Journal of Petsonality and Social Psychology

1980, Vol. 38, No. 4, ó68-678

Influence of Extraversion and Neuroticism on Subjective

Well-Being: Happy and Unhappy People
Paul T. Costa, Jr. and Robert R. McCrae
Gerontology Research Center
National Institute on Aging, National Institutes of Health
Baltimore, Maryland

Three studies are reported that examine the relations between personality
and happiness or subjective well-being. It is argued that (a) one set of traits
influ- ences positive affect or satisfaction, whereas a different set of traits
influences negative affect or dissatisfaction ; (b) the former set of traits can
be viewed as components of extraversion, and the latter as components of
neuroticism; and
(c) personality differences antedate and predict differences in happiness over
a period of 10 years, thus ruling out the rival hypothesis that temporary
moods or states account for the observed relations. A model of individual
differences in happiness is presented, and the separate and complementary
roles of trait and adaptation-level theories in explaining happiness are
discussed.

Nowhere is the relevance of psychology to

and extends it. Data are provided in support
human concern more evident than in studies of several parts of the model.
of happiness or subjective well-being. Interest
in measuring the quality of life has lead re-
searchers (Andrews & Withey, 1976 ; Brad- Conceptualizing and Measuring Ha ppiness
burn & Caplovitz, 1965 ; Campbell, Converse, Researchers have used a variety of
& Rodgers, 1976 ; Cantril, 1965 ) to conduct measures that indicate something of the nature
national surveys of happiness and to examine and diversity of conceptions of subjective well-
the influence of social-structural or demo- being. Gurin, Veroff, and Feld ( 1960) adopted
graphic variables on perceived well-being. the most straightforward and intuitive method
Adaptation-level (AL) theory (Helson, 1964) of assessing happiness: They asked subjects
has been applied to explain individual percep- if they were “very happy,” “pretty happy,”
tions of happiness (Brickman & Campbell, or “not too happy.” To obtain a more dif-
197 1 ; B rickman, Coates, & Janoff-Bulman, ferentiated estimate of life satisfaction, Camp-
1978) . As a result of these studies, a number bell, Converse, and Rodgers ( 197ó) required
of issues have been clarified and a few unex- subjects to rate their satisfaction within each
pected findings replicated. The present article of 10 areas of life—job, marriage, family, and
is an attempt to summarize the state of cur- so on. Campbell ( 1976) regarded this strategy
rent knowledge on personality and happiness as a “cognitive” appraisal of life satisfaction,
and to offer a model of happiness that clarifies since it avoided any direct reference to feel-
ings or affects and allowed the subject to
assess his or her satisfaction according to his
This research was supported in part by the Coun- or her own standards and expectations.
cil for Tobacco Research—U.S.A., Inc., Grant 1085R2
and the Medical Research Service of the Veterans Cantril ( 1965 ) used a so-called self-anchor-
Administration (Normative Aging Study) . ing scale, in which individuals defined a
Requests for reprints should be sent to Paul T. “best life” and a “worst life” for themselves,
Costa, Jr., Chief, Section on Stress and Coping, and then rated their present life on this best-
Gerontology Research Center, National Institute on to-worst scale. A difíerent approach that has
Aging, National Institutes of Health, Baltimore City
Hospitals, Baltimore, Maryland 21224. attained considerable use by researchers was
developed by B radburn and his colleagues at

In the public domain

668
 INFLUENCE OF EXTRAVERSION AND NEUROTICISM ON HAPPINESS 669

the National Opinion Research Center (Brad-

burn, 1969 ; Bradburn & Caplovitz, 1965 ) . ing a general evaluation of the respondents'
Instead of asking for cognitive judgments of life-as-a-whole from an absolute perspective
the adequacy of one's life, he measured feel- tend to cluster together.... Measures that
ings or affects. Following the old idea that tap life-as-a-whole less generally ... show
happiness is the sum of pleasures minus positive relationships of varying strength to
pains, Bradburn developed an Affect Balance the core cluster” (Andrews & Withey, 1976,
Scale by subtracting negative afíects (bore- p. 76) .
dom, loneliness, depression) experienced dur- In marked contrast to the apparent ease of
ing the past two weeks from positive afects measurement in the domain of subjective
(pride, excitement, pleasure) . well-being are the conceptual enigmas that
Despite the variety of approaches used to have emerged in the course of research. The
this point, the scope and limits of the domain first of these is the meager relation between
of measures relevant to subjective well-being objective and subjective indicators of happi-
or happiness have not yet been established. ness or well-being. Common sense suggests
In particular, scales originated in clinical that wealth, youth, and social privilege should
contexts to measure dysphoric affect, hope- contribute substantially to happiness, and
lessness, or insecurity may measure what we much research has been devoted to an investi-
recognize within normal limits to be simple gation of this hypothesis. Yet Campbell
unhappiness. The phrase “psychological well- ( 1976) reports that only 17 % of life satis-
being” itself carries with it the connotation of faction is predictable from 10 demographic
mental health and has been so interpreted by indicators in a national probability sample.
many researchers (e.g., Moriwaki, 1974 ; Rob- Similarly, Andrews and Withey ( 1976), also
inson, 1969 ) . using national probability samples, account
This array oí alternative operationalizations for only 8% of the variance in life satisfac-
of happiness would be almost unmanageable tion using age, family cycle stage, family in-
were it not for one happy circumstance: The come, education, race, and sex as predictors,
scales all show reasonably high intercorrela- either singly or in combination. An even more
tions. Bradburn ( 1969) found that the Nega- dramatic instance of the apparent irrelevance
tive Affect Scale (NAS), the Positive Affect of objective circumstances to subjective well-
Scale (PAS), and especially the Affect Balance being is provided by Brickman, Coates, and
Scale (ABS) scores correlated significantly Janoff-Bulman ( 1978) , who report that lot-
with avowals of “very happy,” “pretty tery winners were no happier than controls on
happy,” or “not too happy.” Campbell present and estimated future happiness. Para-
( 197ó) , in a national probability sample, plegics, although somewhat less happy than
showed that his index containing judgments of controls, did not differ from lottery winners
life satisfaction in several areas of life corre- or controls in estimation of future happiness.
lated highly (r = .57) with an index of gen- The second problem is a paradox that has
eral affect based on semantic differential rat- never been fully explained. In 1969, Brad-
ings of life on such scales as interesting— burn reported that when positive and nega-
boring and enjoyable-miserable. tive affects are independently measured, the
Moriwaki ( 1974) demonstrated the close items form two independent clusters. Al-
kinship of morale scales to the subjective though positive and negative carry the strong
well-being domain when she reported a corre- mathematical suggestion of being opposite,
lation of .61 between the ABS and the Rosow Bradburn's PAS and NAS were not opposite
Morale Scale in a small sample of elderly (negatively correlated) but independent, vir-
persons. And in the most comprehensive study tually uncorrelated. Despite this, both positive
of the subject, Andrews and Withey ( 197 b) and negative affects were found to be associ-
correlated 68 measures and indices of happi- ated with overall estimates of happiness.
ness (including the Gurin, Cantril, and Brad- Using a slight modification of the three Brad-
burn scales) in five national probability sam- burn scales, Lowenthal, Thurner, and Chiri-
ples. They concluded that measures “involv- boga ( 1975 ) confirmed that positive and
nega- tive affect were independent
predictors of
 PAUL T. COSTA, JR. AND ROBERT R. McCRAE

global happiness. Similarly, Costa and Mc-

(Palmore & Kivett, 1977 ; Wolk & Telleen,
Crae (Note 1) found a median correlation of 1976) .
only — .11 between PAS and NAS across four
The prominence of the Bradburn scales has
administrations. Andrews and Withey ( 1976)
begun to encourage investigators to look for
also replicated the finding, using both the
personality correlates of positive and negative
three Bradburn scales and “cognitive” items
aNect separately, and some revealing trends
that required respondents to assess the
have begun to emerge. Moriwaki ( 1974) re-
“good” and “bad” aspects of their life sepa-
ported that a nine-item mental health scale
rately. It is not surprising that pleasant emo-
was significantly related to NAS but not to
tions enhance life satisfaction or that un-
PAS. Beiser ( 1974) found that reports of
pleasant emotions diminish it, but the re-
psychophysiological disorders were associated
peated observation that the pleasantness and
with a negative affect factor but not with a
unpleasantness of one's life are uncorrelated
positive affect factor in his instruments. On
is a puzzling phenomenon the explanation Io
the other hand, role planning and social par-
which is of considerable theoretical impor-
ticipation were associated with positive but
tance.
not negative affect factors. Recently, Bradburn
(Note 2) has pointed out a similar trend in his
Personalit y Correlal es oJ Sub)ective data. He reports that positive affect exclu-
Well-Being sively is related to social interest, sociability,
and activity and that negative affect only is
Personality descriptions of happy persons associated with psychosomatic symptoms,
generally resemble descriptions of psychologi- anxiety, poor role adjustment, and worries.
cal and social adjustment. Smith ( 1961 ), for These findings suggest a hypothesis that
example, lists as correlates of happiness: may explain the independence of positive and
optimism, warmth, emotional stability, socia- negative affect and meaningfully organize the
bility, and self-insight. Wessman and Ricks body of evidence on personality and happi-
( 1966) , in their intensive study of a small ness. It is hypothesized that one set of dispo-
sample of Harvard and Radcliffe students, sitions is responsible for positive affect or
point to large negative correlations with the satisfaction, whereas another, independent set
Minnesota Multiphasic Personality Inventory of dispositions influences negative affect or
D scale and the 16 Personality Factor (PF) dissatisfaction.
0 or “guilt-prone” scale in justifying their In Study 1 the relation between four mea-
conclusion that characteristically happier peo- sures of happiness and seven personality dis-
ple are well-adjusted, high in ego strength, positions hypothesized to be related to posi-
and high in self-esteem, as well as being tive or negative affect is examined. In Study
socially involved. 2 an attempt is made to clarify and organize
Wilson ( 1967) , in his studies, found social the results by testing the original hypothesis
and family adjustment and self—ideal congru- using measures of the broader dimensions of
ence to be important correlates of happiness extraversion (E) and neuroticism (N) . Fi-
and concluded his review of the literature by nally, in Study 3 happiness is predicted from
asserting that “happiness is consistently re- E and N data obtained 10 years previously.
lated to successful involvement with people”
(p. 304) . At the unhappy pole, a number of Study 1
investigators have found signs of psycho-
pathology or neuroticism among unhappy From the large number of traits with re-
people. Veroff, Feld, and Gurin ( 1962 ) re- ported associations to subjective well-being,
ported worry, anxiety, and psychosomatic some emerge as more likely to be associated
concerns among the correlates of unhappiness, uniquely with one side or the other of the
affect balance formula. Some of the specific
as did Bradburn and Caplovitz ( 1965 ) . Fi-
facets of temperament articulated by Buss
nally, self-rated health has also recurred as an
and Plomin ( 1975 ) fall into this category.
important predictor of subjective well-being
Buss and Plomin trace negative affects (par-
 INFLUENCE OF EXTRAYERSION AND NEUROTICISM ON HAPPINESS

ticularly fear and anger) to strength of emo- and surety or confidence” (p. 24) in a number of
tional drive, and they speculate that “If areas of life. Knutson reports validation studies in
there is temperamental input into individual which psychiatric patients scored significantly
differences in positive emotions, it is likely to lower than normals ; personal security was also
found to be positively related to occupational status.
be activity (low elation) or sociability (for A shortened, 16-item form of the Personal Security
friendliness and warmth) ” (p. 57). Some evi- Inventory was given as part of the fourth
dence supporting this idea is offered in terms mailing.
of daily level of mood ratings, and the divi- Those subjects who came to the study for medical
examinations during the data collection period were
sion of traits agrees generally with Bradburn's asked to complete the Life Satisfaction Index, For
(Note 2) observations. Study 1 tests the each of nine areas (work, health, money, appear-
hypothesis that the temperamental traits ance, self-respect, getting along with others, love,
of emotionality, fearfulness, hostility, and sex, and religious faith) subjects rated their satis-
faction on a S-point scale. Internal consistency
im- pulsivity will be associated with lower (coefficient alpha) for the summed score was .84 in
levels of happiness and especially with high our sample of 172.
negative affect, and that temperamental Scales from the EASI-III Temperament Survey
traits of so- ciability and activity will be (Emotionality/Activity/Sociability/Impulsivity ; Buss
associated with higher levels of happiness & Plomin, 1975) were included in the second mail-
ing. Each scale consists of five items and was de-
and, particularly, with positive affect. veloped through factor analysis and rating valida-
tion. Two- to three-month test-retest reliabilities
Method averaged .79 in a sample of 32 women ; self-reports
correlated .51 on the average with spouse ratings in
Subj «cts. The data reported were collected as a sample of 137 couples. Scales hypothesized to relate
part of a project on smoking and personality con- to NAS included general emotionality, fear, anger,
ducted in collaboration with the Normative Aging and poor inhibition of impulse ; scales hypothesized
Study, an interdisciplinary longitudinal study of to relate to PAS included sociability, tempo, and
health and aging in men (Bell, Rose, & Damon, vigor.
1972) . Participants, volunteers screened for health Analyses. Pearson correlations among the four
and geographical stability, ranged in age from 35 to happiness measures were used to examine evidence
85 at tht time of this research. The sample consists for convergent validity. Correlations were then com-
largely of white veterans, with all but the lowest puted between happiness and temperament mea-
socioeconomic groups well represented. A subsample sures. Finally, correlations between temperaments
of 1,100 men was contacted by mail and war asked and the components of ABS were calculated for each
to complete a series of four questionnaires mailed at of the four time points to examine the hypothesized
intervals of 3 months in 1976. Response rates were relations of the measured traits to PAS and NAS
79Qo, 82 Jo, 73Jo, and 549c for the four mailings. over four replications.
Additional information was obtained on a sub-
sample of 172 subjects visiting the study center for Results and Discussion
regular medical examinations during the data collec-
tion period. Results are based on all available cases,
with specific us given in each table.
Correlations between the three Bradburn
3fcosures. Four measures of happiness were col- scales and the alternative happiness measures
lected. The principal measure was Bradburn's (1969) at four times follow the pattern reported in
scales, which yielded scores for PAS, NAS, and the the literature: PAS and NAS scores are sig-
difference of these scores, ABS. The Bradburn mea-
sures were obtained at each of the four quarterly nificantly related to happiness measures in 23
mailings. An ABS sum score was calculated by sum- of 24 cases, but in every case ABS is more
ming the ABS scores over the four administrations. highly correlated with the Beck scale, the
The hopelessness Scale (Beck, Weissman, Lester, Knutson inventory, and the index than either
& Trexler, 1974) was given in the third question- of its components is. Of the Bradburn scales,
naire mailing. The scale was developed to assess
hopelessness or pessimism in clinical populations. the ABS thus appears to measure happiness
In a sample of 294 hospitalized patients, internal best.
con- sistency was found to be .93. Correlations with Table 1 presents the intercorrelations of the
clinicians' ratings of hopelessness ranged from .62 ABS Sum and the three other happiness
to
.74. Little evidence of the test's applicability to
mea- sures and suggests that different strategies or
nonpsychiatric populations has yet been provided. instruments for measuring subjective well-
The Personal Security Inventory (Knutson, 1952) being produce similar results. All correlations
war designed to measure personal security as the are significant and are generally high enough
subjective evaluation of "... success, to suggest convergent validity for the mea-
satisfaction,
672 PAUL T. COSTA, JR. AND ROBERT R. McCRAE

Table 1
Interco rrefatiotis oJ Happincss Measures will influence primarily positive affect, is
tested in Table 3.
For three of the temperament scales—gen-
Measure 1 2 3 4 eral emotionality, anger, and poor inhibition
1. ABS sum — .64** .61** .40**
of impulse—it is clear that only the negative
2. Personal security 524 — .59** .32** side of affect balance is substantially related.
3. Hopelessness• 529 552 — . t g• In each case, correlations with NAS are higher
4. Life Satisfaction than with the corresponding ABS score. The
IndeX ' 9’ 15
’ — pattern for the fear scale is not so clear: At
Note. us are given below the diagonal. ABS = each time, fear is more closely related to NAS
Aifect Balance Scale. than to PAS, but it does show a consistent
• Scale reflected to “hopef ulness.” effect on lowering PAS as well.
• p < .os. •• p < .ooi. Similarly, two of the positive temperament
scales—tempo and vigor—are associated with
sures. The Life Satisfaction Index appears to PAS but not with NAS, as predicted. Socia-
be a weaker indicator of happiness in this bility is more closely related to PAS than to
group, but Campbell's ( 1976) correlation of NAS, as predicted, but it also appears to
.57 between a similar index and a happiness have a consistent effect on NAS.
measure strengthens the argument for treat- Thus five of the seven scales appear to
ing the index as a measure of happiness. influence happiness by their impact on only
The temperamental correlates of happiness, one side of the affect balance equation. Two
shown in Table 2, are also consistent with the others show their primary effect on the hy-
literature. Happiness is positively associated pothesized affect component but also show
with sociability and activity and negatively some influence on the other as well.
associated with emotionality and impulsivity.
All of the 28 correlations are in the pre- Study 2
dicted direction, and 27 of them are statisti-
cally significant. The median correlation is The hypothesis that some traits influence
.24, a value that compares favorably with positive affect and some influence negative
multi plc correlations of .41 (Campbell, Con- affect was generally supported by the results
verse, & Rodgers, 1976) or .28 (Andrews & of Study 1. It is possible to take these results
Withey, 1976) reported by survey researchers one step further by noting that these traits
when demographic characteristics are used to have an internal organization and coherence.
predict life satisfaction. To those familiar with factor models of per-
The hypothesis that the scales of general sonality, the list of traits provided by Study
emotionality, fear, anger, and poor inhibition 1 (and much previous research) begins to
of impulse influence primarily negative affect, take the shape of two established dimensions
whereas sociability, tempo, and vigor scales of personality: extraversion (E) and neuroti-

Table 2
Correlations oJ Ternperamenl Scales With Happiness Measures

General
emotion- Poor
inhibition Socia-
Measure ality Fear Anger of impulse bility Tempo Vigor
ABS sum (529) —.33'** —.40’** —.21*** —.22'*• .32*'*
Hopelessness• (757) —.33*** —.41*** —.l9*•* —.23•** .25•** .13** .28*’*
Personal security (563) —.25*** —.40•** —.09• —.16*" .32*** .09•• .18*‘•
Life Satisfaction Index (149) —.21** —.32**’ —.12 —.l5• .24”* .12•* .28‘•*
/}A€3/A¥¥

Note. nz are given in parentheses. ABS = Affect Balance Scale.

• Scale reflected to “hopef ulness.”
‘ Q < .05. *’ § < .01. *'* § < .001.
 INFLUENCE OF EXTRAVERSION AND NEUROTICISM ON HAPPINESS
673
Table J
Correlations oJ Temperament ScaJes with Bradburn ScaJes at Four Times

General inhibition
Time It emotional it Fear Anger nf impulse Sociability Tempo Vigor
y
Positive Affect Scale
1 823 —.07** —.18’** —.00 —.03 .23*'* .17*** .23”*
2 903 —.06’ —.20*** —.02 —.06* .24*'* .21*** .24***
3 757 —.13*** —.22*** —.02 —.08* .24*•* .14**’ 23'•*
4 566 —.05 —.17*** .00 —.04 .22**’ .19•" :29’**

Negative Affect Scale

1 823 .30*** .26*** .24'** .21*" — .13*'• .05 — .06’
2 903 .38*** .34*** .26'** .28*'• — .20"' .05 — .06’
J 757 .34"‘ .28'** .21"' .22’** — .20“’ .02 — .06
4 566 .32"* .33"‘ .20"' .24”’ — .1J”' .04 —.08’

Affect Balance Scale

1 823 —.25*'* —.29*** —.17*** —.16*** .24*•* .08* .19***
2 903 —.29•** —.36*•' —.18’** —.23•*• .29*** .10*** .20***
3 75? —.28*'* —.31••* —.14*** —.18•*• .28**• .08• .18••‘
4 566 —.23*** —.32*•• —.12•** —.18*•* .23*'• .10*’ .24'**

• p < .os. •’ p < .0i. ••• p < .0oi.

cism (N ) . And indeed, factor analyses (Costa full N and E scores in a sample of 969 men. These
& McCrae, in press) showed that the EASI- 30 items were used as short-form N and E cluster
III scales of general emotionality, fear, anger, scales and were included in the third mailing. Addi-
and poor inhibition of impulse defined an N tionally, the standard Form A of the Eysenck Per-
sonality Inventory (EPI: Eysenck & Eysenck, 1964)
factor, whereas sociability, tempo, and vigor was included in the fourth mailing. The theoretically
formed part of an E factor. It is now possible independent dimensions of E and N were empirically
to propose a model of the relations between uncorrelated in the present sample (r .00, n 808
personality and happiness. Extraversion, to- for short-form 16 PF scales ; r m —.0z, a = 576 for
EPI scales) . Evidence of convergent validity is seen
gether with its component traits of sociabil- in the correlation of .65 (ti = 549) between the 16
ity, tempo, and vigor, predisposes individuals PF and EPI E measures and .68 (o = 553) between
toward positive affect, whereas neuroticism the two N measures.
(and hence general emotionality, impulsivity,
fear, and anger) predisposes individuals Results and Discussion
toward negative affect. The simplest test of
the model is direct correlation of measures of E Table 4 shows the Pearson correlations of
and N with happiness measures. E and N measures with the Bradburn scales
at four times, 3 months apart.
Method In all eight (Time x Measures) cases,
neuroticism or anxiety is more strongly corre-
3utijccts. Study 2 employed the same sample and
procedures as Study 1. lated with NAS than with either PAS or ABS.
Measures. Two measures of N and E were ad- In all eight cases, extraversion is more
ministered to the subjects by mail, as in Study 1. strongly correlated with PAS than with NAS
Cluster analysis of the Cattell Sixteen Personality or, in six of the eight cases, than with ABS.
Factor Questionnaire (16 PF) scales (Costa ñ Mc-
Crae, 197h) had shown an anxiety or N cluster and When E and N measures are correlated with
an E cluster that closely resembled the major second- the three alternative operationalizations of
order factors reported for the 16 PF (Cattell, Eber, happiness—hopelessness, personal security,
& Tatsuoka, 1970) . Multiple regression was used to and the Life Satisfaction Index—11 of the 12
identify the 15 items in Form A that best predicted correlations are statistically significant, and
674 PAUL T. COSTA, JR. AND ROBERT R. McCRAE

Table 4
Correlations of 16 P F and E PI Scales With Bradburn Scales tit Four Times

16 PF Short-Form EPJ scales

scales
Time N E N E

Positive Affect Scale

— .11*’* .16*** 753 - 11** .16*** 554
— .06 .22*** 757 - .08” .21*** 559
— .16”’* .19*** 808 .17*** 549
— .10** .25*** 556 .27*** 575

Negative Aifect Scale

1 .29”* — .03 753 .35”* — .01 554
.41*** —.04 757 .38*** — .01 559
.40•** —.13*** 808 .39’** — .05 549
4 .34'** —.12** 556 .43*’* — .07’ 575

Affect Balance Scale

1 _ 22s»+
.12"' 753 - .31 ***
2 — .31**' .17•** 757 - .J2*” .15*** 559
— .34*"
.20’** 808 - .J4*’* .15** 549
— .39’** .22*** 575
Note. PF — Cattell Sixteen Personality Factor Questionnaire. EPI = Eysenck Personality Inventory.
N = neuroticism. E = extraversion.
* p < .05. ** p < .01. •** p < .001.

all are in the predicted direction. Thus, E

heading of E come sociability, warmth, in-
and N not only influence the experience of
volvement with people, social participation,
positive or negative affect ; they also show
and activity. Under N come such character-
consistent correlations with measures of hap-
istics as ego strength, guilt proneness, anxiety,
piness that do not depend on direct reports
psychosomatic concerns, and worry. Extra-
of affective experience.
verted traits contribute to one's positive en-
Most factorial-based trait systems recognize
joyment or satisfaction in life, although they
N and E as the broadest and most pervasive
do not generally appear to reduce the un-
dimensions of personality. Eysenck (Eysenck
pleasantness of adverse circumstances. Neu-
& Eysenck, 1969 ) has devoted most of his
rotic traits predispose one to suffer more
research to an investigation of these two di-
acutely from one's misfortunes, but they do
mensions directly. Cattell ( 1973) sees them
not necessarily diminish one's joy or pleasures.
as second-order traits and has attempted to
measure the more molecular, first-order as-
pects of personality that form them. Guilford Study 5
( 1976) prefers to call the E cluster social Studies 1 and 2 made the causal assumption
actitit y and the N cluster emotional health, that personality influenced happiness or
but the similarity of these schemes is beyond subjective well-being—an interpretation that
question. The bulk of the literature on the simple correlations cannot themselves sustain.
personality correlates of happiness can be Although it appears unlikely that temporary
summarized by saying that more extraverted states of happiness would substantially alter
and more adjusted people are happier. The personality, it is plausible to argue that short-
characteristics listed somewhat indiscrimi- term moods or states may affect responses to
nately under the heading of “psychological personality measures. Perhaps an individual in
and social adjustment” can now be broken an upbeat mood will respond like an extravert,
apart into two discrete groups. Under the whereas the person who is temporarily de-
 INFLUENCE OF EXTRAVERSION AND NEUROTICISM ON HAPPINESS
675
pressed will score high on neuroticism. The EXTIIA¥£RSI0N:
long-term stability of E and N (Costa & Mc-
Crae, 1977, in press; Moss & Susman, in
press) argues against this interpretation, but a stsr:
more direct test is given by an examination of
the predictive relations between personality
measures and levels of subjective well-being
obtained 10 years later. Over this long a time
span, any systematic bias introduced by Fig4fre J. A model of personality influences on posi-
temporary moods or states should be elimi- tive and negative affect on subjective well-being.
nated. Predictive relations between E, N, and
happiness would thus strengthen the con- The personality traits found to be corre-
temporaneous evidence for the proposed model. lates of happiness have been grouped under
the headings of E and N. The direct
Method outcomes of these dispositions, according to
Subi•‹‹•. Subjects were a subset of those de- the model, are positive affect and negative
scribed in Study 1 who had been given the 16 PF affect, re- spectively. These two components
between 1965 and 1967. Data were available for are sub- jectively “balanced” by the individual
234 men.
Measures. An N and an E cluster score were to ar- rive at a net sense of subjective well-
obtained from analysis of combined A and B forms being, which may be measured as morale, life
of the 16 PF (Costa & McCrae, 1976) . (These scores satis- faction, hopefulness, or simply
formed the criterion in the selection of items for happiness.
the short-form 16 PF scales described in Study 2.)
These clusters resemble the second-order factors
Although it has been known for some time
reported by Cattell ; evidence for their validity and that positive and negative affect were inde-
stability is reported elsewhere (Costa & McCrae, pendent contributions to global happiness, no
i977). one has ever provided a fully satisfactory ex-
planation low this phenomenon. It is clear
Results and Discussion that there must be two independent sources
N cluster scores were significantly related of variation, two sets of causes operating to
to NAS (r —— .39, p < .001) and to ABS (r produce the two independent effects. In his
= —.30, § < .001) but not to PAS (r = initial attempt at an explanation, Bradburn
—.08, ni). E cluster scores, by contrast, were looked for objective sources. He suggested
not related to NAS (r = .03, nâ) but were that the situations that contribute to positive
related to PAS (r - .23, p < .001 ) and ABS afect are separate from those that contribute
(r —— .14, p < .03). Knowing an individual's to negative affect. It is easy to find examples
standing on these two personality dimensions that support the plausibility of this position.
allows a prediction of how happy the person Thus, poor health makes one unhappy, but
will be 10 years later. These data efectively good health is taken for granted, not seen as
rule out the alternative explanation that a source of positive joy. Conversely, a hobby
associations between happiness and personal- may bring us considerable pleasure, but few
ity result solely from the mediating effect of of us have hobbies that carry much potential
temporary moods or states. This finding is for pain.
also impressive as indirect evidence of the Plausible as the objective sources idea is, it
enduring effects of these dimensions of person- rests more on speculation than on fact. Some
ality. of the available facts even contradict it. For
example, we might expect that when we ask
General Discussion subjects to rate their level of satisfaction with
separate parts of their lif job, marriage,
A Model oJ Happiness money, religion, and so on—there would be
Figure 1 presents a model of happiness that little correlation between them, even though
accounts for the correlational data reported all might contribute to overall happiness.
here and in the literature. Instead, such items tend to intercorrelate
sub- stantially in our own data and that of
others
d7d PAUL T. COSTA, JR. AND ROBERT R. McCRAE

(e.g., Campbell, Converse, & Rodgers, 1976) .

show less zest, vigor, or enthusiasm is a
Regardless of the area of life, people tend to
thorny question, conceptual rather than
be either satisfied or dissatisfied. The two
empirical in nature. This caution applies
sources of variation must lie trithin the per-
particularly to social gerontologists (Lemon,
son, and the dimensions of E and N are prime
Bengtson, & Peterson, 1972; Neugarten,
candidates. Havighurst, & Tobin, 1961 ) who have used as
Figure 1 calls attention to the separability criteria of adjustment in old-age measures
of satisfaction (rom dissatisfaction, a phe- that include “zest” as a component of life
nomenon somewhat foreign to “common satisfaction.
sense” notions of happiness. We tend to
assume that these two components are
opposites and that more of one means less of Ha ppiness, Personalit y, and Adaptation Level
the other. The data show that reality is more
Few would argue against the position that,
complex. The differ- ence between common for normal people, the major determinant of
sense and the model is seen most clearly when
momentar y happiness is the specific situation
the traits of E and N are considered in in which the individual finds himself or her-
combination. Low N intro- verts and high N self, Social slights hurt our feelings, tooth-
extraverts may have similar levels of life aches make us miserable, compliments raise
satisfaction or happiness, but they achieve
our spirits, eating a good meal leaves us satis-
this result in utterly different ways. The former fied. The contribution of personality to any
are seldom depressed but just as seldom
one of these feelings is doubtless small. Yet
elated. The latter are prone to both extremes over time, the small but persistent effects of
and reach “average” satisfac- tion only because
traits emerge as a systematic source of varia-
there is as much satisfaction as dissatisfaction tion in happiness, whereas situational deter-
in their lives. In some re- spects, the two
minants that vary more or less randomly tend
groups are similar, but future studies should to cancel each other out (ct. Epstein, 1977) .
also be sensitive to the many differences On the other hand, the finding that traits
between individuals who may show the same predict happiness more successfully than such
level of subjective well-being.
enduring objective conditions as health,
Finally, Figure 1 points out a need for con- wealth, sex, or race is more problematic. We
ceptual clarification of the relation between all believe that we would be happier if we
happiness and mental health. Many research-
had more vigor, money, or power. Survey
ers consider the Bradburn scales a measure of
research data, however, show that these cir-
mental health, and our finding that the ABS
cumstances have very limited impact on sub-
is strongly related to two clinically validated
jective estimates of well-being. Brickman has
scales—the Beck Hopelessness Scale and the
proposed that adaptation-level theory can
Knutson Personal Security Inventory—might
account for these facts. Brickman and Camp-
reinforce this notion. If these scales were only
bell ( 197 I ) state as the fundamental postu-
associated with N, there would be little reason
to object to considering them as measures of late of AL theory that “the subjective expe-
adjustment, since N is clearly a conceptual rience of stimulus input is a function not of
correlate of mental illness. But the happiness the absolute level of that input but of the
scales also reflect E, whose conceptual relation discrepancy between the input and past levels”
to mental health is by no means unequivocal. (p. 287) . As applied to happiness, this means
The use of happiness scales as criteria of that the standards by which people judge the
adjustment portrays introverts as less men- pleasantness or unpleasantness of events or
tally healthy than extraverts; scientific re- circumstances are not absolute but relative, set
searchers should consider whether they wish and reset by the positive and negative experi-
to so penalize introverts. The independence of ences of the individual. According to this view,
E and N argues that introverts are no more habituation makes extreme circumstances (like
prone to anxiety, depression, or anger than are great wealth or great poverty) appear more
extraverts. Whether they should be considered normal to the individual concerned, who
lower in mental health simply because they comes to take advantages for granted or
learns to
 INFLUENCE OF EXTRAVERSION AND NEUROTICISM ON HAPPINESS

live with misfortunes. Additionally, AL theory Reference Notes

predicts that in contrast to extreme events,
more mundane experiences will be devalued. 1. Costa, P. T., Jr., & McCrae, R. R. The refatiotii
In this way, highly favorable circumstances between smoking motives, personalit y, and ]eel-
deprive the individual of many routine plea- ings, Progress R eporl III. Boston: University of
sures. Brickman found that lottery winners Massachusetts at Boston, 1977.
2. Bradburn, N. The measurement o f ps ycliologicol
took less pleasure in such small matters as well-being. In Jack Ellinson ( C h ha i re) ,alth
watching television than did control subjects gools and health indicators. Symposium
(although the correlative prediction that para- presented at the meeting of the American
plegics would take more pleasure in Association for the Advancement of Science,
mundane activities was not confirmed) . Denver, 1977.
AL theory can thus be used to explain the
small magnitude of the effects of objective References
circumstances on well-being. As a theory of
happiness, however, it fails to account for the Andrews, F. M., & Withey, S. B. Socia/ indicators o J
large observable individual differences in hap- well-being : Americans' perceptions o J ii/e qualit y.
New York: Plenum, 1976.
piness. Indeed, if happiness were solely the
Beck, A. T„ Weissman, A., Lester, D., & Trexler, L.
outcome of processes of adaptation, we would The measurement of pessimism: The Hopelessness
expect that all individuals would answer that Scale. Journal o J Consulting and Clinical Ps ychol-
they were “neutral” on the dimension of hap- og y, 1974, 42, 861—865.
piness. Brickman and Campbell speak gloom- Beiser, M. Components and correlates of mental
ily about the “hedonic treadmill” and give the well-being. Journal o f ?feofi£ and Social Beha ior,
1974, 15, 320-32 7.
impression that no one can remain happy for
Bell, B., Rose, C. L., & Damon, A. The normative
long. Yet Gurin reports that 35 % of the aging study: An interdisciplinary and longitudinal
population considers itself to be “very happy,” study of health and aging. A ging and Human De-
and the stability coefficients of well-being velopment, i 912, 3, S- i 1.
measures, which range from .4 to .5 (An- B radburn, N. M. The structure o f psychological well-
being. Chicago: Aldine, 1969.
drews & Withey, 1976 ; Costa & McCrae, Note B radburn, N. M., & Caplovitz, D. Re ports on hap-
1 ; Palmore & Kivett, 1977) , indicate that piness: A pilot stud y o J behavior related to mental
people tend to stay at the same relative level health. Chicago: Aldine, 1965.
of happiness over long periods of time. This B rickman, P., & Campbell, D. T. Hedonic relativism
relative stability in well-being is most easily and planning the good society. In M. M. Appley
(Ed.) , A daptoiioii Jevsi theory : A iytiipos*uwi.
interpreted as an outcome of the stability of
New York: Academic Press, 1971.
personality dimensions that underlie charac- B rickman, P., Coates, D., & Jan off-Bulman, R.
teristic levels of happiness. Lottery winners and accident victims: Is happiness
AL theory predicts that subjective judg- relative ? Journal o f Personalit y and Sociaf Ps y-
cliolog y, 1978, 3d, 917-92 7.
ments will be a function of the discrepancy of
Buss, A. H., & Plomin, R. A tetm9cratneat theory o J
present stimuli from a neutral point deter- personalit y de elopment. New York: Wiley, 1973.
mined by past experience. In the case of Campbell, A. Subjective measures of well-being.
happiness, it appears that this formulation is American Psycho/ogist, 1976, 31, 117—124.
insufficient. Constants representing the contri- Campbell, A., Converse, P. E., & Rodgers, W. L.
The qualit y o J American Ii/s: Perce ptions, erafuo-
butions of E and N must be added to the
tions, and sat is factions. New York: Russell Sage
value predicted by AL theory. People will Foundation, 1976.
adapt to changing circumstances, which in the Cantril, H. The poHrrs o/ £u»iati concerns. New
long run will neither add to nor detract from B runswick, N.J.: Rutgers University Press, 1965.
their happiness. But throughout these changes, Cattell, R. B . Personalit y and mood b y question-
tioire. San Francisco: J ossey-Bass, 1973.
the absolute advantages of being more extra- Cattell, R. B., Eber, H. W., & Tatsuoka, M. M.
verted or less neurotic will continue. We may T'he handb aok for Ihe Sixteen Personalit y Foci or
all be on hedonic treadmills, but the tread- Quesfiotirioire. Champaign, Ill.: Institute for Per-
sonality and Ability Testing, 1970.
mills of adjusted extraverts are much happier
Costa, P. T., Jr., & McC rae, R. R. Age differences
places to be. in personality structure: A cluster analytic ap-
proach. Journal o f G eront olog y, 1976, 3J, 564-570.
678 PAUL T. COSTA, JR. AND ROBERT R. McCRAE

Costa, P. T., Jr., & McC rae, R. R. Age differences in

personality structure revisited: Studies in validity, Four stages o J li]e. San Francisco: J ossey-Bass,
1975.
stability, and change. A ging and Human Develop-
tnent, 1977, 8, 261-2 75.
Moriwitki, S. Y. The affect balance scale: A valid-
Costa, P. T., Jr., & McCrae, R. R. Still stable after ity study with aged samples. Journal o] Geran-
! olo g y, 1974, 29, 73—78.
all these years: Personality as a key to some issues
in aging. In P. B. Baltes & 0. G. Brim, Jr. (Eds.) , Moss, H. A., & Susman, E. J. Constancy and change
Li/e matt development and behavior (Vol. 3) . in personality development. In O. G. Brim, Jr., &
J. Kagan (Eds.) , Constanc y and cJeiige iti £utiioti
New York: Academic Press, in press.
Epstein, S. Traits are alive and well. In P. Magnus- development. Cambridge, Mass.: Harvard Uni-
versity Press, in press.
son & N. S. Endler (Eds.) , Personalit y at the
crossroads: Current ‘usues in interactional ps y- Neugarten, B. L., Havighurst, R. J., & Tobin, S. S.
ecology. Hillsdale, N.J.: Erlbaum, 197 7. Measurement of life satisfaction. Journal o J Geron-
Eysenck, H. J., & Eysenck, S. B. G. 3fonoof o/ tie Iology, 1961, Jd, 134—143.
Eysenck personolit y invent or y. San Diego, Calif. : Palmore, E., & Kivett, V. Change in life satisfaction :
Educational & Industrial Testing Service, 1964. A longitudinal study of persons aged 46—70.
Eysenck, H. J., & Eysenck, S. B. G. Personalit y Journal o f G erontolog y, 1977, 32, 311—316.
structure and measurement. San Diego, Calif.: Robinson, J. P. Life satisfaction and happiness. In
Robert R. Knapp, 1969. J. P. Robinson & P. R. Shaver (Eds.) , Measures
Guilford, J. S., Zimmerman, W. S., & Guilford, J. o J social psychological attitudes. Ann Arbor, Mich.:
P. The Gu’i1Jord-Zimmerman temperament surge y Institute for Social Research, 1969.
handbook : Twent y- five years o] research and Smith, H. C. Personalil y adjustment. New York:
McGraw-Hill, 1961.
oppJicatioti. San Diego, Calif. : Educational & In-
dustrial Testing Service, 1976. Veroff, J., Feld., & Gurin, G. Dimensions of sub-
Gurin, G., Veroff, J., & Feld, S. Americans view their jective adjustment. Journal o J Abnormal and So-
mental health. New York: Basic Books, 1960. cial Ps ycholo g y, 19h2, d4, 192—205.
Helson, H. Adaptation-level tlieary. New York: Wessman, A. E., & Ricks, D. F. Mo od and person-
Harper & Row, 1964. ality. New York: Holt, Rinehart & Winston,
Knutson, A. Personal security as related to station in 1966.
life. Ps ychologicol M onogra phs, 1952, dd{4, Whole Wilson, W. Correlates of avowed happiness. Ps ycho-
No. 336) pp. 1—31. logical Bulletin, 1967, 67, 294-306.
Lemon, B. W., Bengtson, V. L., & Peterson, J. A. An Wolk, S., & Telleen, S. Psychological and social
exploration of the activity theory of aging: Ac- correlates of life satisfaction as a function of
tivity types and life satisfaction among in-movers residential constraint. Journal o J G «rout olo g y,
1976, 5/, 89-98.
to a retirement community. Journal o J Geron-
tO/0g'y, 1972, 2Z, S 11—523.
Lowenthal, M. F., Thurner, M., & Chiriboga, D.
Received April 4, 1979 •