands placed on a sup- tirely for the sex-distress relationship. We be-
porter are likely to lead to distress when they lieve that women are not only exposed to more
come on top of an already demanding set of acute stresses than men, but also to more of the
role responsibilities. This pressure is likely to daily stresses that are associated with routine
be exacerbated by a personal style that pro- role functioning. A consideration of the latter
motes emotional involvement and subordina- will be necessary before we can comprehen-
tion of personal needs to the needs of others. sively assess the part played by stress.
Even in the absence of the demands created by Nonetheless, the work presented here has
being a supporter, finally, deep personal con- taken us considerably closer to a synthesis of
cern for a person in the midst of a life crisis is research on role stress and life events. We
almost certainly capable of promoting personal have documented for the first time that
distress. women are exposed more than men to acute
life stresses which are centrally associated with
their nurturant roles, and that this role-related
difference is one important source of the men-
SUMMARY
tal health advantage of men. Deeper interpre-
We began by noting that most research on the tation of this exposure and vulnerability to
sex-distress relationship has emphasized the network events will require a detailed under-
central importance of male and female roles. standing of the ways in which networks make
We described how this research has for the demands on their female members during times
most part been indirect, based on the compari- of crisis. This sort of understanding is probably
son of men and women within different status a good deal easier to develop than an under-
configurations rather than on a detailed con- standing of the small day-to-day role demands
sideration of role demands and resources. We that differentiate the situations of men and
also described how, in response to this limited women.
sort of progress in the consideration of chronic We have also pinpointed a direction for fu-
stresses and fulfillments within role domains, ture research that is likely to provide clearer
some researchers have begun to study sex dif- insights into the differentiation of male and fe-
ferences in exposure and response to life male role stresses than we have been able to
events. glean so far from the analysis of ongoing life
In the elaboration of this research on life- situations. We are directed to an analysis of
event effects we have documented that the life-event effects on the network of the victim,
theoretical considerations about role demands on empathic reactions, on support, and on the
that have informed thinking about ongoing emotional costs of care giving. These are im-
stresses and fulfillments in the lives of men and portant areas for investigation for they appar-
women emerge in a new form when we look ently hold the key to understanding the perva-

This content downloaded from 92.241.95.11 on Thu, 01 Sep 2016 14:59:48 UTC
All use subject to http://about.jstor.org/terms
630 AMERICAN SOCIOLOGICAL REVIEW

sive mental health disadvantage of women in works in Town and City. Chicago: Univer-
our society today. sity of Chicago Press.
Fox, John W.
1980 "Gove's specific sex-role theory of mental
REFERENCES illness: a research note.' Journal of Health
and Social Behavior 21:260-67.
Al-Issa, Ihsan
Gilligan, Carol
1982 "Gender and adult psychopathology." Pp.
1982 In a Different Voice: Psychological Theory
84-103 in Ihsan Al-Issa (ed.), Gender and
and Women's Development. Cambridge:
Psychopathology. New York: Academic.
Harvard University Press.
Aneshensel, Carol S., Ralph R. Frerichs and Virginia
Gove, Walter R.
A. Clark
1972 "The relationship between sex roles, mari-
1981 "Family roles and sex differences in de-
tal status, and mental illness." Social
pression." Journal of Health and Social Be-
Forces 51:34-44.
havior 22:379-93.
1978 "Sex differences in mental illness among
Belle, Deborah
adult men and women." Social Science and
1982 "The stress of caring: women as providers
Medicine 12B: 187-98.
of social support." Pp. 496-505 in Leo
Gove, Walter R. and Michael R. Geerken
Goldberger and Shlomo Breznitz (eds.),
1977 "The effect of children and employment on
Handbook of Stress: Theoretical and Clini-
the mental health of married men and
cal Aspects. New York: Free Press.
women." Social Forces 56:66-76.
Brody, Elaine M.
Gove, Walter R. and Michael Hughes
1981 " 'Women in the middle' and family help to
1984 Overcrowding in the Household. New
older people." The Gerontologist 21:
York: Academic Press.
471-80.
Iams, Howard M. and Arland Thornton
Brown, George W. and Tirril Harris
1975 "Decomposition of differences: a cau-
1978 Social Origins of Depression: A Study of
tionary note." Sociological Methods and Re-
Psychiatric Disorder in Women. New York:
search 3:341-52.
Free Press.
Kessler, Ronald C.
Campbell, Angus, Phillip Converse and Willard
1979 "Stress, social status, and psychological
Rodgers
distress." Journal of Health and Social Be-
1976 The Quality of American Life: Perceptions,
havior 20:259-72.
Evaluations, and Satisfactions. New York:
1982 "A disaggregation of the relationship be-
Russell Sage.
tween socioeconomic status and psycho-
Cohler, B. and M. Lieberman
logical distress." American Sociological
1980 "Social relations and mental health among
Review 47:752-64.
three European ethnic groups." Research
1983 "Methodological issues in the study psy-
on Aging 2:445-69.
chosocial stress." Pp. 267-341 in Howard
Comstock, G. W. and K. J. Helsing
B. Kaplan (ed.), Psychosocial Stress;
1976 "Symptoms of depression in two com-
Trends in Theory and Research. New York:
munities." Psychological Medicine 6:551-
Academic Press.
63.
Kessler, Ronald C. and David F. Greenberg
Dohrenwend, Barbara S.
1981 Linear Panel Analysis: Quantitative Models
1973 "Social status and stressful life events."
of Change. New York: Academic Press.
Journal of Personality and Social Psychol-
Kessler, Ronald C., Jane D. McLeod and Elaine
ogy 9:203-14.
Wethington
1977 "Anticipation and control of stressful life
forth- "The costs of caring: a perspective on the
events: An exploratory analysis." Pp.
com- relationship between sex and psychological
135-86 in J. S. Strauss, H. M. Babigian and
ing distress." In Irwin G. Sarason and Barbara
M. Rolf (eds.), The Origins and Course of
R. Sarason (eds.), Social Support: Theory,
Psychopathology. New York: Plenum.
Research and Applications. The Hague:
Dohrenwend, Bruce P. and DeWitt Crandell Martinus Nijhof.
1970 "Psychiatric symptoms in community,
Kessler, Ronald C. and James A. McRae, Jr.
clinic and mental hospital groups." Ameri-
1982 "The effect of wives' employment on the
can Journal of Psychiatry 126:1611-15. mental health of married men and women."
Dunkel-Schetter, C. E. and C. B. Wortman
American Sociological Review 47:217-27.
1982 "The interpersonal dynamics of cancer: 1984 "A note on the relationships of sex and
problems in social relationships and their
marital status with psychological distress."
impact on the patient." In H.S. Friedman In James Greenley (ed.), Community and
and M.R. DiMatteo (eds.), Interpersonal Mental Health, Volume III. Greenwich,
Issues in Health Care. New York: Aca- CT: JAI.
demic Press.
Langner, Thomas S.
Eisenberg, Nancy and Randy Lennon 1962 "A twenty-two item screening score of psy-
1983 "Sex differences in empathy and related chiatric symptoms indicating impairment."
capacities." Psychological Bulletin 94: Journal of Health and Human Behavior
100-31. 3:269-76.
Fischer, Claude S. Makosky, Vivian P.
1982 To Dwell Among Friends: Personal Net- 1980 "Stress and the mental health of women: a

This content downloaded from 92.241.95.11 on Thu, 01 Sep 2016 14:59:48 UTC
All use subject to http://about.jstor.org/terms
UNDESIRABLE LIFE EVENTS 631

discussion of research and issues. Pp. Rosenthal, Robert


111-28 in Marcia Guttentag, Susan Salasin 1978 "Combining results of independent
and Deborah Belle (eds.), The Mental studies." Psychological Bulletin 85:185-93.
Health of Women. New York: Academic Shrout, Patrick E.
Press. 1981 "Scaling of stressful life events." Pp. 29-47
Menaghan, Elizabeth G. in Barbara S. Dohrenwend and Bruce P.
1978 "The effect of family transitions on marital Dohrenwend (eds.), Stressful Life Events
experience.' Unpublished doctoral disser- and Their Contexts. New York: Prodist.
tation, Department of Human Development Southwood, K. E.
University of Chicago. 1978 "Substantive theory and statistical interac-
Miller, J. B. tion: five models." American Journal of
1976 Toward a New Psychology of Women. Sociology 83:1154- 1203.
Boston: Beacon. Stroebe, Margaret S. and Wolfgang Stroebe
Myers, J. K., J. J. Lindenthal and M. P. Pepper 1983 "Who suffers more'? Sex differences in
1974 "Social class, life events and psychiatric health risks of the widowed." Psychological
symptoms: a longitudinal study." Pp. 191- Bulletin 93:279-301.
206 in Bruce P. Dohrenwend and Barbara Surtes, P. G. and J. G. Ingham
S. Dohrenwend (eds.), Stressful Life 1980 "Life stress and depressive outcome: appli-
Events: Their Nature and Effects. New cation of a dissipation model to life events."
York: Wiley. Social Psychiatry 15:21-31.
Paykel, E. S. Thoits, Peggy A.
1978 "Contribution of life events to causation 1981 "Undesirable life events and psycho-
of psychiatric illness." Psychological physiological distress: a problem of opera-
Medicine 8:245-53. tional confounding." American Socio-
Pearlin, Leonard I. logical Review 46:97-109.
1975 "Sex roles and depression." Pp. 191-207 in
Veroff, Joseph, Elizabeth Douvan and Richard
Nancy Datan and Leon H. Ginsberg (eds.),
Kulka
Proceedings of Fourth Life-Span Devel-
1981 The Inner American: Ae, Self-Portrait from
opmental Psychology Conference: Norma-
1957 to 1976. New York: Basic.
tive Life Crises. New York: Academic
Press. Wallerstein, J. and J. Kelly
Pearlin, Leonard I. and Carmi Schooler 1980 Surviving the Breakup. New York: Basic.
1978 "The structure of coping." Journal of Weissman, Myrna M. and Jerome K. Myers
Health and Social Behavior 22:337-56. 1978 "Rates and risks of depressive symptoms in
Radloff, Lenore S. a United States urban community." Acta
1975 "Sex differences in depression: the effects Psychiatrica Scandinava 57:219-31.
of occupational and marital status." Sex Wheaton, Blair
Roles 1:249-65. 1982 "Uses and abuses of the Langner index: a
1977 "The CES-D scale: A self-report depression re-examination of findings on psychological
scale for research in the general popula- and psychophysiological distress." Pp.
tion." Applied Psychological Measurement 25-54 in David Mechanic (ed.), Symptoms,
1:385-401. Illness Behavior, and Help-Seeking. New
Radloff, Lenore S. and M. K. Monroe York: Prodist.
1978 "Sex differences in helplessness-with im-
Wortman, C.B. and D. Lehman
plications for depression." Pp. 199-221 in
L.S. Hansen and R.S. Rapoza (eds.), forth- "Reactions to victims of life crises: support
Career Development and Counselling of com- attempts that fail." In Irwin B. Sarason and
Women. Springfield, IL: Charles Thomas. ing Barbara R. Sarason (eds.), Social Support:

Radloff, Lenore S. and Donald S. Rae Theory, Research and Application. The
Hague: Martinus-Nijhof.
1981 "Components of the sex difference in de-
pression." Pp. 77-110 in Roberta G. Sim- Zung, W.
mons (ed.), Research in Community and 1965 "A self-rating depression scale." Archives
Mental Health. Greenwich, CT: JAI Press. of General Psychiatry 12:63-70.

This content downloaded from 92.241.95.11 on Thu, 01 Sep 2016 14:59:48 UTC
All use subject to http://about.jstor.org/terms

You might also like