Journal of Personality and Social Psychology Copyright I 87 by ihc American Psychological Association, Inc.

1987. Vol. 52, No. 1. 11-17 0022-3514/87/S00.75

Narcissism: Theory and Measurement

Robert A. Emmons
University of Illinois at Urbana-Champaign

Lack of a suitable measuring device hampered the empirical study of narcissism until Raskin and
Hall (1979) developed the Narcissistic Personality Inventory (NPI). The NPI possesses desirable
psychometric properties, and in this article I used the scale in a variety of studies. Factor analysis of
the scale replicated the four-factor solution found by Emmons (1984): Leadership/Authority, Self-
Absorption/Self-Admiration, Superiority/Arrogance, and Exploitiveness/Entitlemcnt. The Exploi-
tiveness/Entitlement subscale was found to correlate with measures of pathological narcissism and
affective intensity and variability. The relevance of Linville's( 1982) theory of self-complexity-affect
intensity for understanding aspects of narcissism is outlined. Implications of the study of narcissism
for attribution theorv and research are discussed.

Although the 1970s were characterized as the "me genera- comes. Greenwald (1980) referred to this phenomenon as be-
tion," interest in narcissism shows no signs of abatement in the neffectance and included it among two other cognitive biases
1980s. Three dominant trends can be noted. One trend focuses (egocentricity and cognitive conservatism) with narcissistic
on narcissism as a cultural or societal entity, contending that overtones that characterize the cognitive processes of indi-
society is becoming increasingly narcissistic (Lasch, 1979; Maz- viduals.
lish, 1982; Nelson, 1977; Stern, 1980). One has only to look at The third trend focuses on narcissism as a clinical entity. Cur-
the popularity of such books as The An of Being Selfish and rent psychoanalytic perspectives of narcissism can be found in
Looking Out for Number One to see that a major segment of the writings of Kemberg (1976, 1980) and Kohut (1976).
society has become increasingly self-absorbed. Wallach and Differences between the two regarding the etiology and treat-
Wallach (1983) traced the impact of various psychological ment of the narcissistic personality have resulted in a lively de-
schools of thought (Freudian, neo-Freudian, humanistic) on the bate (Millon, 1981). Kernberg sees narcissism developing as a
increasing prevalence of selfishness and egoism in society today. consequence of parental rejection or abandonment. This paren-
The implications of such a trend should not be underestimated. tal-devaluation hypothesis states that because of cold and re-
It has been suggested that continuous self-seeking may lessen jecting parents, the child defensively withdraws and comes to
an individual's willingness to pursue common social objectives believe that it is only himself or herself that can be trusted and
(Kanfer, 1979). Also, the potential for social conflict may in- relied on and therefore loved. Kernberg adheres to a stage model
crease as a result of this trend. For example, Fichten (1984) of libidinal development where difficulties arise when there is
found that attributions in distressed marital partners reflected regression in the developmental sequence of undifferentiated li-
narcissistic or egotistic biases. Furthermore, such conditions as bido followed by autoeroticism, narcissism, and then object
racism, sexism, and nationalism can be viewed as examples of a love, with narcissistic individuals not reaching the final stage.
narcissistic tendency manifested at group levels. Fromm (1973) Kohut, on the other hand, does not see narcissistic libido as
spoke of group narcissism as a sublimation of individual narcis- being transformed into object love, but rather sees it as follow-
sism. The individual satisfies his own narcissistic cravings by ing its own course of development into adulthood. Kohut's the-
belonging to and identifying with a group, such as a political or ory is actually a developmental theory of the self, where patho-
religious group. logical narcissism can result from failure to idealize the parents
A second trend that can be noted is social psychology's bur- because of rejection or indifference. Yet a third recent theory
geoning literature on a phenomenon known as the self-serving has been espoused by Millon (1981) and is what he calls a social-
bias (Harvey & Weary, 1984; Snyder, Stephan, & Rosenfield, learning theory of narcissism. This view sees narcissism devel-
1978). This refers to the tendency for people to accept responsi- oping not as a response to parental devaluation but rather as a
bility for successful outcomes and to deny blame for failed out- consequence of parental overvaluation. The child is treated as
a special person, provided with a lot of attention, and led by
parents to believe he or she is lovable and perfect. According to
Millon (1981), such unrealistic overvaluation will lead to self-
I would like to thank Randy Larsen for his assistance in data collec- illusions that "cannot be sustained in the outer world" (p. 165).
tion and analysis, and Laurie Emmons and Laura Faynor for their help
Often the child is either the firstborn or is an only child, which
in data coding. I am also indebted to an anonymous reviewer whose
contributes to the abundance of attention and special treat-
insightful comments helped to improve the quality of this article.
Correspondence concerning this article should be addressed to Rob- ment.
ert A. Emmons, who is now at the Department of Psychology, Michigan Although both the cultural and psychoanalytic approaches
State University, Psychology Research Building, East Lansing, Michi- are rich in theoretical speculations, they are both fraught with
gan 48824. difficulties. Anyone familiar with psychoanalytic formulations
12 ROBERT A. EMMONS

knows that assumptions and conjecture are often treated as ab- out reciprocation, and (e) interpersonal exploitiveness (Ameri-
solute truths, when often in reality they could not even be sub- can Psychiatric Association, 1980). Although the inventory is
jected to empirical scrutiny. The cultural view would hold that based on the DSM-III criteria, it is only extreme manifestations
there is something unique about this period in history to distin- of those behaviors that constitute pathological narcissism, and
guish it as the age of narcissism. However, if narcissistic traits the assumption is that when exhibited in less extreme forms
are formed in childhood, and there is ample agreement that these behaviors are reflective of narcissism as a personality trait.
they are, how can society be fostering narcissism in its mem- This assumption seems warranted because social critics such as
bers? Although Lasch's (1979) intent was to state a cultural Lasch (1979) have argued that narcissistic personality charac-
trend, Mazlish (1982) held that it is unjustified to describe an teristics are prevalent in the general population. Fischer (1984)
entire culture with a single clinical concept. Also, it is unclear refers to this form of narcissism as subclinical narcissism. The
as to whether the prevalence of narcissism has actually in- NPI is to date the only objective self-report inventory of narcis-
creased or whether narcissistic individuals are simply more visi- sism as a normal personality trait. Its creation has opened the
ble today (Dervin, 1982). Millon (1981) states the problems as- door for the empirical investigation of narcissism.
sociated with both approaches quite succinctly: "The viability
of the narcissistic personality does not stand or fall on the vagar-
Previous Studies Using the NPI
ies of the future of psychoanalysis. . . nor does its validity rest
on the passing character of contemporary life styles" (p. 165). Several studies have now been conducted with the NPI.
With all of the current interest in narcissism, it is unfortunate Raskin and Hall (1981) reported an 8-week alternate-form reli-
that empirical research on narcissism has lagged so far behind. ability of .72, and these authors also found that scores on the
Little progress can be expected to be made in this area unless NPI were positively related to Eysenck's extraversion and psy-
testable hypotheses are formulated and subjected to empirical choticism scales. Raskin (1980) found that there was a small
scrutiny. This article is an initial attempt to provide a founda- but significant correlation between narcissism and creativity,
tion on which an empirical theory of narcissism can be built. using the Barren Symbolic Equivalents Test. The same author
(Raskin, 1981) found that NPI scores were positively related to
the use of first-person singular pronouns and negatively related
The Measurement of Narcissism
to the use of first-person plural pronouns. Emmons(1981), in-
There have been several attempts to construct an individual vestigating the relation between narcissism and sensation seek-
difference measure of narcissism, and the results have been ing, obtained significant correlations between the NPI and dis-
mixed. Many of these have been projective instruments, such as inhibition, experience seeking, and boredom susceptibility. La-
the Thematic Apperception Test (TAT; Grayden, 1958; Harder, Vopa (1981) found that narcissism was positively related to
1979; Young, 1959) and the Rorschach (Exner, 1969; Harder, Machiavellianism in women but not in men and also found that
1979; Urist, 1977). Ashby, Lee, and Duke (1979) reported the NPI scores were uncorrelated with scores on the Marlowe-
development of an MMPI Narcissistic Personality Disorder Crowne Social Desirability Scale. Emmons (1984) factor ana-
Scale (NPD), consisting of 19 items from the MMPI. Solomon lyzed the NPI and uncovered four separate factors that he la-
(1982) found that the NPD distinguished between individuals beled Exploitiveness/Entitlement, Leadership/Authority, Supe-
with healthy and pathological self-esteem. The Millon Clinical riority/Arrogance, and Self-Absorption/Self-Admiration. He
Multiaxial Inventory (MCMI; Millon, 1982) contains a narcis- also found that all of the factors except Exploitiveness/Entitle-
sistic personality subscale, but its validity has yet to be estab- ment were highly correlated with self-esteem. The total NPI
lished. Phares and Erskine (1984) have developed a 28-item score was also positively associated with the need for unique-
scale designed to measure the construct of selfism within a soci- ness, extraversion, and acting. Peer ratings of narcissism were
al-learning framework. Individuals differ in selflsm in the extent found to correlate highly with self-reported NPI scores. Wat-
to which they construe situations that present problems in need son, Grisham, Trotter, and Biderman (1984) found that scores
satisfaction in either egotistical or nonegotistical terms. Phares on the NPI, particularly the Exploitiveness/Entitlement sub-
and Erskine prefer the term selfism over narcissism because scale, correlated negatively with two measures of empathy. A
they consider selfism to be an attitudinal rather than a motiva- negative correlation between that subscale and a measure of so-
tional construct. However, a purely cognitive construct fails to cial desirability was also observed; however, neither the full scale
take into account the emotional, motivational, and interper- nor any of the other subscales were significantly related to social
sonal processes underlying narcissistic behaviors (Masterson, desirability. Watson, Hood, and Morris (1984) reported that
1981). NPI scores were negatively correlated with intrinsic religious
Raskin and Hall (1979) constructed the Narcissistic Person- values (indicative of transcending self-centered needs) as mea-
ality Inventory (NPI), a 54-item, forced-choice questionnaire sured by the Allport and Ross (1967) religious-orientation mea-
designed to measure individual differences in narcissism as a sure. Finally, Prifitera and Ryan (1984) found that NPI scores
personality trait. The construction of the inventory was based distinguished between narcissistic and nonnarcissistic psychiat-
on the Diagnostic and Statistical Manual (DSM-III) criteria for ric patients. Thus, evidence for the reliability and validity of the
the narcissistic personality disorder. These criteria include (a) a NPI in both normal and pathological samples has emerged
grandiose sense of self-importance and uniqueness, (b) preoc- from a number of different sources.
cupation with fantasies of unlimited success, power, beauty, or The purposes of the following studies are (a) to attempt to
ideal love, (c) exhibitionistic—requires constant attention and replicate the factor structure uncovered by Emmons (1984); (b)
admiration, (d) entitlement-expectation of special favors with- to examine the relation between the NPI and various measures
 NARCISSISM 13

of pathological narcissism, selfism, and egocentricity; (c) to test Table 1

some aspects of certain theoretical formulations of narcissism, NPI Items and Factor Loadings
such as CattelFs (1957) and Murray's (1938); and (d) to provide
Factor
further validational evidence for the NPI. Three studies are re-
ported that address in turn each of these three substantive is-
Item L/A S/S S/A E/E

1 see myself as a good leader. .72 .11 -.16 -.03

Study 1 I would prefer to be a leader. .70 .08 -.16 -.19
I really like to be the center of
It has been suggested (Comrey, 1973) that factor-analytic re- attention. .65 .01 .00 .00
1 like having authority over other
sults based on dichotomous items may be especially unstable
people. .60 .18 -.06 -.16
owing to possible extreme item-endorsement splits. Therefore, I would be willing to describe
the NPI was once again factor analyzed to determine whether myself as a strong personality. .60 .01 .03 .13
the factors uncovered by Emmons (1984) would once again I have a natural talent for
influencing people. .58 -.08 .05 .17
emerge in a different sample.
I like to be the center of attention. .55 .01 .09 -.08
I am assertive. .49 .03 .12 .20
Method People always seem to recognize
my authority. .44 .14 .20 -.10
Several different groups of University of Illinois undergraduates (N = 1 like to look at my body. .08 .66 -.14 -.04
I like to look at myself in the
388) were administered the NPI along with several other personality
mirror. .09 .59 -.13 -.07
tests. Three hundred eight students were enrolled in introductory psy-
1 am an extraordinary person. .02 .57 .07 .05
chology and were participating in order to fulfill a course requirement. I like to display my body. .02 .54 -.03 .07
The other 80 subjects were enrolled in a semester-long course and re- I have good taste when it comes to
search project on mood and personality. They received 3 hours of couise beauty. .02 .51 -.04 -.20
credit for participating. Twenty-six of the subjects did not complete all I think I am a special person. .14 .50 .00 .15
of the items, and their data were eliminated from further analyses, leav- I like to be complimented. .23 .40 -.28 .14
ing a total sample of 362. Questionnaires were completed in group set- I am going to be a great person. .06 .36 .30 -.07
I know that I am good because
tings.
everyone keeps telling me so. .05 .35 .15 -.09
Everybody likes to hear my
stories. .19 -.24 .56 -.04
Results and Discussion
I usually dominate any
conversation. .09 -.12 .54 -.22
Interitem correlations (phi coefficients) were computed, and
1 can make anybody believe
the resulting correlation matrix was subjected to a principal- anything. .09 -.07 .52 .01
axes factor analysis. The number of factors to be extracted was 1 am a born leader. .27 .07 .48 .00
determined by a joint consideration of Kaiser's eigenvalue cri- I can read people like a book. -.08 .19 .48 .00
I am apt to show off if I get the
terion and the scree plot of eigenvalues. Using these criteria,
chance. .24 -.04 .44 -.08
four factors emerged, which were then rotated obliquely. The People can learn a great deal from
four factors accounted for 70% of the variance. Oblique rota- me. -.23 .27 .40 -.04
tion was used on the assumption that the various hypothetical 1 always know what I am doing. -.23 .28 .39 .30
I can usually talk my way out of
factors should be related to each other, as it is their combination
anything. .32 -.17 .38 -.11
that defines the trait of narcissism. The NPI items and their Superiority is something you are
respective factor loadings are presented in Table 1. Only items born with. .02 -.02 .36 .01
with loadings greater than .35 are included. Because the format I would do almost anything on a
dare. .31 -.17 .35 -.02
of the NPI is that of forced choice, each item actually consists
I expect a great deal from other
of a dyad of statements. For simplicity, only the narcissistic al- people. .02 .04 .12 .56
ternative of each dyad is shown. I am envious of other people's
The pattern of factor loadings is virtually identical to those good fortune. .00 .02 -.05 .55
I insist upon getting the respect
uncovered previously (Emmons, 1984), the only difference be-
that is due me. -.05 .02 .05 .52
ing the percentage of variance accounted for by each factor. The I will never be satisfied until I get
factors are labeled, in order, Leadership/Authority, Self-Ab- all that I deserve. .03 .17 .28 .42
sorption/Self-Admiration, Superiority/Arrogance, and Exploi- I have a strong will to power. .39 .14 .12 .41
I get upset when people don't
tiveness/Entitlement. The factors reflect to some degree the
notice how I look when I go out
DSM-III criteria for the narcissistic personality disorder, partic- in public. .19 .19 .02 .38
ularly exploitation and superiority. It is interesting to note that I find it easy to manipulate
there is no mention of leadership in DSM-III, though several people. .21 .09 .20 .35
I am more capable than other
items loading on Leadership/Authority showed the highest cor-
people. -.14 .25 .29 .35
relations with the total scale score and this factor accounted for
the most variance.
Note. L/A = Leadership/Authority. S/S = Self-absorption/Self-admira-
To examine intercorrelations among the factors, subscale tion. S/A = Superiority/Arrogance. E/E = Exploitiveness/Entitlement.
scores were computed for each of the four factors by summing Variance accounted for by the factors = 28,16,15, and 11, respectively.
14 ROBERT A. EMMONS

Table 2 Results and Discussion

Factor Intercorrelations
Correlations between the NPI, the three objective measures
Factors L/A S/S S/A of narcissism, and the number of responses in each SFSC cate-
_ gory are shown in Table 3. This table shows that the total NPI
L/A
score correlated highest with the Selfism scale, or what might
S/S .16 —
S/A .57 .40 be considered the measure of normal cognitive narcissism. In-
E/E .45 .40 —
.44 terestingly, the NPI subscale that correlated most strongly with
the MCMI and NPDS was Exploitiveness/Entitlement. These
AWc. L/A = Leadership/Authority. S/S = Self-Absorption/Self-Admi- results support Emrnons' (1984) conjecture that the Exploitive-
ration. S/A - Superiority/Arrogance. E/E - Exploitiveness/Entitle-
ment. ness/Entitlement items tap the maladaptive and possibly patho-
logical aspects of narcissism. Three of the four NPI subscales
correlated significantly with the Selfism scale, that is, all except
for Leadership/Authority. Turning to the SFSC, the total NPI
the items in that scale. The correlations ranged from. 16 to .57, score and three of the four subscales correlated significantly
with the average correlation being .42. The subscale corre- with the self-focus responses, as predicted. However, only Lead-
lations are given in Table 2. ership/Authority correlated significantly negatively with self-fo-
The internal consistencies of the full scale and each subscale cus negative responses, and only Superiority/Arrogance was
(or factor) were assessed via Cronbach's coefficient alpha (Cron- significantly related to the external-world-focus responses. In-
bach, 1951). The coefficients obtained were .87, .69, .81, .70, terestingly, the self-focus negative responses were positively as-
and .68 for the total scale and Factors 1 through 4, respectively, sociated with the Exploitiveness/Entitlement subscale, suggest-
and these are satisfactory (Nunnally, 1978). ing once again that this factor is tapping the maladaptive aspects
The factorial structure of the NPI can be considered repli- of narcissism. It does appear that egocentricity, as measured by
cated in this study. It appears that narcissism, as measured by SFSC responses, is a prime component of narcissism, particu-
the NPI, consists of four moderately correlated factors, tapping larly the Superiority/Arrogance subscale. At the same time,
the domains of leadership, self-admiration, superiority, and in- these results indicate that the Leadership/Authority component
terpersonal exploitiveness. may represent healthier aspects of narcissism.

Study 2 Study 3

Given that the NPI has a stable factorial structure, the next Several theorists (Cattell, 1957; Murray, 1938) have charac-
step was to examine the correlations between it and alternative terized narcissistic individuals as emotionally intense, reacting
measures of the construct. These included three objective mea- strongly to events and exhibiting greater fluctuations in their
sures: the Narcissistic Personality subscale of the MCMI (Mil- moods. According to DSM-III (American Psychiatric Associa-
Ion, 1982); the Narcissistic Personality Disorder Scale (NPD; tion, 1980), included among the diagnostic criteria for the Nar-
Solomon, 1982); and the Selfism scale (Phares & Erskine, cissistic Personality Disorder are "marked feelings of rage, infe-
1984). A projective test, Exner's (1973) Self-Focus Sentence riority, shame, humiliation, or emptiness in response to criti-
Completion Blank (SFSC), designed and validated to measure cism, indifference to others, or defeat" (p. 317). The hypothesis
egocentricity as a response style, was also administered. The
SFSC consists of 30 sentence stems, and the subject is asked to
complete the thought begun in each. The responses can be
Table 3
scored as reflecting self-focus, negative self-focus, and external-
Correlations Between the NPI and Other
world focus, as well as in other ways depending upon content.
Measures of Narcissism
It was hypothesized that scores on the NPI would correlate posi-
tively with the self-focus responses and negatively with both the NPI factors
self-focus negative responses and the external-world-focus re-
Measures Total L/A S/S S/A E/E
sponses.
MCMI .27* .25* .07 .48" .31*
Method NPDS .12 -.09 .09 -.04 .32**
Selfism .45" .01 .25* .48" .33"
Subjects were 48 undergraduates who completed the four measures SFSC category
and the NPI in class for extra credit. For the SFSC, subjects were in- Self-focus .33" .32" .15 .60" .29*
structed to read each stem and to complete the thought with the first Negative self-focus -.13 -.38" -.18 -.08 .28*
response that came to mind. The forms were scored independently by External-world focus -.08 -.11 .10 -.42" -.23*
two raters, both of whom were blind to the subjects' NPI scores. Each
Note. N = 48. NPI = Narcissistic Personality Inventory. MCMI = Mil-
item was scored according to Exner's (1973) criteria. The agreement
Ion Clinical Multiaxial Inventory. NPDS = Narcissistic Personality Dis-
rate of response assignment between the two raters was .84, a value sim-
order Scale. SFSC = Self-Focus Sentence Completion. L/A = Leader-
ilar to the reliability coefficients reported in Exner(1973), and items on ship/Authority. S/S = Self-Absorption/Self-Admiration. S/A = Superi-
which raters did not agree were excluded from further analysis. Subjects ority/Arrogance. E/E = Exploitiveness/Entitlement.
were also administered the NPI, which was scored in the usual fashion */><.05.
for the total and subscale scores.
 NARCISSISM 15

Table 4 correlates significantly with the variability of both positive and

Correlations Between the NPI and Mood Variability negative affect, indicating that narcissistic individuals do expe-
and Emotional Intensity rience day-to-day fluctuations in their moods. Of the four sub-
scales, Self-Admiration/Self-Absorption and Exploitiveness/
NPI factors
Entitlement are positively associated with mood variability.
Daily
measures Total L/A S/S S/A E/E Looking at emotional intensity, only the Exploitiveness/Entitle-
ment subscale is significantly correlated with both measures of
PA variability .27 .16 .22* .12 .30" intensity. This indicates that individuals who have adopted an
NA variability .31" .18 .20* .11 .26' interpersonal style of exploitation react with strong emotion to
Intensity .18 .18 -.03 -.17 .34**
AIM .01 experiences in their lives. It is apparent that any general state-
.12 .15 -.11 .39**
ment regarding emotional reactivity as it relates to narcissism
Note. N = 62. NPI = Narcissistic Personality Inventory. PA = positive needs to be restricted to the interpersonal-exploitation domain.
affect. NA = negative affect. AIM = affect intensity measure.
*/><.05.
General Discussion

Additional evidence for the validity of narcissism as a normal

personality trait and the NPI as a measure of that trait has been
that narcissistic individuals typically experience intense emo-
provided in these studies. The factorial structure of the NPI
tions and greater mood variability was tested in this study. Daily
uncovered by Emmons (1984) was replicated in Study 1. Nar-
mood reports were collected on a group of individuals over an
cissism, rather than being a unidimensional construct, consists
extended period of time.
of four moderately correlated factors tapping the domains of
leadership, self-admiration, superiority, and interpersonal ex-
Method ploitiveness. Only the Exploitiveness/Entitlement subscale was
Subjects were 62 (38 women, 24 men) undergraduates enrolled in a found to correlate significantly with two measures of pathologi-
semester-long course and research project on mood and personality. cal narcissism. This finding supports previous claims that this
They received 3 hours of course credit for participating. The subjects factor represents the maladaptive aspects of the trait, indicating
filled out mood reports daily for 42 consecutive days. The mood report that interpersonal maneuvers may be especially troublesome for
consisted of a number of monopolar affect adjective scales, including narcissistic individuals.
both positive (happy, pleased, joyful, and enjoyment/fun) and negative
(unhappy, depressed, anxious, frustrated, and angry/hostile) emotions.
Subjects indicated the extent to which they had felt each of these emo- Narcissism and Affect Extremity
tions during the day on a 6-point scale ranging from I (not at all) to 6
Some of the more interesting findings were those regarding
(extremely much). Initially, composite positive- and negative-affect
affect intensity and variability. Many theorists (Cattell, 1957;
scores were computed by summing and averaging the positive- and neg-
ative-affect words separately. These composite affect scales have reliabil- Kernberg, 1980; Murray, 1938) have maintained that mood
ities that approach .90 (Diener & Emmons, 1984). In past research swings and affective extremity are prototypical characteristics
(Diener, Larsen, Levine, & Emmons, 1985), emotional intensity has of narcissistic individuals. Empirical support for this claim was
been described as the strength with which subjects experienced their found in my research. The question becomes, why should nar-
dominant affect. Intensity is calculated by taking the mean positive- cissism be related to affect intensity and variability? Current
affect score for each subject on days when positive affect exceeded nega- work on self-complexity and affect (Linville, 1982, 1985) pro-
tive affect (positive-affect intensity) and taking the mean negative-affect vides a clue. Self-complexity refers to the capacity to differen-
score on days when negative affect exceeded positive affect (negative-
tiate among aspects of the self, such as one's professional self,
affect intensity). A number of studies have shown that these two inten-
one's social self, and one's family self. Linville has found that
sity scores correlate very highly with each other (rs range from .65 to
greater self-complexity results in less extreme and variable
.77). Thus, a composite intensity score based on the sum of mean positi-
ve- and negative-affect intensity was computed for each subject. The moods and fewer fluctuations in self-appraisal. Individuals with
Affect Intensity Measure (AIM; Larsen, 1984), which assesses the simple cognitive representations of themselves are more ex-
strength or intensity with which individuals typically experience emo- treme and variable in their moods than individuals who have
tions, was also administered. complex self-representations. For example, individuals whose
Variability was operationalized as the standard deviation of the mean concept of self is relatively simple may see themselves in only
positive- and negative-affect scores. This indicates the degree to which one social role (as a psychology professor, for example). Thus,
subjects fluctuated day to day in their average levels of positive and nega- affective reactions because of success or failure in this profes-
tive affect. Positive and negative affect are independent in people's lives
sional domain, such as having an article accepted or rejected,
(Diener & Emmons, 1984); thus, the variability of positive and negative
will spill over into other areas of the person's life. On the other
affect were treated and analyzed separately. Subjects were also adminis-
hand, if the person views himself as not only a psychology pro-
tered the NPI, along with a battery of other personality questionnaires
that were included for the purpose of another study. fessor but also a father, husband, or high school football coach,
affect resulting from success or failure in one domain is less
likely to spill over into other domains, resulting in overall milder
Results and Discussion
affective reactions. In the latter case the person maintains dis-
Table 4 shows the correlations between the NPI and the daily tinctions among various aspects of the self (complex self-repre-
measures of affect intensity and variability. The total NPI score sentation), whereas in the former case professional aspects are
16 ROBERT A. EMMONS

closely tied to all aspects of the person's self-image (simple self- tic individuals were viewed as having more positive characteris-
representation). Linville (1982) pointed out that with increas- tics than were low-narcissistic individuals. Also, in attribu-
ing age, a person's self-concept and self-evaluation become tional-style therapy (Layden, 1982), individuals with a self-
more differentiated. This may explain why affective-intensity blaming tendency (e.g., depressed persons) are taught in effect
scores decline with age (Diener, Sandvik, & Larsen, 1985). to adopt a more narcissistic attributional style. This does not
Links between mood swings and positive and negative self- necessarily mean egocentric or conceited, as Layden points out,
appraisals have been noted by Diener (1984). That is, when peo- but rather it is simply recognizing one's own contribution to
ple are in a happy mood, they report feeling better about them- positive outcomes without overtly exaggerating one's accom-
selves than when they are unhappy. How are these observations plishments to others. Where the distinction is between healthy
related to narcissism? Although narcissism has been found to and pathological narcissism is difficult to specify. Clearly the
correlate with high self-esteem (Emmons, 1984), some theorists exploitiveness/entitlement interpersonal style causes individu-
(cf. Kernberg, 1980) have argued that the superficial appear- als some difficulty. Emmons (1984) found that this factor was
ance of self-assurance masks a deeper narcissistic vulnerability, related to neuroticism, social anxiety, and the interpersonal
especially to failure and criticism. Thus, it would be expected styles aggressive/sadistic and rebellious/distrustful. Watson et
that narcissistic individuals would react strongly after both pos- al. (1984) found this aspect of narcissism to be related to a lack
itive and negative experiences, in accordance with their feelings of empathy. Perhaps narcissistic individuals exploit and manip-
of self-worth. Narcissistic individuals may have simple cogni- ulate others to increase their sense of self-worth. Lastly, studies
tive representations of themselves, resulting in their showing are needed on the relation between subjective well-being, nar-
more extreme swings in mood following success or failure. Add cissism, defensiveness, and self-esteem in both healthy and
to this the observation that narcissistic behavior has been con- pathologically narcissistic individuals. Such efforts will help to
sidered a relatively immature style of responding (Kernberg, expand the nomological network surrounding the construct of
1980; Plutchik, Kellerman, & Conte, 1979), and it is immature narcissism and will aid in differentiating normal and pathologi-
individuals who have a simple cognitive representation of them- cal manifestations of the trait.
selves; the pieces begin to fall nicely into place. Whether narcis-
sistic individuals actually do have a simple cognitive representa-
References
tion of themselves will have to be empirically documented in
future research. Allport, G. W., & Ross, J. M. (1967). Personal religious orientation and
prejudice. Journal of Personality and Social Psychology. 25.432-443.
American Psychiatric Association (1980). Diagnostic and statistical
Relevance for Attribution Theory
manual of mental disorders: DSM-I11. Washington, DC: Author.
One of the most widespread and robust findings in attribu- Ashby, H. U., Lee, R. R., & Duke, E. H. (1979, August). A narcissistic
tion theory is that individuals have the tendency to take credit personality disorder MMPl scale. Paper presented at the meeting of
the American Psychological Association, New \brk.
for successful outcomes and to deny blame for failed outcomes.
Cattell, R. B. (1957). Personality and motivation structure and measure-
This phenomenon has been referred to as the self-serving bias
ment. New York: World Book.
(Harvey & Weary, 1984), attributional egotism (Snyder et al., Comrey, A. L. (1973). A first course in factor analysis. New York: Aca-
1978), and beneffectance (Greenwald, 1980). This process is be- demic Press.
lieved to serve a self-esteem enhancement or protection func- Cronbach, L. J. (1951). Coefficient alpha and the internal structure of
tion. One might expect such egotistical attributions to be par- tests. Psychometrika, 16, 297-334.
ticularly prevalent among narcissistic individuals, given that Dervin, D. (1982). Steve and Adam and Ted and Dr. Lasch: The new
their self-esteem is especially vulnerable and that they may be culture and the culture of narcissism. The Journal of Psychohistory,
motivated to enhance their self-esteem. Those working in the 9,355-371.
Diener, E. (1984). Subjective well-being. Psychological Bulletin, 95,
domain of attribution theory have begun to examine the role of
542-575.
motivational processes and individual differences in attribu-
Diener, E., & Emmons, R. A. (1984). The independence of positive and
tions (Harvey & Weary, 1984). Given the widespread nature of
negative affect. Journal of Personality and Social Psychology, 47,
attributional egotism, the role of narcissism might prove espe- 1105-1117.
cially useful as a moderator variable, as there are individual Diener, E., Larsen, R. J., Levine, S., & Emmons, R. A. (1985). Intensity
differences in the extent to which people make egotistical attri- and frequency: Dimensions underlying positive and negative affect.
butions. The causal attributions of narcissistic individuals may Journal of Personality and Social Psychology, 48, 1253-1265.
also reflect self-presentational concerns (Harvey & Weary, Diener, E., Sandvik, E., & Larsen, R. J. (1985). Age and sex effects for
1984), such as the desires to obtain approval from others or to emotional intensity. Developmental Psychology, 21, 495-507.
appear modest. A potential avenue for future research would be Emmons, R. A. (1981). Relationship between narcissism and sensation
seeking. Psychological Reports, 48, 247-250.
the examination of the attributional styles of individuals who
Emmons, R. A. (1984). Factor analysis and construct validity of the
differ in the trait of narcissism.
Narcissistic Personality Inventory. Journal of Personality Assess-
Lastly, attention needs to be drawn to the distinction between
ment, 48, 291-300.
normal and pathological narcissism. No one would argue that Exner, J. E. (1969). Rorschach responses as an index of narcissism.
a certain amount of narcissism reflected in a healthy sense of Journal of Personality Assessment, )3, 324-330.
self-worth and self-confidence is desirable. Both Cattell (1957) Exner, J. E. (1973). The self-focus sentence completion: A study of ego-
and Fromm (1973) have pointed to the positive aspects of nar- centricity. Journalq/'PersonalityAssessment, 37, 437-455.
cissism. Interestingly, Fischer (1984) found that high-narcissis- Fichten, C. S. (1984). See it from my point of view: Videotape and attri-
 NARCISSISM 17

buiions in happy and distressed couples. Journal of Social and Clini- Murray, H. A. (1938). Explorations in personality. New York: Oxford
cal Psychology, 2. 125-142. University Press.
Fischer, C. H. (1984, April). Correlates of subclinical narcissism in col- Nelson, M. C. (1977). The narcissistic condition: A fact of our lives and
lege males and females. Paper presented at the meeting of the South- limes. New York: Human Sciences Press.
em Society for Philosophy and Psychology, Columbia, SC. Nunnally, J. (1978). Psychometric theory. New York: McGraw-Hill.
Fromm, E. (1973). The anatomy of human destruaiveness. New York: Phares, E. J., & Erskine, N. (1984). The measurement of sclfism. Edu-
Holt, Rinehart & Winston. cational and Psychological Measurement, 44, 597-608.
Grayden, C. (1958). The relationship between neurotic hypochondria- Plutchik, R., Kellerman, H., & Conte, H. R. (1979). A structural theory
sis and three personality variables: Feelings of being unloved, narcis- of ego defenses and emotion. In C. E. Izard (Ed.), Emotions in person-
sism, and guilt feelings (Doctoral dissertation, New York University, ality andpsychopathology (pp. 229-257). New York: Plenum Press.
1958). Dissertation Abstracts International, IS, 2209-2210. Prifitera, A., & Ryan, J. J. (1984). Validity of the Narcissistic Personality
Greenwald, A. G. (1980). The totalitarian ego: Fabrication and revision Inventory in a psychiatric sample. Journal of Clinical Psychology, 40,
of personal history. American Psychologist, 35. 603-618. 140-142.
Harder, D. W. (1979). The assessment of ambitious-narcissistic charac- Raskin, R. N. (1980). Narcissism and creativity: Are they related? Psy-
ter style with three projective tests: The early memories, TAT, and chological Repons. 46, 55-60.
Rorschach. Journal of Personality Assessment, 43, 23-32. Raskin, R. N. (1981). An exploration of the relationship between narcis-
Harvey, J. H., & Weary, G. (1984). Current issues in attribution theory sism and the use of first-person singular and first-person plural pro-
and research. Annual Review of Psychology, 35, 427-459. nouns in a free speech situation. Unpublished manuscript, University
Kanfer, F. H. (1979). Personal control, social control, and altruism. of California at Berkeley.
American Psychologist, 34, 231-239. Raskin, R. N., & Hall, C. S. (1979). A narcissistic personality inventory.
Kernberg, O. (1976). Borderline conditions and pathological narcissism. Psychological Reports. 45, 590.
New York: Jason Aronson. Raskin, R. N., & Hall, C. S. (1981). The narcissistic personality inven-
Kernberg, O. (1980). Internal world and external reality New York: Ja- tory: Alternate form reliability and further evidence of construct va-
son Aronson. lidity. Journal of Personality Assessment, 45, 159-162.
Kohut, H. (1976). The restoration of the self. New York: International Solomon, R. S. (1982). Validity of the MMPI narcissistic personality
Universities Press. disorder scale. Psychological Reports. 50, 463-466.
Larsen, R. J. (1984). Affect intensity as an individual difference charac- Snyder, M. L., Stephan, W. G., & Rosenfield, D. (1978). Attributional
teristic. Unpublished doctoral dissertation, University of Illinois at egotism. In J. H. Harvey, W. Ickes, & R. F. Kidd (Eds.), New direc-
Urbana-Champaign. tions in attribution research (Vol. 2, pp. 91-117). Hillsdale, NJ: Erl-
Lasch, C. L. (1979). The culture of narcissism. New York: Norton. baum.
LaVopa, L. A. (1981). Relationship between narcissism and Machiavel- Stem, A. (1980). Me: The narcissistic American. New York: Ballantine
lianism. Unpublished paper, University of Southern Maine, Portland. Books.
Layden, M. A. (1982). Attributional style therapy. In C. Antaki & Urist, J. (1977). The Rorschach test and the assessment of object re-
C. Brewin (Eds.), Attributions and psychological change (pp. 63-82). lations. Journal of Personality Assessment. 41. 3-9.
London: Academic Press. Wallach, M. A., & Wallach, L. (1983). Psychology's sanction for
Linville, P. W. (1982). Affective consequences of complexity regarding selfishness: Theerror of egoism in theory and therapy. San Francisco:
the self and others. In M. S. Clark & S. T. Fiske (Eds.), Affect and Freeman.
cognition: The Seventeenth Annual Carnegie Symposium on Cogni- Watson, P. J., Grisham, S. Q, Trotter, M. V, & Biderman, M. D. (1984).
tion(pp. 79-109). Hillsdale, NJ: Erlbaum. Narcissism and empathy: Validity evidence for the Narcissistic Per-
Linville, P. W. (1985). Self-complexity and affective extremity: Don't sonality Inventory. Journal of Personality Assessment, 48, 301 -305.
put all of your eggs in one cognitive basket. Social Cognition, 3, 94- Watson, P. J., Hood, R. W., & Morris, R. J. (1984). Religious orienta-
120. tion, humanistic values, and narcissism. Review oj Religious Re-
Masterson, J. F. (1981). The narcissistic and borderline disorders. New search. 25, 257-264.
York: Brunner/Mazel. Young, M. F. (1959). An investigation of narcissism and correlates of
Mazlish, B. (1982). American narcissism. The Psychohistory Review, narcissism in schizophrenics, neurotics, and normals (Doctoral dis-
10. 185-202. sertation, Temple University, 1958). Dissertation Abstracts, 20, 3394.
Millon, T. (1981). Disorders of personality. New York: Wiley.
Millon, T. (1982). Millon Clinical Multiaxial Inventory Manual. Min- Received August 3, 1984
neapolis, MN: National Computer Systems. Revision received December 31, 1984 •