Assessment

http://asm.sagepub.com/

What Does the Narcissistic Personality Inventory Really Measure?

Robert A. Ackerman, Edward A. Witt, M. Brent Donnellan, Kali H. Trzesniewski, Richard W. Robins and Deborah A. Kashy
Assessment 2011 18: 67 originally published online 27 September 2010
DOI: 10.1177/1073191110382845

The online version of this article can be found at:

http://asm.sagepub.com/content/18/1/67

Published by:

http://www.sagepublications.com

Additional services and information for Assessment can be found at:

Email Alerts: http://asm.sagepub.com/cgi/alerts

Subscriptions: http://asm.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://asm.sagepub.com/content/18/1/67.refs.html

>> Version of Record - Feb 3, 2011

Proof - Oct 14, 2010

Proof - Sep 27, 2010

What is This?

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

 Assessment

What Does the Narcissistic Personality 18(1) 67­–87

© The Author(s) 2011
Reprints and permission: http://www.
Inventory Really Measure? sagepub.com/journalsPermissions.nav
DOI: 10.1177/1073191110382845
http://asm.sagepub.com

Robert A.   Ackerman1, Edward A. Witt1, M. Brent Donnellan1,

Kali H.Trzesniewski2, Richard W. Robins3, and Deborah A. Kashy1

Abstract
The Narcissistic Personality Inventory (NPI) is a widely used measure of narcissism. However, debates persist about its
exact factor structure with researchers proposing solutions ranging from two to seven factors. The present research
aimed to clarify the factor structure of the NPI and further illuminate its nomological network. Four studies provided
support for a three-factor model consisting of the dimensions of Leadership/Authority, Grandiose Exhibitionism, and
Entitlement/Exploitativeness. The Leadership/Authority dimension was generally linked to adaptive outcomes whereas the
other two dimensions, particularly Entitlement/Exploitativeness, were generally linked to maladaptive outcomes. These
results suggest that researchers interested in the psychological and behavioral outcomes associated with the NPI should
examine correlates at the facet level. In light of the findings, we propose a hierarchical model for the structure of the NPI
and provide researchers with a scoring scheme for this commonly used instrument.

Keywords
narcissism, Narcissistic Personality Inventory, factor structure, construct validity, grandiosity, entitlement

Rooted in mythology and psychoanalytic thought, the con- Campbell, Foster, & Finkel, 2002; Locke, 2009; Raskin,
struct of narcissism has attracted a considerable amount of Novacek, & Hogan, 1991). Critically, much of this work
attention in psychology over the past several decades (e.g., uses NPI total scores that simply aggregate responses to all
Cain, Pincus, & Ansell, 2008; Miller & Campbell, 2008; items on the measure (see Miller & Campbell, 2008, p.
Pincus & Lukowitsky, 2010). The vast majority of research in 456). This approach is problematic given that such a sum-
social/personality psychology uses various forms of the Nar- mary score might conflate various dimensions of personality.
cissistic Personality Inventory (NPI; Raskin & Hall, 1979) to Accordingly, a better understanding of the structure of the
assess this construct (Cain et al., 2008). However, there are NPI can help researchers identify whether there are differ-
increasing concerns about the conceptual underpinnings and ent constructs embedded within the instrument that might
psychometric properties of this measure (e.g., Brown, Budzek, be generating these seemingly paradoxical results.
& Tamborski, 2009; Cain et al., 2008; Corry, Merritt, Mrug, & A deeper understanding of the NPI’s dimensional struc-
Pamp, 2008; Pincus & Lukowitsky, 2010; Trzesniewski, Don- ture will also provide researchers with clear “targets” for
nellan, & Robins, 2008). Indeed, Cain et al. (2008) noted that future scale development work. Moreover, on a purely
the NPI contains a “confusing mix of adaptive and maladap- practical level, it will provide a scoring scheme that can be
tive content” (p. 643). These misgivings parallel earlier used in future studies. Indeed, we suspect that researchers
discussions about the factor structure of the NPI and the pos- will be reluctant to abandon the NPI given its widespread
sibility that the various dimensions embedded within the use in social/personality psychology. In light of these con-
instrument may have differential relations with criterion vari- siderations, the goal of the present set of studies is to clarify
ables (e.g., Bradlee & Emmons, 1992; Emmons, 1984, 1987; the factor structure of the NPI and to evaluate how the
Raskin & Terry, 1988; Ruiz, Smith, & Rhodewalt, 2001;
Watson & Morris, 1991).
As it stands, research in social/personality psychology 1
Michigan State University, East Lansing, MI, USA
has generated a complicated picture of the correlates of nar- 2
University of Western Ontario, London, Ontario, Canada
cissism. The construct seems to be associated with 3
University of California–Davis, USA
psychological health and resilience on the one hand (e.g.,
Corresponding Author:
Sedikides, Rudich, Gregg, Kumashiro, & Rusbult, 2004; Robert A.   Ackerman, Department of Psychology, Michigan State
Wallace, Ready, & Weitenhagen, 2009) and aggression and University, Psychology Building, East Lansing, MI 48824-1116, USA
impaired interpersonal relationships on the other hand (e.g., Email: [email protected]

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

68 Assessment 18(1)

dimensions embedded within it are related to existing con- of psychological dysfunction characterized by fragile self-
structs that have been linked with narcissism. esteem, emotional instability, and internalizing pathology.
As can be seen, narcissism has been conceptualized in
several distinct ways in the existing literature. This diver-
A Brief History of the Constructs sity can create confusion as to which attributes should be
Associated With Narcissism emphasized on inventories designed to assess narcissism.
One of the earliest mentions of narcissism in psychological Pincus and Lukowitsky (2010) have suggested that the NPI
contexts comes from the British sexologist Havelock Ellis serves as a measure of normal narcissism rather than patho-
(1898) who wrote about a “Narcissus-like” tendency logical narcissism (but see Miller, Gaughan, Pryor, Kamen,
(p. 280) to engage in autoerotic activity. However, it was & Campbell, 2009). One basis for this judgment is the rela-
psychoanalytic theorists such as Kohut (1971) and Kern- tive independence of scores on the NPI and their recently
berg (1975) who made narcissism a central component of developed Pathological Narcissism Inventory (r = .13;
their theorizing regarding normal and abnormal personality Pincus, Ansell, Pimentel, Cain, Wright, & Levy, 2009).
development (see also Freud, 1914/1986). Although these This contention is also consistent with observations that the
theorists offered somewhat divergent perspectives on the NPI is negatively related to internalizing problems and
etiology of narcissism, Kernberg (1998) argued that both strongly related to the normal personality trait domain of
psychoanalytic accounts shared the belief that narcissists Extraversion (Emmons, 1984; Pincus & Lukowitsky, 2010;
possess abnormal self-structures and exhibit pathological Rhodewalt & Morf, 1995; Trzesniewski et al., 2008; Watson
self-esteem regulation strategies (see also Rhodewalt & & Biderman, 1993). Therefore, an important concern for
Peterson, 2009). Kernberg (1998) contended that the chief researchers using the NPI is that it might not be maximally
manifestations of narcissism include a sense of superiority, informative about the maladaptive aspects of personality
grandiosity, and self-absorption, along with exhibitionism, associated with narcissism. Nonetheless, as will be argued
envy, exploitativeness, and instability in mood. Likewise, later, we believe that the NPI may capture some maladap-
Kohut (1966) believed that narcissists possess a self- tive elements of personality traditionally associated with
concept characterized by grandiosity and overambition as narcissism. The major problem with the NPI might be the
well as exhibitionist drives. Millon (1996), drawing on a use of the total score, which may conflate various adaptive
social learning perspective, described the manifestations of and maladaptive personality dimensions into a composite.
narcissism in a similar fashion. He argued that narcissists
are self-absorbed and possess beliefs of superiority and
entitlement. Thus, these various accounts of narcissism A Brief History of the NPI and its
coalesce around themes of self-absorption, exhibitionism, Dimensional Structures
arrogance, and feelings of entitlement. The original version of the NPI (Raskin & Hall, 1979) was
An important theme in recent discussions concerns the developed from the description of Narcissistic Personality
distinction between normal and pathological forms of narcis- Disorder (NPD) anticipated to be included in the Diagnos-
sism. These are thought to be two distinct constructs (Pincus tic and Statistical Manual of Mental Disorders, Third
& Lukowitsky, 2010). Normal narcissism reflects the strate- Edition (DSM-III; American Psychiatric Association,
gies used to promote a positive self-image and facilitate 1980). As noted by Pincus and Lukowitsky (2010), these
agency by otherwise psychologically well-adjusted individu- criteria largely emphasize the grandiose expressions of
als (Pincus & Lukowitsky, 2010). For instance, psychologically pathological narcissism rather than the vulnerable expres-
healthy individuals may assert dominance in social hierar- sions. Raskin and Hall (1979) developed 223 rationally
chies, hold positive illusions about the self, show adaptive keyed items to capture the attributes associated with NPD.
self-enhancement, and strive for success in achievement- They used a forced-choice response format such that par-
related contexts. These attributes are not conventionally ticipants had to choose between a narcissistic alternative
understood to reflect problematic aspects of personality. In and a non-narcissistic alternative for each item (e.g., “I
contrast, pathological narcissism is characterized by mal- really like to be the center of attention” vs. “It makes me
adaptive self-regulation processes that cause significant uncomfortable to be the center of attention”; Raskin & Hall,
distress and impairment (Pincus & Lukowitsky, 2010). 1979, p. 590). Raskin and Hall (1979, 1981) later refined
Pincus and Lukowitsky (2010) further identified two primary the item pool to the 40-item forced-choice measure that is
ways in which pathological narcissism is expressed: grandi- widely used today (Raskin & Terry, 1988).
osity and vulnerability. Grandiosity is reflected in arrogant Several research groups have explored the underlying
attitudes, inflated self-esteem, and interpersonal behaviors structure of the NPI item pool and found mixed results in
characterized by exploitativeness, entitlement, and exhibi- terms of the identification of a replicable and robust dimen-
tionism. Vulnerability, on the other hand, reflects expressions sional structure (see Table 1 in Corry et al., 2008). Two early

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

Ackerman et al. 69

Table 1. Pattern Coefficients for the Three-Factor Exploratory Factor Analysis Solution (Study 1)
Item Narcissistic Response I II III
10 I see myself as a good leader −.84 −.01 .33
33 I would prefer to be a leader .76 .03 .01
36 I am a born leader .72 .02 .06
32 People always seem to recognize my authority −.68 .02 −.02
11 I am assertive .64 .00 −.07
1 I have a natural talent for influencing people .63 −.04 −.03
12 I like having authority over people .62 .01 .19
27 I have a strong will to power .52 −.00 .28
34 I am going to be a great person .46 −.09 −.23
5 If I ruled the world, it would be a much better place −.42 .04 −.03
40 I am an extraordinary person −.42 .35 .30
19 I like to look at my body .02 .75 .20
7 I like to be the center of attention −.20 .71 −.00
15 I like to display my body .08 .71 .01
30 I really like to be the center of attention .23 −.69 .07
29 I like to look at myself in the mirror −.10 −.67 −.02
20 I am apt to show off if I get the chance −.04 .54 −.17
26 I like to be complimented −.03 .53 .19
4 I know that I am good because everybody keeps telling me so −.13 .46 −.07
38 I get upset when people don’t notice how I look when I go out in public −.02 −.44 .40
28 I like to start new fads and fashions .00 .42 −.09
25 I will never be satisfied until I get all that I deserve .11 −.02 .63
24 I expect a great deal from other people .02 .01 .56
14 I insist upon getting the respect that is due to me .03 −.03 .47
13 I find it easy to manipulate people .42 .00 .46
9 I think I am a special person −.38 .36 .26
6 I can usually talk my way out of anything .38 −.04 .36
8 I will be a success .37 −.08 −.06
39 I am more capable than other people .37 .00 .32
35 I can make anybody believe anything I want them to −.36 .06 −.12
16 I can read people like a book .34 .03 .09
31 I can live my life in any way I want to .34 −.07 −.04
3 I would do almost anything on a dare .24 −.18 .16
37 I wish somebody would someday write my biography .15 −.28 .03
2 Modesty doesn’t become me .12 −.22 .36
18 I want to amount to something in the eyes of the world −.20 .20 −.01
22 I rarely depend on anyone else to get things done −.25 −.10 .21
17 I like to take responsibility for making decisions −.26 .03 .21
23 Everybody likes to hear my stories −.32 .19 −.00
21 I always know what I am doing .39 .09 .06
Note. Coefficients equal to or greater than |.40| are in boldface. Latent factor correlations: r between I and II = −.45, r between I and III = .19, and
r between II and III = −.23.

analyses yielded two different solutions. Using principal two-dimensional (i.e., Power and Exhibitionism) and three-
components analysis (PCA) and exploratory factor analysis dimensional (i.e., Power, Exhibitionism, and Specialness)
(EFA), Emmons (1984, 1987) proposed a four-dimension solutions for the NPI using a combination of PCA and confir-
solution: (a) Exploitativeness/Entitlement, (b) Leadership/ matory factor analytic (CFA) approaches on the same data
Authority, (c) Superiority/Arrogance, and (d) Self-Absorp- set. However, as Corry et al. (2008) pointed out, the Kubarych
tion/ Self-Admiration. Raskin and Terry (1988), on the other et al. analysis may not have adequately addressed the statisti-
hand, identified a seven-component solution using PCA: (a) cal complexities involved in factor analyzing dichotomous
Authority, (b) Self-Sufficiency, (c) Superiority, (d) Exhibi- responses. Given this limitation, Corry et al. (2008) con-
tionism, (e) Exploitativeness, (f) Vanity, and (g) Entitlement. ducted an EFA of the NPI item pool using methods that were
More recently, Kubarych, Deary, and Austin (2004) identified appropriate for the dichotomous nature of the NPI items (i.e.,

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

70 Assessment 18(1)

conducting an EFA based on the tetrachoric correlation matrix). Rhodewalt & Morf, 1995), and higher self-esteem (Brown
In the end, they identified two robust factors—Leadership/ et al., 2009; Emmons, 1984; Rhodewalt & Morf, 1995;
Authority and Exhibitionism/Entitlement. Watson & Biderman, 1993; Watson et al., 1992). Further-
more, Leadership/Authority is associated with indices of
psychological adjustment such as lower social anxiety
Criterion-Related Validity of the NPI Dimensions (Emmons, 1984; Watson & Biderman, 1993), neuroticism
The likelihood that there are multiple dimensions embedded (Emmons, 1984; Rhodewalt & Morf, 1995), personal dis-
within the NPI item pool makes it somewhat difficult to tress (Watson et al., 1992; Watson & Morris, 1991),
interpret total scores based on this instrument (e.g., Trzesn- depression (Watson & Biderman, 1993), and anxiety
iewski et al., 2008). The NPI total score seems to capture (Watson & Biderman, 1993). In short, there appears to be a
some amalgamation of self-perceived confidence, leadership dimension of personality embedded within the NPI that
ability, and social potency, as well as potentially socially captures psychological resilience and social potency. These
toxic elements of personality such as a sense of entitlement elements of personality form the core of what Barry et al.
and a willingness to exploit others. Accordingly, the different (2003) call adaptive narcissism and what Watson and
dimensions of personality located within the NPI may have Morris (1991) term adaptive self-functioning. These ele-
differential relations with the criterion variables that are a ments also likely represent what Pincus and Lukowitsky
central focus of much of the research on narcissism such as (2010) label normal narcissism and form the basis for
aggression, hostility, self-enhancement, and self-esteem claims regarding links between narcissism and psychologi-
(e.g., Barry, Frick, & Killian, 2003; Brown et al., 2009; cal health (e.g., Sedikides et al., 2004).
Emmons, 1987; John & Robins, 1994; Rhodewalt & Morf,
1998; Ruiz et al., 2001; Trzesniewski et al., 2008).
Supportive evidence that the NPI measures multiple The Present Studies
dimensions of personality with a differentiated set of exter- In light of the previous issues regarding the factor structure
nal correlates is consistently found in studies examining and construct validity of the NPI dimensions, the goals of
two of the four dimensions identified by Emmons (1984, the present research were to better understand the structure
1987), Exploitiveness/Entitlement and Leadership/Authority. of the NPI and to further elaborate the nomological network
Individuals with higher scores on Exploitiveness/Entitle- of the dimensions of personality embedded within this mea-
ment are more likely to be self-conscious (Watson & sure. Additional factor analytic work is necessary given the
Biderman, 1993), report larger actual-ideal discrepancies ambiguities in the literature regarding the NPI. In Study 1,
(Rhodewalt & Morf, 1995), possess lower self-esteem we conducted our own EFA of the NPI using a very large
(Brown et al., 2009), and exhibit lower levels of empathy data set. To address some potential limitations of previous
and social desirability (Watson, Grisham, Trotter, & Biderman, research, we used EFA instead of PCA, performed analyses
1984; Watson, Little, Sawrie, & Biderman, 1992; Watson & on the tetrachoric correlation matrix, and used the scree plot
Morris, 1991). Higher levels of Exploitiveness/Entitlement and interpretability of the factors as a guide in our selection
have also been found to be connected to increased mood of a factor solution. Thus, we follow contemporary recom-
variability and emotional intensity (Emmons, 1987), and mendations for conducting factor analytic work. To be sure,
neuroticism (Emmons, 1984), as well as higher scores on it is now widely recognized that the K1 heuristic (i.e.,
the Narcissistic Personality Disorder Scale (Emmons, 1987; extracting all factors with an eigenvalue greater than 1.0) is
Watson et al., 1984). Such findings suggest a dimension of rarely an optimal strategy for determining the number of
personality embedded within the NPI item pool that is char- dimensions given its general tendency to lead to an overex-
acterized by emotional reactivity and brittleness, and traction of factors (see Fabrigar, Wegener, MacCallum, &
accompanied by a lack of concern for others. These attri- Strahan, 1999).
butes generally correspond to what Barry et al. (2003) The primary difference between our approach to factor
identified as maladaptive narcissism in their modifications analysis and the one adopted by Corry et al. (2008) con-
of the NPI for use with children and adolescents. Taken as a cerns the importance of the internal consistency of the
whole, such findings suggest that the NPI does capture scales. Corry et al. (2008) placed a premium on identifying
some maladaptive aspects of personality. dimensions that would yield scales with relatively high
In contrast, the Leadership/Authority dimension argu- levels of internal consistency. Although internal consis-
ably represents the more adaptive aspects of personality tency is certainly important, we believe that this emphasis
captured by the NPI summary scores. Individuals who score may have led to an underextraction of key factors. It might
higher on this dimension report a greater degree of self- be that the NPI includes a relatively small number of items
awareness (Watson & Biderman, 1993), a lower actual–ideal that nonetheless measure theoretically important dimen-
self-discrepancy (Emmons, 1984; Raskin & Terry, 1988; sions of personality associated with maladaptive aspects of

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

Ackerman et al. 71

narcissism. Thus, there are compelling reasons to conduct provide a benchmark for evaluating the fit of our structural
additional factor analytic work on the NPI with an eye models in Studies 2 and 3.
toward recovering the dimensions of narcissism most
emphasized in the current literature. After identifying the
most robust and theoretically compelling factor solution in Method
this large data set, we used CFA in Studies 2 and 3 to evalu- Participants, Measure, and Procedure. Participants
ate how well this solution replicated across two independent were 19,001 college students (64.0% women) from a large
samples. As Briggs and Cheek (1986) point out, an impor- university in northern California who completed the 40-item
tant criterion for the acceptance of any factor solution is the NPI (Raskin & Terry, 1988) as part of a prescreening ques-
ability to duplicate the structure across different samples. tionnaire in exchange for course credit between 2002 and
Our second objective was to refine the nomological net- 2007. The sample consisted primarily of European Ameri-
work of the NPI dimensions uncovered in Study 1, and can (38.0%) and Asian American participants (39.3%),
confirmed in Studies 2 and 3. Study 2 evaluated their associa- most of whom were 18 (29.8%), 19 (26.3%), 20 (17.2%), or
tions with constructs frequently linked to narcissism (e.g., 21 (12.8%) years of age. All questionnaires were completed
psychopathy, self-esteem, and self-control). Study 3 evaluated online through a secure departmental website.
convergent associations between the NPI dimensions and psy-
chological entitlement (Campbell, Bonacci, Shelton, Exline, &
Bushman, 2004), as well as pathological narcissism measured Results and Discussion
by the newly developed Pathological Narcissism Inventory Exploratory Analyses. We used Mplus 6.0 to perform a
(PNI; Pincus et al., 2009). Finally, Study 4 evaluated the asso- series of EFAs on the NPI-40 because it offers algorithms
ciations between the NPI dimensions and self-reports and that are appropriate for factor analyzing dichotomous vari-
informant reports of the Big Five traits, as well as their associa- ables. We rotated the factors using the oblique geomin
tions with several variables related to self-reported college procedure, which is often recommended by methodologists
adjustment. Thus, we pursue a more comprehensive assess- (Browne, 2001). Nine initial eigenvalues were above 1.0
ment of the criterion-related validity of the NPI dimensions (first 10 eigenvalues: 9.443, 2.823, 2.577, 1.732, 1.703,
than was conducted by Corry et al. (2008). 1.311, 1.242, 1.154, 1.122, and 0.996). However, a close
examination of the scree plot revealed a noticeable bend
between the third and the fourth factors, thus suggesting a
Study 1 three-factor solution. Specifically, the difference in eigen-
In Study 1, we conducted an EFA of the NPI-40, using a values between the third and fourth factors was .845, which
sample of 19,001 college students previously described in was considerably larger than the difference between the
Trzesniewski et al. (2008). The primary goal of these analyses second and third factors (.246), the fourth and fifth factors
was to identify one or more replicable factor solutions. Two (.029), the fifth and sixth factors (.392), the sixth and sev-
different considerations were used to decide on a factor solu- enth factors (.069), and the seventh and eighth factors
tion in our exploratory analyses. First, we examined the scree (.088). In short, evidence for extracting two-, four-, five-,
plot to obtain a rough idea of the number of large dimensions six-, and seven-factor solutions was less compelling than
embedded within the NPI. We used this approach rather than the evidence for extracting three factors.
extracting all eigenvalues greater than 1.0 given the previ- We also examined whether we could recover the two-,
ously described limitations associated with this rule of thumb three-, four-, and seven-dimension solutions proposed by
(see Fabrigar et al., 1999). Second, we examined factor load- previous researchers. The Corry et al. (2008) two-factor solu-
ings to identify the solutions that were most consistent with tion was more or less recovered in these analyses. Out of the
previous research and interpretable in light of past theoretical nine items that made up their original Leadership/Authority
work. Our factor analytic decisions were generally consistent factor, eight of them exhibited loadings greater than .40 on a
with the considerations used by Corry et al. (2008) with the single factor. Moreover, 9 out of 14 items on their Exhibition-
exception that we did not place as high of a premium on the ism/Entitlement factor exhibited loadings greater than .40 on
internal consistency of the scales that emerged from the analy- a second factor. In contrast, the canonical three-factor
ses. This was motivated by our expectation that some Kubarych et al. (2004) solution, the four-factor Emmons
important dimensions embedded within the NPI may only be (1984) solution, and the seven-factor Raskin and Terry
measured by a relatively small number of items (see Raskin & (1988) solution were not as easily identifiable. The seven-
Terry, 1988), and this can have a detrimental impact on alpha factor solution even contained a factor that did not have
coefficients as this statistic depends on the number of items in substantial loadings for any of the NPI items.
a measure. In addition to exploratory analyses, we conducted In light of the above findings, we were left to decide
CFAs of the previously reported NPI solutions in Study 1 to between a two- and three-factor solution. These two respective

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

72 Assessment 18(1)

Table 2. CFA Fit Indexes of NPI Solutions (Study 1)

Model χ2 (df) CFI RMSEA TLI
Corry et al. (two-factor) 17999.22 (229) .867 .064 .853
Kubarych et al. (two-factor) 34762.32 (628) .829 .053 .818
Kubarych et al. (three-factor) 31919.33 (626) .843 .051 .833
Emmons (four-factor) 31250.44 (428) .830 .062 .815
Raskin and Terry (seven-factor) 29364.28 (719) .868 .046 .857
Note. CFA = confirmatory factor analysis; NPI = Narcissistic Personality Inventory; CFI = comparative fit index; RMSEA = root mean square error of
approximation; TLI = Tucker–Lewis index; WLSMV = weighted least square with mean and variance adjustment. Each analysis used the WLSMV method
of estimation. N = 18,998 for Corry et al. (2008) solution; N = 19,001 for Kubarych et al. (2004) two- and three-factor solutions, Emmons (1984, 1987)
four-factor solution, and Raskin and Terry (1988) seven-factor solution. Differences in sample size reflect different patterns of missing data.

solutions appeared to be the cleanest and most easily inter- estimation as implemented in Mplus 6.0. Table 2 shows that
pretable in the context of existing research with the NPI. The none of these solutions produced consistently good-fit sta-
two-factor solution was similar to the Corry et al. (2008) tistics using conventional rules of thumb. This is not
solution and it seemed to correspond reasonably well with uncommon for omnibus measures of personality and is
the Barry et al. (2003) distinction between adaptive and mal- likely due to the presence of a large number of correlated
adaptive narcissism (see also Watson & Morris, 1991). item residuals and cross-loadings (see Corry et al., 2008).
Consistent with our oblique rotation methods, the two latent In light of this and broader concerns about the usefulness of
factors were correlated (r = .47). However, given recent theo- established conventions used to interpret model fit indexes
retical considerations that grandiosity and entitlement are (e.g., Marsh, Hau, & Wen, 2004), Hopwood and Donnellan
important but somewhat distinct elements of maladaptive (2010) have suggested that researchers interpret CFA fit
narcissism (e.g., Brown et al., 2009), we ultimately favored indexes within a local context by using the fit of the models
the three-factor solution uncovered in these analyses, as this that are commonly proposed for an inventory (or for similar
solution preserved this distinction. Pattern loadings for this inventories) for interpreting new results. They argued that
three-factor solution are reported in Table 1. this will give researchers an idea of the range of model fit
The first factor was similar to the Corry et al. (2008) statistics that might be expected in subsequent research.
Leadership/Authority factor as it was indicated by items These “local” fit statistics might then serve as more appro-
related to self-perceived leadership ability, social potency, priate benchmarks for the interpretation of model fit rather
and to a lesser extent, dominance. We therefore used the than the often-cited conventions (e.g., comparative fit index
Corry et al. label for this factor. The second factor was indi- values of .95 or higher) or the use of the exact fit test.
cated by items reflecting a combination of self-absorption, Accordingly, the model fit statistics reported in Table 2 pro-
vanity, superiority, and exhibitionistic tendencies. Accord- vide a rubric for the interpretation of fit statistics for the
ingly, this factor seemed to illustrate the features of self-love NPI, which we will use as a point of comparison for replica-
and theatrical self-presentation emphasized in early writ- tion of our three-factor structure in Studies 2 and 3.
ings on narcissism. We labeled this factor Grandiose
Exhibitionism. The third factor consisted of items capturing
entitled beliefs and behaviors related to interpersonal con- Study 2
texts, such as a sense of deserving respect and a willingness In Study 2, we used CFA to replicate the three-factor solution
to manipulate and take advantage of others. This dimension uncovered in Study 1. We also evaluated the nomological
seemed to correspond to the socially toxic elements of nar- network of these three NPI dimensions by testing their
cissism that are frequently discussed in the literature. To relations to constructs of theoretical relevance to narcis-
maintain consistency with the terminology used in the lit- sism—self-esteem, self-control, antisocial behavior, basic
erature, we used the label of Entitlement/Exploitativeness.1 motivational systems, and the two remaining members of the
A table with the correlations between the different scales “Dark Triad” of personality (i.e., psychopathy and Machia-
for each of the NPI solutions is available on request. vellianism; see Paulhus & Williams, 2002). Scales were
Confirmatory Analyses. In addition to the extensive set of constructed by selecting those items for each factor in Table
EFA analyses, we used CFA techniques to formally evaluate 1 that exhibited pattern loadings for their factor at or above
the fit of previously identified factor solutions discussed in .40. In the cases where an item exhibited loadings greater
the literature (e.g., Corry et al., 2008). To account for the than .40 on more than one factor, we assigned the item to the
dichotomous nature of the NPI indicators, we used WLSMV factor with the larger of the two loadings. This was relevant
(weighted least square with mean and variance adjustment) for only two items. Items with negative loadings on the

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

Ackerman et al. 73

Leadership/Authority factor were reverse scored, and items obscures some important differences with respect to rela-
with positive loadings on the Grandiose Exhibitionism factor tions between the NPI scales and feelings of self-worth.
were reverse scored. Leadership/Authority was represented Specifically, it appears that the Leadership/Authority factor
by 11 items (Items 1, 5, 10, 11, 12, 27, 32, 33, 34, 36, and 40), shows a moderate positive relation with self-esteem
Grandiose Exhibitionism by 10 items (Items 4, 7, 15, 19, 20, whereas Exhibitionism/Entitlement shows little relation
26, 28, 29, 30, and 38), and Entitlement/Exploitativeness by with self-esteem.
4 items (Items 13, 14, 24, and 25). Given concerns over the Last, given the suggestion by Vazire and Funder (2006)
internal consistency of the NPI scales and issues with how that narcissism is linked to poor self-control, we evaluated
alpha is linked to the number of items in a scale, we also connections between the NPI factors and constructs related
report the average interitem correlations for the NPI scales to to behavioral control. One potential limitation of the exist-
provide a broader context for evaluating the internal consis- ing work examining links between narcissism and
tency of our scales. impulsivity is that it has largely used the NPI total score
Patrick and Bernat (2009) proposed a dual process model (but see Fulford, Johnson, & Carver, 2008; Rhodewalt &
of psychopathy wherein the clinical syndrome stems from Morf, 1995). Accordingly, we examined relations between
deficits in two separate underlying neurobiological systems: facets of the NPI and several measures linked to self-con-
one that manifests itself as trait fearlessness and another that trol and basic motivational systems, including the
manifests itself as a tendency toward externalizing problems. Behavioral Activation System (BAS) and Behavioral Inhi-
These two dimensions have been operationalized using mea- bition System (BIS), the construct of self-control discussed
sures of Fearless Dominance and Impulsive Antisociality by Tangney, Baumeister, and Boone (2004), and an out-
(e.g., Benning, Patrick, Hicks, Blonigen, & Krueger, 2003; come of impulsivity—counterproductive school behaviors.
Witt, Donnellan, & Blonigen, 2009; Witt, Donnellan, Bloni- The BIS and BAS analyses are particularly informative as
gen, Krueger, & Conger, 2009). Empirical research suggests Tracy and Robins (2003, figure 1) proposed a temperament-
that the NPI total score is more strongly related to Fearless based account of narcissism in which high avoidance (i.e.,
Dominance than Impulsive Antisociality (e.g., Witt & high BIS scores) and high approach (i.e., high BAS scores),
Donnellan, 2008; Witt, Donnellan, Blonigen, Krueger, et al., in conjunction with shame-inducing early childhood expe-
2009); however, there are some indications that this statisti- riences, serve to provide the foundation for the development
cal effect is driven by the putatively adaptive factors of the of narcissism.
NPI (see table 4 of Witt & Donnellan, 2008). On the other
hand, the more socially toxic elements of the NPI (e.g., Enti-
tlement and Exploitativeness) are more strongly linked with Method
Impulsive Antisociality than Fearless Dominance. Given this Participants. Participants were 353 college students
past work, we expected the Leadership/Authority factor to be (70.8% women) from a large Midwestern university who
more strongly related to Fearless Dominance and the Grandi- participated in exchange for course credit or extra credit.
ose Exhibitionism and Entitlement/Exploitativeness factors The sample consisted primarily of European American par-
to be more strongly related to Impulsive Antisociality. More- ticipants (81.9%), the majority of whom were 18 (28.6%),
over, we expected that Machiavellianism will be more 19 (37.1%), 20 (19.3%), or 21 (10.2%) years of age. All
strongly linked with the maladaptive facets of personality questionnaires were completed online through a secure
assessed by the NPI given the findings for Impulsive website maintained by the Psychology Department.
Antisociality.
In addition to the Dark Triad, we investigated connec- Measures
tions between facets of the NPI and constructs linked with NPI-40 (Raskin & Terry, 1988). We calculated the NPI total
positive self-evaluations. Researchers frequently report a score (M = 0.41, SD = 0.17; α = .84, average interitem r =
modest to moderate positive association between self- .12) and the subscales for the present three-factor solution:
esteem and NPI total scores (e.g., around .30; Trzesniewski Leadership/Authority (M = 0.48, SD = 0.27; α = .78, aver-
et al., 2008). However, Trzesniewski et al. (2008) reported age interitem r = .25), Grandiose Exhibitionism (M = 0.39,
differential correlates between self-esteem and some of the SD = 0.25; α = .72, average interitem r = .21), and Entitle-
Raskin and Terry (1988) dimensions such that self-esteem ment/Exploitativeness (M = 0.26, SD = 0.27; α = .46,
had a very small negative correlation with their Entitlement average interitem r = .18).2 Because we used the mean of
subscale from the NPI (r = −.04) and very small positive participants’ responses to represent their scores on each of
correlations with their Exhibitionism (r = .08) and Exploit- the subscales, these sample means indicate the average pro-
ativeness (r = .09) subscales. In contrast, self-esteem had portion of items on each subscale that participants endorsed.
larger positive associations with Raskin and Terry’s Author- The Leadership/Authority subscale showed moderate asso-
ity (r = .33) and Self-Sufficiency (r = .32) scales. Thus, the ciations with Grandiose Exhibitionism (r = .43) and
overall correlation for NPI total scores and self-esteem Entitlement/Exploitativeness (r = .31), and Grandiose

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

74 Assessment 18(1)

Exhibitionism showed a moderate association with Entitle- ranged from 1 (strongly disagree) to 5 (strongly agree).
ment/Exploitativeness (r = .21). Higher scores corresponded to higher levels of the trait (M =
Psychopathy. The two dimensions of Fearless Dominance 3.13, SD = 0.57; α = .83).
and Impulsive Antisociality were measured by the Interna- Counterproductive school behaviors. A modified version of
tional Personality Item Pool (IPIP)-based scales developed by the 11-item Bennett and Robinson (2000) workplace devi-
Witt, Donnellan, and Blonigen (2009; i.e., “Participants ance measure was used to assess counterproductive school
responded using a 5-point scale that ranged from 1 [strongly behaviors. We replaced the word “work” with “school” for
disagree] to 5 [strongly agree].”). Fearless Dominance cap- the seven Interpersonal Deviance items. We also modified
tures an interpersonally dominant orientation characterized by their third item from “Made an ethnic, religious, or racial
thrill seeking and a lack of anxiety (20 items; M = 3.33, remark at work” to “Made a derogatory ethnic, religious, or
SD = 0.47; α = .74) whereas Impulsive Antisociality captures racial remark at school.” Participants responded to each
a general susceptibility to deviance (20 items; M = 2.38, SD = item on a 5-point scale that ranged from 1 (never) to 5 (fre-
0.55; α = .74). Witt, Donnellan, and Blonigen (2009) reported quently). Higher scores indicated more frequent reports of
good convergence between these scales (i.e., rs > .65) and the misbehavior at school (M = 2.25, SD = 0.54; α = .82).
more commonly used scales in the Psychopathic Personality
Inventory–Revised (Lilienfeld & Widows, 2005).
Machiavellianism. The 20-item “Kiddie Mach” (Christie Results and Discussion
& Geis, 1970) was used to assess Machiavellianism. The Confirmatory Factor Analysis of the Three-Factor Solu-
Kiddie Mach is written in simpler and more direct language tion. To account for the dichotomous nature of the NPI
and therefore takes less time to complete than the MACH- indicators, we used WLSMV estimation within Mplus 6.0 to
IV (e.g., the item “Most people cannot be easily fooled” evaluate the three-factor solution. Standardized estimates of
captures the same content of the MACH-IV item of the factor loadings were obtained by fixing the variance of
“Barnum was wrong when he said that there’s a sucker born each factor to one; moreover, covariances were freely speci-
every minute”). Participants responded using a 5-point fied between the three latent factors. Model fit was not
scale that ranged from 1 (strongly disagree) to 5 (strongly acceptable by existing conventions given that the comparative
agree). Higher scores reflected higher levels of Machiavel- fit index (CFI) and Tucker–Lewis index (TLI) values were less
lianism (M = 2.78, SD = 0.38; α = .74). than .90: χ2(272, n = 353) = 626.822, p < .001, CFI = .869,
Self-esteem. The 10-item Rosenberg scale (1965) was TLI = .855, root mean square error of approximation
used to measure self-esteem. Participants responded using a (RMSEA) = .061. However, the fit indexes were in line with
5-point scale that ranged from 1 (strongly disagree) to 5 those reported in Table 2. Examination of the modification
(strongly agree). Higher scores reflected higher levels of indexes revealed that certain items with similar item content
self-esteem (M = 3.74, SD = 0.58, α = .86). had correlated residuals (e.g., “I like to display my body” and
Behavioral Inhibition System (BIS)/Behavioral Activation “I like to look at my body”), a phenomenon reported by others
System (BAS). The 24-item Carver and White (1994) mea- (e.g., Corry et al., 2008; Kubarych et al., 2004).
sure was used to assess these two systems. Participants Consequently, we specified covariances between residu-
responded to each item on a 5-point scale that ranged from als for Items 7 and 30, 15 and 19, 4 and 26, 19 and 29, 10
1 (strongly disagree) to 5 (strongly agree). Items were and 33, and 34 and 40. This improved model fit indexes into
scored such that higher scores on the BIS (7 items; M = ranges that were more or less acceptable by existing con-
3.70, SD = 0.58; α = .76) and BAS (13 items; M = 3.59, SD = ventions: χ2(266, n = 353) = 460.320, p < .001, CFI = .928,
0.44; α = .81) scales suggest higher activity of that respec- TLI = .919, RMSEA = .045. More important, the fit was
tive system. Furthermore, the BAS scale was scored to favorable compared with the other structural models
reflect its subscales of Drive (i.e., a persistent approach reported in Table 2. Pattern loadings from this model are
toward goals; 4 items; M = 3.20, SD = 0.67; α = .74), reported in Table 3. The item loadings for the Leadership/
Reward Responsiveness (i.e., a tendency to experience pos- Authority factor ranged from .30 to .79 (average factor
itive affect on anticipation or completion of goal attainment; loading ≈ .64), the loadings for Grandiose Exhibitionism
5 items; M = 3.98, SD = 0.43; α = .72), and Fun Seeking ranged from .34 to .84 (average factor loading ≈ .54), and
(i.e., the seeking out of and spontaneous engagement in the loadings for Entitlement/Exploitativeness ranged from
new experiences; 4 items; M = 3.49, SD = 0.61; α = .65). .38 to .76 (average factor loading ≈ .54). An EFA with
Self-control. Self-control was measured with the 13-item geomin rotation revealed that most of the items that were
scale developed by Tangney et al. (2004). This scale scored on the scales had their largest loadings on their
assessed participants’ general ability to inhibit impulses and respective factors (i.e., 8 out of 11 Leadership/Authority
work toward long-term goals in a wide variety of domains. items, 7 out of 10 Grandiose Exhibitionism items, and 4 out
Participants responded to each item on a 5-point scale that of 4 Entitlement/Exploitativeness items).

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

Ackerman et al. 75

Table 3. Confirmatory Factor Analysis Standardized Pattern Loadings for the Three-Factor Model (Studies 2 and 3)
Item Leadership/Authority Grandiose Exhibitionism Entitlement/Exploitativeness
Study 2 Study 3 Study 2 Study 3 Study 2 Study 3
1 .72 .51
5 .51 .58
10 .69 .65
11 .73 .43
12 .77 .71
27 .69 .65
32 .79 .71
33 .74 .74
34 .30 .32
36 .64 .76
40 .43 .49
4 .54 .45
7 .71 .62
15 .45 .61
19 .34 .56
20 .66 .70
26 .54 .34
28 .39 .61
29 .46 .38
30 .84 .64
38 .43 .46
13 .76 .58
14 .38 .58
24 .46 .39
25 .57 .62

Note. Table only includes rows for those items that load on the current three-factor solution. Items 10, 32, 5, 40, 28, 4, 26, 20, 15, 7, and 19 were reverse
scored for all analyses. Correlations between latent factors for Study 2 were as follows: Leadership/Authority and Grandiose Exhibitionism (r = .63);
Leadership/Authority and Entitlement/Exploitativeness (r = .56); and Grandiose Exhibitionism and Entitlement/Exploitativeness (r = .43). Correlations
between latent factors for Study 3 were as follows: Leadership/Authority and Grandiose Exhibitionism (r = .54); Leadership/Authority and Entitlement/
Exploitativeness (r = .33); and Grandiose Exhibitionism and Entitlement/Exploitativeness (r = .38).

Correlates of the NPI Total Score and NPI Scales. Table 4 Consistent with our concerns about the interpretation of
reports the correlations between the NPI scales and the crite- the NPI total score, however, an inspection of the NPI scale
rion variables. The NPI total score exhibited moderate to correlations demonstrated that the overall pattern of asso-
strong positive associations with the Fearless Dominance and ciations was not always consistent across each of the NPI
BAS drive scales. The NPI total score showed slightly weaker scales. For example, the Leadership/Authority scale was
positive links with Impulsive Antisociality, Machiavellianism, unrelated to Impulsive Antisociality and Machiavellianism.
Self-Esteem, total BAS, BAS Reward Responsiveness, BAS A somewhat different pattern emerged when considering
Fun Seeking, and Counterproductive School Behaviors. In the other dimensions. The Grandiose Exhibitionism scale,
addition, the NPI total score was negatively associated with for instance, was modestly associated with Impulsive Anti-
the BIS scale. Self-Control was the only variable with no sociality, Machiavellianism, and Counterproductive School
detectable overall association with the NPI. Using only the Behaviors. On the other hand, the Entitlement/Exploitative-
NPI total score, we would therefore conclude that individuals ness scale was more strongly associated with Impulsive
with higher levels of “narcissism” have higher levels of self- Antisociality and Machiavellianism, which is noteworthy
esteem, are more socially potent, possess temperaments that in light of its relatively low internal consistency.
predispose them to be more sensitive to rewards and less sensi- We also conducted a series of multiple regressions to deter-
tive to signs of punishment, and engage in higher rates of mine the unique relations of the three NPI scales with the
deviant behaviors at school. criterion variables. Table 5 presents the 11 multiple regression

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

76 Assessment 18(1)

Table 4. Correlations (r) Between the Narcissistic Personality Inventory (NPI) Total Score, the NPI Scales of Leadership/Authority,
Grandiose Exhibitionism, and Entitlement/Exploitativeness, and Variables Relevant to Narcissism (Study 2)
NPI Scale Scores
NPI Total Score Leadership/ Grandiose Entitlement/
Authority Exhibitionism Exploitativeness
Psychopathy: Fearless dominance .49 .52 .38 −.01
Psychopathy: Impulsive antisociality .21 .03 .20 .37
Machiavellianism .19 .04 .15 .35
Self-esteem .27 .35 .15 −.12
Behavioral inhibition system (BIS) −.26 −.27 −.05 −.13
Behavioral activation system (BAS) .33 .29 .28 .09
BAS: Drive .40 .37 .27 .17
BAS: Reward responsiveness .11 .13 .15 −.08
BAS: Fun seeking .23 .14 .21 .08
Self-control .05 .16 −.05 −.11
Counterproductive school behaviors .24 .13 .20 .24
Note. Entries in boldface are significant at p < .05.

analyses. In each analysis, we used the NPI subscales of Summary

Leadership/Authority, Grandiose Exhibitionism, and Entitle-
ment/Exploitativeness to predict the criterion variable of Taken together, these results provide evidence that the positive
interest (e.g., self-control). To reduce redundancy with the pre- relations between the NPI total score and agency-related con-
vious analyses, we place much of our focus here on those structs (i.e., Fearless Dominance and BAS) are primarily driven
relations that differed from their zero-order counterparts. As by the Leadership/Authority and Grandiose Exhibitionism con-
Table 5 shows, the Leadership/Authority scale exhibited tent of the NPI. The Entitlement/Exploitativeness scale does
negative associations with the Impulsive Antisociality and not appear to explain these relations; in fact, once the other NPI
Machiavellianism scales. Moreover, the connections factors are controlled for, the Entitlement/Exploitativeness
between the Leadership/Authority scale and the Counter- scale has a negative relation with Fearless Dominance. In
productive School Behaviors and BAS Fun Seeking scales addition, the positive relation between the NPI total score
became nonsignificant once the effects of Grandiose Exhi- and self-esteem, as well as the negative relation between
bitionism and Entitlement/Exploitativeness were controlled. the NPI total score and BIS, appear to be mostly driven by
These findings provide evidence that the unique aspects of the Leadership/Authority content. Finally, the positive
the Leadership/Authority scale capture a mostly adaptive associations between the NPI total score and Impulsive
dimension of personality. Antisociality, Machiavellianism, and counter-productive
The Grandiose Exhibitionism and Entitlement/Exploit- school behaviors appear to be driven by the more maladap-
ativeness scales were also shown to provide an informative tive contents of the NPI—Grandiose Exhibitionism and
pattern of relations within a multiple regression context. As Entitlement/Exploitativeness.
Table 5 shows, the Grandiose Exhibitionism scale became
uniquely associated with less behavioral regulation and essen-
tially unrelated to self-esteem. In addition, links between the Study 3
Grandiose Exhibitionism scale and the Fearless Dominance, In Study 3, we again evaluated the fit of the three-factor solu-
BAS total score, and BAS drive scales became attenuated. The tion for the NPI using CFA methods. This replication is
Entitlement/Exploitativeness scale was found to uniquely pre- important given the model modifications made in Study 2.
dict less interpersonal dominance and a tendency to experience We also assessed the convergent validity of the three NPI
less positive affect after completing goals. Furthermore, Enti- factors using the Psychological Entitlement Scale (Campbell
tlement/Exploitativeness became unrelated to BAS drive and et al., 2004) and the PNI (Pincus et al., 2009). Campbell et al.
more strongly inversely related to self-esteem. (2004) developed the Psychological Entitlement Scale to

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

 Table 5. Multiple Regression Analyses Using the NPI Scales of Leadership/Authority, Grandiose Exhibitionism, and Entitlement/Exploitativeness to Predict Variables Relevant to
Narcissism (Study 2)
Leadership/Authority Grandiose Exhibitionism Entitlement/Exploitativeness
              b SE β b SE β b SE β Model F R2
Psychopathy:
Fearless dominance 0.87 .09 .49 0.40 .09 .21 −0.37 .08 −.21 59.22a .34
Psychopathy:
Impulsive antisociality −0.35 .11 −.17 0.41 .12 .19 0.79 .11 .38 24.65b .18
Machiavellianism −0.18 .08 −.13 0.20 .08 .13 0.51 .08 .36 18.82c .14
Self-esteem 0.91 .12 .42 0.06 .13 .03 −0.55 .11 −.25 26.00c .18
Behavioral inhibition system
(BIS) −0.62 .13 −.29 0.20 .13 .09 −0.12 .12 −.06 10.24a .08
Behavioral activation system
(BAS) 0.34 .09 .21 0.34 .10 .19 −0.04 .09 −.02 14.64a .11
BAS: Drive 0.73 .14 .29 0.36 .15 .13 0.11 .13 .04 20.63a .15
BAS: Reward responsiveness 0.20 .10 .12 0.22 .10 .13 −0.24 .09 −.15 5.80a .05
BAS: Fun seeking 0.11 .14 .05 0.46 .14 .19 0.06 .13 .03 5.92b .05
Self-control 0.54 .13 .26 −0.28 .13 −.12 −0.35 .12 −.16 7.88a .06
Counterproductive school
behaviors 0.00 .12 .00 0.34 .12 .16 0.40 .11 .20 9.96c .08

Note. NPI = Narcissistic Personality Inventory; b = unstandardized regression coefficient; SE = standard error of the unstandardized regression coefficient; β = standardized regression coefficient.
Entries in boldface are significant at p < .05 (i.e., “Differences in the degrees of freedom error for the models reflect different patterns of missing data.”).

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

a. A model with the following degrees of freedom: (3, 348).
b. A model with the following degrees of freedom: (3, 347).
c. A model with the following degrees of freedom: (3, 349).

77
78 Assessment 18(1)

provide researchers with a more face-valid and reliable mea- M = 2.26, SD = 1.01; α = .91), Hiding the Self (7 items; M =
sure of narcissistic entitlement and we expect this measure to 2.86, SD = 0.94; α = .79), and Devaluing (7 items; M = 1.88,
converge with the NPI Entitlement/Exploitativeness scale. SD = 0.98; α = .84). Wright, Lukowitsky, Pincus, and Conroy
The PNI was designed to assess pathological narcissistic (IN PRESS) recently provided evidence for the existence of
traits associated with vulnerability and grandiosity. These two higher order factors of Narcissistic Grandiosity (indicated
analyses will provide important information as to which by the Exploitativeness, Self-Sacrificing Self-Enhancement,
aspects of the NPI correspond with these newly developed and Grandiose Fantasy subscales) and Narcissistic Vulnera-
measures. This work will also help establish whether any bility (indicated by the Entitlement Rage, Contingent
aspects of the NPI are related to pathological narcissism as Self-Esteem, Hiding the Self, and Devaluing subscales) that
captured by the PNI. help explain the observed covariation between the PNI sub-
scales. We therefore additionally scored the instrument for
these broader dimensions: Narcissistic Grandiosity (18 items;
Method M = 2.81, SD = 0.69; α = .85) and Narcissistic Vulnerability
Participants. Participants were 332 college students (34 items; M = 2.30, SD = 0.75; α = .93). Participants
(81.6% women) from a large Midwestern university who responded to each question on a 6-point Likert-type scale that
participated in exchange for course credit or extra credit. ranged from not at all like me (coded as a 0) to very much like
Most students reported being European American/White me (coded as a 5). Each subscale was scored such that higher
(84.9%). The majority reported being 18 (33.4%), 19 scores corresponded to higher levels of the construct.
(33.7%), 20 (15.1%), or 21 (10.5%) years of age. All ques-
tionnaires were completed online through a secure website
maintained by the Psychology Department. Results and Discussion
Confirmatory Factor Analysis of the Three-Factor Solu-
Measures tion. We specified the three-factor solution using the same
NPI-40 (Raskin & Terry, 1988). Along with the NPI total modifications described in Study 2. When judged relative
score (M = 0.39, SD = 0.15; α = .80, average interitem r = to the fit statistics presented in Table 2, this model showed
.09), scales were created according to the present three- indications of reasonable fit, χ2(266, n = 332) = 423.993,
factor solution: Leadership/Authority (M = 0.46, SD = 0.25; p < .001, CFI = .922, TLI = .912, RMSEA = .042. Pattern
α = .75, average interitem r = .21), Grandiose Exhibition- loadings are reported in Table 3. An EFA with geomin rota-
ism (M = 0.35, SD = 0.24; α = .71, average interitem r = tion again revealed that most of the items scored on the
.20), and Entitlement/Exploitativeness (M = 0.21, SD = scales had their largest loadings on their respective factors
0.25; α = .44, average interitem r = .17). The Leadership/ (i.e., 9 out of 11 Leadership/Authority items, 7 out of 10
Authority subscale showed a moderate association with Grandiose Exhibitionism items, and 3 out of 4 Entitlement/
Grandiose Exhibitionism (r = .34) and a weaker association Exploitativeness items).3 In general, the results from Study
with Entitlement/Exploitativeness (r = .16). Grandiose 3 mirrored Study 2 suggesting that the three-factor model
Exhibitionism also showed a modest association with Enti- was replicable.
tlement/Exploitativeness (r = .17). Correlations Between the NPI Scales and Existing Mea-
Psychological Entitlement Scale (Campbell et al., 2004). This sures of Narcissism. Table 6 shows correlations between the
scale assesses the entitlement facet of narcissism. Participants NPI scales and existing measures of narcissism.4 The NPI total
responded to nine items on a 5-point scale ranging from 1 score showed moderate convergence with the Psychological
(strongly disagree) to 5 (strongly agree). Example items Entitlement Scale but only a modest association with the PNI
included, “Great things should come to me” and “I deserve total score. This suggests areas of important distinction
more things in my life.” Each participant’s score was the between some of the elements of “normal” narcissism assessed
mean across items, with higher scores signaling greater by the NPI and elements of “pathological” narcissism assessed
levels of psychological entitlement (M = 2.66, SD = 0.58; α by the PNI. An examination of the correlations between the
= .85, average interitem r = .38). NPI scale scores and the subscales of the PNI suggested a
Pathological Narcissism Inventory (Pincus et al., 2009). This complex pattern of associations.
52-item measure was designed to assess the more pathologi- Consistent with previous analyses showing that Leader-
cal characteristics associated with narcissism (M = 2.48, SD = ship/Authority was largely unrelated to maladaptive
0.66; α = .93). This inventory contains seven subscales: personality traits, this scale did not have many positive
Exploitativeness (5 items; M = 2.35, SD = 0.92; α = .76), Self- associations with the PNI subscales. In fact, the NPI Lead-
Sacrificing Self-Enhancement (6 items; M = 3.05, ership/Authority scale was found to be negatively related to
SD = 0.80; α = .70), Grandiose Fantasy (7 items; M = 2.93, PNI Contingent Self-Esteem, indicating that individuals
SD = 0.97; α = .85), Entitlement Rage (8 items; M = 2.26, with higher levels of Leadership/Authority show a reduced
SD = 0.90; α = .83), Contingent Self-Esteem (12 items; tendency to experience self-esteem as fluctuating and

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

Ackerman et al. 79

Table 6. Correlations (r) Between the NPI Total Score, the NPI Scales of Leadership/Authority, Grandiose Exhibitionism, and
Entitlement/Exploitativeness, the Psychological Entitlement Scale, and Measures of “Pathological” Narcissism (Study 3)
NPI Scale Scores Psychological
Entitlement
Leadership/ Grandiose Entitlement/
Scale
NPI Total Score Authority Exhibitionism Exploitativeness
Psychological Entitlement Scale .41 .26 .32 .36 —
Pathological Narcissism Inventory .14 −.02 .13 .30 .28
Narcissistic vulnerability .05 −.10 .07 .32 .24
  Entitlement rage .31 .08 .30 .44 .46
  Contingent self-esteem −.04 −.19 .06 .23 .14
  Hiding the self −.07 −.06 −.15 .05 −.02
  Devaluinga .02 −.07 −.01 .31 .19
Narcissistic grandiosity .29 .15 .20 .16 .26
  Exploitativeness .40 .27 .18 .31 .25
  Self-sacrificing self-enhancementa .11 .06 .13 −.02 .12
  Grandiose fantasya .17 .05 .16 .09 .22
Note. NPI = Narcissistic Personality Inventory. The Psychological Entitlement Scale used here refers to the measure developed by Campbell et al.
(2004). The Pathological Narcissism Inventory used here refers to the measure developed by Pincus et al. (2009). Entries in boldface are significant at
p < .05.
a.Variables for which the difference in dependent correlations between the Psychological Entitlement Scale and the NPI scale score for Entitlement/
Exploitativeness was significant at .05.

dependent on external sources. One notable exception to between dependent correlations revealed that whereas the
this pattern included the moderate positive association with Psychological Entitlement Scale demonstrated stronger
PNI Exploitativeness. This likely explains the modest posi- relations with Grandiose Fantasy and Self-Sacrificing
tive association between Leadership/Authority and the Self-Enhancement, the NPI Entitlement/Exploitativeness
composite PNI Narcissistic Grandiosity scale. subscale showed a stronger relation with the PNI Devaluing
In contrast to Leadership/Authority, the Grandiose Exhi- subscale. Thus, Campbell et al.’s (2004) scale appears to
bitionism and Entitlement/Exploitativeness scales had more exhibit stronger associations with some of Pincus et al.’s
consistent associations with the PNI subscales (see Table 6). (2009) scales of narcissistic grandiosity than the NPI scale
The NPI Grandiose Exhibitionism scale, for instance, was developed here. Additional hierarchical regression analyses
positively associated with PNI Narcissistic Grandiosity and using the Psychological Entitlement Scale and the NPI
all the PNI subscales pertaining to narcissistic grandiosity. Entitlement/Exploitativeness scale to predict each of the
Moreover, Grandiose Exhibitionism was positively associ- PNI subscales corroborated this finding.6 These analyses
ated with the PNI subscale of Entitlement Rage and negatively illustrate that both scales may not assess the exact same
associated with the PNI subscale of Hiding the Self. How- construct. Indeed, the NPI Entitlement/Exploitativeness
ever, with the exception of PNI Entitlement Rage, correlations scale appears to be capturing a somewhat more vulnerable
between the NPI Grandiose Exhibitionism scale and the PNI aspect of personality than the Campbell et al. scale.
subscales were generally modest in size. Like Grandiose
Exhibitionism, the NPI Entitlement/Exploitativeness scale
was positively related to PNI Narcissistic Grandiosity and Study 4
the PNI subscales of Exploitativeness and Entitlement Rage.5 Study 4 extends the previous studies by evaluating connec-
Perhaps more interestingly, NPI Entitlement/Exploitative- tions between the NPI and both self- and informant ratings
ness was also positively linked with the PNI Contingent of the Big Five personality traits. Using self-reports of the
Self-Esteem and Devaluing subscales, thus demonstrating a Big Five traits, Corry et al. (2008) found that Leadership/
more notable connection between this NPI scale and vulner- Authority was linked with higher levels of Extraversion and
able expressions of narcissistic pathology. Conscientiousness and lower levels of Neuroticism and
Comparing the NPI Entitlement/Exploitativeness Agreeableness, whereas Exhibitionism/Entitlement was
Scale and the Psychological Entitlement Scale. The related to higher levels of Extraversion and lower levels of
final column of Table 6 presents the correlations between Agreeableness. We therefore expected to obtain a similar
the Psychological Entitlement Scale and the PNI total and pattern with self-reports and informant reports of the Big
subscale scores. There were a few areas of divergence Five with the expectation that the effect sizes would be
between these correlations and the correlations for the NPI smaller for informant reports. In addition, we evaluated the
Entitlement/Exploitativeness scale. A test of the difference intrapersonal and interpersonal correlates of the different

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

80 Assessment 18(1)

scales of the NPI in the context of college adjustment and the last week, how much have you” and then asked to
roommate relationships. respond to a series of statements on a 7-point Likert-type
scale ranging from 1 (not at all) to 7 (a great deal). Sample
items included, “Liked your classes” and “Missed your
Method friends from high school.” An overall college adjustment
Participants and Procedures. Participants included 200 score was created by taking the mean of all the items (with
roommates (34 men; 93 complete pairs of roommates)7 from the appropriate items reverse scored; M = 4.27, SD = 0.71;
a large Midwestern university. A majority of these were first- α = .79) such that higher scores indicated higher levels of
year students (72.9%) and European American (73.8%). overall college adjustment.
Participants’ ages ranged from 18 to 25 years (M = 18.44, SD = Roommate Relationship Satisfaction. This measure, origi-
1.00). A member of the pair was recruited from the psychol- nally developed by Hendrick (1988) for use with romantic
ogy subject pool and offered credit for participation. These relationships, was modified in the present study to assess
participants were then asked to bring in their roommates to individuals’ overall levels of satisfaction with their room-
participate in the study in exchange for coupons for free ice- mates. This 7-item measure included items such as, “How
cream. Participants arrived in the lab with their roommate but well does your roommate meet your needs?” and “How
were sent to separate rooms where they completed the entire good is your relationship with your roommate compared to
battery of measures. Participants knew their roommates for most?” Participants responded on a 5-point Likert-type
less than a month (24.8%), 1 to 2 months (35.5 %), 3 to 4 scale that ranged from 1 (poor/hardly at all) to 5 (extremely
months (3.3%), 5 to 6 months (0.9%), 6 months to 1 year well/excellent). Each participant’s score was the mean
(6.1%), or more than a year (22.4%). across items, with higher scores indicating greater relation-
ship satisfaction (M = 4.21, SD = 0.86; α = .93).
Measures Negative Roommate Behavioral Interactions. This question-
NPI-40 (Raskin & Terry, 1988). Along with the NPI total score naire was based on a measure used by Donnellan, Assad,
(M = 0.42, SD = 0.16; α = .82, average interitem r = .10), sub- Robins, and Conger (2007) to assess negative interactions
scales were created according to the three-factor solution: with romantic partners. The current version included 15
Leadership/Authority (M = 0.54, SD = 0.25; α = .75, average items that assessed the frequency of negative interactions
interitem r = .22), Grandiose Exhibitionism (M = 0.39, SD = with one’s roommate. Participants were asked, “During the
0.23; α = .66, average interitem r = .16), and Entitlement/ past week when you and your roommate have spent time
Exploitativeness (M = 0.21, SD = 0.25; α = .47, average inter- talking or doing things together, how often did you . . .?”
item r = .18). The Leadership/Authority subscale showed Participants then indicated the frequency with which they
modest to moderate associations with Grandiose Exhibition- engaged in a variety of behavioral interactions with their
ism (r = .35) and Entitlement/Exploitativeness (r = .25), and roommates on a 7-point Likert-type scale that ranged from 1
Grandiose Exhibitionism showed a modest association with (always) to 7 (never). Sample items included, “Get angry at
Entitlement/Exploitativeness (r = .15). him/her” and “Insult or swear at him/her.” Items were scored
Big Five Inventory (BFI). The BFI (John, Naumann, & Soto, so that higher scores indicated greater amounts of negative
2008) assessed individuals’ self-ratings of Extraversion (M = behavioral interactions (M = 2.33, SD = 0.70; α = .83).
3.61, SD = 0.69; α = .80), Neuroticism (M = 3.00, SD = 0.65;
α = .73), Conscientiousness (M = 3.56, SD = 0.57; α = .71),
Agreeableness (M = 4.00, SD = 0.59; α = .78), and Openness Results and Discussion
(M = 3.52, SD = 0.64; α = .80). It was also modified to assess Relations Between NPI Scales and the Big Five Trait
individuals’ perceptions of their roommates’ levels of Extra- Domains. As Table 7 shows, the NPI total score demonstrated
version (M = 3.53, SD = 0.88; α = .86), Neuroticism (M = significant associations with self-reports and informant reports
2.55, SD = 0.73; α = .83), Conscientiousness (M = 3.51, SD = of Extraversion and Agreeableness. This pattern fits well with
0.75; α = .85), Agreeableness (M = 3.93, SD = 0.79; α = .89), Paulhus’s (2001) conceptualization of narcissists as “disagree-
and Openness (M = 3.28, SD = 0.62; α = .81). Participants able extraverts.” This pattern of results is largely consistent
responded to all 88 personality items on a 5-point Likert-type with those for the Leadership/Authority and Grandiose Exhi-
scale that ranged from 1 (strongly disagree) to 5 (strongly bitionism scales of the NPI. In contrast, neither self- nor
agree). Each participant’s score for each of the scales was the informant-reported Extraversion related to the NPI Entitle-
mean of the responses, with higher scores indicating higher ment/Exploitativeness scale, but the correlations between this
levels of the construct. scale and Agreeableness were particularly strong and negative.
College Adjustment Test. This 19-item questionnaire, Thus, it seems as if a considerable portion of the disagreeable-
developed by Pennebaker, Colder, and Sharp (1990), ness associated with the NPI is likely driven by this facet. The
assessed the degree to which individuals had adjusted to the Entitlement/Exploitativeness scale was also positively corre-
college transition. Participants were given the stem, “Within lated with self-reports of Neuroticism, perhaps in part because

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

Ackerman et al. 81

Table 7. Correlations (r) between the Narcissistic Personality Inventory (NPI) Total Score, the NPI Scales of Leadership/Authority,
Grandiose Exhibitionism, and Entitlement/Exploitativeness, and the Big Five traits (Study 4)
NPI Scale Scores
NPI Total Score Leadership/ Grandiose Entitlement/
Authority Exhibitionism Exploitativeness
Extraversion
Self-report .44 .42 .40 .02
Informant report .20 .18 .18 .04
Agreeableness
Self-report −.35 −.25 −.15 −.51
Informant report −.20 −.17 −.09 −.23
Conscientiousness
Self-report .09 .11 −.01 −.01
Informant report −.14 −.04 −.12 −.19
Neuroticism
Self-report −.05 −.06 −.08 .26
Informant report .00 .00 .00 .07
Openness
Self-report .11 .09 .01 −.03
Informant report −.08 −.07 −.09 −.15
Note. Entries in boldface are significant at p < .05. NPI scores were based on self-report data.

that domain of personality is also linked with anger and hostil- the overall NPI reported engaging in a greater frequency of
ity. Finally, individuals reporting greater levels of Entitlement/ negative behavioral interactions with their roommate. A
Exploitativeness were perceived by their roommates to be more interesting pattern of results was found for the sub-
colder and quarrelsome, more disorganized and careless, less scales, consistent with the previously identified nomological
reliable and determined, and less artistically inclined. Although network for these dimensions.
not shown, multiple regressions revealed a generally similar The APIM analyses revealed that persons with higher
pattern of independent associations. levels of Grandiose Exhibitionism reported greater levels of
Actor-Partner Interdependence Model Analyses of college adjustment (see Table 8). Perhaps of more interest,
College Adjustment and Roommate Relationship individuals reporting higher levels of Entitlement/Exploit-
Quality. The Actor-Partner Interdependence Model (APIM; ativeness were found to express lower levels of college
Kenny, Kashy, & Cook, 2006) provides a useful method- adjustment, as well as have roommates who expressed
ological tool for assessing intrapersonal and interpersonal lower levels of college adjustment. This latter finding sug-
effects of narcissism in roommate data. In particular, the gests that Entitlement/Exploitativeness has a socially toxic
APIM permits researchers to derive estimates of the asso- effect on the adjustment of one’s roommate. Individuals’
ciation between one person’s predictor variable and her or levels of Entitlement/Exploitativeness were also associated
his own outcome variable (i.e., an actor effect), as well as with lower levels of relationship satisfaction for both them-
the association between one person’s predictor variable and selves and their roommates. In addition, individuals with
her or his partner’s outcome variable (i.e., a partner effect). higher levels of Entitlement/Exploitativeness reported
For example, we can test for an actor effect of narcissism on engaging in a higher frequency of negative behavioral inter-
relationship satisfaction, which asks whether individuals’ actions with their roommates. All in all, these findings
levels of narcissism are associated with their own levels of suggest that the Entitlement/Exploitativeness facet of the
relationship satisfaction. We can also test for a partner NPI is the dimension with more robust interpersonal and
effect of narcissism on relationship satisfaction, which asks intrapersonal correlates. Indeed, the only partner effects
whether individuals’ levels of narcissism are associated that we observed were for the Entitlement/Exploitativeness
with their roommates’ levels of relationship satisfaction. scale.
Multilevel modeling (MLM) was used to carry out the
APIM analyses. Table 8 presents the results for college
adjustment and the roommate relationship variables as out- General Discussion
comes. Analyses involving the NPI total score only revealed Disagreements about the precise factor structure of the NPI
one significant actor effect for negative behaviors (b = have generated confusion as to how the inventory should be
0.67, SE = .32, p < .05). Individuals with higher scores on scored and interpreted, as well as whether researchers

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

82 Assessment 18(1)

Table 8. Actor-Partner Interdependence Models With College Adjustment and Roommate Relationship Variables (Study 4)
Leadership/Authority Grandiose/ Exhibitionism Entitlement/Exploitativeness
Actor Partner Actor Partner Actor Partner
b SE b SE b SE b SE b SE b SE
College adjustment .39 .21 −.14 .21 .48 .24 .27 .24 −.94 .20 −.50 .20
Relationship satisfaction .19 .25 .03 .25 .05 .30 .44 .30 −.74 .25 −.53 .25
Negative behaviors .11 .21 .19 .21 −.05 .25 −.06 .25 .66 .21 .28 .21
Note. b = unstandardized regression coefficient, SE = standard error of the unstandardized regression coefficient. Entries in boldface are significant at
p < .05.

should rely exclusively on NPI total scores. The current personality and therefore should be excluded from invento-
studies ultimately provided support for a robust three-factor ries designed to measure maladaptive personality features.
solution consisting of Leadership/Authority, Grandiose In other words, what is captured in this dimension may
Exhibitionism, and Entitlement/Exploitativeness. In addi- largely reflect self-perceptions of assertiveness, social
tion to demonstrating that the three-factor structure holds potency, and other adaptive self-enhancement tendencies,
across multiple data sets, we presented evidence in support all of which bear little resemblance to constructs that are
of the utility and validity of these three dimensions by considered pathological by clinical psychologists.
showing their differential relations with a wide range of On the other hand, the elements of personality captured
theoretically relevant constructs. We now discuss the gen- by the Leadership/Authority domain are consistent with
eral implications of our findings for research and theorizing more recent discussions of narcissism by social/personality
about the multifaceted construct of narcissism. psychologists that have emphasized the agentic, approach-
Based on the current results, we believe that the concep- oriented component of narcissism and the possibility that
tual model depicted in Figure 1 provides a reasonable narcissistic tendencies may be adaptive in some contexts
perspective on the structure of the NPI and its correlates. A (Robins, Tracy, & Shaver, 2001; Sedikides & Luke, 2008).
considerable amount of variance in the NPI captures ele- Still another possibility is that Leadership/Authority may
ments of personality linked with confidence, assertiveness, be better construed as an outcome of narcissistic processes.
and beliefs of leadership potential. These elements cohere Indeed, it could be that self-perceptions of leadership abil-
into a factor that emerges in nearly all published structural ity are better thought of as a consequence of traits, such as
analyses of the NPI. An important point is that this dimen- grandiosity and exploitativeness, or even the result of adap-
sion does not seem to have consistent associations with tive strategies associated with normal narcissism, such as
socially toxic elements of personality, with the exception of assertiveness and social potency.
a modest negative correlation with Agreeableness and a In the end, it is up to the field to decide whether Leader-
moderate positive correlation with the PNI Exploitativeness ship/Authority is a core aspect of narcissism (normal or
scale. For example, our Leadership/Authority scale is unre- pathological). Therefore, as is reflected in Figure 1, we
lated to impulsive and antisocial aspects of psychopathy, present Leadership/Authority as one dimension of person-
Machiavellianism, and with most aspects of personality ality within the NPI that is generally separable from the
associated with pathological narcissism. Instead, this other facets of personality embedded within this inventory.
dimension is positively correlated with self-esteem and a Keeping this dimension of the NPI distinct from the other
reduced propensity toward internalizing psychopathology. facets will help clarify which aspects of personality are
Accordingly, this dimension is likely to be a reason why driving the observed associations with different criterion
scores on the NPI have been linked to indices of psycho- variables. This practice will likely isolate one of the more
logical health and adjustment (e.g., Sedikides et al., 2004). salient differences between social/personality and clinical
The fact that the Leadership/Authority factor is gener- conceptualizations of narcissism.
ally related to positive outcomes and unrelated to As Figure 1 shows, we also believe that the NPI captures
pathological narcissism raises questions as to whether such some of the socially noxious and socially toxic elements of
a dimension should be included in a measure of narcissism. personality associated with the DSM characterization of
After all, narcissism is one third of the so-called dark triad narcissism. Importantly, in our model, we make a distinc-
of personality (Paulhus & Williams, 2002), and it is com- tion between Grandiose Exhibitionism and Entitlement/
monly understood as a personality disorder. One possibility Exploitativeness because of the conceptual and empirical
is that Leadership/Authority assesses adaptive aspects of distinctions provided by Brown et al. (2009). Of the two

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

Ackerman et al. 83

NPI
Total
Adaptive or Score
Normal Maladaptive
Narcissism Narcissism

Leadership/ Exhibitionism/
Authority Entitlement

Grandiose Entitlement
Exhibitionism Exploitativeness

“Adaptive” Aspects of the “Socially Toxic” Aspects of

NPI the NPI
Social Potency (+) Social Potency (+)
Extraversion (+) Antisocial tendencies (+)
Extraversion (+)
Global Self-esteem (+) Machiavellianism (+)
Drive/Goal persistence (+)
BIS activation (-) Contingent self-esteem (+)
Entitlement-rage (+)
Drive/Goal persistence (+) Exploitativeness (+)
Contingent self-esteem (-) Devaluing Others (+)
Entitlement-rage (+)
Neuroticism (+)
Agreeableness (-)
College adjustment (-)
Relationship Quality (-)

Figure 1. Proposed higher order structure for the narcissistic personality inventory and initial nomological network of corresponding
dimensions
Note. NPI = Narcissistic Personality Inventory; BIS = Behavioral Inhibition System.

dimensions, Entitlement/Exploitativeness seems to have To be sure, the average interitem correlation for the Entitle-
more consistent and stronger associations with maladaptive ment/Exploitativeness scale was close to .20 in the four
outcomes. Interpersonally, this dimension (but not Grandi- studies reported here. According to Briggs and Cheek (1986),
ose Exhibitionism) was also connected with several the optimal average interitem correlation coefficient for a
pernicious outcomes assessed in our study of roommate scale should be between .20 and .40 (this is the level of asso-
relationships. In short, we believe there are compelling rea- ciation found for scales that are neither too disparate nor too
sons to separate these maladaptive dimensions and we homogeneous, respectively). Thus, based on this consider-
suspect that Entitlement/Exploitativeness will account for ation, the items in our scale might verge on acceptable levels
many of the explicitly negative outcomes linked with the of intercorrelation. It is also important to be precise about the
NPI. consequences of low internal consistency. The concern is that
We should also acknowledge some concerns with our this will attenuate relations with criterion variables. Nonethe-
Entitlement/Exploitativeness scale. The most notable issue is less, we found theoretically meaningful relations with
the low alpha coefficient. Part of the explanation for this is criterion variables reflecting socially toxic characteristics.
that the subscale consists of only four items. The NPI simply This pattern corresponds well with recent evidence by
does not have many items that assess this dimension (see McCrae, Kurtz, Yamagata, and Terracciano (IN PRESS) sug-
Raskin & Terry, 1988). Thus, the concern might boil down to gesting that the internal consistency of personality scales is
the total number of items rather than the content of the scale. only weakly related to their validity.

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

84 Assessment 18(1)

Taken together, the current set of studies reinforces the current research consistent with previous studies. More-
claim that the NPI is a multidimensional instrument. This is over, given that our factor solution is based on the 40-item
consistent with the view expressed by its creators in 1988. forced-choice NPI, we recommend that researchers con-
However, this multidimensionality is often overlooked by sub- tinue using this format until other alternative response
sequent researchers (including ourselves) who have focused formats for the NPI become more widely used and similarly
on NPI total scores. Our broad concern is that an exclusive validated.
reliance on the NPI total score is likely to lead to an imprecise Although we believe that the present set of studies repre-
understanding of narcissism because it conflates adaptive and sents one of the most comprehensive evaluations of the
maladaptive aspects of personality (Barry et al., 2003). As structure of the NPI, there are limitations to the current
Briggs and Cheek (1986) cautioned nearly 25 years ago, “. . . it work. The most notable limitations are the exclusive reli-
is unacceptable to continue using a total score alone when to ance on college student samples and the use of predominantly
do so deliberately ignores distinctions that are conceptually self-report data to establish the network of criterion-related
meaningful and empirically useful” (p. 129). associations surrounding the NPI. Future research should
The current studies also provide at least one example of aim to replicate our findings with more diverse populations
how the NPI total score may generate findings that have the and behavioral measures as outcomes.
potential to be misleading. Consider that some researchers In conclusion, the present findings provide support for
have suggested that approach-related motivations are a cen- the concerns articulated by Brown et al. (2009) over the use
tral feature of narcissism (e.g., Foster & Trimm IV, 2008). of the NPI for the next generation of studies on narcissism.
However, our investigation suggests that this may only be We are sympathetic to their view regarding the possible
the case for the Leadership/Authority and Grandiose Exhi- advantages of replacing the NPI with better and more direct
bitionism dimensions of narcissism. In fact, the Entitlement/ measures of the personality attributes associated with nar-
Exploitativeness dimension was shown to be unrelated to cissism (e.g., p. 963). It is undeniable that there are
self-reports and informant reports of extraversion, and even limitations with respect to the psychometric properties of
negatively related to the reward responsiveness subscale of the NPI. Nonetheless, we suspect that the field will be
the BAS. This dimension of personality therefore appears reluctant to simply abandon this measure, and the present
to have little to do with approach-related motivational sys- study is important as it identifies one potentially useful way
tems, yet it was the dimension with the most consistent of parsing the NPI items into meaningful subscales. Our
links with maladaptive criterion variables. Thus, there are work also highlights some of the constructs embedded
hints, at least in these data, that approach-related motiva- within the NPI that should be the targets for future scale
tions are only linked to certain facets of narcissism and development work. The bottom line is that we believe
those dimensions have little to do with psychological enti- increased attention to the three replicable facets identified
tlement and exploitativeness. in the present research will provide important benefits for
Ultimately, we suggest that researchers who use the NPI future studies regarding the nature and correlates of the
routinely conduct subscale analyses. Given that all three NPI.
dimensions are positively interrelated, there may be some
situations in which the subscales do not yield a distinct pat- Acknowledgment
tern of results and thus total score reporting is more efficient. Christopher Hopwood provided helpful comments on a previous
Nonetheless, we suspect that there will be many cases draft.
where subscale analyses yield more precise psychological
insights into the correlates and consequences of narcissism. Declaration of Conflicting Interests
The present findings also show that countervailing relations The authors declared no potential conflicts of interest with respect
between the NPI subscales and criterion variables can gen- to the authorship and/or publication of this article.
erate a null effect for the total score, which could mislead
researchers into believing that narcissism has no relation Funding
with the criterion-related variables. The authors received no financial support for the research and/or
Researchers may initially consider using these results to authorship of this article.
justify administering a reduced pool of the forced-choice
NPI items to participants given that only 25 items are used Notes
in our three-factor solution. This practice might save a few 1. The Corry et al. (2008) factor structure seemed to be embedded
minutes of participant time but we think that this strategy within the three-factor solution. To quantify this, we calculated
would be less than ideal at this point. Instead, we recom- scores for the Corry et al. factors using the items reported in
mend that researchers continue to administer all 40 items their article, and created scales for the three-factor solution by
because this approach will afford the most flexibility in selecting those items with the highest pattern loadings for each
terms of choices of existing factor solutions and will keep factor that were greater than .40. Our Leadership/Authority

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

Ackerman et al. 85

scale was strongly correlated with the Corry et al. Leadership/ p < .001. Hierarchical regression analyses with Entitlement/
Authority scale (r = .94). Moreover, the Grandiose Exhibition- Exploitativeness in the first step and the Psychological En-
ism and Entitlement/Exploitativeness scales were both correlat- titlement Scale entered in the second step showed that this
ed with the Corry et al. Exhibitionism/Entitlement scale (r = .89 measure accounted for significant incremental variance in the
and r = .55, respectively), and were only modestly correlated PNI total score, ∆R2 = .033, ∆F(1, 329) = 12.19, p = .001; the
with each other (r = .23). These findings suggest that our solu- PNI Narcissistic Grandiosity scale, ∆R2 = .048, ∆F(1, 329)
tion splits the Corry et al. Exhibitionism/Entitlement factor into = 17.18, p < .001; the PNI Narcissistic Vulnerability scale,
two reasonably separate dimensions whereas it retains a factor ∆R2 = .018, ∆F(1, 329) = 6.74, p = .01, and the PNI subscales
that is isomorphic with their Leadership/Authority dimension. of Exploitativeness, ∆R2 = .021, ∆F(1, 329) = 7.71, p = .006;
2. Given the low reliability of the Entitlement/Exploitativeness Self-Sacrificing Self-Enhancement, ∆R2 = .02, ∆F(1, 329) =
scale, we considered relaxing the scale-inclusion criteria to 6.78, p = .01; Grandiose Fantasy, ∆R2 = .042, ∆F(1, 329) =
patterns loadings >.30 to increase the number of items; un- 14.51, p < .001; and Entitlement Rage (∆R2 = .105, ∆F(1, 329) =
fortunately, most of the items that fulfilled this criterion had 48.99, p < .001.
substantial cross-loadings with one of the other factors, and so 7. Some roommate pairs had missing data from one member of
we decided to retain the initial four-item scale. the pair.
3. Across both EFAs in Studies 2 and 3, we found that Item 40
exhibited a substantially higher loading on Grandiose Exhibi- References
tionism than Leadership/Authority. To anticipate the degree to American Psychiatric Association. (1980). Diagnostic and statis-
which removing Item 40 from Leadership/Authority and plac- tical manual of mental disorders (3rd ed.). Washington, DC:
ing it in the Grandiose Exhibitionism scale would change the Author.
results, we computed new variables for Leadership/Authority Barry, C. T., Frick, P. J., & Killian, A. L. (2003). The relation of
and Grandiose Exhibitionism with these modifications and narcissism and self-esteem to conduct problems in children: A
correlated them with the original scales in Study 3 (rs = .99 preliminary investigation. Journal of Clinical Child and Ado-
and .98, respectively). The placement of Item 40 will likely lescent Psychology, 32, 139-152.
have little bearing on the results. Bennett, R. J., & Robinson, S. L. (2000). Development of a mea-
4. Multiple regression analyses were also conducted between the sure of workplace deviance. Journal of Applied Psychology,
NPI scales and these other measures. Overall, the pattern of asso- 85, 349-360.
ciations between the NPI scales and the criterion variables did not Benning, S. D., Patrick, C. J., Hicks, B. M., Blonigen, D. M., &
diverge substantially from their zero-order relations. There were Krueger, R. F. (2003). Factor structure of the Psychopathic
only three exceptions. First, when controlling for Leadership/Au- Personality Inventory: Validity and implications for clinical
thority and Entitlement/Exploitativeness, Grandiose Exhibition- assessment. Psychological Assessment, 15, 340-350.
ism was no longer related to PNI Exploitativeness. Second, when Bradlee, P. M., & Emmons, R. A. (1992). Locating narcissism
controlling for Grandiose Exhibitionism and Entitlement/Ex- within the interpersonal circumplex and the five-factor model.
ploitativeness, Leadership/Authority was no longer significantly Personality and Individual Differences, 13(7), 821-830.
related to PNI Narcissistic Grandiosity. Finally, when controlling Briggs, S. R., & Cheek, J. M. (1986). The role of factor analysis in
for Grandiose Exhibitionism and Entitlement/Exploitativeness, the development and evaluation of personality scales. Journal
Leadership/Authority became significantly negatively related to of Personality, 54, 106-148.
PNI Narcissistic Vulnerability. Brown, R. P., Budzek, K., & Tamborski, M. (2009). On the mean-
5. Given the overlap in content, it is potentially notable that the NPI ing and measure of narcissism. Personality and Social Psy-
Entitlement/Exploitativeness scale was only moderately (rather chology Bulletin, 35, 951-964.
than strongly) correlated with the PNI subscale of Exploitative- Browne, M. W. (2001). An overview of analytic rotation in explor-
ness. Nevertheless, after correcting for the attenuation due to un- atory factor analysis. Multivariate Behavioral Research, 36,
reliability, the correlation between the scales becomes .54. 111-150.
6. Hierarchical regression analyses with the Psychological En- Cain, N. M., Pincus, A. L., & Ansell, E. B. (2008). Narcissism at
titlement Scale entered in the first step and Entitlement/Ex- the crossroads: Phenotypic description of pathological narcis-
ploitativeness entered in the second step showed that Entitle- sism across clinical theory, social/personality psychology, and
ment/Exploitativeness accounted for significant incremental psychiatric diagnosis. Clinical Psychology Review, 28, 638-
variance in the PNI total score, ∆R2 = .046, ∆F(1, 329) = 656.
17.22, p < .001; the PNI Narcissistic Vulnerability scale, ∆R2 = Campbell, W. K., Bonacci, A. M., Shelton, J., Exline, J. J., &
.064, ∆F(1, 329) = 24.02, p < .001, and the PNI subscales Bushman, B. J. (2004). Psychological entitlement: Interper-
of Contingent Self-Esteem ∆R2 = .036, ∆F(1, 329) = 12.61, sonal consequences and validation of a self-report measure.
p < .001; Exploitativeness, ∆R2 = .057, ∆F(1, 329) = 21.30, Journal of Personality Assessment, 83, 29-45.
p < .001; Devaluing, ∆R2 = .065, ∆F(1, 329) = 23.72, p < Campbell, W. K., Foster, C. A., & Finkel, E. J. (2002). Does
.001; and Entitlement Rage, ∆R2 = .084, ∆F(1, 329) = 39.23, self-love lead to love for others? A story of narcissistic game

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

86 Assessment 18(1)

playing. Journal of Personality and Social Psychology, 83, Kernberg, O. F. (1975). Borderline conditions and pathological
340-354. narcissism. New York, NY: Jason Aronson.
Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behav- Kernberg, O. F. (1998). Pathological narcissism and narcissistic
ioral activation, and affective responses to impending reward personality disorder: Theoretical background and diagnos-
and punishment: The BIS/BAS scales. Journal of Personality tic classification. In E. F. Ronningstam (Ed.), Disorders of
and Social Psychology, 67, 319-333. narcissism: Diagnostic, clinical, and empirical implications
Christie, R., & Geis, F. L. (1970). Studies in Machiavellianism. (pp. 29-51). Northvale, NJ: Jason Aronson.
New York, NY: Academic Press. Kohut, H. (1966). Forms and transformations of narcissism. In
Corry, N., Merritt, R. D., Mrug, S., & Pamp, B. (2008). The factor C. B. Strozier (Ed.), Self psychology and the humanities:
structure of the Narcissistic Personality Inventory. Journal of Reflections on a new psychoanalytic approach (pp. 124-160).
Personality Assessment, 90, 593-600. New York, NY: W. W. Norton.
Donnellan, M. B., Assad, K. K., Robins, R. W., & Conger, R. D. Kohut, H. (1971). The analysis of the self. New York, NY: Interna-
(2007). Do negative interactions mediate the effects of nega- tional Universities Press.
tive emotionality, communal positive emotionality, and con- Kubarych, T. S., Deary, I. J., & Austin, E. J. (2004). The Narcis-
straint on relationship satisfaction? Journal of Social and sistic Personality Inventory: Factor structure in a non-clinical
Personal Relationships, 24, 557-573. sample. Personality and Individual Differences, 36, 857-872.
Ellis, H. (1898). Auto-eroticism: A psychological study. Alienist Lilienfeld, S. O., & Widows, M. R. (2005). Psychopathic Person-
and Neurologist, 19, 260-299. ality Inventory–Revised: Professional manual. Lutz, FL: Psy-
Emmons, R. A. (1984). Factor analysis and construct validity of chological Assessment Resources.
the Narcissistic Personality Inventory. Journal of Personality Locke, K. D. (2009). Aggression, narcissism, self-esteem, and the
Assessment, 48, 291-300. attribution of desirable and humanizing traits to self versus
Emmons, R. A. (1987). Narcissism: Theory and measurement. others. Journal of Research in Personality, 43, 99-102.
Journal of Personality and Social Psychology, 52, 11-17. Marsh, H. W., Hau, K. T., & Wen, Z. (2004). In search of golden
Fabrigar, L. R., Wegener, D. T., MacCallum, R. C., & Strahan, E. rules: Comment on hypothesis-testing approaches to setting cut-
J. (1999). Evaluating the use of exploratory factor analysis in off values for fit indexes and dangers in overgeneralizing Hu and
psychological research. Psychological Methods, 4, 272-299. Bentler’s (1999) findings. Structural Equation Modeling, 11, 320.
Foster, J. D., & Trimm, R. F., IV. (2008). On being eager and unin- McCrae, R. R., Kurtz, J. E., Yamagata, S., & Terracciano, A. (IN
hibited: Narcissism and approach-avoidance motivation. Per- PRESS). Internal consistency, retest reliability, and their impli-
sonality and Social Psychology Bulletin, 34, 1004-1017. cations for personality scale validity. Personality and Social
Freud, S. (1986). On narcissism: An introduction. In A. P. Mor- Psychology Review.
rison (Ed.), Essential papers on narcissism (pp. 17-43). New Miller, J. D., & Campbell, W. K. (2008). Comparing clinical and
York, NY: New York University Press. (Original work pub- social-personality conceptualizations of narcissism. Journal of
lished 1914) Personality, 76, 449-476.
Fulford, D., Johnson, S. L., & Carver, C. S. (2008). Common- Miller, J. D., Gaughan, E. T., Pryor, L. R., Kamen, C., & Campbell,
alities and differences in characteristics of persons at risk for W. K. (2009). Is research using the Narcissistic Personality
narcissism and mania. Journal of Research in Personality, 42, Inventory relevant for understanding narcissistic personality
1427-1438. disorder? Journal of Research in Personality, 43, 482-488.
Hendrick, S. (1988). A generic measure of relationship satisfac- Millon, T. (1996). Disorders of personality: DSM-IV and beyond.
tion. Journal of Marriage and the Family, 50, 93-98. New York, NY: Wiley.
Hopwood, C. J., & Donnellan, M. B. (2010). How should the Patrick, C. J., & Bernat, E. M. (2009). Neurobiology of psychopa-
internal structure of personality inventories be evaluated? Per- thy: A two-process theory. In G. G. Berntson & J. T. Cacioppo
sonality and Social Psychology Review, 14, 332-346. (Eds.), Handbook of neuroscience for the behavioral sciences
John, O. P., Naumann, L. P., & Soto, C. J. (2008). Paradigm shift (pp. 1110-1131). New York, NY: Wiley.
to the integrative Big Five trait taxonomy: History, measurement, Paulhus, D. L. (2001). Normal narcissism: Two minimalist
and conceptual issues. In O. P. John, R. W. Robins, & L. A. Pervin accounts. Psychological Inquiry, 12, 228-230.
(Eds.), Handbook of personality psychology: Theory and research Paulhus, D. L., & Williams, K. M. (2002). The dark triad of per-
(3rd ed., pp. 114-158). New York, NY: Guilford Press. sonality: Narcissism, Machiavellianism and psychopathy.
John, O. P., & Robins, R. W. (1994). Accuracy and bias in self- Journal of Research in Personality, 36, 556-563.
perception: Individual differences in self-enhancement and the Pennebaker, J. W., Colder, M., & Sharp, L. K. (1990). Accelerat-
role of narcissism. Journal of Personality and Social Psychol- ing the coping process. Journal of Personality and Social Psy-
ogy, 66(1), 206-219. chology, 58, 528-537.
Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006). Dyadic data Pincus, A. L., Ansell, E. B., Pimentel, C. A., Cain, N. M., Wright,
analysis. New York, NY: Guilford Press. A. G. C., & Levy, K. N. (2009). Initial construction and

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011

Ackerman et al. 87

validation of the pathological narcissism inventory. Psycho- Tangney, J. P., Baumeister, R. F., & Boone, A. L. (2004). High
logical Assessment, 21, 365-379. self-control predicts good adjustment, less pathology, better
Pincus, A. L., & Lukowitsky, M. R. (2010). Pathological narcis- grades, and interpersonal success. Journal of Personality, 72,
sism and narcissistic personality disorder. Annual Review of 271-322.
Clinical Psychology, 6, 421-446. Tracy, J. L., & Robins, R. W. (2003). “Death of a (narcissistic)
Raskin, R. N., & Hall, C. S. (1979). A Narcissistic Personality salesman”: An integrative model of fragile self-esteem. Psy-
Inventory. Psychological Reports, 45, 590. chological Inquiry, 14, 57-62.
Raskin, R. N., & Hall, C. S. (1981). The Narcissistic Personality Trzesniewski, K. H., Donnellan, M. B., & Robins, R. W. (2008). Is
Inventory: Alternate form reliability and further evidence of con- “generation me” really more narcissistic than previous genera-
struct validity. Journal of Personality Assessment, 45, 159-162. tions? Journal of Personality, 76, 903-918.
Raskin, R. N., Novacek, J., & Hogan, R. (1991). Narcissistic self- Vazire, S., & Funder, D. C. (2006). Impulsivity and the self-
esteem management. Journal of Personality and Social Psy- defeating behavior of narcissists. Personality and Social Psy-
chology, 60, 911-918. chology Review, 10, 154-165.
Raskin, R. N., & Terry, H. (1988). A principal components analysis Wallace, H. M., Ready, C. B., & Weitenhagen, E. (2009). Narcis-
of the Narcissistic Personality Inventory and further evidence sism and task persistence. Self and Identity, 8, 78-93.
of its construct validity. Journal of Personality and Social Psy- Watson, P. J., & Biderman, M. D. (1993). Narcissistic Personality
chology, 54, 890-902. Inventory factors, splitting and self-consciousness. Journal of
Rhodewalt, F., & Morf, C. C. (1995). Self and interpersonal cor- Personality Assessment, 61, 41-57.
relates of the Narcissistic Personality Inventory. Journal of Watson, P. J., Grisham, S. O., Trotter, M. V., & Biderman, M. D.
Research in Personality, 29, 1-23. (1984). Narcissism and empathy: Validity evidence for the
Rhodewalt, F., & Morf, C. C. (1998). On self-aggrandizement and Narcissistic Personality Inventory. Journal of Personality
anger: A temporal analysis of narcissism and affective reac- Assessment, 48, 301-305.
tions to success and failure. Journal of Personality and Social Watson, P. J., Little, T., Sawrie, S. M., & Biderman, M. D. (1992).
Psychology, 74, 672-685. Measures of the narcissistic personality: Complexity of rela-
Rhodewalt, F., & Peterson, B. (2009). Narcissism. In M. R. Leary tionships with self-esteem and empathy. Journal of Personality
& R. H. Hoyle (Eds.), Handbook of individual differences in Disorders, 6, 434-449.
social behavior (pp. 547-560). New York, NY: Guilford Press. Watson, P. J., & Morris, R. J. (1991). Narcissism, empathy and
Robins, R. W., Tracy, J. L., & Shaver, P. R. (2001). Shamed into social desirability. Personality and Individual Differences, 12,
self-love: Dynamics, roots, and functions of narcissism. Psy- 575-579.
chological Inquiry, 12, 230-236. Witt, E. A., & Donnellan, M. B. (2008). Furthering the case for the
Rosenberg, M. (1965). Society and the adolescent self-image. MPQ-based measures of psychopathy. Personality and Indi-
Princeton, NJ: Princeton University Press. vidual Differences, 45(3), 219-225.
Ruiz, J. M., Smith, T. W., & Rhodewalt, F. (2001). Distinguishing Witt, E. A., Donnellan, M. B., & Blonigen, D. M. (2009).
narcissism and hostility: Similarities and differences in inter- Using existing self-report inventories to measure the psycho-
personal circumplex and five-factor correlates. Journal of Per- pathic personality traits of fearless dominance and impulsive
sonality Assessment, 76, 537-555. antisociality. Journal of Research in Personality, 43,
Sedikides, C., & Luke, M. (2008). On when self-enhancement 1006-1016.
and self-criticism function adaptively and maladaptively. In Witt, E. A., Donnellan, M. B., Blonigen, D. M., Krueger, R. F., &
E. C. Chang (Ed.), Self-criticism and self-enhancement: The- Conger, R. D. (2009). Assessment of fearless dominance and
ory, research, and clinical implications (pp. 181-198). Wash- impulsive antisociality via normal personality measures: Con-
ington, DC: American Psychological Association. vergent validity, criterion validity, and developmental change.
Sedikides, C., Rudich, E. A., Gregg, A. P., Kumashiro, M., & Journal of Personality Assessment, 91(3), 265-276.
Rusbult, C. (2004). Are normal narcissists psychologically Wright, A. G. C., Lukowitsky, M. R., Pincus, A. L., & Conroy, D. E.
healthy? Self-esteem matters. Journal of Personality and (IN PRESS). The higher order factor structure and gender invari-
Social Psychology, 87, 400-416. ance of the Pathological Narcissism Inventory. Assessment.

Downloaded from asm.sagepub.com at UNIV CALIFORNIA DAVIS on September 22, 2011