You are on page 1of 8

PERSONALITY PROCESSES AND INDIVIDUAL

DIFFERENCES

Gender Differences in Negative Affect and Well-Being:


The Case for Emotional Intensity
Frank Fujita, Ed Diener, and Ed Sandvik
University of Illinois

Affect intensity (AI) may reconcile 2 seemingly paradoxical findings: Women report more negative
affect than men but equal happiness as men. AI describes people's varying response intensity to
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

identical emotional stimuli. A college sample of 66 women and 34 men was assessed on both
positive and negative affect using 4 measurement methods: self-report, peer report, daily report,
and memory performance. A principal-components analysis revealed an affect balance component
and an AI component. Multimeasure affect balance and AI scores were created, and t tests were
computed that showed women to be as happy as and more intense than men. Gender accounted for
less than 1% of the variance in happiness but over 13% in AI. Thus, depression findings of more
negative affect in women do not conflict with well-beingfindingsof equal happiness across gender.
Generally, women's more intense positive emotions balance their higher negative affect.

When researchers ask, "Which gender experiences more un- out arguing either for or against this position, we will examine
happiness?'' the answer is usually that women experience more evidence that suggests that the consideration of additional fac-
negative emotions than men. For example, in a nationwide sur- tors is also important.
vey in 1960, American women reported having more emotional Paradoxically, in most surveys, women report as much over-
problems than American men (Gurin, Veroff, & Feld, 1960). all happiness as do men. In the same 1960 American national
Women report more of the unpleasant affect symptoms of de- sample in which women expressed having more emotional dif-
pression than do men (Comstock & Helsing, 1976; Eaton & ficulties, they reported nearly the same overall level of happi-
Kessler, 1981; Frerichs, Aneshensel, & Clark, 1981; Husaini, ness as did men (Gurin et al., 1960). To the item "Taking all
Neff, & Stone, 1979; Levitt & Lubin, 1975; Murrell, Himmel- things together, how would you say things are these days—
farb, & Wright, 1983; Pearlin, 1975; Rosenfeld, 1980; Silver- would you say that you are very happy, pretty happy, or not too
man, 1968). Women report more dissatisfaction with their happy?" women's responses were similar to men's (Bradburn &
marriages and their health than men, and women state that Caplovitz, 1965, p. 9). In a European sample, women reported
they "have had less than their share of happiness in life" more being more satisfied with their overall life than did men (Haa-
often than do men (Campbell, 1981, p. 130). When interrupted vio-Mannila, 1971). In a cross-cultural study, Alex Michalos
while at home, school, work, or leisure, more women report (1987) surveyed 6,000 college students in 23 countries, and his
moods of unhappiness than do men (Cameron, 1975). These data revealed that women reported being as happy as or hap-
findings raise questions about the cause of women's unhappi- pier than men. Larson (1978) reviewed the well-being literature
ness. and found "No consistent sex differences in well-being for older
To explain women's unhappiness, Ellen McGrath, the chair persons on any type of measure." Acrossfivelarge quality-of-
of the 1987 American Psychological Association Task Force on life surveys conducted between 1972 and 1978, women consis-
Women and Depression, asked rhetorically, "How could any tently reported that they were happier than men, although the
woman not be depressed, at least sometimes, living in our magnitude of the difference was not considered large (Herzog,
current society?" (Landers, 1988). McGrath's implication is that Rodgers, & Woodworth, 1982). Thus, women frequently report
women's unhappiness is caused by their societal status. With- being just as happy as men, and women frequently report being
depressed at twice the rate that men do.
Given these seemingly contradictory findings, one possible
Preparation of this article was supported in part by National Insti- explanation would be that men and women are equally happy.
tute of Mental Health Training Grant MH-15140. We gratefully ac- Perhaps both genders experience emotional difficulty at about
knowledge the perceptive comments of Karla Brock and anonymous
reviewers on a draft of this article and the invaluable help of Dennis
the same rate, but men are more reluctant than women to admit
Gallagher during data collection. that they have a problem and to seek help for it. Studies that
Correspondence concerning this article should be addressed to Ed have attempted to verify this reticence in men, however, have
Diener, University of Illinois, Psychology Department, 603 East Dan- failed to do so (Amenson & Lewinsohn, 1981; Bryson & Pilon,
iel Street, Champaign, Illinois 61820. 1984; Clancy & Gove, 1974; King & Buchwald, 1982).
Journal of Personality and Social Psychology 1991, Vol. 61, No. 3, 427-434
Copyright 1991 by the American Psychological Association, Inc. 0022-3514/91/!

427
428 F. FUJITA, E. DIENER, AND E. SANDVIK

An alternative explanation would be that women are actually in this context refers to the percentage of time spent in an affec-
more unhappy than men. Given this explanation, women who tive state. Although affect intensity is uncorrelated with he-
are having emotional difficulties would report in surveys that donic level (Larsen & Diener, 1987), often items that tap both
they are happy because of a need to conform to societal norms constructs are used to measure negative affect. Thus, an item
—one of which is being happy. This explanation, however, like "I cry often" depends on both hedonic level and affect
makes it difficult to account for women's self-reported higher intensity. It is desirable, therefore, to separately measure affect
negative affect. If women falsely report being happy only to intensity and hedonic level when investigating gender differ-
appear being happier, they would not accurately report experi- ences in negative affect.
encing high levels of negative affect. Also, Diener, Sandvik, Pa- In the present study, we divided participants by gender and
vot, and Gallagher (in press) have shown that to the extent that measured hedonic level and affect intensity using four different
people with a high need for approval report higher levels of measurement procedures. Most of the research cited in the in-
happiness, they are actually happier. We propose that gender troduction used more representative and larger samples than
differences in affect intensity can explain the paradoxical pres- were used in the present study; unfortunately, however, pre-
ence of both the greater prevalence of negative affect and the vious research has often used a single self-report item as the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

equal (or greater) overall happiness reported by women. dependent variable. In addition, past research efforts have
Affect intensity is the individual difference variable that taken measures of either positive affect or negative affect, but
refers to one's response intensity to a given level of emotion-pro- seldom of both. Because our study includes multiple measures
voking stimulation (Larsen & Diener, 1987). People who experi- of positive and negative affect—most of which are not self-re-
ence high levels of negative emotional intensity also tend to port—greater confidence can be placed in thesefindings.Also,
experience high levels of positive emotional intensity (Larsen & the use of multiple measures helps to assure that artifacts do not
Diener, 1987). Diener, Colvin, Pavot, and Allman (in press) bias the results.
show that there are emotional costs to having very intense posi-
tive emotions. One of the costs is the experiencing of more
intense negative emotions. Method
If people who experience high levels of positive affect also
experience high levels of negative affect, it is possible that these Participants
people would report much more distress while also reporting Participants were 136 students at the University of Illinois who ob-
equal levels of overall well-being. Thayer and Miller (1988) pre- tained course credit to take part in extended psychological research
sent strong evidence of this independence of hedonic level (af- and to learn the method of scientific inquiry and empirical measure-
fect balance) and affect intensity. We propose that, in general, ment. The students completed many self-report measures, gathered
women are more affectively intense than men. This allows them observer-report data, completed memory performance measures of
to experience both more joy and more sorrow. These gender affect intensity, and produced daily reports of their activities, experi-
differences in affective experience can be used as a basis for ences, and emotions over a 42-day time period. The participants from
explaining the gender differences in depression found in com- whom data were analyzed included 34 men and 66 women. The data
munity surveys; moreover, affect intensity is equally proficient from other participants were not used because these data were incom-
plete. In addition, data from 1 woman and 5 men were completely
in explaining women's equal overall well-being as compared to removed from the data set because these data were largely incomplete
men because affect intensity is uncorrelated with hedonic level or showed evidence of being falsified.
(Larsen & Diener, 1987).
Often, when well-being is measured in large nationally repre-
sentative samples, it is with a single self-report item. One of the Measures
chief complaints about this type of measure is that the true
effects of well-being are hopelessly confounded with a self-pre- Three affect intensity measures and three hedonic level measures
sentational bias. When people say that they are happy because were used. The three affect intensity measures were based on three
they wish to appear normal, even when they are experiencing a very different methods, as were the three hedonic level measures. Af-
fect intensity was measured with the self-report Affect Intensity Mea-
considerable amount of distress, well-being is difficult to mea- sure (AIM), a modified AIM for observer reports, and a memory per-
sure. In the present study, positive intensity, negative intensity, formance measure of affect intensity. Hedonic level was measured us-
and global happiness were measured in four different ways. ing a self-report summary emotion adjectives scale, a modified version
Our self-report measures are subject to the possible confounds of this scale for observer reports, and a daily emotion adjectives scale
that plague earlier research (Goldsmith & Walters, 1989). The that was averaged over a 42-day time period.
observer report and the daily report, however, are free from AIM. The AIM (Larsen & Diener, 1987) is a 40-item instrument
confounding stemming from any weak form of social desirabil- that measures how intensely participants feel emotions. The AIM asks
ity bias in which the participant tells a "white lie" to make participants to answer questions like "When I feel happy it is a strong
himself or herself appear better. The memory performance type of exuberance," and "When I am nervous I get shaky all over."
measure is also free of any but the strongest form of bias (i.e., the Notice that these items do not assess how frequently the participant
participant repressing unhappy memories.) Thus, in our study experiences an emotion, but rather the intensity of the emotion when
experienced. The AIM yields both a positive and negative affect inten-
the gender differences were tested across measures that differ in sity score. Larsen & Diener (1987) report 1-, 2-, and 3-month test-retest
their susceptibility to a social desirability artifact. reliabilities of 0.80. These authors report that the AIM is not correlated
Another difficulty with previous research is that the effects of with either happiness or life satisfaction. Finally, the authors report
affect intensity and hedonic level are confounded. Hedonic level that a positive correlation between positive affect intensity and nega-
GENDER AND AFFECT INTENSITY 429

tive affect intensity is found in diverse samples (Larsen & Diener, Table 1
1987). Coefficient Alpha on Four Measurement Methods
Observer AIM. In addition to the self-report AIM, the AIM was also of Emotional Intensity
modified for this study for observer use by changing the instructions so
that the observer answered questions as he or she imagined the partici- Breakdown of measures
pant would answer them. From 7 to 11 observer reports were collected
for each participant. A minimum of three reports from college friends Positive Negative
and three reports from the participant's family were collected. These Measure affect affect
reports were averaged and used as the observer AIM score. In a multi-
Memory performance intensity .76 .68
trait-multimethod study, Larsen & Diener (1985) showed convergent
Observer intensity (AIM) .77 .70
validity across methods of measuring affect intensity and hedonic level Self-report intensity (AIM) .91 .74
and discriminant validity between affect intensity and hedonic level Daily hedonic level
using similar measures. (daily adjectives) .91 .86
Memory performance intensity measure. Participants were asked to Observer hedonic level
recall positive and negative emotional experiences in a timed period. (summary adjectives) .94 .86
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

In each memory category, participants were given either 2 or 3 min to Self-report hedonic level
This document is copyrighted by the American Psychological Association or one of its allied publishers.

write down as many memories of events as they could. The time de- (summary adjectives) .93 .82
pended on the category involved. The four memory categories tested
were lifetime happy events, lifetime unhappy events, last year happy Note. #=100.
events, and last year unhappy events. A simple count of the number of
positive memories recalled in the fixed time period was used for the
positive affect intensity measure. The negative intensity measure was
constructed using a simple count of negative memories recalled. The affect measure, on which women reported more positive affect
rationale was that more intense emotional experiences can be recalled than men did.
with greater ease; therefore, the recall of greater numbers of emotional Next, correlation matrixes were computed for the complete
episodes suggests greater emotional intensity. sample and separately within gender. The complete sample
Hedonic level measures. Twice during the semester, participants correlation matrix is presented in Table 3. The within-method,
estimated how frequently they experienced various emotions on a scale across-valence correlations are underlined. In the complete
of 0% to 100%. Each emotion word rating was then standardized across sample matrix, a significant positive correlation was found be-
subjects. The positive hedonic level measure was constructed by averag- tween the positive intensity and negative intensity measures
ing four positive emotion adjectives (happy, pleased, enjoyment/fun,
within measurement method. Also, significant negative corre-
and joyful). The negative hedonic level measure was constructed by
averaging four negative emotion adjectives (depressed/blue, worried/ lations were found between the hedonic level measures within
anxious, unhappy, and hostile/angry). Like the AIM, this measure was measurement method. This finding was expected because af-
also modified so that observers answered as if they were the partici- fect intensity has been found to be unitary across valence,
pants. Also like the AIM, 7-11 observer reports were collected, with a whereas positive and negative hedonic level make up a bipolar
minimum of 3 reports from college friends and 3 reports from the affect balance dimension. All affect intensity measures corre-
participant's family. These reports were averaged to yield observer he- lated positively with each other across valence and methods. In
donic level scores. addition, all of the hedonic level measures correlated in the
Daily mood report. For 42 consecutive days, participants returned a hypothesized direction across method.
form that asked them to rate on a 0 to 6 scale (0 = not at all, 6 =
extremely much) how much of several positive and negative emotions The men-only correlation matrix is presented in Table 4.
they had experienced that day. Each participant completed the entire Most of the correlations were in the expected direction. How-
form for both positive and negative emotions each day. Scores for each ever, one of the intensity measures correlated negatively within
emotion token were averaged across the 42 days for each participant measurement method (although insignificantly), contrary to
and then standardized across participants. The positive affect measure predictions. A few of the affect intensity measure correlations
was constructed by averaging four positive affect adjectives (happy, across method were also in the direction not hypothesized, but
pleased, enjoyment/fun, and joyful), and the negative affect measure these correlations were not significant. All hedonic level corre-
was constructed by averaging four negative affect adjectives (depressed/ lations, however, were in the expected direction.
blue, worried, unhappy, and hostile).
The women-only correlation matrix is presented in Table 5.
All within-measurement-method correlations were significant
Results and in the expected direction. All across-method affect inten-
sity measure correlations were positive, as predicted, and all
We first computed internal reliability coefficients. These are across-method hedonic level correlations were in the expected
presented in Table 1. As can be seen, the reliabilities were ac- directions.
ceptable for the affect intensity measures and proved very We submitted the complete sample correlation matrix to a prin-
strong for the hedonic level measures. cipal-components analysis to verify the existence of the two hy-
Means, standard deviations, and independent t tests for the pothesized factors, affect balance and intensity. Four compo-
12 affect measures are presented in Table 2. On five of the six nents were greater than one, with eigenvalues of 2.8,2.5,1.3, and
intensity measures (both positive and negative), the women's 1.2. Only the first two components were retained and were ro-
mean level was found to be more intense than the men's mean tated using a varimax rotation to simplify interpretation. In addi-
level. On only one of the six hedonic level measures did women tion, the application of the scree criterion would suggest
differ from men. This difference was on the self-report positive the presence of only two components. The varimax rotated
430 F. FUJITA, E. DIENER, AND E. SANDVIK

Table 2 variables by standardizing each measure and summing across


Descriptive Statistics on 12 Measures of Affect methods. The six intensity measures (positive self-report, nega-
Intensity and Balance tive self-report, positive observer report, negative observer re-
port, positive memory, and negative memory) were summed to
Breakdown of measures create a single multimethod affect intensity score. The coeffi-
Measure Positive Negative cient alpha of the multimethod intensity score was 0.70. This is
properly interpreted as a validity coefficient, the amount of
Memory intensity multimethod agreement about affect intensity. The six hedonic
Men level measures were combined by addition, after the signs of the
M 19.853 15.912 negative hedonic level measures were reversed, to create a sin-
SD 7.644 7.145
Women gle multimethod affect balance score. This is analogous to sub-
M 23.136* 19.227* tracting the negative affect from the positive affect to create an
SD 6.968 6.302 affect balance score. The coefficient alpha of the multimethod
t 2.159 2.380 affect balance score was also .70. This also is properly inter-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Observer intensity preted as a validity coefficient, because it is the amount of mul-


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Men
M 81.647 45.097 timethod agreement about affect balance. To compare the
SD 7.679 4.158 genders on the multimethod intensity variable, we performed
Women an independent sample t test yieldinga value of 3.90, p < .001.
M 86.270* 48.337** Likewise, to compare the genders on the multimethod affect
SD 9.784 5.280 balance variable, an independent sample t test was performed,
t 2.399 3.113
Self-report intensity yielding a nonsignificant value of 1.05, p > .10. Converting t
Men values into R2 or percentage of variance explained shows that
M 79.412 45.853 gender accounts for less than 1% of the variance in affect bal-
SD 12.517 7.394 ance, whereas it accounts for over 13% of the variance in affect
Women intensity.
M 86.379* 48.424
SD 13.017 7.553
t 2.568 1.624
Daily hedonic level Discussion
Men
M -0.168 0.672 The results of this study offer support for the idea that
SD 3.946 4.029 women in the United States experience emotion more strongly
Women than do men. The t tests provide evidence that women differ
M 0.165 -0.099 from men in the intensity of their emotions—both positive and
SD 3.616 3.124
negative emotions. However, there is little evidence of affect
t 0.423 1.058
Observer hedonic level balance differences between the genders. The principal-compo-
Men nents analysis, used as a condensation of the complete sample
M -0.423 -1.048 correlation matrix, provides evidence that affect intensity is
SD 3.273 3.027 unitary across affect valence. This means that ifa person experi-
Women ences strong negative emotions, that person is more likely to
M 0.711 0.198
SD 3.630 3.517 experience strong positive emotions. More specifically, in the
t 1.530 1.756 total sample correlation matrix, the positive and negative affect
Self-report hedonic level intensity measures correlated significantly and positively within
Men method. This gives stronger support for the idea that the people
M -1.055 0.379
SD 5.120 5.992 who experience strong positive emotions are the same people
Women who experience strong negative emotions. There were a few
M 1.532* -0.267 unexpected correlations in the men-only sample correlation
SD 5.858 4.980 matrix, although these may be attributable to the relatively
t 2.180 0.573 small sample of men.
Note, n for men = 34; n for women = 66. Hedonic level measures are The results of these studies help clarify past anomalies. Al-
sums of standardized items. though women scored significantly higher on almost all of the
*p<.05. **p<.01. intensity measures, they did not differ from men on most of the
hedonic level measures. Based on these findings, one would
predict that, if researchers only collect negative emotion data
component loadings are listed in Table 6. The first component and the items used tap both hedonic level and affect intensity,
(balance) was composed of the variables hypothesized to be they would find women experiencing more negative affect than
hedonic level variables, and the positive and negative variables men. One would also predict that, if researchers collect data
loaded with opposite signs. The second component (intensity) that balances the positive affect against the negative affect, then
was composed of the variables hypothesized to be affect inten- the gender differences will disappear. Thus, the paradox of past
sity variables, and the variables all had positive factor loadings. research is replicated; women score higher on negative affect
We constructed multimethod affect balance and intensity even though they are not lower on global happiness. This para-
GENDER AND AFFECT INTENSITY 431
Table 3
Complete Sample Correlation Matrix
Measure 1 2 3 4 5 6 7 8 9 10 11 12

Memory intensity
1. Positive 1.000
2. Negative .598 1.000
Observer intensity
3. Positive .206 .250 1.000
4. Negative .133 .229 .336 1.000
Self-intensity
5. Positive .189 .295 .518 .212 1.000
6. Negative .145 .175 .189 .444 .344 1.000
Daily hedonic level
7. Positive .168 .055 .349 -.058 .274 -.121 1.000
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

8. Negative -.079 -.091 -.124 .041 -.115 .247 -.370 1.000


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Observer hedonic level


9. Positive .139 -.210 .217 -.205 .020 -.038 .198 -.164 1.000
10. Negative -.120 .122 -.088 .462 -.058 .182 -.254 .053 -.475 1.000
Self-hedonic level
11. Positive .170 -.228 .219 -.051 .070 -.084 .290 -.186 .564 -.309 1.000
12. Negative -.136 .128 -.001 .207 .107 .251 -.182 .199 -.223 .369 -.416 1.000

Note. N = 100. Underlined correlations are within method, across hedonic level.

dox is resolved in our data, however, because the college women lated to the experience of negative affect, depression. In an
also scored higher on all positive affect intensity measures. excellent review, Nolen-Hoeksema (1987) explored many causal
Therefore, it appears that more intense positive emotions tend explanations for the gender differences in clinical depression
to counterbalance the more intense negative emotions when and proposed her own theory. We may use a parallel theory to
subjects report their global happiness. These results are highly apply to community survey findings of gender differences in
consistent with data based on national samples. For example, depression, which are mostly driven by differences in experi-
Wood, Rhodes, and Whelan (1989) reviewed a number of stud- enced negative affect (Newmann, 1984). That is, Nolen-Hoek-
ies in which women reported more positive affect than men; sema (1987) outlined a theory to explain gender differences in
greater reports of negative affect by women are many (e.g., clinical depression; we will use a similar theory to attempt to
Gurin et al., 1960). Thus, the present study using diverse mea- understand gender differences in negative affect—the differ-
sures converges with simple self-report studies that used more ences that create the gender differences found in community
heterogeneous subject samples. surveys of depression.
It may be profitable to consider these findings in a field re- Nolen-Hoeksema (1987) discussed the possibility that gender

Table 4
Men-Only Correlation Matrix
Measure 1 2 3 4 5 6 7 8 9 10 11 12

Memory intensity
1. Positive 1.000
2. Negative .622 1.00
Observer intensity
3. Positive .001 .147 1.000
4. Negative -.102 -.057 -.174 1.000
Self-intensity
5. Positive .071 .297 .283 -.124 1.000
6. Negative -.109 .035 .126 .268 .428 1.000
Daily hedonic level
7. Positive .219 -.121 .329 -.216 .041 -.140 1.000
8. Negative -.095 -.045 -.044 .320 -.169 .234 -.321 1.000
Observer hedonic level
9. Positive -.002 -.118 .116 -.346 -.039 -.177 .452 -.277 1.000
10. Negative -.075 .104 -.146 .338 -.000 .221 -.488 .120 -.588 1.000
Self-hedonic level
11. Positive -.168 -.349 .223 -.091 -.085 -.188 .569 -.125 .348 -.463 1.000
12. Negative .054 .278 .110 .143 .180 .203 -.056 .065 -.342 .354 -.368 1.000
Note, n = 34. Underlined correlations are within method, across hedonic level.
432 F. FUJITA, E. DIENER, AND E. SANDVIK

Table 5
Women-Only Correlation Matrix
Measure 1 2 3 4 5 6 7 8 9 10 11 12

Memory intensity
1. Positive 1.000
2. Negative .548 1.000
Observer intensity
3. Positive .239 .238 1.000
4. Negative .153 .279 .431 1.000
Self-intensity
5. Positive .182 .228 .569 .253 1.000
6. Negative .236 .204 .171 .483 .264 1.000
Daily hedonic level
7. Positive .128 .152 .363 -.014 .397 -.124 1.000
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

8. Negative -.031 -.086 -.136 -.045 -.045 .296 -.404 1.000


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Observer hedonic level


9. Positive .168 -.330 .214 -.238 -.011 -.014 .066 -.079 1.000
10. Negative -.208 .076 -.131 .470 -.152 .132 -.163 .049 -.487 1.000
Self hedonic level
11. Positive .276 -.268 .161 -.136 .061 -.096 .154 -.197 .631 -.322 1.000
12. Negative -.248 .054 -.034 .287 .095 .302 -.263 .296 -.151 .410 -.445 1.000

Note, n = 66. Underlined correlations are within method, across hedonic level.

differences may be one of the following: artifactually produced, that men will distract themselves when they experience a simi-
biologically or genetically determined, psychoanalytically lar unhappy event. In her review, Nolen-Hoeksema detailed one
caused, the result of sex-roles, or a product of learned helpless- of her earlier studies that showed that men report doing things
ness. She found support for each of these hypotheses to be to distract themselves when they are in a depressed mood,
equivocal, with the least support existing for the hypothesis that whereas women report focusing their attention on their de-
gender differences are artifacts. Increasing support was found pressed mood. This tendency, Nolen-Hoeksema hypothesized,
for the other theories listed, with the most support existing for makes the probability of clinical depression greater for women.
the learned helplessness theory. Nolen-Hoeksema went on to We view Nolen-Hoeksema's (1987) theory as applying more
propose a new theory. She posited that it is the way that women directly to findings in the community depression—which is
respond to a predepressive episode that causes the gender dif- conceptually distinct from clinical depression—literature,
ferences in clinical depression. Her hypothesis is that women where gender differences in depression are properly viewed as
ruminate about the causes of their predepressive episode, but gender differences in the amount of affective distress (New-
mann, 1984, 1986). Newmann (1984) dissects the Psychiatric
Evaluation Research Interview, a commonly used community
depression scale, and shows that the gender differences found
Table 6
in community surveys are due to differences in the experienced
Principal-Components Analysis: Rotated Loadings
sadness items and are not due to gender differences on the more
Balance Intensity clinical items. Although there are genuine gender differences in
Variable component component clinical unipolar depression, we see Nolen-Hoeksema's (1987)
rumination theory as applying more directly to findings of
Memory intensity
gender differences in community depression, or negative affect.
Positive .555
Negative .654 Thus, we agree that women generally ruminate about the
Observer intensity causes of their momentary sadness, which results in increased
Positive .695 sadness, whereas men generally distract themselves when they
Negative .601 experience a similar unhappy event. We posit, however, that
Self-report intensity further theoretical and empirical work needs to be done to
Positive .706
Negative .539 determine if gender differences in experienced negative affect
Daily hedonic level are the cause of gender differences in clinical depression.
Positive .574 An important theoretical question is, "What mechanism al-
Negative -.429
Observer hedonic level lows high affectively intense people to experience more negative
Positive .705 affect than low affectively intense people?" In 1987, Larsen and
Negative -.695 his colleagues showed that people with high affect intensity use
Self-report hedonic level the cognitive operations that Beck (1976) proposed (Larsen,
Positive .722
Negative -.604 Diener, & Cropanzano, 1987). Beck's operations are personaliz-
ing, overgeneralizing, and selective abstraction. Personalizing
Note. Loadings of less than .40 are not presented. is a cognitive operation in which a person interprets events in a
GENDER AND AFFECT INTENSITY 433

self-referential manner. Overgeneralizing is the cognitive opera- ence characteristic called emotional control. Emotional con-
tion in which a person takes a single event as being representa- trol-possibly identical to low affect intensity—is made up of
tive of the events in the world at large. Selective abstraction is a four factors: rehearsal, emotional inhibition, aggression con-
cognitive operation in which a person focuses his or her atten- trol, and benign control. Future research will discover whether
tion on the specific, emotion-provoking aspects of events. A affect intensity and emotional control differentially contribute
person who interprets events in a self-referential manner, takes to higher rates of depression. Finally, another research objective
single events as being representative of the world at large, and could be to collect longitudinal data on affect intensity, depres-
focuses attention on emotion-provoking aspects of events will sion-related cognitive operations, and depression. Then, one
probably experience more intense emotional outcomes than a would be able to construct a path analysis from gender to affect
person who does not. A review of these cognitive operations can intensity to cognitive operations to reported negative affect.
lead to an understanding of how people with high affect inten- This would determine if the causal path that is easiest to imag-
sity might be able to experience both more joy and more sad- ine is the one actually operating in society.
ness. The present study proposed an alternative perspective for
In such a review, Larsen et al. (1987) found that these pro- viewing the higher rates of sadness among women. Although
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

cesses were used in relation to both negative and positive stim- women may be at a greater risk of experiencing depressive
uli. Flett, Boase, McAndrews, Pliner, and Blankstein (1986) symptoms, women may also profit from greater affective inten-
also found that high AIM scorers have more vivid recall of their sity; a strong case can be made that women experience more
emotions. More vivid recall of past emotional experiences intense joy than men. Alternatively, men may be more inhib-
could contribute to the ruminative style proposed by Nolen- ited, incapable of reacting positively to pleasant events, and,
Hoeksema (1987). Of course, it is possible that the causal arrow therefore, may suffer a more drab emotional life. Similarly, it
could point in the opposite direction. It may be the case that the can be argued that men tend to be cut off from valuable emo-
ruminative style could influence encoding and recall of emo- tional information to a greater extent than are women. It is also
tion-laden events, leading to the cognitive operations reviewed important to note that there is no evidence that greater affective
above, and in turn, leading to greater affect intensity. intensity interferes with job performance or any other responsi-
Some researchers might be tempted to think that the use of bilities or capabilities. In general, we caution researchers
such cognitive operations is a "cognitive error." In a review of the against limiting their focus only to gender differences that per-
social judgment literature, however, Funder (1987) presents the petuate the depiction of women as more "fragile" creatures
thesis that although participants can often be forced to make than men. Specifically, we recommend against an exclusive
"errors" in the laboratory, the underlying processes that cause focus on the greater amounts of negative affect in women. We
the "errors" in the laboratory are not mistakes with respect to a recommend instead that the entire emotional spectrum of both
broader, more realistic frame of reference. Following from women and men be used as the focus of future research.
Funder's (1987) logic, we posit that people with higher affect
intensity may come to be that way because higher affect inten-
sity is generally more adaptive for them in their environments. References
The paradox that was the focus of this study seems to have
been clarified. The reason that women can be generally both Amenson, C. S., & Lewinsohn, P. M. (1981). An investigation into the
more depressed and more happy than men is that they gener- observed sex difference in prevalence of unipolar depression. Jour-
ally experience all emotions more vividly. Realizing the limits of nal of Abnormal Psychology, 90,1-13.
the present study, we offer it as a heuristic idea for future re- Beck, A. T. (1976). Cognitive therapy and the emotional disorders. New
search. One avenue of future research would be to replicate the %rk: International Universities Press.
present findings with a larger, more heterogeneous sample. Bradburn, N. M., & Caplovitz, D. (1965). Reports on happiness. Chi-
Diener, Sandvik, and Larsen (1985) report AIM gender differ- cago: Aldine.
ences across the life span, but thesefindingsare based solely on Bryson, S. E., & Pilon, D. J. (1984). Sex differences in depression and
the method of administering the Beck Depression Inventory. Journal
self-report data. Another area for research would be to sample
of Clinical Psychology, 40, 529-534.
emotions from around the Russell circumplex of emotion Cameron, P. (1975). Mood as an indicant of happiness: Age, sex, social
(1980) to determine if gender differences in affect intensity are class and situational differences. Journal of Gerontology, 30, 216—
consistent across emotions. Are there emotions that women 224.
experience less intensely than men? Do women report greater Campbell, A. (1981). The sense of well-being in America. New York:
amounts of the low-activation emotions (e.g., contented and McGraw-Hill.
calm) or only the more active emotions (e.g., elated, ecstatic, and Clancy, K., & Gove, W (1974). Sex differences in mental illness: An
joyful)? An additional line of research would be to trace the analysis of response bias in self-reports. American Journal of Sociol-
development of gender differences in affect intensity. For exam- ogy, 80, 205-216.
ple, Malatesta, Culver, Tessman, and Shepard (1989) report Comstock, G. W, & Helsing, K. J. (1976). Symptoms of depression in
that mothers react in more emotional ways to female toddlers two communities. Psychological Medicine, 6, 551-563.
Diener, E., Colvin, R., Pavot, W, & Allman, A. (in press). The psychic
than to male toddlers. Yet another avenue of research would be
costs of intense positive affect. Journal of Personality & Social Psy-
to measure the extent to which the gender differences in depres- chology.
sion are accounted for by greater female affect intensity and the Diener, E., Sandvik, E., & Larsen, R. J. (1985). Age and sex effects for
extent to which affect intensity does not account for these find- emotional intensity. Developmental Psychology, 21, 542-546.
ings. Roger and Nesshoever (1987) propose an individual differ- Diener, E., Sandvik, E., Pavot, W, & Gallagher, D. (in press). Response
434 E FUJITA, E. DIENER, AND E. SANDVIK

artifacts in the measurement of subjective well-being. Social Indica- Levitt, E. E., & Lubin, B. (1975). Depression. New York: Springer.
tors Research. Malatesta, C. Z., Culver, C, Tessman, J. R., & Shepard, B. (1989). The
Eaton, W W, & Kessler, L. G. (1981). Rates of symptoms of depression development of emotion expression during thefirsttwo years of life.
in a national sample. American Journal of Epidemiology, 114, 528— Monographs of the Society for Research in Child Development. 54,
538. (1-2, Serial No. 219).
Flett, G. L., Boase, P., McAndrews, M. P., Pliner, P., & Blankstein, K. R. Michalos, A. C. (1987). Final progress report on global report on student
(1986). Affect intensity and the appraisal of emotion. Journal ofRe- well-being: Applications ofmultiple discrepancies theory. Guelph, On-
search in Personality, 20, 447-459. tario: University of Guelph.
Frerichs, R. R., Aneshensel, C. S., & Clark, V A. (1981). Prevalence of Murrell, S. A., Himmelfarb, S., & Wright, K. (1983). Prevalence of
depression in Los Angeles County. American Journal of Epidemiol- depression and its correlates in older adults. American Journal of
ogy, 113, 691-699. Epidemiology, 117,173-185.
Funder, D. C. (1987). Errors and mistakes: Evaluating the accuracy of Newmann, J. P. (1984). Sex differences in symptoms of depression:
social judgment. Psychological Bulletin, 101, 75-90. Clinical disorder or normal distress? Journal of Health and Social
Goldsmith, R. E., & Walters, H. (1989). A validity study of the Affect Behavior, 25,136-159.
Intensity Measure. Journal of Social Behaviorand Personality, 4,133— Newmann, J. P. (1986). Gender, life strains, and depression. Journal of
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

140. Health and Social Behavior, 27,161-178.


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Gurin, G., Veroff, J., & Feld, S. (1960). Americans view their mental Nolen-Hoeksema, S. (1987). Sex differences in unipolar depression:
health: A nationwide interview survey. New \brk: Basic Books. Evidence and theory. Psychological Bulletin, 101, 259-282.
Haavio-Mannila, E. (1971). Satisfaction with family, work, leisure and Pearlin, L. I. (1975). Sex roles and depression. In N. Datan & L. Gins-
life among men and women. Human Relations, 24, 585-601. berg (Eds.), Life-span developmental psychology: Normative life crises
Herzog, A. R., Rodgers, W L., & Woodworth, J. (1982). Subjective well- (pp. 191-207). San Diego, CA: Academic Press.
being among different age groups. University of Michigan (Ann Ar- Roger, D, & Nesshoever, W (1987). The construction and preliminary
bor): Institute for Social Research. validation of a scale for measuring emotional control. Personality
Husaini, B. A., Neff, J. A., & Stone, R. H. (1979). Psychiatric impair- and Individual Differences, 8, 527-534.
ment in rural communities. Journal of Community Psychology, 7, Rosenfeld, S. (1980). Sex differences in depression: Do women always
137-146. have higher rates? Journal ofHealth and Social Behavior, 21, 33-42.
King, D. A., & Buchwald, A. M. (1982). Sex differences in subclinical Russell, J. A. (1980). A circumplex model of affect. Journal ofPersonal-
depression: Administration of the Beck Depression Inventory in ity and Social Psychology, 39, 1161-1178.
public and private disclosure situations. Journal of Personality and Seidlitz, L. (1990). Subjective well-being and memory for life events.
Social Psychology, 42, 963-969. Unpublished masters thesis. University of Illinois, Urbana.
Landers, S. (1988, October). Facts confound myths of women, depres- Silverman, C. (1968). The epidemiology ofdepression. Baltimore: John
sion. APA Monitor, p. 5. Hopkins Press.
Larsen, R. J., & Diener, E. (1985). A multitrait-multimethod examina- Thayer, J. F, & Miller, M. L. (1988). Further evidence of the indepen-
tion of affect structure: Hedonic level and emotional intensity. Per- dence of hedonic level and emotional intensity. Personality and Indi-
sonality and Individual Differences, 6, 631-636. vidual Differences, 9, 425-426.
Larsen, R. X, & Diener, E. (1987). Affect intensity as an individual Wood, W, Rhodes, N., & Whelan, M. (1989). Sex differences in positive
difference characteristic: A review. Journal of Research in Personal- well-being: A consideration of emotional style and marital status.
ity, 21,1-39. Psychological Bulletin, 106, 249-264.
Larsen, R. J., Diener, E., & Cropanzano, R. S. (1987). Cognitive opera-
tions associated with individual differences in affect intensity. Jour-
nal of Personality and Social Psychology, 53,161-11 A. Received November 22,1989
Larson, R. (1978). Thirty years of research on the subjective well-being Revision received June 20,1990
of older Americans. Journal of Gerontology, 33, 109-125. Accepted April 9,1991 •

You might also like