You are on page 1of 18

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/338753754

Replicable factor structure and correlates of an alternate version of the


narcissistic personality inventory

Preprint · January 2020

CITATIONS READS

0 80

2 authors:

Matthew Brown Kasey Stanton


The University of Western Ontario Virginia Polytechnic Institute and State University
13 PUBLICATIONS   116 CITATIONS    42 PUBLICATIONS   338 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Dissociative Experiences and Self-Destructive Behaviours View project

All content following this page was uploaded by Matthew Brown on 05 February 2020.

The user has requested enhancement of the downloaded file.


Journal of Psychopathology and Behavioral Assessment
https://doi.org/10.1007/s10862-020-09790-y

Replicable Factor Structure and Correlates of an Alternate Version


of the Narcissistic Personality Inventory
Matthew F. D. Brown 1 & Kasey Stanton 1,2 & David Watson 3

# Springer Science+Business Media, LLC, part of Springer Nature 2020

Abstract
The Narcissistic Personality Inventory (NPI) traditionally has been administered using 40 pairs of items with a forced-choice
response format. However, recent NPI research indicates that it may be advantageous to administer only the item from each pair
that is in the narcissistic, keyed direction (i.e., using a “single-stimulus” format) in conjunction with a Likert style response rating
scale (Ackerman et al. 2016). This same research and other prior studies also suggest that the NPI items assess distinct grandiose
narcissistic traits (e.g., manipulativeness, leadership), such that focusing solely on total scores for analyses may be problematic.
We extend this previous work by examining the NPI’s factor structure and correlates in both clinically-oriented community (N =
700) and undergraduate (N = 527) samples when this measure is administered using a single-stimulus format and a Likert style
rating scale. Our results indicated that five replicable NPI factors (i.e., Manipulativeness, Grandiose Fantasies, Vanity,
Leadership/Authority, Superiority) could be identified across samples. Subscales modeling these factors showed distinctive
patterns of personality and psychopathology relations. For example, some subscales (e.g., Leadership/Authority, Superiority)
showed strong positive relations with agentic aspects of extraversion and robust negative relations with neuroticism, but other
subscales (e.g., Manipulativeness, Grandiose Fantasies) showed comparatively weaker associations with these personality traits.
These results highlight a potentially more useful administration format for the NPI, and also underscore need to consider the
distinctive nature of NPI subscales in future research using this measure.

Keywords Narcissism . Assessment . Factor analysis . Psychopathology

A voluminous literature spanning clinical, social/personality, (Krizan and Herlache 2018) organizes these distinct personal-
and industrial-organizational psychology focuses on the broad ity dimensions and suggests that narcissistic traits can be
construct of narcissism. Integrating these different perspec- categorized into three domains of entitlement, grandiosity,
tives has proven challenging for the field and has resulted in and vulnerability. Krizan and Herlache (2018) describe
debate regarding how narcissism should be defined and mea- entitlement/self-importance as core to narcissism, with gran-
sured (Wright and Edershile 2018). Much of the challenge diosity and vulnerability being other broad dimensions rele-
with conceptualizing narcissism relates to integrating both vant to narcissism. Other recent articles providing integrative
maladaptive (e.g., contingent self-esteem) and adaptive nar- summaries of the narcissism literature have reached similar
cissistic dimensions (e.g., leadership, authority, motivation for conclusions regarding which narcissistic trait dimensions are
achievement and success) within a single integrated frame- core versus peripheral to this broad construct (Miller et al.
work. The recently proposed Narcissism Spectrum Model 2017; Wright and Edershile 2018).
For decades, various iterations of the Narcissistic
Personality Inventory (NPI; Raskin and Hall 1979) have been
* Matthew F. D. Brown
mbrow253@uwo.ca
used widely in clinical and social/personality research. This
measure has been shown to primarily assess grandiose narcis-
1
sistic traits (e.g., seeing oneself as a good leader; being
Department of Psychology, The University of Western Ontario, 1151
exhibitionistic; Cain et al. 2008; Krizan and Herlache 2018).
Richmond Street, London, Ontario N6A 3K7, Canada
2
The traditional response format of the NPI forces participants
Psychology Department, Virginia Polytechnic Institute and State
to choose between one of two oppositely keyed items (e.g.,
University, 121 Williams Hall, Blacksburg, VA 24061, USA
3
“am essentially a modest person” or “modesty does not be-
Department of Psychology, The University of Notre Dame, 390
come me”), and a 40-item version of the NPI has become most
Corbett Family Hall, Notre Dame, IN 46556, USA
J Psychopathol Behav Assess

widely used in the literature (Raskin and Terry 1988). Leadership, Exhibitionism, Vanity, Manipulativeness, and
However, despite the widespread use of the NPI over multiple Superiority.
decades, researchers have raised concerns about the concep- Furthermore, and importantly, limited research has exam-
tual underpinnings and psychometric properties of the instru- ined relations for NPI subscales when this measure is admin-
ment (Miller et al. 2017; Pincus and Lukowitsky 2010). First, istered using a Likert-style response format rather than the
researchers have had difficulty identifying the NPI’s optimal traditional forced-choice format. Miller et al. (2018) examined
factor structure, as evidence exists for two- (Corry et al. 2008), this issue directly by evaluating the degree to which patterns
three- (Ackerman et al. 2011; Wetzel et al. 2016), four- of correlates for NPI subscales based on Ackerman et al.’s
(Emmons 1984), and five-factor NPI structures (Ackerman (2011) three-factor structure of Leadership/Authority,
et al. 2016). Notably, the three-factor structure of Grandiose Exhibitionism, and Entitlement/Exploitativeness
Leadership/Authority, Grandiose Exhibitionism, and differed when using a Likert-style versus forced-choice for-
Entitlement/Exploitativeness reported by Ackerman et al. mat. These researchers reported that the forced-choice and
(2011) has become most widely used (Ackerman et al. 2016; Likert versions of the NPI performed relatively similarly re-
Miller et al. 2017; Stanton et al. 2017). However, this progress garding their relations with external variables. For example,
aside, concerns have been raised about the forced-choice re- they found that all NPI subscale scores—regardless of the
sponse format of the NPI; namely, that different response pairs rating scale used—were associated with immodesty, high ex-
included within the same item do not necessarily represent traversion, antagonism, and attention-seeking across both rat-
opposite ends of the same targeted dimension (Ackerman ing types. Examination of Miller et al.’s (2018) results also
et al. 2016). For example, one NPI item asks participants to point to the importance of examining relations between the
choose between response options related to being able to “talk NPI and external variables at the subscale level. To illustrate,
my way out of anything” and being willing to “accept the although all NPI subscales were negatively related to agree-
consequences of my behavior.” As another example, it may ableness, Entitlement/Exploitativeness subscale scores were
be problematic to force participants to choose between more strongly related to low agreeableness than were scores
“insisting upon getting respect” and “usually being able to on other subscales. Similarly, although the NPI total score was
get respect.” strongly related to dominance (e.g., rating oneself as domi-
Due to concerns with the original forced-choice format of neering, assertive), NPI item content related to leadership/
the NPI, some researchers have favoured adopting (a) a Likert authority (e.g., “am a natural leader”) appears to show much
style rating format when administering the NPI as well as (b) stronger ties to dominance than other NPI content. Finally,
administering only the 40 statements that are in the keyed Miller et al. (2018) noted that the Likert version of the NPI
direction (e.g., for an item assessing attention seeking, admin- may be preferable when using NPI subscales assessing
istering a statement about “liking attention” but not a matched exhibitionism/entitlement item content in future research, as
statement about preferring to “blend in with the crowd”). this format appears to yield subscales showing higher internal
Administering only items in the keyed direction—referred to consistency and improved convergent and discriminant valid-
as administering the NPI with a “single-stimulus” format— ity patterns compared to administration with forced-choice.
offers several practical benefits, including briefer, more
straightforward administration (Ackerman et al. 2016).
However, use of a single-stimulus administration format also The Present Study
has been found to influence the NPI’s factor structure
(Ackerman et al. 2016; Boldero et al. 2015; c.f. Wetzel et al. Primary Study Aims Taken together, prior research indicates
2016). that a single-stimulus rather than forced-choice administration
For instance, Ackerman et al. (2016) report data from sev- format may be optimal when administering the NPI
eral large samples of adults indicating that the NPI’s factor (Ackerman et al. 2016; Miller et al. 2018). However, only
structure may be more nuanced when using a single- one study has examined the item-level structure of the NPI
stimulus versus forced-choice administration format. when administered using a single-stimulus format (Ackerman
Specifically, these researchers found that although three fac- et al. 2016), and to date, limited research has explicated the
tors best represent the NPI’s structure when using forced- personality and psychopathology relations of the NPI when it
choice administration, five factors best reflect its structure is administered using alternate formats.
when using a single-stimulus administration format. Thus, in the current study, our primary aim was to examine
Ackerman et al. (2016) reported that when using a single- the NPI’s factor structure when it is administered using a
stimulus administration format with either a dichotomous single-stimulus format. We also administered the NPI using
(true/false) or 5-point Likert (i.e., strongly disagree to strongly a 5-point Likert response format ranging from 1 (strongly
agree) response scale, a subset of 23 of the NPI’s 40 items can disagree) to 5 (strongly agree) here given evidence indicating
be used to model these five dimensions, which they labeled that this response scale may be advantageous when examining
J Psychopathol Behav Assess

the NPI’s factor structure and correlates (Ackerman et al. Method


2016; Miller et al. 2018). Notably, we conducted these analy-
ses in both an undergraduate sample and a clinically-oriented Participants
community sample (i.e., individuals indicating that they were
receiving current psychotherapy and/or taking psychotropic Undergraduates The first sample consisted of undergraduates
medication). This provided another significant extension of from the psychology department research subject pool at a
previous NPI research, which focused on adult samples but Midwestern university in the United States. These participants
did not collect treatment-related information from participants completed all study measures online and were provided
(Ackerman et al. 2016). Thus, we were able to determine the course credit for participation. Our original sample size
degree to which our factor analytic findings replicated across consisted of 556 undergraduates. However, four validity items
samples in which many individuals were likely (i.e., commu- (e.g., “Please answer ‘very much like me’”) were included in
nity adults reporting current mental health treatment) versus our study protocol, and we removed data for 29 participants
relatively unlikely (e.g., undergraduates) to report high levels who incorrectly answered two or more of these unambiguous
of psychopathology. In conducting our factor analyses, our items. Thus, our final sample size used for analyses was 527
goal was to identify the largest number of interpretable factors undergraduates. Of these 527 participants, the majority iden-
that replicated across samples, consistent with prior NPI struc- tified as female (71.2%), and the mean age was 19.2 years
tural research (Ackerman et al. 2016). (SD = 1.5). The majority of these undergraduates identified
As a second related aim, we created subscales modeling the as Caucasian (74.0%), with 9.1%, 8.2%, 3.8%, and 3.6%
emergent factors from our structural analyses and examined identifying as Hispanic/Latinx, Asian/Asian American,
their relations with other personality and psychopathology var- Black/African American, and Multiracial, respectively (the
iables. It is important to note here that researchers have chal- small remaining percentage identified as other ethnicities or
lenged the degree to which some NPI content—at least when chose not to provide this information). In total, 7.6% of the
administered with a force-choice format—is pathological in sample endorsed receiving current psychotherapy, and 11.6%
nature. For example, it has been argued that select NPI content reported receiving current psychotropic medication.
(e.g., “am assertive”) represents relatively adaptive content in-
dicative of achievement motivation and normal range extraver- Clinically-Oriented Community Sample Our second sample
sion more so than pathological narcissism (e.g., Stanton et al. consisted of adults recruited from Amazon Mechanical Turk
2017). However, whether similar arguments can be raised (see Arditte et al. 2016; McCredie and Morey 2019), which
when the NPI is administered using alternate formats has we refer to as the “community sample.” Prior to completing
remained relatively unexplored (e.g., it may be possible that the full study, participants were screened for current psychiat-
using a Likert style response scale influences the degree to ric treatment status using a brief survey including items
which item content appears more or less adaptive). assessing receiving (a) current psychotherapy and (b) medica-
tion to treat mental health issues. Participants had to positively
Hypotheses Based on the nature of the NPI’s item content and endorse at least one of these items to be invited to complete the
the results presented by Ackerman et al. (2016), we predicted full study, and our brief screening measure also included other
that factors related to leadership (e.g., “am assertive”), exhibi- items (e.g., items assessing religiosity and physical activity
tionism (e.g., “like to be the center of attention”), and manip- levels) so that it was not obvious that current psychiatric treat-
ulativeness (e.g., “can talk my way out of anything”) likely ment status would be required to be endorsed.
would emerge, as might other potential factors. However, we We screened 6271 participants using this brief screening
were unsure of the exact nature of factors and how many rep- measure. However, only 1250 (19.9%) of these original 6271
licable factors would emerge in our analyses given the limited participants who were screened endorsed current psychothera-
prior item-level research administering the NPI and utilizing py and/or psychotropic medication and were therefore invited
different types of respondents (Ackerman et al. 2016). to complete the full study. Some individuals invited to com-
Next, we predicted that item content related to manipula- plete the full study elected not to do so, as only 800 participants
tiveness would associate robustly with indicators of psychop- proceeded to complete the full study. Furthermore, 100 of these
athy and (low) agreeableness, and that such associations 800 participants who elected to complete the full study provid-
would be stronger than those for other emergent dimensions. ed incorrect responses to two or more of four unambiguous
Furthermore, we anticipated that leadership and exhibitionism validity items (e.g., “Please answer ‘very much like me’”;
item content would associate robustly with agentic aspects of n = 28), failed to respond to nearly all study items, or did not
extraversion such as assertiveness and exciting seeking. As report currently receiving psychotherapy and/or psychotropic
stated, we were particularly interested in determining the de- medication when treatment status was reassessed in the full
gree to which NPI content assessing leadership and related study; therefore, data for these 100 participants were removed
traits appeared maladaptive versus adaptive. prior to conducting final analyses.
J Psychopathol Behav Assess

Thus, we focused on a final sample of 700 participants here spectrum, and obsessive-compulsive and related disorders.
who reported current psychiatric treatment status, completed The Mania and Euphoria scales from the IDAS-II have dem-
the full study, and provided valid data (also see Stanton et al. onstrated good internal consistency (α’s from .72–.86; Watson
(2019) for additional details regarding this sample). We did et al. 2012), as well as strong convergent validity with other
not collect information regarding specific psychiatric diagno- indicators of bipolar disorder (Watson et al. 2012).
ses in either the screening measure or the full study protocol, Participants indicated the extent to which they experienced
both of which were completed by participants online. At the each IDAS-II symptom over the past 2 weeks using a 5-
time of the study, 62% of the sample reported receiving cur- point scale ranging from 1 (not at all) to 5 (extremely).
rent psychotherapy, and 88% of the sample reported receiving
psychotropic medication. The majority of the sample identi- Pathological Narcissism Inventory (PNI) The PNI assesses sev-
fied as female (68.4%), and the average age was 32.8 (SD = en specific narcissism facets defining broad dimensions of
10.1). In total, 84.4% of the sample identified as Caucasian Grandiosity (i.e., Grandiose Fantasy, Exploitativeness, and
(4.3%, 3.4%, 3.4%, 3.0% identified as Multiracial, Asian/ Self-Sacrificing Self-Enhancement) and Vulnerability (i.e.,
Asian American, Black/African-American, and Hispanic/ Contingent Self-Esteem, Hiding the Self, Devaluing, and
Latinx, respectively; the small remaining percentage endorsed Entitlement Rage; Pincus et al. 2009; Wright et al. 2010).
other ethnicities or chose not to provide this information). Here, we focus on these broad Grandiosity and Vulnerability
dimensions in our analyses, rather than scores for all individual
Measures PNI subscales, due to reporting a large number of relations for
emergent NPI factors. However, we did not include scores from
Narcissistic Personality Inventory (NPI) As discussed, partici- the PNI Exploitativeness subscale in our analyses when creating
pants completed the 40-item Narcissistic Personality a PNI Grandiosity dimension because this subscale is defined by
Inventory (NPI; Raskin and Terry 1988) with a single- items included within the NPI (e.g., both measures include the
stimulus administration format. Again, participants rated same item about being capable of “reading others like a book”).
themselves on each item using a 5-point response scale rang- Participants rated themselves on the PNI items using a 0 (not at
ing from 1 (strongly disagree) to 5 (strongly agree). all like me) to 5 (very much like me) rating scale.

120-Item International Personality Item Pool-NEO (IPIP-120) Personality Inventory for DSM-5 (PID-5) The PID-5 (Krueger
The IPIP-120 (Maples et al. 2014a) is a five-factor model et al. 2012) is a comprehensive measure used to assess the trait
personality inventory used to measure the domains of neurot- domains and facets described in Section III (Emerging Models
icism, extraversion, openness, agreeableness, and conscien- and Measures) of the DSM–5. Scores on the PID-5 shows
tiousness. This measure also includes six lower order facet strong convergence with clinician ratings of personality pa-
scales within each domain (e.g., facet scales for neuroticism thology, and dimensions assessed via this measure also show
assessing Anxiety, Anger, Depression, Self-Consciousness, incremental predictive validity above and beyond DSM-IV/
Immoderation, and Vulnerability). Participants were asked to DSM-5 conceptualizations of personality disorders (Few
rate themselves on each item using a 5-point scale ranging et al. 2013; Krueger et al. 2012; Wright et al. 2012).
from 1 (very inaccurate) to 5 (very accurate). In the current Participants completed only the PID-5 Emotional Lability
study, we assessed only the four domains of neuroticism, ex- scale (seven items; e.g., “my emotions are unpredictable”)
traversion, agreeableness, and conscientiousness. for this study, responding to the items using a 4-point scale
ranging from 0 (very false or often false) to 3 (very true or
Externalizing Spectrum Inventory-Brief Form (ESI-BF) often true).
Participants completed both the Drug Use (six items; e.g.,
“taken an illegal drug that gave me a rush”) and Alcohol Short Dark Triad (SD3) The Short Dark Triad (SD3; Jones and
Use (nine items; “like having a drink of alcohol to relax”) Paulhus 2014) assesses Machiavellianism (e.g., “not wise to
scales from the ESI-BF (Patrick et al. 2013). Participants share secrets”), psychopathy (e.g., “like to get revenge”), and
responded to these items using a 4-point scale ranging from narcissism (e.g., “know that I am special”) using 9-item scales
0 (false) to 3 (true). for each construct. No time frame is specified for these items,
as they are intended to provide general self-reported ratings of
Expanded Version of the Inventory for Depression and affect, behavior, and cognition. Previous research has found
Anxiety Symptoms (IDAS-II) Participants completed the good internal consistency (α > .78) for all SD3 scales, as well
Euphoria (five items; e.g. “elated for no reason”) and Mania as good convergent validity for all individual scales (e.g.,
(five items; “thoughts raced”) scales from the IDAS-II Jones and Paulhus 2014; Maples et al. 2014b). Participants
(Watson et al. 2012), a widely used measure assessing symp- were asked to rate each item using a 5-point Likert scale rang-
tom dimensions spanning the depressive, anxiety, bipolar ing from 1 (disagree strongly) to 5 (agree strongly).
J Psychopathol Behav Assess

Overview of Study Analyses Results

Prior to conducting other study analyses, we computed de- Descriptive Statistics


scriptive statistics and internal consistency estimates for per-
sonality and psychopathology measures. We also calculated Tables 1 and 2 present descriptive statistics and internal con-
independent samples t-tests to determine whether mean scale sistency estimates for personality and psychopathology mea-
and subscale scores differed across samples for various study sures, respectively. Undergraduates tended to score signifi-
measures. cantly higher on indices of grandiose narcissism (e.g., SD3
To address our primary study goal, we conducted explor- Narcissism and PNI Grandiosity), extraversion, agreeableness
atory factor analyses (EFAs) of the NPI items in both study and conscientiousness. Conversely, the community adults
samples. To help guide us in determining the number of fac- generally had significantly higher mean scores on indicators
tors to extract in our datasets, we conducted both parallel of vulnerable narcissism, psychopathy, Machiavellianism,
analyses (O’Connor 2000) and Velicer’s (1976) minimum av- drug use, and indices of emotional instability (e.g., neuroti-
erage partial (MAP) test using principal components analyses cism, PID-5 Emotional Lability).
of the 40 NPI items in both datasets. In parallel analysis,
observed eigenvalues from a principal components analysis Factor Analyses of the Narcissistic Personality
in the actual dataset are compared to the eigenvalues from Inventory
random datasets with the same sample size and number of
variables. For Velicer’s MAP test, mean squared partial corre- Determining the Number of Factors Parallel analyses con-
lations are calculated for a range of factor solutions consisting ducted in both the community (Ncases = 700, Nvars = 40,
of an increasing number of factors, with the optimal solution Ndatasets = 1000, percent = 95) and undergraduate datasets
producing the lowest mean value. (Ncases = 527, Nvars = 40, Ndatasets = 1000, percent = 95) indi-
Both parallel analyses and the MAP test were conducted cated that across both samples up to six factors could be ex-
with SAS 9.4 software. We used principal components analy- tracted. Specifically, the eigenvalues for the sixth principal
ses for these procedures because the use of principal factor component were larger than their random counterparts in both
analyses tends to overestimate the number of factors to be samples (Community eigenvalues: 1.44 vs. 1.34;
extracted (Timmerman and Lorenzo-Seva 2011), even though Undergraduate eigenvalues: 1.41 vs. 1.29), but the eigen-
we report results from principal factor analyses following the values for the seventh were not (Community eigenvalues:
presentation of these initial analyses. Also note that although 1.10 vs. 1.27; Undergraduate eigenvalues: 1.307 vs. 1.308).
we considered the information provided by both procedures in The MAP test in the community sample indicated that the
selecting the number of factors to extract in each dataset, our mean squared partial correlation was lowest in the seven-
primary goal when conducting these EFAs was to extract the factor solution (.0128). However, the MAP test results in the
largest number of interpretable and sufficiently defined (i.e., undergraduate sample indicated that the lowest mean squared
each factor defined by several or more items) factors in each partial correlation was in the five-factor solution (.0128).
dataset.
After conducting EFAs in both datasets, we then conducted Description of the Factors Given that the parallel analysis and
analyses of factor similarity in both the community and un- MAP test results differed in some ways (e.g., parallel analyses
dergraduate samples (Finn 1986). Comparability coefficients suggested that as many as six factors could be extracted in the
are derived by (a) generating regression-based scoring weights community dataset, but the MAP test suggested a seven-factor
for each factor in each solution, and (b) applying these weights solution was optimal), we considered a range of different fac-
to participants’ actual scores in each dataset (Everett and tor solutions in each sample, focusing on identifying the larg-
Entrekin 1980; Finn 1986; Gorsuch 1983). Comparability co- est number of interpretable dimensions in each sample. Based
efficients ≥ |.90| provide strong evidence that the same factor on our parallel analyses, the MAP test analyses, and prior
emerged across different solutions (Everett 1983). research (Ackerman et al. 2016), we anticipated that we would
We then created subscales to model emergent factors from be able to identify at most six or seven factors defining the NPI
the EFAs and examined their correlations with one another items in each dataset (e.g., because parallel analysis suggested
and with other personality and psychopathology measures. that at most, six meaningful factors could be extracted in each
All correlations reported are Pearson correlations, and we con- dataset). All EFAs were conducted using principal factor anal-
sider the statistical significance of correlations at a p < .001 ysis, rotating factors using an oblique promax rotation (pow-
level due to examining and reporting many correlations. We er = 3). With this approach, we found that a five-factor solu-
also focus on general patterns of correlations within and across tion yielded the largest number of well-defined, meaningful
subscales rather than focusing on the significance of individ- dimensions in both samples. Solutions with four or fewer
ual correlations. factors also yielded interpretable sets of dimensions.
J Psychopathol Behav Assess

Table 1 Descriptive statistics and


internal consistency estimates for Measure Undergraduates (N = 527) Community (N = 700)
personality domain and facet
measures α Mean SD α Mean SD

IPIP Neuroticism
Domain .91 67.19 15.25 .92 81.49b 17.63
Anxiety .77 13.27 3.58 .84 15.39b 3.82
Anger .86 10.50 3.66 .90 12.79b 4.28
Depression .86 10.14 3.91 .90 14.00b 4.42
Self-Consciousness .72 11.59 3.22 .75 13.24b 3.74
Immoderation .80 11.46 3.51 .82 13.49b 3.99
Vulnerability .78 10.21 3.24 .81 12.57b 3.89
IPIP Extraversion
Domain .91 83.48a 14.15 .92 64.95 16.81
Friendliness .87 14.32a 3.47 .87 11.05 4.16
Gregariousness .87 11.78a 4.24 .84 8.38 3.93
Assertiveness .82 13.49a 3.12 .87 10.73 3.95
Activity Level .71 15.76a 2.80 .75 11.49 3.62
Excitement Seeking .70 14.50a 3.40 .81 10.75 3.74
Cheerfulness .77 15.24a 2.86 .75 12.55 3.45
IPIP Agreeableness
Domain .86 91.06a 11.15 .86 88.06 13.13
Trust .89 14.50a 3.40 .90 11.95 3.98
Morality .71 15.24 2.86 .76 15.16 3.38
Altruism .73 16.98a 2.17 .71 15.74 2.91
Cooperation .73 16.21a 2.97 .75 15.65 3.53
Modesty .72 12.35 3.13 .78 14.50b 3.64
Sympathy .67 15.79a 2.60 .77 15.06 3.43
IPIP Conscientious
Domain .90 88.89a 12.73 .90 80.57 15.10
Self-Efficacy .76 16.23a 2.09 .81 14.26 3.21
Orderliness .82 13.60a 3.89 .78 12.41 3.91
Dutifulness .69 16.12a 2.31 .72 15.42 2.93
Achievement Striving .66 16.96a 2.17 .74 14.44 3.34
Self-Discipline .87 11.75a 3.90 .88 10.68 4.26
Cautiousness .89 14.24a 3.83 .91 13.35 4.45
a
= indicates that the undergraduate sample mean is significantly greater than the community sample mean at
p < .001, b = indicates that the community sample mean is significantly greater than the undergraduate sample
mean at p < .001. IPIP = International Personality Item Pool

However, solutions with six or more factors yielded one or which we labeled Superiority. Next, a factor consisting of item
more factors that were very narrow (i.e., fewer than three item content assessing being able to influence and deceive others
markers) and difficult to interpret. For example, when we emerged in both samples (i.e., Factor II in the community
extracted six factors in the community and undergraduate sample; Factor III in the undergraduate sample) and was there-
datasets, one factor had only one clear item marker (i.e., factor fore labelled Manipulativeness. Factor III in the community
loading ≥ .40 and cross-loadings ≤ .30). Therefore, we focused sample and Factor V in the undergraduate sample represented
on the five-factor structure in each sample. a somewhat narrow dimension that was defined strongly by
Tables 3 and 4 present the full set of factor loadings for the items assessing liking one’s own physical appearance; it was
five-factor structure in the community and undergraduate labeled Vanity. Items assessing having an assertive and social-
samples, respectively. In both samples a factor (i.e., Factor I ly dominant personality style loaded strongly on Factor IV in
in the community data; Factor IV in the undergraduate data) the community sample and Factor I in the undergraduate sam-
defined by items reflecting a superior sense of self emerged, ple; we labelled this dimension Leadership/Authority. Finally,
J Psychopathol Behav Assess

Table 2 Descriptive statistics and


internal consistency estimates for Measure Undergraduates (N = 527) Community (N = 700)
psychopathology measures
α Mean SD α Mean SD

NPI
Manipulativeness .77 14.09 3.58 .84 13.90 4.55
Grandiose Fantasies .45 7.68 2.13 .59 7.38 3.43
Vanity .83 8.06a 2.85 .87 6.40 3.15
Leadership/Authority .80 22.52a 4.56 .87 18.38 6.37
Superiority .80 14.53a 2.51 .87 12.11 3.81
SD3
Narcissism .65 26.12a 4.58 .77 22.14 5.90
Psychopathy .74 17.62 4.74 .78 19.68b 6.10
Machiavellianism .79 25.58 5.50 .82 27.11b 6.05
PNI
Grandiosity .88 3.12a .83 .87 2.94 .94
Vulnerability .95 2.34 .84 .94 2.56b .88
ESI-BF
Alcohol .93 15.70 8.71 .90 16.26 8.17
Drug Use .77 3.21 4.09 .86 8.16b 6.03
PID-5
Emotional Lability .91 7.08 5.03 .94 10.73b 5.95
IDAS-II
Mania .86 12.21 4.94 .90 11.58 5.65
Euphoria .83 10.93a 4.49 .87 9.50 4.69
a
= indicates that the undergraduate sample mean is significantly greater than the community sample mean at
p < .001, b = indicates that the community sample mean is significantly greater than the undergraduate sample
mean at p < .001. NPI = Narcissistic Personality Inventory, SD3 = Short Dark Triad, PNI = Pathological
Narcissism Inventory, ESI-BF = Externalizing Spectrum Inventory-Brief Form, PID-5 = Personality Inventory
for DSM-5, IDAS-II = Expanded Version of the Inventory for Depression and Anxiety Symptoms

a factor strongly defined by items assessing grandiose to define each respective factor, and scores on these subscales
thoughts and wishes emerged across samples (Factor V in were used for our correlational analyses. In determining which
the community sample; Factor II in the undergraduate data), items were good indicators of latent factors, we first identified
which we called Grandiose Fantasies. those items with (a) loadings on their primary factor ≥ |.40| in
both study samples and (b) cross loadings ≤ |.30| on other factors
Factor Structure Similarity We were able to compute ten factor across samples. Note that in some cases, we relaxed these selec-
scores in each dataset to model our factors (i.e., five scores for tion criteria slightly (e.g., item 4 in the Grandiose Fantasies
the community sample factors and five for the undergraduate subscale met both selection criteria in the undergraduate sample,
sample factors) and to conduct two sets of factor similarity tests but only met one selection criterion in the community sample).
(i.e., examining similarity with the undergraduate dataset used Returning briefly to Table 2, undergraduates had significantly
for scoring and a second examination using the community higher scores on the NPI Leadership/Authority, Superiority, and
dataset used for scoring). All factors from the five-factor solu- Vanity subscales compared to the community adults (ps < .001
tion were highly comparable with their respective factors (e.g., for all comparisons). There were no significant differences be-
the same basic Superiority factor emerged in both datasets) tween samples on the NPI Manipulativeness and Grandiose
regardless of which dataset was used for scoring, as all coeffi- Fantasies subscales. See Tables 3 and 4 for which items were
cients were ≥ .90 and eight of 10 coefficients were ≥ .95. used to score each subscale. In Table 5, we report Pearson cor-
relations for subscales modeling the five factors in both the
Subscale Creation and Associations community and undergraduate samples. Note that correlations
all were positive and most of them were small to moderate in
Subscales modeling the dimensions from the five-factor NPI strength (rs ranged from .31 to .60 in the community sample and
structure were created by summing scores on items specified .26 to .46 in the student sample).
J Psychopathol Behav Assess

Table 3 Five-factor structure promax-rotated item loadings in community participants

Item Content Factor I Factor II Factor III Factor IV Factor V

34. I am going to be a great person 0.76a −0.10 0.15 0.07 −0.02


8. I will be a success 0.72a −0.12 0.14 0.15 −0.06
9. I think I am a special person 0.72a −0.12 0.16 −0.04 0.23
40. I am an extraordinary person 0.67a −0.01 0.15 0.02 0.25
31. I can live my life in any way I want to 0.50 0.09 −0.02 −0.02 0.09
39. I am more capable than other people 0.48 0.19 0.01 0.00 0.26
21. I always know what I am doing 0.35 0.11 −0.06 −0.02 0.07
14. I insist on getting the respect that is due me 0.30 0.17 −0.04 0.13 0.28
27. I have a strong will to power 0.29 0.15 −0.01 0.29 0.24
22. I rarely depend on anyone else to get things done 0.26 0.17 −0.16 0.03 0.03
18. I want to amount to something 0.25 −0.08 −0.11 0.17 0.25
13. I find it easy to manipulate people −0.08 0.81b 0.04 −0.07 0.16
6. I can usually talk my way out of anything −0.06 0.79b 0.04 0.02 0.04
35. I can make anybody believe anything I want 0.03 0.71b 0.03 −0.12 0.30
1. I have a natural talent for influencing people 0.14 0.59b 0.09 0.23 −0.04
16. I can read people like a book 0.07 0.55b 0.04 0.11 −0.18
32. People always seem to recognize my authority 0.27 0.41 0.01 0.32 0.10
23. Everybody likes to hear my stories 0.21 0.30 0.08 0.18 0.05
19. I like to look at my body 0.13 0.08 0.87c −0.04 −0.10
15. I like to display my body −0.03 0.00 0.82c 0.03 0.08
29. I like to look at myself in the mirror 0.20 0.12 0.82c −0.09 −0.09
20. I am apt to show off if I get the chance −0.07 0.07 0.41 0.23 0.31
28. I like to start new fads and fashions 0.03 0.06 0.33 0.15 0.29
7. I like to be the center of attention −0.25 −0.10 0.27 0.73d 0.26
33. I would prefer to be a leader 0.22 0.06 −0.13 0.66d 0.15
30. I really like to be the center of attention −0.20 −0.07 0.33 0.64 0.31
10. I see myself as a good leader 0.40 0.09 −0.03 0.63d −0.11
11. I am assertive 0.27 0.08 −0.02 0.52d −0.09
12. I like having authority over people 0.09 0.11 −0.17 0.52d 0.34
36. I am born a leader 0.36 0.25 −0.04 0.51d −0.01
17. I like to take responsibility for making decisions 0.28 0.11 0.01 0.46d −0.20
26. I like to be complimented 0.18 −0.25 0.21 0.26 0.23
25. I will never be satisfied until I get all that I deserve 0.25 0.07 −0.05 −0.08 0.64e
37. I wish somebody would write my biography 0.20 −0.10 0.09 0.13 0.52e
38. I get upset when people don’t notice how I look −0.05 −0.04 0.46 −0.08 0.50
5. If I ruled the world it would be a much better place 0.29 0.12 −0.07 −0.02 0.38
4. I know that I am good because people tell me 0.29 0.14 0.19 −0.10 0.35e
2. Modesty doesn’t become me −0.05 0.18 0.16 0.05 0.34
24. I expect a great deal from other people 0.14 0.14 −0.08 0.11 0.34
3. I would do almost anything on a dare −0.17 0.25 0.19 0.08 0.32

N = 700. Some items are paraphrased versions of the original NPI items. Loadings ≥ |.40| are bolded. a = item was used to score the Superiority subscale,
b
= item used to score the Manipulativeness subscale, c = item used to score the Vanity subscale, d = item used to score the Leadership/Authority
subscale, e = item used to score the Grandiose Fantasies subscale

NPI Subscale Associations with Personality Domains samples, as shown in Tables 6 and 7, respectively. Note that we
and Facets also examined whether correlations for any specific NPI subscale
with other variables were significantly different in magnitude
Overall, patterns of correlations for the NPI subscales with per- than correlations for all other NPI subscales (e.g., examining
sonality were similar across the community and undergraduate whether any NPI subscale associated significantly more strongly
J Psychopathol Behav Assess

Table 4 Five-factor structure


promax-rotated item loadings in Item Content Factor I Factor II Factor III Factor IV Factor V
undergraduates
10. I see myself as a good leader 0.80a −0.17 −0.06 0.31 −0.06
33. I would prefer to be a leader 0.76a 0.01 0.01 0.16 −0.11
11. I am assertive 0.67a −0.19 0.09 0.03 0.08
36. I am born a leader 0.65a −0.04 0.20 0.16 −0.09
12. I like having authority over people 0.53a 0.27 0.14 −0.05 −0.08
7. I like to be the center of attention 0.53a 0.38 −0.05 −0.24 0.13
32. People always seem to recognize my authority 0.48 −0.04 0.42 0.10 −0.04
17. I like to take responsibility 0.35a −0.20 0.07 0.26 0.17
37. I wish somebody would write my biography −0.03 0.63b 0.11 0.05 −0.18
38. I get upset when people don’t notice me −0.24 0.55 0.24 −0.08 0.11
25. I will never be satisfied −0.15 0.54b 0.13 0.15 0.05
18. I want to amount to something −0.09 0.49 −0.01 0.18 −0.08
30. I really like to be the center of attention 0.46 0.49 −0.07 −0.29 0.12
26. I like to be complimented 0.03 0.49 −0.21 0.11 0.14
20. I am apt to show off if I get the chance 0.05 0.47 0.06 −0.06 0.30
4. I know that I am good because people tell me 0.00 0.41b 0.20 0.15 −0.05
5. If I ruled the world it would be a better place 0.04 0.34 0.23 0.27 −0.19
24. I expect a great deal from other people −0.02 0.33 −0.04 0.10 −0.04
23. Everybody likes to hear my stories 0.07 0.28 0.17 0.15 0.07
35. I can make anybody believe anything I want 0.00 0.18 0.72c 0.04 0.03
13. I find it easy to manipulate people 0.13 0.11 0.67c −0.07 0.04
6. I can usually talk my way out of anything 0.15 0.06 0.59c 0.05 −0.05
1. I have a natural talent for influencing people 0.39 −0.13 0.53c −0.01 0.06
16. I can read people like a book 0.03 0.08 0.45c 0.04 0.07
22. I rarely depend on anyone else −0.14 −0.14 0.36 0.36 −0.03
2. Modesty doesn’t become me 0.00 0.15 0.26 −0.06 0.23
3. I would do almost anything on a dare −0.04 0.23 0.26 −0.21 0.08
34. I am going to be a great person 0.14 0.18 −0.09 0.70d 0.01
8. I will be a success 0.15 0.11 −0.13 0.66d 0.05
9. I think I am a special person 0.21 0.25 −0.20 0.62d 0.08
40. I am an extraordinary person 0.05 0.28 0.07 0.59d 0.14
39. I am more capable than other people −0.14 0.35 0.26 0.39 −0.01
21. I always know what I am doing 0.00 −0.16 0.30 0.35 0.01
31. I can live my life in any way I want to 0.09 −0.02 0.16 0.29 0.12
27. I have a strong will to power 0.20 0.22 0.21 0.22 0.08
14. I insist on getting the respect that is due me 0.12 0.19 0.14 0.21 0.09
15. I like to display my body −0.03 −0.05 0.11 −0.02 0.81e
19. I like to look at my body −0.03 −0.11 0.01 0.18 0.90e
29. I like to look at myself in the mirror −0.03 0.08 −0.04 0.17 0.75e
28. I like to start new fads and fashions 0.19 0.18 0.10 −0.12 0.20

N = 527. Some items are paraphrased versions of the original NPI items. Loadings ≥ |.40| are bolded. a = item was
used to score the Leadership/Authority subscale, b = item used to score the Grandiose Fantasies subscale, c = item
used to score the Manipulativeness subscale, d = item used to score the Superiority subscale, e = item used to
score the Vanity subscale

with extraversion domain scores than all other subscales in the In some ways, the NPI subscales showed similarities in
same dataset) for these and all subsequent correlational analyses their personality relations. For example, all NPI subscales
presented. These comparisons were conducted using the tended to associate positively with extraversion domain and
Williams modification to the Hotelling test for dependent corre- acet scores across samples; similarly, all NPI subscales showed
lations with a common index (Weaver and Wuensch 2013). robust negative associations with the Modesty facet of
J Psychopathol Behav Assess

Table 5 Correlations between the


narcissistic personality inventory Manipulativeness Grandiose Vanity Leadership/ Superiority
subscales Fantasies Authority

Manipulativeness – .47 .31 .56 .42


Grandiose Fantasies .37 – .43 .52 .55
Vanity .24 .26 – .40 .44
Leadership/Authority .44 .34 .32 – .60
Superiority .26 .33 .36 .46 –

N = 700 in the community sample, and N = 527 in the undergraduate sample. All correlations are significant at
p < .001. Correlations for the community sample are presented above diagonal, and correlations for the under-
graduate sample are presented below the diagonal

agreeableness. However, there were also important differences and Superiority. Furthermore, NPI Grandiose Fantasies was
across the NPI subscales in terms of their correlational patterns. moderately associated with PNI Grandiosity and Vulnerability
For example, NPI subscales such as Manipulativeness and in both samples, even though all other subscales correlated ≤
Grandiose Fantasies displayed stronger negative relations with |.30| with these PNI measures across samples. Overall, sub-
scores on the agreeableness domain (rs for both subscales ≥ scales such as NPI Manipulativeness and Grandiose Fantasies
|.35| across samples) compared to NPI Superiority (r < |.25| also displayed a larger number of robust correlations with
across samples). Additionally, compared to other subscales other psychopathology indicators compared to some other
such as NPI Leadership/Authority and Manipulativeness, subscales. Specifically, NPI Manipulativeness and Grandiose
which had many strong personality correlates, NPI Vanity Fantasies had seven and 11 total correlations with other psy-
showed few robust personality associations. This was true in chopathology ≥ |.30| across samples, respectively; those cor-
the undergraduate sample especially, where in fact, NPI Vanity relational patterns represent an interesting juxtaposition to
had only one personality correlate ≥ |.30| in magnitude (i.e., r = NPI Superiority, which had only three total psychopathology
−.43 with the Modesty facet scale from agreeableness). correlates ≥ |.30| across samples.
Other specific correlation comparisons further under-
score these differences. For example, extraversion domain
scores were correlated significantly more strongly with Discussion
NPI Leadership/Authority in both the community
(r = .74) and undergraduate sample (r = .62) than with The primary goals of this study were to (a) examine the item-
any other NPI subscale in both samples (rs ranged from level structure of the NPI and (b) determine the degree to
.15 to .59). Additionally, the Assertiveness extraversion which subscales modeling NPI factors showed divergent pat-
facet was strongly correlated with NPI Leadership/ terns of personality and psychopathology relations. These
Authority in both the community (r = .81) and the under- aims were intended to extend the limited prior work focusing
graduate sample (r = .73); again, this correlation with NPI on the NPI’s psychometric properties when it is administered
Leadership/Authority was significantly stronger than the with a single-stimulus and Likert style response format. As we
correlation for the Assertiveness facet scale with any other discuss in detail next, our results indicate that a five-factor NPI
NPI subscale across samples (rs ranged from .17 to .48). structure replicated across our datasets. Furthermore, both our
structural and correlational results highlight significant hetero-
NPI Subscale Psychopathology Associations geneity in the NPI’s content. Thus, in discussing the implica-
tions of our key findings, we give special emphasis to the
We discuss the correlational results for both the community and importance of recognizing this item-level heterogeneity when
undergraduate samples together because results generally were using the NPI in future research.
similar across samples, as shown in Tables 8 and 9, respectively.
Once again, there were some common trends observed for the The Structure of an Alternate Version
five NPI subscales, as they all correlated moderately to strongly of the Narcissistic Personality Inventory
with SD3 Narcissism across samples (all rs > .40).
Once more, however, the NPI subscales displayed distinc- Our results indicated that five replicable NPI factors (i.e.,
tive relations from one another in at least some ways. For Manipulativeness, Grandiose Fantasies, Vanity, Leadership/
example, NPI Manipulativeness tended to display much stron- Authority, Superiority) could be identified across samples.
ger relations with SD3 Machiavellianism and Psychopathy These findings are consistent with other research indicating
scores than did subscales such as NPI Leadership/Authority that the NPI contains a heterogeneous range of item content
J Psychopathol Behav Assess

Table 6 Correlations for NPI


subscale scores with personality NPI Subscale
domains and facets in the
community sample Manipulativeness Grandiose Vanity Leadership/ Superiority
Fantasies Authority

Neuroticism
Domain −.19 −.09 −.27 −.42 −.43
Self-Consciousness −.34 −.24 −.25 −.60* −.41
Depression −.17 −.14 −.30 −.38 −.49*
Vulnerability −.29 −.11 −.22 −.44 −.41
Immoderation .03 .07 −.19 −.11 −.34*
Anxiety −.22 −.15 −.21 −.39 −.17
Anger .10 .14 −.01 .02 −.10
Extraversion
Domain .47 .45 .45 .74* .59
Assertiveness .45 .34 .27 .81* .48
Excitement Seeking .38 .48 .42 .47 .34
Friendliness .34 .29 .32 .51 .43
Cheerfulness .25 .24 .36 .40 .39
Activity Level .32 .31 .26 .46 .45
Gregariousness .30 .32 .38 .54 .53
Agreeableness
Domain −.41 −.45 −.29 −.31 −.22
Modesty −.45 −.54 −.50 −.63 −.65
Morality −.62* −.47 −.33 −.38 −.28
Cooperation −.37 −.40 −.25 −.32 −.16
Trust .06 .02 −.11* .13 .17
Altruism −.04 −.10 −.02 .14 −.07
Sympathy −.08 −.21 .10 −.09 .20
Conscientiousness
Domain .06 −.06* .08 .23 .29
Dutifulness −.13 −.22 −.09 .04 .48
Cautiousness −.19 −.27 −.09 .44 .43
Self-Efficacy .35 .17 .22 .45 .01*
Achievement .22 .16 .14 .41* .04
Striving
Self-Discipline .09 .00 .09 .23 .22
Orderliness −.03 −.04 .08 .02 .09

N = 700. Correlations ≥ |.30| are bolded. * = Correlation is significantly higher/lower than any other correlation in
the same row at p < .001. IPIP = International Personality Item Pool, NPI = Narcissistic Personality Inventory

that is not captured using total scores (Ackerman et al. 2016; Vanity, and Superiority dimensions. Furthermore, six of the
Foster et al. 2016; Stanton et al. 2017). Our results partially seven items (i.e., items 10, 11, 12, 17, 33, and 36) used to
replicated and extended previous research by Ackerman et al. score the Leadership/Authority dimension that emerged in
(2016) when administering the NPI using a single-stimulus our results also were clear indicators of Ackerman et al.’s
response format and Likert style response scale. Specifically, (2016) Leadership factor. It is interesting that these results
Table 10 presents a comparison of the composition of NPI replicated quite well across samples, especially given that
subscales based on the results in our samples with subscales these samples differed considerably regarding demographic
that could be scored based on findings presented by Ackerman variables such as treatment status. The largest departure be-
et al. (2016). As Table 10 shows, both our results and those tween our findings and Ackerman et al.’ (2016) was that a
presented by Ackerman et al. (2016) indicated that the same Grandiose Fantasies factor emerged in our datasets, whereas
set of NPI items can be used to model Manipulativeness, their analyses found evidence for a distinct Exhibitionism
J Psychopathol Behav Assess

Table 7 Correlations for NPI


subscale scores with personality NPI Subscale
domains and facets in the
undergraduate samples Manipulativeness Grandiose Vanity Leadership/ Superiority
Fantasies Authority

Neuroticism
Domain −.07 .12* −.15 −.30 −.37
Self-Consciousness −.24 −.04 −.20 −.53* −.36
Depression −.05 .05 −.22 −.27 −.42
Vulnerability −.17 .09* −.10 −.27 −.32
Immoderation .01 .11 −.10 −.09 −.19
Anxiety −.09 .07* −.11 −.18 −.21
Anger .18 .23 .07 .02 −.09
Extraversion
Domain .25 .15 .27 .62* .36
Assertiveness .29 .19 .17 .73* .36
Excitement Seeking .30 .21 .28 .41* .25
Friendliness .19 .08 .25 .47* .30
Cheerfulness .14 .05 .24 .35 .36
Activity Level .04 .03 .00 .29 .20
Gregariousness .15 .11 .23 .43 .13
Agreeableness
Domain −.48 −.38 −.25 −.28 −.16
Modesty −.35 −.37 −.43 −.52 −.58
Morality −.57* −.37 −.26 −.29 −.15
Cooperation −.39 −.32 −.15 −.21 −.03
Trust −.24 −.15 .04 −.01 .13
Altruism −.13 −.13 −.11 .05 .10
Sympathy −.13 −.13 −.08 −.05 −.06
Conscientiousness
Domain −.16 −.18 −.06 .07 .20
Dutifulness −.29 −.25 −.17 −.09 .10*
Cautiousness −.22 −.23 −.16 −.17 −.03
Self-Efficacy .10 −.02 .10 .26 .41*
Achievement −.06 −.01 .00 .18 .27
Striving
Self-Discipline −.07 −.09 .00 .14 .15
Orderliness −.09 −.09 .01 .05 .09

N = 527. Correlations ≥ |.30| are bolded. * = Correlation is significantly higher/lower than any other correlation in
the same row at p < .001. IPIP = International Personality Item Pool, NPI = Narcissistic Personality Inventory

factor. The Grandiose Fantasises factor in our results and the consistent with key trait profiles of grandiose narcissism iden-
Exhibitionism factor identified by Ackerman et al. (2016) tified in some prior research (e.g., Paulhus 2001).
were defined strongly by only three and four items each, re- Additionally, the various NPI subscales showed robust asso-
spectively, none of which were overlapping (again, see ciations with SD3 Narcissism, but comparatively weaker as-
Table 10). sociations with PNI Grandiosity and Vulnerability, except in
the case of NPI Grandiose Fantasies, which showed moderate
positive relations across samples. This finding makes sense
NPI Subscale Correlates: Further Evidence
given that previous research has found the PNI to have diver-
for Distinctive NPI Facets
gent patterns of relations compared to the NPI in some cases
(Pincus et al. 2009). Relatedly, early work conceptualizing the
At a broad level, the NPI item content appears to assess a
Dark Triad primarily used the NPI as a proxy for narcissistic
“disagreeable and extraverted” personality style, which is
J Psychopathol Behav Assess

Table 8 Correlations for NPI


subscales with psychopathology Criterion Measure NPI Subscale
measures in the community
sample Manipulativeness Grandiose Vanity Leadership/ Superiority
Fantasies Authority

Narcissistic Traits
SD3 Narcissism .44 .46 .46 .68 .58
PNI Grandiose .22 .41* .14 .22 .18
PNI Vulnerability .24 .35 .05 .04 −.06
Externalizing Traits
SD3 Machiavellianism .49 .39 .17 .21 .13
SD3 Psychopathy .43 .36 .25 .23 .07*
ESI-BF Alcohol Use .14 .06 .22 .12 .01
ESI-BF Drug Use .15 .02 .13 .04 −.01
Mood Symptoms
IDAS-II Euphoria .29 .26 .27 .34 .28
IDAS-II Mania .23 .23 .11 .20 .17
PID-5 Emotional .06 .20* .01 −.03 −.09
Lability

N = 700. Correlations ≥ |.30| are bolded. ESI-BF = Externalizing Spectrum Inventory-Brief Form, IDAS-II =
Expanded Version of the Inventory of Depression and Anxiety Symptoms, PID-5 = Personality Inventory for
DSM-5, PNI = Pathological Narcissism Inventory, SD3 = Short Dark Triad

content (Paulhus and Williams 2002), and previous work has assessing primarily grandiosity and to a lesser degree entitle-
shown SD3 Narcissism to be strongly correlated to the NPI ment, with limited coverage of vulnerable narcissistic traits
total score (Jones and Paulhus 2014). These findings also (Krizan and Herlache 2018).
align with prior research from the Narcissistic Spectrum However, coupled with our factor analytic results, our cor-
Model perspective, wherein the NPI is described as a measure relational analyses suggest that limiting analyses to total

Table 9 Correlations for NPI


subscales with psychopathology Criterion Measure NPI Subscale
measures in the undergraduate
sample Manipulativeness Grandiose Vanity Leadership/ Superiority
Fantasies Authority

Narcissistic Traits
SD3 Narcissism .58 .65 .48 .77 .65
PNI Grandiosity .26 .44* .17 .29 .27
PNI Vulnerability .18 .37* .11 .09 .03
Other Externalizing Traits
SD3 Machiavellianism .49 .46 .19 .26 .17
SD3 Psychopathy .46 .48 .30 .31 .16
ESI-BF Drug Use .26 .16 .12 .12 .10
ESI-BF Alcohol Use .22 .18 .11 .12 .10
Other Psychopathology
IDAS-II Euphoria .34 .40 .34 .37 .33
IDAS-II Mania .26 .27 .16 .23 .17
PID-5 Emotional .01 .11 .00 −.10 −.12
Lability

N = 527. Correlations ≥ |.30| are bolded. ESI-BF = Externalizing Spectrum Inventory-Brief Form, IDAS-II =
Expanded Version of the Inventory of Depression and Anxiety Symptoms, PID-5 = Personality Inventory for
DSM-5, PNI = Pathological Narcissism Inventory, SD3 = Short Dark Triad
J Psychopathol Behav Assess

Table 10 Subscale item


compositions of the current study Current Study Ackerman et al. (2016)
compared to those identified in
Ackerman et al. (2016) Manipulativeness Manipulativeness
1. I have a natural talent for influencing people 1. I have a natural talent for influencing people
6. I can usually talk my way out of anything 6. I can usually talk my way out of anything
13. I find it easy to manipulate people 13. I find it easy to manipulate people
16. I can read people like a book 16. I can read people like a book
35. I can make anybody believe anything I want 35. I can make anybody believe anything I want
Superiority Superiority
8. I will be a success 8. I will be a success
9. I think I am a special person 9. I think I am a special person
34. I am going to be a great person 34. I am going to be a great person
40. I am an extraordinary person 40. I am an extraordinary person
Vanity Vanity
15. I like to display my body 15. I like to display my body
19. I like to look at my body 19. I like to look at my body
29. I like to look at myself in the mirror 29. I like to look at myself in the mirror
Leadership/Authority Leadership
7. I like to be the center of attention ** 10. I see myself as a good leader
10. I see myself as a good leader 11. I am assertive
11. I am assertive 12. I like having authority over people
12. I like having authority over people 17. I like to take responsibility for making decisions
17. I like to take responsibility for making 32. People always seem to recognize my authority
decisions
33. I would prefer to be a leader 33. I would prefer to be a leader
36. I am born a leader 36. I am born a leader
Grandiose Fantasies Exhibitionism
4. I know that I am good because everyone keeps 7. I like to be the center of attention
telling me so **
25. I will never be satisfied until I get all that I 20. I am apt to show off if I get the chance
deserve **
37. I wish somebody would someday write my 30. I really like to be the center of attention
biography ** 38. I get upset when people don’t notice how I look when I
go out in public

** = items that are unique to the current study subscales. Items from Ackerman et al. (2016) were selected based
on structural analyses of the Narcissistic Personality Inventory items when administering this measure using a
single-stimulus response format

scores may be problematic. This is because different NPI sub- in many ways. Notably, Leadership/Authority demonstrated
scales modeling factors from our item-level analyses appear to several strong and interesting personality links. Individuals
show distinctive patterns of personality and psychopathology who endorsed NPI items assessing leadership content (e.g.,
relations that would be masked when focusing solely on NPI “see myself as a good leader”, “am born a leader”) also de-
total scores. Next, we review these patterns of correlates for scribed themselves as very extraverted and assertive.
each NPI subscale to highlight these differences. Following Furthermore, NPI Leadership/Authority was correlated with
that, we discuss the degree to which the NPI subscales appear (low) levels of neuroticism, showing an especially strong cor-
to assess content that is maladaptive in nature, which repre- relation with the Self-Consciousness facet of this domain
sents an important consideration given ongoing debate regard- (e.g., “am easily intimidated”). Additionally, Leadership/
ing the degree to which various NPI item content (e.g., lead- Authority was strongly associated with low levels of modesty.
ership, assertiveness) captures pathological narcissism. As was the case for NPI Leadership/Authority, higher
scorers on the NPI Superiority subscale also described them-
Leadership/Authority and Superiority We discuss the correla- selves as having low levels of neuroticism, being immodest,
tional profiles of NPI Leadership/Authority and Superiority and being extraverted; however, NPI Superiority did not show
together given that they showed similarities in their correlates as strong of an affinity with scores on the Assertiveness facet
J Psychopathol Behav Assess

scale from the extraversion domain as did NPI Leadership/ study found that undergraduate students had higher mean
Authority. Another replicable finding was that Superiority scores on these subscales compared to adults from the
scores were linked weakly to other psychopathology overall, clinically-oriented community sample. This is consistent with
as this subscale showed only a few noteworthy correlations some prior work using the NPI administered with a forced-
with other psychopathology across samples, two of which choice format (Stanton et al. 2017). However, at the same
were with SD3 Narcissism. time, scores on both subscales also converged strongly with
scores on some other measures of grandiose narcissistic traits
Vanity Next, compared to subscales such as NPI Leadership/ (e.g., SD3 Narcissism) and showed robust negative links to
Authority and Superiority, the Vanity subscale showed com- modesty. Thus, given that high scores on these subscales seem
paratively weaker associations with neuroticism and extraver- adaptive in some ways, but not others, we anticipate that there
sion and their respective facets. In fact, Vanity showed rela- will continue to be debate regarding the degree to which
tively few robust personality associations across samples (e.g., scores on these subscales are adaptive versus maladaptive.
it had only one personality correlate ≥ |.30| in the undergrad- Of course, many of these issues also are related to broader
uate sample). These relatively weak patterns of correlates may discussion in the field regarding the degree to which external-
be due, at least in part, to the Vanity subscale consisting of izing features present without distress are pathological (see
only three items assessing a narrow range of content focused Miller et al. 2017; Stanton and Zimmerman 2018).
on liking one’s appearance (e.g., “like to look at myself in the Regarding the degree to which other NPI subscales
mirror”); in contrast, other NPI subscales appear to capture emerged as adaptive versus maladaptive, scores on the Vanity
more general attitudes and patterns of behavior (e.g., seeing subscale appeared similar to those for Leadership/Authority
oneself as a success, as influential, etc.). and Superiority in many ways. Specifically, high scorers on
this subscale described themselves as immodest and extravert-
Manipulativeness and Grandiose Fantasies Manipulativeness ed to some degree, although Vanity was linked relatively weak-
and Grandiose Fantasies also are discussed together. Their ly to other personality traits. Next, Manipulativeness and
associations were similar to one another in many ways, and Grandiose Fantasies clearly appear to be maladaptive; as stat-
these two subscales appear to have markedly different person- ed, scores on these subscales were linked strongly and nega-
ality and psychopathology profiles compared to the other NPI tively to agreeableness indicators in many cases, as well as
subscales. Specifically, high scorers on these subscales de- robustly and positively to indicators of traits such as
scribed themselves as especially disagreeable, and relatedly, Machiavellianism and psychopathy. Additionally, scores on
reported high levels of dark traits as assessed by other exter- the Grandiose Fantasies subscale were moderately positively
nalizing measures such as the SD3. In particular, associated with vulnerability, and also did not show some other
Manipulativeness appears closely tied to being dishonest “adaptive” correlational trends that Leadership/Authority and
much more so than other NPI subscales such as Superiority, Superiority displayed (e.g., moderate to strong correlations
which is not surprising considering the nature of the items with low levels of neuroticism). Consequently, we anticipate
used to construct the NPI Manipulativeness subscale. that researchers will continue to view the content assessed
Additionally, we should note that although Grandiose within Manipulativeness and Grandiose Fantasies subscales
Fantasies and Manipulativeness showed similar relations in as indicative of pathological narcissism.
many ways, Grandiose Fantasies was somewhat more strong- We agree with conclusions drawn by Ackerman et al.’
ly tied to PNI Vulnerability than were Manipulativeness and (2016) regarding the future use of the NPI to assess narcissis-
other NPI subscales. tic traits. Although the use of NPI subscales represent an im-
provement compared to total scores for analyses, using mea-
The NPI: To What Degree is Pathological Narcissism sures that were created to model specific entitlement, grandi-
Captured? Implications for Future Narcissism ose, and vulnerable traits comprising the broader narcissism
Research domain may be preferable (e.g., the Five-Factor Narcissism
Inventory [FFNI]; Glover et al. 2012) to using ad hoc sub-
As discussed, previous research has questioned the degree to scales modeling NPI factors. There were many parallels be-
which leadership-related content included in the NPI reflects tween our factor analytic results and those presented by
pathological narcissism. This content that has been challenged Ackerman and colleagues as stated (e.g., Manipulativeness
primarily is subsumed within the NPI Leadership (e.g., “am a factors emerging across samples). However, there also were
born leader”) and Superiority (e.g., “will be a success”) sub- key differences in our studies and this prior work (e.g., a
scales. Subscales modeling these factors showed robust neg- Grandiose Fantasies factor emerging in our study only); more-
ative associations with indicators of neuroticism across sam- over, the NPI items provide limited coverage of entitlement
ples, which suggests that high scores on these subscales ap- and very little coverage of narcissistic vulnerability (Krizan
pear adaptive in at least some ways. In addition, the current and Herlache 2018).
J Psychopathol Behav Assess

Limitations and Conclusion Ethical Standards This research received approval from the University
of Notre Dame Institutional Review Board. All procedures performed in
studies involving human participants were in accordance with the ethical
The current study has several notable strengths. In particular, standards of the institutional and/or national research committee and with
we were able to establish the robustness of many key findings the 1964 Helsinki declaration and its later amendments or comparable
across both community and student samples. At the same ethical standards.
time, however, the findings of the current research need to
be interpreted considering several limitations. First, our study
focused solely on self-report measures of narcissistic traits, as References
well as self-reports of personality traits, other externalizing
Ackerman, R. A., Witt, E. A., Donnellan, M. B., Trzesniewski, K. H.,
traits, and mood symptoms. Future research would benefit
Robins, R. W., & Kashy, D. A. (2011). What does the narcissistic
from a multi-method approach, which could include incorpo- personality inventory really measure? Assessment, 18, 67–87.
rating interview ratings and/or informant ratings of personality Ackerman, R. A., Donnellan, M. B., Roberts, B. W., & Fraley, R. C.
and psychopathology. We also did not include a forced-choice (2016). The effect of response format on the psychometric properties
of the narcissistic personality inventory: Consequences for item
version of the NPI to directly compare similarities and differ-
meaning and factor structure. Assessment, 23, 203–220.
ences between response formats, given that a large body of Arditte, K. A., Çek, D., Shaw, A. M., & Timpano, K. R. (2016). The
research already has explicated the structure and correlates of importance of assessing clinical phenomena in mechanical turk re-
the NPI when administered in this manner; nonetheless, it search. Psychological Assessment, 28, 684–691.
could have been interesting to examine head-to-head compar- Boldero, J. M., Bell, R. C., & Davies, R. C. (2015). The structure of the
narcissistic personality inventory with binary and rating scale items.
isons for both NPI administration formats when examining Journal of Personality Assessment, 97, 626–637.
factor structures and patterns of correlations. We also used Cain, N. M., Pincus, A. L., & Ansell, E. B. (2008). Narcissism at the
two convenience samples, and we hope that future research crossroads: Phenotypic description of pathological narcissism across
will examine these issues in other sample types (e.g., samples clinical theory, social/personality psychology, and psychiatric diag-
nosis. Clinical Psychology Review, 28, 638–656.
of individuals with high levels of personality pathology).
Corry, N., Merritt, R. D., Mrug, S., & Pamp, B. (2008). The factor struc-
Finally, our data were cross-sectional in nature, which pre- ture of the narcissistic personality inventory. Journal of Personality
cluded tests of temporal stability regarding both factor struc- Assessment, 90, 593–600.
tures and correlational patterns. Thus, future research would Emmons, R. A. (1984). Factor analysis and construct validity of the
benefit from using longitudinal designs, especially to examine narcissistic personality inventory. Journal of Personality
Assessment, 48, 291–300.
the degree to which NPI subscales predict scores on measures Everett, J. E. (1983). Factor comparability as a means of determining the
such as self-esteem variability that may be implicated in nar- number of factors and their rotation. Multivariate Behavioral
cissism (Wright and Edershile 2018). Research, 18, 197–218.
These limitations aside, our findings have important impli- Everett, J. E., & Entrekin, L. V. (1980). Factor comparability and the
advantages of multiple group factor analysis. Multivariate
cations for the most widely used assessment instrument of
Behavioral Research, 2, 165–180.
narcissism. They provide strong evidence that the NPI is de- Few, L. R., Miller, J. D., Rothbaum, A., Meller, S., Maples, J., Terry, D.
fined by distinct sets of item content showing divergent pat- P., Collins, B., & MacKillop, J. (2013). Examination of the section
terns of personality and psychopathology relations when ad- III DSM-5 diagnostic system for personality disorders in an outpa-
tient clinical sample. Journal of Abnormal Psychology, 122, 1057–
ministered using a single-stimulus administration format and a
1069.
Likert style response scale, consistent with prior research. Finn, S. E. (1986). Structural stability of the MMPI in adult males.
Therefore, if researchers continue to use the NPI in place of Journal of Consulting and Clinical Psychology, 54, 703–707.
other newer measures created to assess distinct narcissism Foster, J. D., Shiverdecker, L. K., & Turner, I. N. (2016). What does the
dimensions (e.g., the PNI and FFNI), we advise considering narcissistic personality inventory measure across the total score con-
tinuum? Current Psychology, 35, 207–219.
the use of NPI subscales modeling specific narcissistic traits
Glover, N., Miller, J. D., Lynam, D. R., Crego, C., & Widiger, T. A.
instead of focusing solely on total scores. (2012). The five-factor narcissism inventory: A five-factor measure
of narcissistic personality traits. Journal of Personality Assessment,
Funding Information This study was partially funded by University of 94, 500–512.
Notre Dame Institute for Scholarship in the Liberal Arts. Gorsuch, R. L. (1983). Factor analysis (2nd ed.). Hillsdale: Erlbaum.
Jones, D. N., & Paulhus, D. L. (2014). Introducing the short dark triad
(SD3): A brief measure of the dark personality traits. Assessment,
Compliance with Ethical Standards 21, 28–41.
Krizan, Z., & Herlache, A. D. (2018). The narcissism spectrum model: A
Conflict of Interest Matthew F. D. Brown, Kasey Stanton and David synthetic view of narcissistic personality. Personality and Social
Watson declare that they have no conflict of interest. Psychology Review, 22, 3–31.
Krueger, R. F., Derringer, J., Markon, K. E., Watson, D., & Skodol, A. E.
Informed Consent Informed consent was obtained from all individual (2012). Initial construction of a maladaptive personality trait model
participants included in this study. and inventory for DSM–5. Psychological Medicine, 42, 1879–1890.
J Psychopathol Behav Assess

Maples, J. L., Guan, L., Carter, N. T., & Miller, J. D. (2014a). A test of the construct validity. Journal of Personality and Social Psychology,
international personality item pool representation of the revised 54, 890–902.
NEO personality inventory and development of a 120-item IPIP- Stanton, K., & Zimmerman, M. (2018). Clinician ratings of vulnerable
based measure of the five-factor model. Psychological Assessment, and grandiose narcissistic features: Implications for an expanded
26, 1070–1084. narcissistic personality disorder diagnosis. Personality Disorders:
Maples, J. L., Lamkin, J., & Miller, J. D. (2014b). A test of two brief Theory, Research, and Treatment, 9, 263–272.
measures of the dark triad: The dirty dozen and short dark triad. Stanton, K., Daly, E., Stasik-O’Brien, S. M., Ellickson-Larew, S., Clark,
Psychological Assessment, 26, 326–331. L. A., & Watson, D. (2017). An integrative analysis of the narcis-
McCredie, M. N., & Morey, L. C. (2019). Who are the turkers? A char- sistic personality inventory and hypomanic personality scale:
acterization of mturk workers using the personality assessment in- Implications for construct validity. Assessment, 24, 695–711.
ventory. Assessment, 26, 759–766. Stanton, K., McArtor, D. B., & Watson, D. (2019). Parsing the hypoman-
Miller, J. D., Lynam, D. R., Hyatt, C. S., & Campbell, K. W. (2017). ic personality: Explicating the nature of specific dimensions defin-
Controversies in narcissism. Annual Review of Clinical ing mania risk. Assessment, 26, 492–507.
Psychology, 13, 1–25. Timmerman, M. E., & Lorenzo-Seva, U. (2011). Dimensionality assess-
Miller, J. D., Gentile, B., Carter, N. T., Crowe, M., Hoffman, B. J., & ment of ordered polytomous items with parallel analysis.
Campbell, K. W. (2018). A comparison of the nomological networks Psychological Methods, 16, 209–220.
associated with forced-choice and Likert formats of the narcissistic
Velicer, W. F. (1976). Determining the number of components from the
personality inventory. Journal of Personality Assessment, 100, 259–
matrix of partial correlations. Psychometrika, 41, 321–327.
267.
O’Connor, B. P. (2000). SPSS and SAS programs for determining the Watson, D., O’Hara, M. W., Naragon-Gainey, K., Koffel, E.,
number of components using parallel analysis and Velicer’s MAP Chmielewski, M., Kotov, R., et al. (2012). Development and vali-
test. Behavior Research Methods, Instruments, & Computers, 32, dation of new anxiety and bipolar symptom scales for an expanded
396–402. version of the IDAS (the IDAS-II). Assessment, 19, 399–420.
Patrick, C. J., Kramer, M. D., Krueger, R. F., & Markon, K. E. (2013). Weaver, B., & Wuensch, K. L. (2013). SPSS and SAS programs for
Optimizing efficiency of psychopathology assessment through comparing Pearson correlations and OLS regression coefficients.
quantitative modeling: Development of a brief form of the external- Behavioral Research Methods, 45, 880–895.
izing spectrum inventory. Psychological Assessment, 25, 1332– Wetzel, E., Roberts, B. W., Fraley, R. C., & Brown, A. (2016).
1348. Equivalence of narcissistic personality inventory constructs and cor-
Paulhus, D. L. (2001). Normal narcissism: Two minimalist accounts. relates across scoring approaches and response formats. Journal of
Psychological Inquiry, 12(4), 228–230. Research in Personality, 61, 87–98.
Paulhus, D. L., & Williams, K. M. (2002). The dark triad of personality: Wright, A. G. C., & Edershile, E. A. (2018). Issues resolved and unre-
Narcissism, machiavellianism, and psychopathy. Journal of solved and pathological narcissism. Current Opinion in Psychology,
Research in Personality, 36(6), 556–563. 21, 74–79.
Pincus, A. L., & Lukowitsky, M. R. (2010). Pathological narcissism and Wright, A. G. C., Lukowitsky, M. R., Pincus, A. L., & Conroy, D. E.
narcissistic personality disorder. Annual Review of Clinical (2010). The higher order factor structure and gender invariance of
Psychology, 6, 421–446. the pathological narcissism inventory. Assessment, 17, 467–483.
Pincus, A. L., Ansell, E. B., Pimentel, C. A., Cain, N. M., Wright, A. G. Wright, A. G. C., Thomas, K. M., Hopwood, C. J., Markon, K. E., Pincus,
C., & Levy, K. N. (2009). Initial construction and validation of the A. L., & Krueger, R. F. (2012). The hierarchical structure of DSM-5
pathological narcissism inventory. Psychological Assessment, 21, pathological personality traits. Journal of Abnormal Psychology,
365–379. 121, 951–957.
Raskin, R. N., & Hall, C. S. (1979). A narcissistic personality inventory.
Psychological Reports, 45, 590.
Publisher’s Note Springer Nature remains neutral with regard to jurisdic-
Raskin, R., & Terry, H. (1988). A principal-components analysis of the
tional claims in published maps and institutional affiliations.
narcissistic personality inventory and further evidence of its

View publication stats

You might also like