Psychological Bulletin .C--,t~ight 1989 by the American Psychological Association, inc.
1989, Vol. 106, No. 2, 249-264 0033-2909/89/$00.75
Sex Differences in Positive Well-Being: A Consideration
of Emotional Style and Marital Status
Wendy Wood, Nancy Rhodes, and Melanie Whelan
Texas A&M University
This article reviews all published studies reporting tests for sex differences in well-being. Women were
found to report greater happiness and life satisfaction than men. This sex difference was explained in
terms of men's and women's social roles: The female (vs. male) gender role specifies greater emo-
tional responsiveness. Furthermore, past role-related experiences provide women with appropriate
skills and attitudes. Women's (vs. men's) greater well-being was also found to hold for married but
not unmarried Ss: For both sexes the married state (vs. unmarried) was associated with favorable
well-being, but the favorable outcomes proved stronger for women than men. Given that most Ss
were married, the overall sex difference in well-being can be attributed to Ss' marital status. These
findings were discussed in the context of prior research on sex differences in negative well-being.
Research on subjective social indicators has demonstrated
that one's objective life circumstances do not necessarily corre-
spond to one's personal experience of well-being. The fact, then,
that men and women in our society differ in terms of a variety
of biological, personality, and situational factors may or may
not result in sex differences in subjective quality of life.
This article examines whether men and women do differ in
evaluations of their life as a whole. The inquiry is limited to
consideration of data on positive welbbeing and excludes find-
ings on negative affect and psychological symptomatology. This
is because positive and negative affect appem; under some cir-
cumstances, to be uncorrelated (Diener, Larson, Levine, & Em-
mons, 1985; Wart, Barter, & Brownbridge, 1983). I Reports of
positive wen-being are best interpreted as indicators of positive
domains of experience, separate from negative aspects of one's
life circumstances.
Prior research on sex differences in subjective life quality has
focused almost exclusively on negative affect and psychological
symptomatology. Consequently, most theories in this area are
tailored to explain the occurrence of men's and women's poor
well-being. This work, and the data on which it is based, is pre-
sented as a frame of reference for interpreting sex differences in
positive well-being.
First, we consider what is represented by judgments of well-
being. Philosophers and psychologists have debated this ques-
tion at length (see, e.g., Diener's, 1984, impressive review of the
This research was supported by Rockefeller Foundation Grant No.
RF84036, Allocation 31, to the first author. The authors thank Monique
Meeks for her assistance with library research, William A. Stock for his
gracious assistance in providing the list of references from his own work,
and Norval Glenn, George Levinger, Jamie Pennebaker, Jeffry Simpson,
and Wendy Stock for their helpful comments on an earlier version of
the manuscript.
Correspondence concerning this article should be addressed to
Wend), Wood, Department of Psychology, Texas A&M University, Col-
lege Station, Texas 77843-4235.
249
literature), but there are several points worth reiterating in the
present context.
Judgments of well-being are presumed to reflect an overall
evaluation of one's life circumstances. Such judgments are
probably holistic, in that they reflect an overall disposition to-
ward or against happiness and satisfaction, as well as particular-
istic, in that they reflect the outcomes of specific life domains,
such as marriage, friendship, and parenting (cf. Diener, 1984).
The holistic perspective reflects one's general approach to life
and is supported by findings that a general positive or negative
outlook permeates evaluations of more specific life domains.
For example, some research suggests that satisfactions with var-
ious domains do not significantly differ in predicting overall life
satisfaction, so that any domain performs about as well as any
other (Andrews & Withey, 1976). The particularistic perspec-
tive is supported by the alternate finding that when comparisons
are made among categories of persons, evaluations of domains
sometimes differ in their implications for aggregate well-being.
For example, happiness with the domain of marriage has been
found to be a more important predictor of overall happiness
Diener and his colleagues (Diener, 1984; Diener et al., 1985) distin-
guished between frequency and intensity of emotional experience to
explain the relation between positive and negative affect. At any given
point in time, the experience of either positive or negative emotion ap-
parently suppresses the experience of the other. Consequently, measures
that tap frequency of emotional experience will tend to reveal that posi-
tive and negative affect are inversely related (e.g., Carroll, Feinberg,
Smouse, Rawson, & Greden, 1981; Wart et al., 1983). Intensity of these
experiences, however, may be related such that people who tend to have
intense positive emotions will also have intense negative ones (e.g., Ep-
stein, 1983). Measures that tap emotional intensity are thus likely to
observe a positive relation between the two c~ of emotion. Finally,
the measures of overall happiness or life satisfaction that are found most
frequently in this review likely reflect both intensity and frequency of
affect. Because frequency and intensity appear to combine in an addi-
tive fashion to produce mean levels of affect, these measures of positive
well-being are likely to be statistically independent of assessments of
negative aspects of experience.
250 w. WOOD, N. RHODES, AND M. WHELAN
with life for women than for men (Glenn, 1975; Gove, Hughes,
& Style, 1983). In evaluating positive well-being, we consider
both of these perspectives.
Gender Roles and Emotional Experience
According to one approach, sex differences in emotions and
social behavior can be understood in terms of the social roles
filled by women and men (Eagly, 1987; Eagly & Wood, 1989;
Williams & Best, 1982). Men's and women's personal history
of enacting social roles is an indirect cause of sex differences
because of the influence that these experiences have on skills
and attitudes. Sex differences also stem from shared beliefs
about the likely and expected behavior of men and women,
termed
gender roles,
which derive from the association between
sex and social roles in the larger society. Thus, sex-differentiated
prior experiences cause men and women to have somewhat
different skills and attitudes, which, in conjunction with gender
roles, cause sex differences in social behavior and emotions.
This explanation of emotional experience in terms of social
roles bears some similarity to "constructivist" approaches that
view emotions as governed by social norms and rules (Averill,
1983).
The roles typically filled by men and women in our society
differ importantly in terms of emotional experiences. Enact-
ment of caretaker roles, which are typically filled by women in
the home (e.g., mother, wife) and in paid employment settings
(e.g., teacher, nurse), is likely to involve sensitivity to the needs
of others and emotional expression. Men's roles are less likely
to emphasize emotional experience. Prior role enactment and
socialization for particular roles thus plausibly will instill be-
liefs and skills that lead women to be more sensitive to their
own and others' emotions than men. In addition, the content of
social stereotypes is informative about gender role expectations
and emotions. Typical women are described as emotionally ex-
pressive, concerned with their own and others' feeling states,
and emotionally labile (Broverman, Vogel, Broverman, Clark-
son, & Rosenkrantz, 1972; Ruble, 1983). Typical men, in con-
trast, are believed to be emotionally stable and not excitable.
Thus women are attributed both greater emotional expressive-
ness and greater sensitivity to internal emotional events.
Given the emphasis on emotional experience in the female
gender role, we anticipate that women will report more extreme
levels of well-being than men. Indeed, women appear to report
greater affective-type disorders than do men, including greater
depression (Goldman & Ravid, 1980; Nolen-Hoeksema, 1987)
and greater personal discomfort and mental disorganization
(Gove & Tudor, 1973). This is not to say that women necessarily
experience greater mental illness than men, because more men
than women exhibit alcohol and drug abuse and personality dis-
orders (Belle & Goldman, 1980; Dohrenwend & Dohrenwend,
1976, 1977; Weissman & Klerman, 1977; however, see Coch-
rane, 1983).
Gender roles specify several possible sources of the sex
differences in negative well-being. If women are more willing
than men to report extreme emotions (cf. Phillips & Segal,
1969), then data collected from community surveys, depression
inventories, and treatment rates may reflect sex differences in
admission of symptoms rather than true symptom rates. Gove
and his colleagues (Clancy & Gove, 1974; Gove & Geerken,
1977; Gove, McCorkel, Fain, & Hughes, 1976) attempted to
assess response bias as an account of sex differences by estimat-
ing men's and women's perceived desirability of symptoms, ten-
dency to say yes and say no, and need for social approval. Statis-
tic,ally controlling for these possible biasing factors did not re-
duce the observed sex difference in reports of symptomatology;
women continued to report higher levels of disorder than did
men. Although these investigations did not directly evaluate the
mechanisms suggested by a gender-role analysis, they are infor-
mative in ruling out some artifactual influences on sex differ-
ences in reports of emotional events.
Several studies have attempted to directly manipulate the fac-
tors underlying willingness to report depressive symptomatol-
ogy (Bryson & Pilon, 1984; King & Buchwald, 1982). Male and
female college students were required to complete depression
inventories under public or private disclosure conditions. The
fact that disclosure setting did not affect reporting for men or
for women in this research suggests that men's reluctance to
publicly recognize symptoms may not underlie reported sex
differences in depression. Yet some caution is appropriate in
accepting this conclusion. Any sex differences in reporting are
likely to depend on a variety of factors, such as the nature of the
audience and the salience of gender-role norms. To what extent
these operate to effect a general tendency for sex differences in
reports of emotional experience remains unclear.
The female gender role also may specify that women be more
attuned to actual emotional experiences than men. Again, this
possibility has been invoked to explain women's higher inci-
dence of certain types of mental illness. For example, treatment
rates for depression may give a picture of excess pathology
among women because, when men and women experience a
comparable level of disturbance, women may seek medical as-
sistance more than men. Women may seek assistance for a vari-
ety of reasons, including the fact that they more readily than
men label low well-being and depressive symptomatology as
emotional problems (Kessler, Brown, & Broman, 1981). Alter-
natively, men and women may experience different psychologi-
cal symptoms; with depression, men have been known to sup-
press overt depressive responses and focus on physical rather
than psychological disorders (Hammen & Padesky, 1977; No-
len-Hoeksema, 1987). 2
Complementing this work on negative aspects of experience,
several lines of research have obtained sex differences in a gen-
eral tendency to indicate extreme emotional responses. Consis-
tent with our expectations, women report more extreme posi-
tive and (nonsignificantly) more extreme negative feelings on
Bradburn's Affect-Balance Scale (Bradburn, 1969), as well as
2 Sex differences in treatment rates could also be due to the differen-
tial belief that psychiatric help is necessary or actually seeking treatment
once a problem has been identified. Research is inconclusive on these
points. Some studies report findings suggestive of no sex difference in
these aspects of help-seeking (Amenson & Lewinsohn, 1981; Belle,
1980; Gove, 1978; Kessler, Brown, & Broman, 1981); others report that
women seek treatment at a higher rate than men (Link & Dohrenwend,
1980; Phillips & Segal, 1969); and still others report that men are hospi-
talized for certain psychological disorders at a higher rate than women
(Tudor, Tudor, & Gore, 1977).
SEX DIFFERENCES IN WELL-BEING
251
on other measures of affect (e.g., Smith & Kleugel, 1982).
Women have also been found to report more extreme levels of
fear, sadness, and joy than men (e.g., Allen & Haccoun, 1976;
Allen & Hamsher, 1974; Balswick & Avertt, 1977), although
this sex difference does not seem to hold for reports of anger
(Allen & Haccoun, 1976; Averill, 1983). Sex differences also
appear in the tendency to endorse extreme categories on scales
tapping intensity of response. An earlier review concluded that,
when significant sex differences appear, women typically exhibit
more extreme responses than men (Hamilton, 1968). Similarly,
at least one prior review of the literature on impression forma-
tion concluded that women use more extreme positive and neg-
ative evaluative terms than do men (Warr, 1971).3
However, two recent reviews examining positive well-being
are not consistent with our argument that the female (rather
than male) gender role involves greater emotional sensitivity
and expressiveness. A recent narrative review included 13 stud-
ies that compared men's and women's judgments of life satisfac-
tion and happiness and concluded that there appears to be no
difference in reported mean levels of positive well-being (Die-
ner, 1984). However, the small number of studies included in the
review, along with the likely conservatism of narrative reviewing
techniques (Rosenthal, 1984), might have led to an underesti-
mate of any sex difference effects. In contrast, a more extensive
meta-analytic review of 93 studies observed a slight tendency
for men to report higher levels of well-being than women (Har-
ing, Stock, & Okun, 1984). The mean correlation between sex
and well-being obtained in this review was r = .04 (95% confi-
dence interval [CI] = .01/.07); controlling for socioeconomic
status of the original participants did not appear to alter this
relation. However, there is also reason to question these conclu-
sions. The review included a number of summary measures of
well-being that assessed psychological adjustment as well as
multiple-item indices that tapped depression and psychoso-
matic complaints. Inclusion of data on negative well-being
would be expected to have a major impact on the findings be-
cause of the large magnitude of sex differences in this domain.
For example, a recent review by Nolen-Hoeksema (1987) re-
ported women to be twice as likely as men to experience depres-
sion. Thus Haring et al.'s (1984) results favoring men plausibly
reflect negative as well as positive well-being.
Marriage and Well-Being
The particularistic view of well-being noted earlier suggests
that overall assessments of happiness and satisfaction with one's
life are derived from evaluations of important life domains. We
examined this perspective on sex differences in well-being by
considering differences between men's and women's evalua-
tions of two primary areas of life: marriage and employment.
Perhaps the most consistent finding concerning the state of
marriage is its association with enhanced positive well-being
and attenuated negative outcomes for both men and women.
Married individuals report lower rates of psychological symp-
toms than do the unmarried, and they seek psychological ser-
vices less frequently (Gove, 1972). The effects associated with
marriage and positive well-being have been obtained with re-
ported happiness, life satisfaction, and aggregate indices of the
occurrence of positive and negative emotions (Bradburn, 1969;
Glenn, 1975; Gove et al., 1983; Haring-Hidore, Stock, Okun,
& Witter, 1985 ).
The benefits associated with marriage can be understood
from the perspective of role accumulation. Potentially, marriage
provides two additional roles, those of spouse and parent. Al-
though multiple role occupancy was, in early work, linked to the
experience of overload, conflicting demands, and psychological
distress (Coser, 1974; Goode, 1960), recent analyses have em-
phasized the positive consequences of multiple role involve-
ment (Marks, 1977; Sieber, 1974; Thoits, 1983, 1986). Accord-
ing to this view, multiple roles enhance perceived time and en-
ergy and confer a variety of rewards, including the privileges
of the various roles, overall status security, resources for status
enhancement and role performance, and personality enrich-
ment and self-gratification. To the extent that marriage repre-
sents an increase in roles for men and women, the enhanced
well-being associated with married (rather than unmarried) in-
dividuals can be understood as one example of the positive
effects associated with multiple-role occupancy?
Sex Differences in Experience of Marriage
The role of wife differs in a number of important ways from
the role of husband, and we anticipate that subjective well-being
varies as a function of these role differences and associated ex-
pectations. Specifically, wives (rather than husbands) are likely
to prove particularly skilled in and to value highly the emo-
tional experiences associated with marriage. Furthermore,
marriage is typically deemed a more important event in our
society for women than for men, as is implied by popular carica-
tures of marital roles (e.g., the eager bride and reluctant groom,
the playboy bachelor and lonely old maid). A diverse set of find-
ings from various literatures provides suggestive support for our
analysis.
The idea that women may be particularly skilled at and may
value emotional aspects of marriage is suggested by the consis-
tent finding that both marriage partners tend to report that
wives possess a better understanding of their husbands than hus-
bands do of their wives (e.g., Campbell, 1981). Direct observa-
tions of conflict resolution in couples suggests that wives often
play the role of emotional specialist. For example, in compari-
son with husbands, wives appear more likely to determine the
characteristic level of negative affect in marriage (Notarius &
Johnson, 1982). In addition, evaluations of the reciprocity of
3 The tendency for women to indicate more extreme judgments may
not be uniform across positive and negative ends of the judgment scale.
Specifically, women's judgments of others conform more closely than
men's to the "Pollyanna effect;' in that women are more likely to infer
additional positive attributes when others are described positively and
are less likely to infer additional negative attributes when others are
described negatively (Kohn & Fiedler, 1961; Wart, 197 l).
4 We recognize that marriage is also a unique role that conveys spe-
cialized advantages and disadvantages. For example, support from one's
spouse may confer unique rewards, such as a particularly effective buffer
to the stress experienced in other roles. Indeed, the greatest differences
between the married and the unmarried have been noted when these
persons are under conditions of economic hardship, social isolation, and
the responsibilities of parenthood (Pearlin & Johnson, 1977).
252 W. WOOD, N. RHODES, AND M. WHELAN
affect within couples suggest that wives are more finely attuned
to the quality of emotional interchange (Levenson & Gottman,
1985).
Women's greater involvement in marriage is apparent from
findings that happiness with marriage is a more important pre-
dictor of global happiness for women than for men (Glenn,
1975; Glenn & Weaver, 1981; Gove et al., 1983). Similarly, in a
previous report of how individual life domains contribute to
overall happiness, married women's top four rankings con-
cerned aspects of their marriage (i.e., being in love, marriage,
partner's happiness, and sex life), whereas married men placed
less emphasis on this aspect of their lives (rankings, in order,
are personal growth, being in love, marriage, job or primary
activity; Freedman, 1978). Additional evidence for women's in-
vestment in close relationships is provided by a review of the
effects of confidant support on the experience of stress (Cohen
& Wills, 1985). Women seemed to benefit from such support
more than men, and the benefits appeared primarily with the
availabi!ity of confiding husbands and boyfriends and not with
other types of confidants. Furthermore, sex differences have
been found in spouses' descriptions of marital relations.
Women in successful marriages characterize the relationship as
involving emotional security (i.e., affection, trust, and caring)
more than men do, whereas men describe their marriage as in-
volving loyalty (i.e., commitment to the future) more than
women do (Reedy, Birren, & Schaie, 1981).
We have argued that the female (rather than male) gender role
emphasizes the emotional aspects of marriage and that wom-
en's (rather than men's) past experiences yield greater invest-
ment and higher skill level concerning such relations. Conse-
quently, when positive well-being is assessed, we anticipate
wives reporting greater emotional advantages associated with
marriage than husbands. However, the available data are highly
inconsistent. A recent meta-analytic review observed that men
benefit more from the married (rather than unmarried) state,
as indexed by reports of positive well-being, than do women
(Haring-Hidore et al., 1985). However, this review drew on the
data set, mentioned earlier, that appears to have included indi-
ces aggregating across positive and negative well-being. Thus
the results are difficult to interpret. More supportive of our pre-
dictions are the results from national probability surveys that
assessed reports of happiness. In a number of investigations
spanning a 14-year period, married women reported greater
happiness than married men (Bernard, 1972; Glenn & Weaver,
1979, 1988).
The evidence relating marriage to sex differences in negative
well-being is more consistent. A picture of the "grim mental
health" of wives was popularized by Betty Friedan (1963) and
Jesse Bernard (1972) and further promoted by social scientists
adopting a feminist perspective (e.g., Seiden, 1976). Indeed, re-
search focusing on negative well-being has found that married
women experience higher rates of psychological disturbance
than married men in treatment rates for psychological distress
(Gove, 1972, i 978) as well as in self-reports ofsymptomatology
(Aneshensel, Frerichs, & Clark, 1981; Fox, 1980; Radloff,
1980). Such findings are consistent with our gender-role analysis
because we anticipate that women will be particularly sensitive
to the emotional distresses as well as rewards associated with
marriage (see Dion & Dion, 1985, Huston & Ashmore, 1986,
and Peplau, 1983, for more detailed discussions of differences
between men's and women's intimate relations).
Our explanation for sex differences in the experience of mar-
riage differs from prior accounts. For example, Bernard (1972)
labeled the apparent inconsistency in findings of elevated levels
of both distress and happiness among married women, in com-
parison with those among married men, the paradox of the
happy housewife. In her view, women are taught to equate the
marital role with happiness, and when positive outcomes are
assessed, the women do not recognize that it is causing mental
distress. A similar argument by Gove (1972) attributed wives'
psychological distress to deficiencies associated with traditional
roles, such as the following: (a) Married women typically have
only one source of personal gratification, their family, whereas
married men can derive rewards from both employment and
family roles; (b) married women are typically responsible for
housework, which is a low-prestige, unrewarding occupation;
(c) housework is also an unstructured activity that allows mar-
ried women to be self-absorbed and detached from the environ-
ment; (d) married women who are employed may experience
role overload because they are also responsible for household
chores, and they tend to hold low-prestige, unrewarding jobs;
and (e) role expectations for married women are diffuse, and
outcomes are highly dependent on others. Furthermore, house-
wives may experience frustration because they are typically not
using the skills and training they acquired at school (Darley,
1976). These views differ from the present analysis in a number
of features, perhaps most notably in implicitly according lesser
validity to the positive than to the negative outcomes associated
with the role of wife.
Sex Differences in Experience of Being Single
The idea that sex differences may exist in the well-being of
the unmarried has received less research attention. Because the
number and type of roles associated with the unmarried state-
for men and women should be highly similar, role theorists have
anticipated few sex differences in well-being for single people
(Gove, 1972). However, given that women in our society report
closer and more intimate same-sex friendships than men (Reis,
Senchak, & Solomon, 1985) and that such relations are impor-
tant to positive well-being, a difference favoring single women's
mental health may be anticipated. Alternately, given that em-
ployment should have a major impact on the well-being of sin-
gle individuals and that men typically have higher status, poten-
tially more rewarding work roles than women, single men may
experience greater well-being than women.
The available data on negative and positive well-being is not
easy to summarize. Although some studies on psychological
treatment rates and self-reports of symptomatology have found
greater disturbance among single men (Goldman & Ravid,
1980; Gove, 1972; Gur.in, Veroff, &Feld, 1960), others have
found greater disturbance among single women (Aneshensel et
al., 1981; Fox, 1980), others no difference (Thoits, 1986), and
still others variability among types of single people, including
divorced, widowed, and never married (Radloff, 1980). In re-
search on happiness, the sex difference appears to have varied
across time. In 1974, never-married men (in comparison with
women) reported lower well-being, but in 1986, single men's
SEX DIFFERENCES IN WELL-BEING 253
well-being apparently increased, and the sex difference reversed
in direction (Glenn & Weaver, 1988).
Present Research
These ideas about the overall sex difference in well-being and
the differential effects associated with marriage for men and
women were tested in a meta-analytic review of previous re-
search. When possible, we examined the effects associated with
one other social role, employment status. Our data generally
were uninformative with respect to this variable, and thus we
consider employment only briefly in this report.
A number of respondent characteristics may plausibly affect
any sex differences obtained. Age is one attribute reported in
most studies, and one previous narrative review concluded that
at younger ages the sex difference in well-being favors women
but that in later years the sex difference favors men (Diener,
1984). Consistent with this reasoning about the elderly, a recent
meta-analytic review observed a slight, positive association be-
tween age and well-being for both men and women except for
older women; among older women, increases in age were associ-
ated with decreases in well-being (Stock, Okun, Haring, & Wit-
ter, 1983). Age of the sample studied is thus evaluated as a possi-
ble moderator of sex effects.
The measurement procedures used to assess subjective well-
being were also evaluated as moderators of any sex difference.
Measures of well-being include life satisfaction, happiness,
morale, positive affect, and direct reports of well-being. The re-
lation among these various measures is open to debate. Life sat-
isfaction is sometimes considered a cognitive assessment of
well-being, in contrast to happiness and positive affect as more
emotional responses (Campbell, 1981). Empirical evaluations
of the structure of well-being have found variously that happi-
ness and life satisfaction both contribute to a single cognitive
dimension, with positive feelings providing a measure of affect
(Andrews & Withey, 1976), or that assessments of happiness
form a separate dimension from a more cognitive measure, la-
beled self-evaluation
(Bryant & Veroff, 1982). An alternate view
is provided by research that has simply correlated the various
measures of positive well-being; this strategy has revealed rea-
sonably high interrelations among the forms of assessment (e.g.,
Lohmann, 1977). In this review we evaluate the evidence for
sex differences separately by type of measure.
Distinguishing among measures in our sample of studies was
not always straightforward. In cases in which respondents' well-
being was evaluated through direct assessments of, for example,
happiness or life satisfaction, the appropriate category was ob-
vious. In many cases, however, multiple-item measures of well-
being were used, and item content was often similar for scales
intended to tap different domains of positive well-being. Even
more problematic, from our perspective, is that the multiple-
item scales often included items that appeared to tap negative
aspects of experience. We decided to include such assessments
in our review because they constitute such a large portion of
the available data on positive well-being, yet we anticipated that
women's (in comparison with men's) greater well-being should
be apparent only with measures that uniformly tap positive as-
pects of experience. It was difficult to anticipate the outcome
for more heterogeneous assessment techniques.
Method
Description of Data Set
This recta-analysis included every published study (in English) that
could be located that reported a measure of positive subjective well-
being for men and women. The final sample consisted of a total of 93
studies that provided a comparison between male and female subjects
for a measure of life satisfaction, happiness, morale, positive affect, or
general well-being. We excluded measures that tapped psychological ad-
justment, mental illness, psychosomatic symptoms, or physical health.
The final sample of studies was drawn from previous literature re-
views of research on subjective well-being (Diener & Griffin, 1984; Hat-
ing et al., 1984) and from computerized searches of
Psychological Ab-
stracts
from 1976 to 1985 and
Sociological Abstracts
from 1963 to 1985
using the key words
life satisfaction, happiness, and well-being.
For each study, an effect size (d) representing mean level of well-being
was calculated from a test of the sex difference. This statistic represents
the magnitude ofan effect and is calculated from the difference between
the means of the male and female groups divided by the within-group
standard deviation assumed to be common to the two populations
(Hedges & Olkin, 1985). Effect sizes associated with greater male sub-
jective well-being were given a positive sign, and those associated with
greater female well-being were given a negative sign. Sufficient informa-
tion was provided to calculate an effect size for 78 separate study sam-
pies. With seven studies, enough information was provided to compute
effect sizes for two different techniques of assessment, and thus the total
sample includes 85 effect size estimates.
In I 1 reports, the effect size for the sex difference was calculated from
an exact F or t. For 26 reports, the effect size was calculated directly
from means and standard deviations. In 13 reports the effect size was
calculated from a correlation coefficient. Proportions were given in 33
reports, and effect sizes were calculated via probit transformations
(Glass, McGaw, & Smith, 1981). In two reports the effect size was calcu-
lated from z scores.
Individual attributes such as educational level, race, and income
tended not to be reported systematically in the sample of studies in our
review, and thus it was not feasible to examine relations between such
attributes and sex differences in well-being. One strategy in such a case
is to aggregate across individual study samples (and thus across different
educational levels, races, etc.) and assume that the aggregated effect size
represents some average or typical outcome. There is reason to believe,
however, that aggregating the study samples in our review would yield a
biased representation of the U.S. population and thus potentially of the
sex-difference effect. Studies of well-being have been conducted dispro-
portionately on the elderly and on those experiencing psychological and
physical disabilities. Furthermore, studies in our sample included sev-
eral non-Western cultures and other specialized populations (e.g., high
academic achievers, institutionalized adults). For this reason we decided
to validate the analyses conducted on the total sample of studies with
separate analyses on only those studies using samples representative of
the U.S. population (n = 18). The validation studies included national
probability surveys as well as random samples from more specific geo-
graphic regions of the United States (i.e., in those cases in which the
sample was representative of a cross-section of the population). These
analyses are reported only when they differ from the analyses on the
total data set.
Variables Coded
The following variables were examined for each study in the sample:
(a) date of publication, (b) mean age of subjects, and (c) type of well-
being measure (life satisfaction, happiness, positive affect, or general
evaluationna broad category assessing morale, general well-bein~ and
other single-item scales that could not be classified into one of the other