Journal of Personality Disorders, 32(1), 87-108, 2018
© 2018 The Guilford Press
From The Pennsylvania State University, University Park, Pennsylvania (S. D., A. L. P.); Michigan State
University, East Lansing, Michigan (H. S. S.); and Texas A & M University, College Station, Texas (M.
Address correspondence to Sindes Dawood, The Pennsylvania State University, 361 Moore Building,
University Park, PA 16802. E-mail: email@example.com
DAWOOD ET AL.
PATHOLOGICAL NARCISSISM AND SELF-INJURY
PATHOLOGICAL NARCISSISM AND NONSUICIDAL
Sindes Dawood, MS, Hans S. Schroder, MA,
M. Brent Donnellan, PhD, and Aaron L. Pincus, PhD
Few studies have examined associations between pathological narcissism
and self-harm, but those that do indicate that narcissistic vulnerability (not
narcissistic grandiosity) relates to self-harm. The current study extends this
literature by investigating how facets of pathological narcissism assessed
by the Pathological Narcissism Inventory relate to speciﬁc nonsuicidal
self-injury (NSSI) behaviors assessed by the Inventory of Statements About
Self-Injury using statistical models appropriate for non-normally distributed
count data. In a sample of 1,023 undergraduate students, results revealed
that facets of both narcissistic vulnerability and narcissistic grandiosity were
differentially related to the endorsement and frequency of speciﬁc NSSI be-
haviors and higher-order latent NSSI factors (repetitive and impulsive), even
after accounting for levels of borderline pathology. The clinical implications
of these results are discussed.
Pathological narcissism is related to both suicide and nonsuicidal self-injury
(e.g., Ansell et al., 2015; Pincus et al., 2009). Suicide refers to the deliber-
ate act of injuring oneself with the intent to die (Posner, Oquendo, Gould,
Stanley, & Davies, 2007), whereas nonsuicidal self-injury (NSSI) refers to
the intentional damage of one’s body without suicidal intent (Muehlenkamp,
2005). NSSI includes behaviors such as hair pulling, self-hitting, cutting,
carving words or pictures into skin, and burning (Klonsky, 2007). Most of
the existing research focuses on relations between global assessments of nar-
cissism such as the diagnosis of narcissistic personality disorder (NPD) and
suicide attempts or presence/absence of NSSI. The current study extends this
research by examining how facets of pathological narcissism relate to speciﬁc
NSSI behaviors using statistical models appropriate for count data with non-
NPD and pathological narcissism (i.e., narcissistic grandiosity and nar-
cissistic vulnerability) are both signiﬁcant risk factors for suicidal behavior
in adolescents, adults, and the elderly. For instance, early research found be-
tween 4.7% and 23% of suicide completers exhibited elevated symptoms of
88 DAWOOD ET AL.
NPD (Apter et al., 1993; Brent et al., 1994). More recent work conﬁrms that
NPD is a risk factor for suicide attempts (García-Nieto, Blasco-Fontecilla, de
León-Martinez, & Baca-García, 2014; Giner et al., 2013; Soloff & Chiap-
petta, 2012) and a unique predictor (among personality disorder diagnoses)
of making multiple suicide attempts over a 10-year period (Ansell et al.,
2015). Further, the presence of comorbid depression is associated with an in-
creased risk of suicide for individuals with NPD (Conner et al., 2001; Heisel,
Links, Conn, van Reekum, & Flett, 2007). However, it is important to note
that narcissistic individuals are at heightened risk for suicide even when not
clinically depressed (Cross, Westen, & Bradley, 2011; Links, 2013; Ronning-
stam & Maltsberger, 1998; Ronningstam, Weinberg, & Malstberger, 2008).
Moreover, NPD is associated with suicide attempts characterized by lower
impulsivity but higher expected lethality (Blasco-Fontecilla et al., 2009), and
these attempts tend to be associated with certain life stressors/changes (e.g.,
being ﬁred from a job, having more arguments with a spouse; Blasco-Fonte-
cilla et al., 2010).
Both narcissistic grandiosity (maladaptive self-enhancement motivation
characterized by an inﬂated self view with concomitant aggrandizing fanta-
sies, a sense of entitlement, and interpersonal exploitativeness) and narcissis-
tic vulnerability (impaired self, emotion, and behavior regulation in response
to self-enhancement failures and lack of recognition and admiration from
others) are related to suicide attempts, but only narcissistic vulnerability ap-
pears associated with NSSI (Miller et al., 2010; Pincus et al., 2009; Thomas
et al., 2012). Consistent with this, symptoms of NPD, which consist pri-
marily of narcissistic grandiosity, are not reliably associated with NSSI in
adolescent and college student samples (Casillas & Clark, 2002; Tromp &
Koot, 2010). However, military recruits with a history of NSSI scored higher
on NPD (and other Diagnostic and Statistical Manual of Mental Disorders,
fourth edition [DSM-IV] personality disorders) than recruits without a his-
tory of NSSI (Klonsky, Oltmanns, & Turkheimer, 2003). In summary, unique
associations between NSSI and narcissistic vulnerability seem well support-
ed, but the evidence linking NSSI and narcissistic grandiosity appears mixed.
The present study aimed to clarify the relationship between pathological
narcissism and NSSI. Prior research pertaining to narcissism and suicidality
has focused almost entirely on associations between narcissism and suicide
or narcissism and the presence/absence of self-harm. There is a need to un-
derstand exactly which features of suicidality and/or NSSI behavior are re-
lated to narcissism, as both are important clinical constructs to examine in
relation to various forms of personality pathology. Recent research shows
that individual differences such as sensation seeking and trait aggression are
related to speciﬁc forms of NSSI (Kleiman et al., 2015; Knorr, Jenkins, &
Conner, 2013). Given that borderline personality disorder (BPD) has also
long been associated with NSSI (e.g., Brown, Comtois, & Linehan, 2002;
Kleindienst et al., 2008; Lieb, Zanarini, Schmahl, Linehan, & Bohus, 2004),
and self-harm is a diagnostic criterion for BPD (American Psychiatric As-
sociation, 2013), this study investigated whether and how distinct facets of
pathological narcissism relate to speciﬁc NSSI behaviors after accounting for
levels of BPD. We expected that facets of both narcissistic vulnerability and
PATHOLOGICAL NARCISSISM AND SELF-INJURY 89
grandiosity would be associated with various NSSI behaviors even control-
ling for symptoms of BPD.
The current sample is drawn from undergraduate students. Accordingly,
we evaluated the prevalence of NSSI in the sample to make sure there were
adequate instances to investigate our research questions. We also consider
how participants with reports of NSSI compare to those who did not endorse
NSSI. Klonsky (2011) used a random-digit dialing procedure to investigate
NSSI in adults from the United States and found that age was among the
strongest demographic predictors of NSSI. Participants under 30 years of age
were more likely to report NSSI than participants over 30 years. Income and
educational history were not statistically signiﬁcant correlates of reports of
NSSI in that sample. This suggests that samples of younger participants are
particularly well suited for studying correlates of NSSI given considerations
of statistical power. Standard concerns about the selective nature of college
samples in terms of education background and socioeconomic status might
not be as signiﬁcant, given that these factors are not strongly associated with
NSSI (e.g., Klonsky, 2011). Moreover, Whitlock, Eckenrode, & Silverman
(2006) found that many NSSI behaviors were undisclosed to clinical profes-
sionals, suggesting the importance of using nonselective samples to study this
clinically relevant phenomenon.
Participants were undergraduate students from a large Midwestern universi-
ty who completed study questionnaires for partial course credit. Data collec-
tion was ongoing throughout the semester, and a total of 1,389 participants
initiated the survey. We included four items to detect inattentive responding
(e.g., “Please choose answer choice 2 to ensure you are paying attention”),
and excluded participants who failed to respond to all four items accurately
(Maniaci & Rogge, 2014; n = 363). An additional three participants were
excluded from study analyses because of outlier NSSI behavior data. Speciﬁ-
cally, count data for rubbing skin against rough surface, wound picking, and
biting were equal to or greater than 1,000,000 for these participants. Thus,
the ﬁnal sample consisted of 1,023 undergraduates (712 female, 268 male,
4 declined to respond, and 39 missed responses; Mage = 19.84, SD = 1.78).
The racial/ethnic makeup of the sample was 75.30% European American,
9.00% African American, 8.90% Asian, 5.00% Latino/Hispanic, 2.7% Bi-
racial, .90% American Indian, and 1.70% Other (participants could pick
more than one category). All participants consented to the study prior to
data collection, and the university’s institutional review board approved all
MEASURES AND ANALYTIC PLAN
Pathological Narcissism. The Pathological Narcissism Inventory (PNI; Pin-
cus, 2013; Pincus et al., 2009) is a 52-item self-report measure of pathologi-
90 DAWOOD ET AL.
cal narcissism. The PNI has seven subscales that capture facets of narcissistic
grandiosity (i.e., exploitativeness, self-sacriﬁcing self-enhancement, grandi-
ose fantasy) and narcissistic vulnerability (i.e., contingent self-esteem, hiding
the self, devaluing, and entitlement rage). All items are scored on a 6-point
scale ranging from 0 (“Not At All”) to 5 (“Very Much Like Me”). A large
body of laboratory (e.g., Fetterman & Robinson, 2010), clinical (e.g., El-
lison, Levy, Cain, Ansell, & Pincus, 2013; Morf et al., 2016), longitudinal
(e.g., Roche, Pincus, Conroy, Hyde, & Ram, 2013), and correlational (e.g.,
Roche, Pincus, Hyde, Conroy, & Ram, 2013; Wright et al., 2013) research
supports the validity of the PNI. In this sample, Cronbach’s alphas ranged
from .72 to .93 for the scales.
Nonsuicidal Self-Injury. The Inventory of Statements About Self-Injury
(ISAS; Klonsky & Glenn, 2009) is a self-report measure that assesses the fre-
quency and functions of nonsuicidal self-injury (NSSI). The ISAS is broken
up into two sections. The ﬁrst section assesses the lifetime frequency of 12
different NSSI behaviors that are enacted on purpose but without suicidal
intent (biting, burning, carving, cutting, hitting/banging body parts, interfer-
ing with wound healing [wound picking], sticking self with needles, pinch-
ing, hair pulling, rubbing skin against rough surfaces, severe scratching, and
swallowing dangerous substances). The second section of the ISAS assesses
13 speciﬁc functions that are commonly related to NSSI (e.g., Affect Regula-
tion, Self-Punishment, Sensation Seeking). Studies have found the ISAS to
be a reliable and valid measure of NSSI frequency and functions in a large
sample of young adults (Glenn & Klonsky, 2011; Klonsky & Glenn, 2009;
Klonsky & Olino, 2008). For the purposes of this study, we focused analyses
on the lifetime frequency of speciﬁc NSSI behaviors and did not consider
functions of NSSI. In general, more research with the ISAS focuses on the
frequency scales than the functions scales (e.g., Brausch & Boone, 2015;
Hamza & Willoughby, 2013; Klonsky, May, & Glenn, 2013; Nicolai, Wiel-
gus, & Mezulis, 2016).
Some researchers also suggest that classifying NSSI behaviors into dis-
tinct types may be helpful in their conceptualization, assessment, and treat-
ment (Favazza, 2012; Favazza & Rosenthal, 1990; Favazza & Simeon, 1995;
Simeon & Hollander, 2008). Their classiﬁcation system divides NSSI behav-
iors into four major categories: stereotypic NSSI (i.e., highly repetitive, ﬁxed,
driven acts without thought that cause tissue damage ranging from mild to
severe to even life threatening, such as hair pulling, hitting/banging the self,
skin picking, or scratching); major NSSI (i.e., highly impulsive or planned
acts that are most severe or life threatening, such as castration, limb amputa-
tion); compulsive NSSI (i.e., repetitive, often ritualistic acts that tend to be
mild to moderate, such as hair pulling, skin picking, nail biting); and impul-
sive NSSI (i.e., behaviors that are not highly repetitive but often episodic and
gratifying, such as skin cutting, skin burning, self-hitting). However, given
that there is overlap between the categories, other researchers (Favaro &
Santonastaso, 1998, 2000; Mathews et al., 2004) have proposed and em-
pirically validated simpler two-factor classiﬁcation systems that resemble the
system proposed by Favazza and colleagues (Favazza & Rosenthal, 1990;
PATHOLOGICAL NARCISSISM AND SELF-INJURY 91
Favazza & Simeon, 1995). That is, one category (regarded as “compulsive”
or “moderate” NSSI) consists of symptoms from both the compulsive and
stereotypic categories, while the second category (regarded as “impulsive”
or “severe”) consists of symptoms from the impulsive and major catego-
ries. Research ﬁnds that episodic forms of NSSI (e.g., cutting, burning) are
more strongly related to speciﬁc precipitating events than are mildly habitual
forms of NSSI (e.g., skin picking, nail biting; Croyle & Waltz, 2007). Based
on theory and previous research, we evaluated the latent structure of NSSI
behaviors in the current sample, discussed below, to identify the best ﬁtting
model to be used to examine associations between facets of pathological
narcissism and higher-order latent factors of NSSI.
Borderline Pathology. The Borderline Personality Questionnaire (BPQ; Poreh
et al., 2006) is an 80-item true-false questionnaire measuring thoughts, feel-
ings, and behaviors associated with BPD. The BPQ was scored without the
suicide/self-mutilation subscale to avoid criterion contamination with the
study’s primary outcome (self-injury). A total score was used, with higher
values reﬂecting higher levels of borderline pathology. Cronbach’s alpha was
.94 for the BPQ.
To further characterize the sample’s psychological symptom and behav-
ior endorsement, we administered a number of other validated self-report
measures, described below.
Antisocial Behavior. Antisocial behavior was measured with the Subtypes
of Antisocial Behavior Questionnaire (STAB; Burt & Donnellan, 2009).
The STAB consists of a 32 items assessing physical aggression (“Felt like
hitting someone”), rule-breaking (“Stole property from work or school”),
and social aggression (“Called someone names behind his/her back”). Each
subscale asks participants to rate the items on a 5-point scale ranging from
“never” to “nearly all the time.” For the current study, the total score was
used and Cronbach’s alpha was .92.
Depression Symptoms. The Patient Health Questionnaire–9 (PHQ-9; Kroen-
ke et al., 2001) is a 10-item measure of depression. On the PHQ-9, partici-
pants indicate the extent to which they experienced each of nine depressive
symptoms (e.g., “Feeling down, depressed, or hopeless”) over the past two
weeks on a 0 (Not at all) to 3 (Nearly every day) rating scale. Items were
summed for analysis, and Cronbach’s alpha was .90.
Posttraumatic Stress Disorder Symptoms. The Posttraumatic Checklist for
DSM-5 (PCL-5; Blevins et al., 2015; Weathers et al., 2013) is a 20-item self-
report measure of Diagnostic and Statistical Manual of Mental Disorders,
ﬁfth edition (DSM-5) symptoms of posttraumatic stress disorder (PTSD) ex-
perienced during the past month. Items (e.g., “In the past month, how much
were you bothered by: repeated, disturbing, and unwanted memories of the
stressful experience?”) are rated on a scale of 0 (Not at all) to 4 (Extremely).
Items on the PCL-5 were summed for analysis, and Cronbach’s alpha was
92 DAWOOD ET AL.
Drug Use. The Drug Abuse Screening Test-10 (DAST-10; Skinner, 1982) is
a 10-item (Yes/No) self-report measure used to screen for drug abuse prob-
lems in treatment-seeking individuals (Cocco & Carey, 1998; Skinner, 1982),
including college students with substance use problems (Taylor, James, Bo-
badilla, & Reeves, 2008). Items were averaged to score the DAST, and Cron-
bach’s alpha was .71.
Alcohol Abuse. Alcohol abuse was measured with items from the Patient-
Reported Outcomes Measurement Information System–Alcohol Use Short
Form (PROMIS-Alc; Pilkonis et al., 2013). Participants responded to seven
items using a 5-point scale ranging from “Never” to “Almost Always” re-
garding alcohol abuse (e.g., “I spent too much time drinking”) in the past
30 days. Because the PROMIS-Alc items target respondents who have had
a drink in the previous 30 days, data were available from 794 respondents.
Items on the PROMIS-Alc were summed for analysis, and Cronbach’s alpha
Psychological Treatment History. Finally, we assessed whether participants
had ever had experience with professional treatment for a mental health
or emotional problem; we speciﬁcally asked about psychological therapy,
psychiatric medication, and psychiatric hospitalization. For the therapy and
medication questions, participants chose one of the following options: “No,
never,” “No, but I would like to,” Yes, in the past,” or “Yes, currently.”
Items were then coded such that “No, never” and “No, but I would like to”
were scored as 0 and the other two options were scored as 1. For the hospi-
talization item, participants were asked if they had ever been hospitalized for
an emotional problem.
Prior to all analyses, we evaluated the normality of the data using SPSS ver-
sion 22 (SPSS Inc., Chicago, IL), and found that all NSSI behavior count data
violated normality assumptions, as they exhibited signiﬁcant positive skew-
ness (M = 18.386; range = 10.937–31.096) and high kurtosis (M = 410.165;
range = 152.315–982.263), and all distributions were zero inﬂated. Figure 1
provides an example of such a distribution using the ISAS pinching behavior
We also used the PSCL package in R (Zeileis, Kleiber, & Jackman, 2008)
to determine the statistical model most appropriate for non-normally distrib-
uted count data, as well as for all regression analyses estimating associations
between narcissism/BPD and different latent NSSI behaviors. Mplus 7.31
(Muthén & Muthén, 2014) was used for all structural equation-modeling
analyses aimed at identifying the best ﬁtting model of the structure of NSSI
to be used in the count-based regression models of NSSI on narcissism and
borderline pathology. We now describe these different analyses in more de-
PATHOLOGICAL NARCISSISM AND SELF-INJURY 93
Model Fit for Regressions. Given the non-normal, zero-inﬂated distribution
of all NSSI behavior count outcomes, six different count-based regression
models (negative binomial, negative binomial hurdle, Poisson, Poisson hur-
dle, zero inﬂated negative binomial, and zero inﬂated Poisson), rather than
linear-based regression models, were estimated for each narcissism/BPD-NS-
SI behavior combination. These count-based models are more appropriate
for the current data because the distributions of rare but clinically relevant
symptoms and behaviors commonly include many zeros (Wright, Pincus, &
The count models differ in terms of the conditional mean and variance
of the outcome variable. Whereas Poisson models assume that both the con-
ditional mean and conditional variance of the distribution are equal, nega-
tive binomial (NB) models do not make this assumption. Rather, NB models
allow the variance to exceed the mean and thus enable the NB distribution
to account for overdispersion in the data. However, when data have far more
zeros (i.e., “excessive” zeros) than can be accounted for by a standard Pois-
son or NB distributional model, a model that can adjust for extra zeros is
needed. A zero-inﬂated Poisson (ZIP) model is designed for this purpose.
The ZIP regression mixes a distribution degenerate at zero with a Poisson
distribution. However, if the data continue to suggest further overdispersion,
a zero-inﬂated negative binomial (ZINB) model can be used. The ZINB re-
gression mixes a distribution degenerate at zero with a negative binomial dis-
tribution. An alternative method to analyze count data that display excessive
zero observations are hurdle models, such as the Poisson hurdle (PH) model
and the negative binomial hurdle (NBH) model (see Loeys, Moerkerke, De
FIGURE 1. Observed nonsuicidal self-injury (NSSI) pinching
94 DAWOOD ET AL.
Smet, & Buysse, 2012). Hurdle models are two-part models that combine a
binary logistic regression model with a zero-truncated (e.g., Poisson or nega-
tive binomial) count model.
In the present study, we examined which of the count-based regression
models (i.e., Poisson, ZIP, PH, NB, ZINB, NHB) best ﬁt our NSSI behavior
count data. Whereas the two nested models—Poisson and NB models—can
be compared using the likelihood ratio tests (LRTs; Long, 1997), zero-in-
ﬂated and hurdle models are non-nested, and therefore require an alterna-
tive test. We used the Vuong (1989) non-nested LRT to compare non-nest-
ed models (ZIP, ZINB, PH, NBH). A signiﬁcant Vuong statistic favors one
model over the other, and a nonsigniﬁcant statistic indicates the two models
Structural Equation Modeling. We used structural equation modeling (SEM)
to identify the best ﬁtting model of the structure of NSSI to be used in the
narcissism/BPD-latent NSSI factor regression analyses. For all SEM analyses,
a log transformation was used to normalize the distribution of the NSSI
behavior count data, and the robust MLR estimator was used in Mplus to
take into account non-normal data and missing values. We also speciﬁed our
indicators as a hurdle model with a normal distribution by using the DATA
TWOPART function in Mplus. Accordingly, all speciﬁc NSSI behavior vari-
ables included binary and continuous aspects of the measurement models.
We evaluated the latent structure of 12 NSSI behaviors by specifying
four different conﬁrmatory factor analysis (CFA) models. We used CFA be-
cause we had a priori hypotheses regarding the number and composition of
the factors. We created composite scores for each NSSI behavior domain.
Model 1 was a one-factor in which all 12 NSSI behaviors were represented
by a single factor. Model 2 was composed of two correlated latent factors,
such that the repetitive NSSI latent factor was composed of biting, pinch-
ing, hair pulling, wound picking, rubbing skin against rough surfaces, severe
scratching, hitting or banging the self, and the impulsive NSSI latent factor
was composed of burning, carving, cutting, sticking self with needles, and
swallowing dangerous substances. Model 3 was identical to Model 2 except
that hitting or banging the self was moved to the impulsive NSSI latent fac-
tor. Finally, Model 4 was identical to these two previous models except that
hitting or banging the self was allowed to cross-load onto both latent fac-
tors. We anticipated that a two-factor model would ﬁt our data better than
the one-factor model, but it was unclear which of the alternative two-factor
models would be the best ﬁtting model because research has yielded mixed
ﬁndings as to what NSSI category hitting or banging the self best belongs
to (e.g., Kleiman et al., 2015; Mathews et al., 2004). Last, the sample-size
adjusted Bayesian information criterion (BIC) was used for model compari-
sons to select the best ﬁtting model of the structure of NSSI. Lower values
represent better ﬁtting models (Kline, 2005).
Regression Analysis. Last, we performed a series of NBH regression analyses
to examine the effects of pathological narcissism on NSSI behaviors when
PATHOLOGICAL NARCISSISM AND SELF-INJURY 95
accounting for borderline pathology. The ﬁrst part of the hurdle regression
model tests the effect of predictors (levels of borderline pathology and facets
of pathological narcissism in this case) on the likelihood (i.e., the presence
versus absence) of experiencing an outcome (e.g., self-injury). If the outcome
has occurred, the second part of the hurdle regression model tests the effect
of the predictors on the frequency of experiencing an outcome among those
speciﬁc individuals. Hurdle models thus provide separate regression coef-
ﬁcients for the zero versus nonzero and the count part of the model. The
exponentiated coefﬁcients for the zero/nonzero regression can be interpreted
as odds ratios, whereas coefﬁcients in the count regression can be interpreted
as rate ratios. Odds/rate ratios greater than 1 indicate a positive associa-
tion between predictors and outcome variables; values less than 1 indicate
a negative association between the predictor and outcome variables, and an
association near 1 indicates no association.
In our ﬁrst set of regression analyses, each of the NSSI higher-order do-
main counts was regressed onto all seven PNI scales and the BPQ scale. In
our second set of regression analyses, each speciﬁc NSSI behavior count was
regressed onto all seven pathological narcissism facet scales and the BPQ
scale.1 These analyses controlled for shared variance among the predictors.
The results of these analyses are described in further detail in the sections
Table 1 provides endorsement rates and descriptive statistics (frequency,
range) on NSSI variables. Although a college student sample was used in this
study, we can see that 35.78% of the total sample reported a history of hav-
ing engaged in any kind of NSSI behavior. Moreover, there was a consider-
able range of psychological distress in this sample, and individuals who had
engaged in NSSI (NSSI group) had signiﬁcantly more self-reported mental
health problems (e.g., borderline pathology, depression severity, substance
use, psychiatric hospitalizations; see Table 2) than did the non-NSSI group.
There were no statistical differences in demographic variables (e.g., gender,
year in school) between the NSSI group and the non-NSSI group. Moreover,
despite a majority of participants not endorsing NSSI, the endorsement rate
and frequency of NSSI that does exist allows for the prediction of speciﬁc
Results from the LRT tests indicated that in every instance the NB model
ﬁt the data signiﬁcantly better than its Poisson counterpart. On the other
hand, a comprehensive comparison of the models (i.e., comparing a nested
1. Each speciﬁc NSSI behavior count was also regressed on Narcissistic Grandiosity (NG), Narcissistic
Vulnerability (NV), and Borderline Pathology simultaneously in the hurdle regression models. Although
NG and NV (unlike Borderline Pathology) tended to be nonsigniﬁcant for most of the results, this did not
appear to be the case when we used facets of NV and NG, suggesting that assessment at the facet level
could be more informative for NSSI predictions.
96 DAWOOD ET AL.
with a non-nested model; comparing two non-nested models) suggested that,
with a few exceptions, the NB, ZINB, and NBH models were equivalent.
However, several NB and ZINB models were nonidentiﬁed. Thus, a hurdle
model with a normal distribution was used when modeling the structure of
NSSI, and a NBH regression model was chosen as the ﬁnal model to test (a)
whether facets of pathological narcissism predict the presence of higher-or-
der NSSI domains and/or speciﬁc NSSI behaviors when accounting for levels
of borderline pathology, and (b) whether those same facets of pathologi-
cal narcissism predict the frequency of higher-order NSSI dimensions and/or
speciﬁc individual NSSI behaviors when accounting for levels of borderline
THE STRUCTURE OF NSSI
As predicted, Model 2—the two-factor model of repetitive NSSI and impul-
sive NSSI with hitting or banging the self in the repetitive NSSI factor—was
the best-ﬁtting model (BIC = 9120.26; Model 1 BIC = 9147.08, Model 3
BIC = 9131.60, Model 4 BIC = 9123.88). For the preferred CFA model (i.e.,
Model 2),2 repetitive NSSI and impulsive NSSI were correlated .88 (p < .001)
with each other for the binary logistic regression aspect of the hurdle mod-
el, whereas these two higher-order NSSI variables were correlated .63 (p <
.001) with each other for the continuous zero-truncated aspect of the hurdle
model. Based on this selected model, we created two higher-order summary
scores reﬂecting repetitive NSSI behaviors (biting, pinching, hair pulling,
wound picking, rubbing skin against rough surfaces, severe scratching, and
hitting or banging the self) and impulsive NSSI behaviors (burning, carving,
cutting, sticking self with needles, swallowing dangerous substances). These
TABLE 1. Endorsement Rates and Descriptive Statistics on Nonsuicidal Self-Injury (NSSI) Variables
Behavior Type Endorsement, n (%) Frequency of endorsement Range
Biting 69 (6.74) 2,315 0–500
Burning 44 (4.30) 301 0–75
Carving 18 (1.76) 250 0–100
Cutting 131 (12.81) 3,773 0–400
Hitting or Banging Self 124 (12.12) 4,901 0–750
Pinching 121 (11.83) 7,628 0–1,000
Hair Pulling 111 (10.85) 4,249 0–500
Rubbing Skin† 59 (5.77) 3,139 0–1,000
Severe Scratching 110 (10.75) 4,059 0–500
Sticking Self With Needles 31 (3.03) 922 0–333
Swallowing Substances‡ 25 (2.44) 1,372 0–1,000
Wound Picking 231 (22.58) 47,322 0–10,000
Total NSSI Behaviors 366 (35.78) 80,231 0–10,108
N = 1,023. † = rubbing skin against rough surface; ‡ = swallowing dangerous substances.
2. The factor loadings for the preferred CFA model are available upon request from the ﬁrst author.
PATHOLOGICAL NARCISSISM AND SELF-INJURY 97
TABLE 2. Demographic and Mental Health Variables Between Individuals With and Without NSSI
NSSI (–) NSSI (+) Effect Size
Variable M SD Mdn Range M SD Mdn Range
% Female 73.20 — — — 71.60 — — — c2 = 0.31, ns
Income Rating 3.49 1.55 — — 3.45 1.51 — — –0.02, ns
Year in School 2.19 1.15 — — 2.07 1.13 — — –0.10, ns
Psychiatric Symptom and History
Borderline Pathology (BPQ) 15.90 12.09 15.89 .00–61.00 24.12 13.63 22.00 .00–62.00 0.64***
Pathological Narcissism (PNI) 2.43 0.65 2.42 .02–4.23 2.66 0.63 2.69 .74–5.00 0.36***
Antisocial Behaviors (STAB) 47.50 12.21 45.00 31.00–115.00 53.84 13.85 52.00 32.00–117.00 0.49***
Depression Severity (PHQ-9) 5.13 4.86 4.00 .00–27.00 9.03 6.20 8.00 .00–27.00 0.73***
Trauma (PCL-5) 15.32 15.05 11.00 .00–74.00 24.26 18.83 20.50 .00–79.00 0.54***
Drug Use (DAST) 0.15 0.10 .10 .00–.70 0.22 0.17 .20 .00–.90 0.54***
Alcohol Abuse (PROMIS-Alc) 12.96 5.39 12.00 5.00–35.00 14.47 6.06 13.00 7.00–34.00 0.27***
Therapy History (%) 21.33 — — — 40.91 — — — c2 = 42.73***
Psychiatric Medication History (%) 8.74 — — — 20.23 — — — c2 = 26.52***
Psychiatric Hospitalization History (%) 4.45 — — — 8.83 — — — c2 = 7.64**
Note. Percentages for therapy, medication, and hospitalization represent any history (current or past) of use of these services. Unless otherwise noted, effect size is Cohen’s d and is scored such that
positive values indicate higher values for the NSSI (+) group. NSSI = nonsuicidal self-injury (–): absent; (+): present; ns for NSSI (–) ranged from 510 to 657 and for NSSI (+), ns ranged from 312
to 366. Income Ratings based on 1 (under $20,000) to 5 (over $100,000) scale. BPQ = Borderline Personality Questionnaire (Poreh et al., 2006); PNI = Pathological Narcissism Inventory; STAB =
Subtypes of Antisocial Behavior (Burt & Donnellan, 2009); PHQ-9 = Patient Health Questionnaire-9 Depression Scale (Kroenke, Spitzer, & Williams, 2001); PCL-5 = Posttraumatic Stress Disorder
Checklist for DSM-5 (Blevins et al., 2015); DAST = Drug Abuse Screening Test-10 (Skinner, 1982); PROMIS-Alc = Patient-Reported Outcomes Measurement Information System-Alcohol Use Short
Form (Pilkonis et al., 2013). ***p < .001. **p < .01. ns = not statistically signiﬁcant. Results did not change when non-parametric Mann-Whitney U test was used to make group comparisons.
98 DAWOOD ET AL.
variables were used as our outcome variables in the following regression
NARCISSISM AND THE HIGHER-ORDER NSSI DIMENSIONS
Coefﬁcients and their signiﬁcance for the NBH model are summarized in
Table 3. Distinct relationships emerge between speciﬁc facets of pathological
narcissism and higher-order dimensions of NSSI even when accounting for
levels of borderline pathology. Speciﬁcally, Grandiose Fantasy is positively
associated with endorsing repetitive NSSI behaviors, but is not signiﬁcantly
associated with the frequency of repetitive behaviors. In contrast, Hiding
the Self is related to an increased frequency of repetitive behaviors, but is
not signiﬁcantly associated with endorsing repetitive NSSI. Finally, Exploit-
ativeness is positively associated with endorsing impulsive NSSI behaviors,
and Entitlement Rage is negatively associated with endorsing impulsive NSSI
behaviors, but neither is associated with their frequency. Additionally, Self-
Sacriﬁcing Self-Enhancement is signiﬁcantly related to a decreased frequency
of impulsive NSSI, but not with the endorsement of impulsive NSSI.
TABLE 3. Summary of Coefﬁcients From Model Regressing Higher-Order Nonsuicidal Self-Injury
Behavior on Facets of Pathological Narcissism and Borderline Pathology
Negative Binomial Hurdle
Nonsuicidal Self-Injury Domain OR (h) p (h) RR (NB) p (NB)
EXP 0.980 0.816 0.890 0.473
SSSE 0.975 0.834 1.437 0.088
GF 1.333 0.012 0.708 0.061
CSE 0.960 0.707 0.886 0.567
HS 1.169 0.141 1.625 0.009
DEV 1.016 0.887 1.526 0.060
ER 0.839 0.136 0.877 0.496
BPD 1.040 0.000 1.036 0.011
EXP 1.276 0.031 0.925 0.719
SSSE 1.078 0.622 0.500 0.028
GF 1.058 0.622 1.325 0.247
CSE 1.209 0.165 0.670 0.127
HS 1.020 0.887 1.244 0.272
DEV 0.942 0.682 1.135 0.580
ER 0.683 0.009 0.819 0.434
BPD 1.068 0.000 1.070 0.000
Note. OR = odds ratio; RR = rate ratio; h = hurdle class; NB = negative binomial. EXP = exploitativeness; SSSE =
self-sacriﬁcing self-enhancement; GF = grandiose fantasies; CSE = contingent self-esteem; HS = hiding the self; DEV =
devaluing; ER = entitlement rage; BPD = borderline personality disorder. All values are standardized. Boldface indicates
p < .05.
PATHOLOGICAL NARCISSISM AND SELF-INJURY 99
ENDORSEMENT AND FREQUENCY OF SPECIFIC NSSI BEHAVIORS
We examined the same model described in Table 3, but this time separate-
ly regressing speciﬁc NSSI behaviors, rather than latent NSSI behavior do-
mains, on facets of pathological narcissism while accounting for borderline
pathology. Table 4 presents these ﬁndings.
Facets of Grandiosity. Exploitativeness is positively associated with endors-
ing burning, but not with its frequency. Self-Sacriﬁcing Self-Enhancement is
signiﬁcantly associated with an increased frequency of wound picking, but
not its endorsement. Grandiose Fantasy is positively associated with endors-
ing hitting or banging the self and negatively associated with endorsing both
swallowing dangerous substances and wound picking, but not their frequen-
cy. Grandiose Fantasy is also associated with a decreased frequency of biting,
hair pulling, severe scratching, and an increased frequency of carving, if these
behaviors were endorsed.
Facets of Vulnerability. Contingent Self-Esteem is negatively associated with
endorsing biting, but positively associated with its frequency if endorsed.
Hiding the Self is positively associated with endorsing severe scratching, as
well as an increased frequency of pinching and a decreased frequency of
carving, if these behaviors were endorsed. Devaluing is positively associated
with endorsing pinching, and an increased frequency of carving and wound
picking. Finally, Entitlement Rage was negatively associated with endorsing
carving, cutting, hitting or banging self, pinching, and severe scratching, but
not their frequency if endorsed. Entitlement Rage was also associated with a
decreased frequency of biting if endorsed.
This is the ﬁrst study to examine associations between distinct facets of
pathological narcissism and speciﬁc nonsuicidal self-injury (NSSI) behaviors
using hurdle regression models. All seven facets of pathological narcissism
were associated with NSSI behaviors, but in varying ways, even when ac-
counting for levels of borderline pathology in the undergraduate sample.
Thus, these results highlight the usefulness of the facet-level assessment of
Overall, ﬁndings supported our hypothesis in showing that when lev-
els of BPD are accounted for, facets of both narcissistic vulnerability and
grandiosity are associated with various NSSI behaviors. Relative to facets of
narcissistic grandiosity, facets of narcissistic vulnerability were more often
signiﬁcantly associated with both the presence and increased frequency of
speciﬁc NSSI behaviors (Table 4), providing further support for narcissis-
tic vulnerability–NSSI associations (Miller et al., 2010; Pincus et al., 2009;
Thomas et al., 2012). Whereas other studies found mixed evidence link-
ing narcissistic grandiosity and NSSI, our ﬁndings demonstrated that certain
facets of narcissistic grandiosity (e.g., Grandiose Fantasy) are signiﬁcantly
100 DAWOOD ET AL.
TABLE 4. Summary of Coefﬁcients from Model Regressing Speciﬁc Self-Injury Behavior on Facets of
Pathological Narcissism and Borderline Pathology
Negative Binomial Hurdle
Nonsuicidal Self-Injury Behavior OR (h) p (h) RR (NB) p (NB)
EXP 1.211 0.243 0.686 0.235
SSSE 0.915 0.684 2.252 0.123
GF 1.049 0.816 0.388 0.007
CSE 0.652 0.032 2.614 0.008
HS 1.353 0.126 1.391 0.337
DEV 1.475 0.074 1.580 0.197
ER 0.722 0.139 0.305 0.001
BPD 1.044 0.000 0.989 0.565
EXP 1.530 0.035 0.442 0.307
SSSE 0.746 0.280 1.158 0.913
GF 1.137 0.620 0.480 0.570
CSE 1.254 0.347 3.822 0.085
HS 0.911 0.701 0.735 0.537
DEV 1.048 0.857 1.314 0.656
ER 0.748 0.259 0.918 0.923
BPD 1.080 0.000 1.038 0.240
EXP 1.272 0.412 0.671 0.290
SSSE 0.959 0.918 0.645 0.229
GF 1.280 0.512 2.103 0.046
CSE 1.036 0.919 0.669 0.199
HS 0.597 0.166 0.216 0.005
DEV 1.535 0.292 5.844 0.010
ER 0.431 0.033 0.637 0.304
BPD 1.083 0.000 1.002 0.945
EXP 1.131 0.317 1.175 0.534
SSSE 1.051 0.770 0.501 0.059
GF 1.008 0.959 1.345 0.331
CSE 1.234 0.161 0.800 0.515
HS 1.208 0.203 1.458 0.094
DEV 0.785 0.133 1.103 0.687
ER 0.660 0.009 0.685 0.194
BPD 1.073 0.000 1.047 0.017
Hitting or Banging Self
EXP 1.048 0.707 1.240 0.351
SSSE 0.983 0.922 2.137 0.069
GF 1.507 0.012 0.935 0.839
CSE 0.855 0.300 1.694 0.074
HS 0.942 0.697 1.066 0.841
DEV 1.379 0.052 0.838 0.610
ER 0.708 0.036 0.603 0.091
BPD 1.059 0.000 1.038 0.038
EXP 1.118 0.377 1.283 0.339
SSSE 1.032 0.856 0.724 0.416
GF 1.250 0.170 0.784 0.456
CSE 0.972 0.853 1.764 0.079
HS 1.049 0.754 1.679 0.046
DEV 1.403 0.040 1.048 0.872
ER 0.645 0.009 0.596 0.059
BPD 1.042 0.000 1.050 0.005
EXP 0.980 0.872 0.853 0.434
SSSE 1.239 0.234 1.902 0.096
GF 1.007 0.965 0.405 0.002
CSE 0.906 0.530 1.187 0.561
PATHOLOGICAL NARCISSISM AND SELF-INJURY 101
Negative Binomial Hurdle
Nonsuicidal Self-Injury Behavior OR (h) p (h) RR (NB) p (NB)
HS 1.206 0.231 1.628 0.057
DEV 1.064 0.712 0.835 0.501
ER 0.813 0.216 0.870 0.595
BPD 1.059 0.000 1.005 0.790
EXP 1.191 0.298 1.152 0.814
SSSE 1.034 0.887 0.455 0.447
GF 0.813 0.338 0.928 0.919
CSE 0.953 0.818 4.290 0.071
HS 1.477 0.053 1.099 0.887
DEV 1.022 0.917 0.715 0.618
ER 1.087 0.700 0.793 0.792
BPD 1.046 0.000 0.979 0.396
EXP 1.129 0.354 0.823 0.389
SSSE 0.810 0.233 1.544 0.196
GF 1.082 0.635 0.532 0.017
CSE 1.080 0.628 1.156 0.622
HS 1.479 0.013 1.145 0.619
DEV 1.159 0.378 1.204 0.575
ER 0.711 0.044 1.455 0.140
BPD 1.047 0.000 0.989 0.552
Sticking Self With Needles
EXP 1.441 0.111 0.770 0.671
SSSE 0.956 0.886 0.899 0.928
GF 0.861 0.606 0.802 0.745
CSE 0.785 0.381 0.445 0.121
HS 0.978 0.936 0.567 0.589
DEV 1.156 0.634 5.778 0.120
ER 0.861 0.624 0.786 0.770
BPD 1.083 0.000 1.052 0.145
EXP 1.502 0.134 0.575 0.602
SSSE 1.303 0.471 3.123 0.608
GF 0.511 0.044 0.676 0.620
CSE 1.179 0.611 0.177 0.154
HS 0.877 0.667 0.964 0.963
DEV 0.658 0.210 2.920 0.138
ER 1.154 0.675 7.246 0.176
BPD 1.061 0.001 1.089 0.047
EXP 1.502 0.666 0.575 0.944
SSSE 1.303 0.261 3.123 0.036
GF 0.511 0.048 0.676 0.284
CSE 1.179 0.948 0.177 0.052
HS 0.877 0.094 0.964 0.111
DEV 0.658 0.847 2.920 0.024
ER 1.154 0.842 7.246 0.558
BPD 1.061 0.000 1.089 0.087
Note. OR = odds ratio; RR = rate ratio; h = hurdle class; NBH = negative binomial hurdle. EXP = exploitativeness;
SSSE = self-sacriﬁcing self-enhancement; GF = grandiose fantasies; CSE = contingent self-esteem; HS = hiding the
self; DEV = devaluing; ER = entitlement rage; BPD = borderline personality disorder. † = rubbing skin against rough
surface; ‡ = swallowing dangerous substances. All values are standardized. Boldface indicates p < .05.
102 DAWOOD ET AL.
related to speciﬁc NSSI behaviors and to speciﬁc higher-order NSSI domains.
Moreover, the fact that most facets of narcissistic vulnerability and grandios-
ity were associated with the endorsement and frequency of NSSI behaviors,
even after controlling for borderline pathology, supports the view that both
expressions of pathological narcissism are important in understanding self-
harming behaviors. Individuals higher in pathological narcissism have high
expectations for themselves and others, but also rely on others for self and
emotional regulation (e.g., Pincus, 2013; Pincus, Roche, & Good, 2015).
Thus, such individuals may be more likely to engage in NSSI behaviors in
response to self-enhancement failures and the lack of needed recognition or
admiration from others (Blasco-Fontecilla et al., 2010; Pincus, Dowgwillo,
& Greenberg, 2016).
Several distinct patterns emerged when speciﬁc NSSI behaviors were
grouped under the two higher-order domains of repetitive NSSI and impul-
sive NSSI. Notably, distinct facets of narcissistic grandiosity and vulnerabil-
ity were associated with both repetitive and impulsive NSSI behaviors. For
instance, Grandiose Fantasy was positively associated with the presence of
repetitive NSSI behaviors. This result makes sense given that individuals ele-
vated on this narcissistic feature are likely to retreat into their private worlds
to cope with negative affect and dysregulation associated with narcissistic
injuries (e.g., Pincus, 2013; Pincus, Cain, & Wright, 2014). Once dysregu-
lated, Grandiose Fantasy in the service of self-enhancement (see Schoenleber
& Berenbaum, 2012) is negatively associated with NSSI frequency, while
continued social isolation (Hiding the Self) is positively associated with fre-
quency. In contrast, Exploitativeness was positively associated with the en-
dorsement of impulsive NSSI behaviors. Exploitativeness is a highly inter-
personal facet of narcissism, and this suggests that episodic social triggers
may be particularly problematic for narcissistic individuals, possibly evoking
impulsive NSSI behaviors.
We believe it is efﬁcient and useful to examine NSSI at the latent level
(i.e., repetitive vs. impulsive) because the data appear to conform well to a
two-factor solution, and focusing on these dimensions produces theoreti-
cally coherent results. Nevertheless, there is still value in focusing on speciﬁc
forms of NSSI behaviors to capture some nuances that might otherwise be
missed. For instance, although Devaluing was not signiﬁcantly associated
with the endorsement or frequency of repetitive or impulsive behaviors when
accounting for levels of borderline pathology, this facet of narcissistic vulner-
ability was particularly associated with the increased frequency of carving on
the skin with a sharp object if it was endorsed. One reason why Devaluing
may be so strongly related to some NSSI behaviors is that it not only reﬂects
the dismissal of others, but also has a strong component of devaluing the
self that involves self-criticism and shame for needing recognition and ad-
miration from disappointing others. Thus, narcissistic individuals may be
more likely to engage in various forms of NSSI to punish themselves when
they experience frustrations and disappointment over these unmet needs. In
contrast to those who experience self-criticism and shame, individuals who
experience anger (Entitlement Rage) evoked by unmet entitled expectations
PATHOLOGICAL NARCISSISM AND SELF-INJURY 103
of others’ interest, attention, and admiration may be less likely to engage in
NSSI as their negative affect is directed toward others rather than the self.
PATHOLOGICAL NARCISSISM, NSSI, AND SUICIDE
Although the underlying motivations for suicide and self-harm are different,
there is research indicating they are related to each other. Engaging in NSSI
increases the risk for suicide attempts (Bryan, Bryan, May, & Klonsky, 2015;
Guan, Fox, & Prinstein, 2012), increases the lethality of suicide attempts
(Andover & Gibb, 2010), and increases the risk of completed suicides (Coo-
per et al., 2005; Hawton, Zahl, & Weatherhall, 2003). The present study
found that facets of both narcissistic vulnerability and narcissistic grandiosi-
ty are related to the endorsement and frequency of NSSI behaviors even after
accounting for levels of borderline pathology. These ﬁndings underscore the
importance of assessing the full range of manifestations of pathological nar-
cissism and corroborate clinical conceptualizations of suicidality in narcis-
sistically disturbed individuals as a process reﬂecting the interplay of grandi-
osity and vulnerability. For example, clinical observations show that patients
with a combination of affect dysregulation and pathological narcissism may
respond to interpersonal stress with unexpected, potentially deadly, self-de-
structive behaviors (Blasco-Fontecilla et al., 2010; García-Nieto et al., 2014;
Pincus et al., 2014, 2016). Such behaviors may be an effort to protect the
self from negative affect (e.g., shame, rage) associated with loss, defeat, and
narcissistic injury while simultaneously giving the narcissistic individual the
feelings of invulnerability and mastery over his or her own life (Ronningstam
& Malstberger, 1998; Stone, 1989). Thus, in addition to routine monitor-
ing for suicidality (Links & Prakash, 2014), the current results indicate the
potential beneﬁts of monitoring for NSSI as well.
LIMITATIONS, CONCLUSIONS, AND FUTURE DIRECTIONS
This study has several limitations that should be taken into consideration
when interpreting the ﬁndings. One potential limitation of this study is that
the sample was a fairly homogenous group of college students. However, this
concern may be mitigated by the fact that self-injurious behaviors are rela-
tively common in college samples, and the prevalence of self-reported history
of NSSI in the current study’s sample was higher or similar to that observed
in other studies of self-harm among college students (e.g., Gratz, 2001; Whit-
lock et al., 2006, 2011). Further, although caution must be exercised in gen-
eralizing our ﬁndings to clinical populations, the fact that our NSSI group
self-reported greater severity of mental health problems (e.g., borderline pa-
thology, depression severity, see Table 2) than the non-NSSI group suggests
to us that this sample provides useful insights into how pathological narcis-
sism is associated with NSSI. Nevertheless, research should replicate and
extend our ﬁndings in diverse clinical (e.g., inpatient, outpatient) and other
nonclinical (e.g., community) samples, and consider taking into account the
co-occurrence of other mental health problems beyond borderline pathology
that are linked with NSSI. A second limitation is that data were based solely
104 DAWOOD ET AL.
on self-report questionnaires. Future studies should consider the inclusion of
standardized clinical interviews for NSSI (e.g., Gratz, Dixon-Gordon, Chap-
man, & Tull, 2015; Linehan, Comtois, Brown, Heard, & Wagner, 2006;
Nock, Holmberg, Photos, & Michel, 2007) and pathological narcissism
(e.g., Gunderson, Ronningstam, & Bodkin, 1990; Loranger et al., 1994).
However, respondents may be more likely to report pathological, embarrass-
ing, or shameful behaviors on anonymous self-report measures than in face-
to-face interviews (e.g., Kaplan et al., 1994; Keel, Crow, Davis, & Mitchell,
2002; Locke & Gilbert, 1995). Finally, given the cross-sectional design of
this study, future work is needed to prospectively examine the presence and
frequency of NSSI behaviors in individuals high in pathological narcissism.
Despite its limitations, the present study found pathological narcissism
was related to self-reported endorsement and frequency of NSSI behaviors
even after accounting for levels of borderline pathology among a large sample
of undergraduate students. Our results show that facets of both narcissistic
vulnerability and narcissistic grandiosity differ in their associations with spe-
ciﬁc NSSI behaviors and higher-order (repetitive and impulsive) NSSI behav-
iors. Also important is that this study used statistical models appropriate for
highly skewed count distributions typical for rare clinical events, methods
that were not used in other correlational studies examining narcissism–NSSI
associations. Not only do the current ﬁndings extend understanding of the
link between narcissistic pathology and NSSI, they also support the use of
lower-order facets of pathological narcissism. The current results also point
to possible psychological and interpersonal mechanisms linking pathological
narcissism and NSSI. Future research into the contexts and motivations for
NSSI in narcissistic individuals is needed.
American Psychiatric Association. (2013). Diag-
nostic and statistical manual of mental dis-
orders (5th ed.). Washington, DC: Author.
Andover, M. S., & Gibb, B. E. (2010). Non-sui-
cidal self-injury, attempted suicide, and sui-
cidal intent among psychiatric inpatients.
Psychiatry Research, 178, 101–105.
Ansell, E. B., Wright, A. G. C., Markowitz, J. C.,
Sanislow, C. A., Hopwood, C. J., Zanarini,
M. C., … Grilo, C. M. (2015). Personality
disorder risk factors for suicide attempts
over 10 years of follow-up. Personality Dis-
orders: Theory, Research, and Treatment, 6,
Apter, A., Bleich, A., King, R. A., Kron, S., Fluch,
A., Kotler, M., & Cohen, D. J. (1993).
Death without warning? A clinical post-
mortem study of suicide in 43 Israeli adoles-
cent males. Archives of General Psychiatry,
Blasco‐Fontecilla, H., Baca‐Garcia, E., Dervic, K.,
Perez‐Rodriguez, M. M., Saiz‐Gonzalez, M.
D., Saiz‐Ruiz, J., ... De Leon, J. (2009). Se-
verity of personality disorders and suicide
attempt. Acta Psychiatrica Scandinavica,
Blasco-Fontecilla, H., Baca-Garcia, E., Duberstein,
P., Perez-Rodriguez, M., Dervic, K., Saiz-
Ruiz, J., ... Oquendo, M. A. (2010). An ex-
ploratory study of the relationship between
diverse life events and personality disorders
in a sample of suicide attempters. Journal of
Personality Disorders, 24, 773–784.
Blevins, C. A., Weathers, F. W., Davis, M. T.,
Witte, T. K., & Domino, J. L. (2015). The
Posttraumatic Stress Disorder Checklist for
DSM-5 (PCL-5): Development and initial
psychometric evaluation. Journal of Trau-
matic Stress, 28, 489-–98.
Brausch, A. M., & Boone, S. D. (2015). Frequency
of nonsuicidal self‐injury in adolescents:
Differences in suicide attempts, substance
use, and disordered eating. Suicide and
Life-Threatening Behavior, 45, 612–622.
Brent, D. A., Johnson, B. A., Perper, J., Connol-
ly, J., Bridge, J., Bartle, S., & Rather, C.
PATHOLOGICAL NARCISSISM AND SELF-INJURY 105
(1994). Personality disorder, personality
traits, impulsive violence, and completed
suicide in adolescents. Journal of the Ameri-
can Academy of Adolescent Psychiatry, 33,
Brown, M. Z., Comtois, K. A., & Linehan, M.
M. (2012). Reasons for suicide attempts
and nonsuicidal self-injury in women with
borderline personality disorder. Journal of
Abnormal Psychology, 111, 198–202.
Bryan, C. J., Bryan, A. O., May, A. M., & Klon-
sky, E. D. (2015). Trajectories of suicide
ideation, nonsuicidal self-injury, and suicide
attempts in a nonclinical sample of military
personnel and veterans. Suicide and Life-
Threatening Behavior, 45, 315–325.
Burt, S. A., & Donnellan, M. B. (2009). Develop-
mental and validation of the Subtypes of
Antisocial Behavior Questionnaire. Aggres-
sive Behavior, 35, 376–398.
Casillas, A., & Clark, L. A. (2002). Dependency,
impulsivity and self-harm: Traits hypoth-
esized to underlie the association between
cluster B personality and substance use dis-
orders. Journal of Personality Disorders,
Cocco, K. M., & Carey, K. B. (1998). Psychomet-
ric properties of the Drug Abuse Screening
Test in psychiatric outpatients. Psychologi-
cal Assessment, 10, 408–414.
Conner, K. R., Cox, C., Duberstein, P. R., Tian,
L., Nisbet, P. A., & Conwell, Y. (2001).
Violence, alcohol, and completed suicide:
A case-control study. American Journal of
Psychiatry, 158, 1701–1705.
Cooper, J., Kapur, N., Webb, R., Lawlor, M., Guth-
rie, E., Mackway-Jones, K., & Appleby, L.
(2005). Suicide after deliberate self-harm: A
4-year cohort study. American Journal of
Psychiatry, 162, 297–303.
Cross, D., Westen, D., & Bradley, B. (2011). Per-
sonality subtypes of adolescents who at-
tempt suicide. Journal of Nervous and Men-
tal Disease, 199, 750–756.
Croyle, K. L., & Waltz, J. (2007). Subclinical self-
harm: Range of behaviors, extent, and as-
sociated characteristics. American Journal
of Orthopsychiatry, 77, 332–342.
Ellison, W. D., Levy, K. N., Cain, N. M., Ansell, E.
B., & Pincus, A. L. (2013). The impact of
pathological narcissism on psychotherapy
utilization, initial symptom severity, and
early-treatment symptom change: A natu-
ralistic investigation. Journal of Personality
Assessment, 95, 291–300.
Favaro, A., & Santonastaso, P. (1998). Impulsive
and compulsive self-injurious behavior in
bulimia nervosa: Prevalence and psycho-
logical correlates. Journal of Nervous &
Mental Disease, 183, 157–165.
Favaro, A., & Santonastaso, P. (2000). Self-injuri-
ous behavior in anorexia nervosa. Journal
of Nervous & Mental Disease, 188, 537–
Favazza, A. R. (2012). Nonsuicidal self-injury:
How categorization guides treatment. Cur-
rent Psychiatry, 11, 21–26.
Favazza, A. R., & Rosenthal, R. J. (1990). Variet-
ies of pathological self-mutilation. Behav-
ioural Neurology, 3, 77–85.
Favazza, A. R., & Simeon, D. (1995). Self-mutila-
tion. In E. Hollander & D. J. Stein (Eds.),
Impulsivity and aggression (pp. 185–200).
London, UK: Wiley.
Fetterman, A. K., & Robinson, M. D. (2010).
Contingent self-importance among patho-
logical narcissists: Evidence from an implic-
it task. Journal of Research in Personality,
García-Nieto, R., Blasco-Fontecilla, H., de León-
Martinez, V., & Baca-García, E. (2014).
Clinical features associated with suicide
attempts versus suicide gestures in an inpa-
tient sample. Archives of Suicide Research,
Giner, L., Blasco-Fontecilla, H., Mercedes Perez-
Rodriguez, M., Garcia-Nieto, R., Giner, J.,
Guija, J. A., … Baca-Garcia, E. (2013). Per-
sonality disorders and health problems dis-
tinguish suicide attempters from completers
in a direct comparison. Journal of Affective
Disorders, 151, 474–483.
Glenn, C. R., & Klonsky, E. D. (2011). One-year
test-retest reliability of the Inventory of
Statements About Self-Injury (ISAS). As-
sessment, 18, 375–378.
Gratz, K.L. (2001). Measurement of deliberate
self-harm: Preliminary data on the Delib-
erate Self-Harm Inventory. Journal of Psy-
chopathology and Behavioral Assessment,
Gratz, K. L., Dixon-Gordon, K., Chapman, A.
L., & Tull, M. T. (2015). Diagnosis and
characterization of DSM-5 nonsuicidal
self-injury disorder using the Clinician-
Administered Nonsuicidal Self-Injury Dis-
order Index. Assessment, 22, 527–539.
Guan, K., Fox, K. R., & Prinstein, M. J. (2012).
Nonsuicidal self-injury as a time-invariant
predictor of adolescent suicide ideation and
attempts in a diverse community sample.
Journal of Consulting and Clinical Psychol-
ogy, 80, 842–849.
Gunderson, J. G., Ronningstam, E., & Bodkin, A.
(1990). The diagnostic interview for narcis-
sistic patients. Archives of General Psychia-
try, 47, 676–680.
Hamza, C. A., & Willoughby, T. (2013). Nonsui-
cidal self-injury and suicidal behavior: A
latent class analysis among young adults.
PLoS One, 8, e59955. doi:10.1371/journal.
106 DAWOOD ET AL.
Hawton, K., Zahl, D., & Weatherhall, R. (2003).
Suicide following deliberate self-harm:
Long-term follow-up of patients who pre-
sented to a general hospital. British Journal
of Psychiatry, 182, 537–542.
Heisel, M. J., Links, P. S., Conn, D., van Reekum,
R., & Flett, G. L. (2007). Narcissistic per-
sonality and vulnerability to late-life suicid-
ality. American Journal of Geriatric Psychi-
atry, 15, 734–741.
Kaplan, M. L., Asnis, G. M., Sanderson, W. C.,
Keswani, L., De Lecuona, J. M., & Joseph,
S. (1994). Suicide assessment. Clinical in-
terview vs. self-report. Journal of Clinical
Psychology, 50, 294–298.
Keel, P. M., Crow, S., Davis, T. L., & Mitchell, J.
E. (2002). Assessment of eating disorders:
Comparison of interview and questionnaire
data from a long-term follow-up study of
bulimia nervosa. Journal of Psychosomatic
Research, 53, 1043–1047.
Kleiman, E. M., Ammerman, B. A., Kulper, D. A.,
Uyeji, L. L., Jenkins, A. L., & McCloskey,
M. S. (2015). Forms of non-suicidal self-in-
jury as a function of trait aggression. Com-
prehensive Psychiatry, 59, 21–27.
Kleindienst, N., Bohus, M., Ludascher, P., Lim-
berger, M. F., Kuenkele, K., Ebner-Priemer,
U. W., et al. (2008). Motives for non-suicid-
al self-injury among women with borderline
personality disorder. Journal of Nervous
and Mental Diseases, 196, 230–236.
Kline, R.B. (2005). Principles and practice of struc-
tural equation modeling (2nd ed.). New
York, NY: Guilford.
Klonsky, E. D. (2007). The functions of deliberate
self-injury: A review of the evidence. Clini-
cal Psychology Review, 27, 226–239.
Klonsky, E. D. (2011). Non-suicidal self-injury in
United States adults: Prevalence, sociode-
mographics, topography and functions.
Psychological Medicine, 41, 1981–1986.
Klonsky, E. D., & Glenn, C. R. (2009). Assessing
the functions of non-suicidal self-injury:
Psychometric properties of the Inventory of
Statements About Self-injury (ISAS). Jour-
nal of Psychopathology and Behavioral As-
sessment, 31, 215–219.
Klonsky, E. D., May, A. M., & Glenn, C. R. (2013).
The relationship between nonsuicidal self-
injury and attempted suicide: Converging
evidence from four samples. Journal of Ab-
normal Psychology, 122, 231–237.
Klonsky, E. D., & Olino, T. M. (2008). Identifying
clinically distinct subgroups of self-injurers
among young adults: A latent class analysis.
Journal of Consulting and Clinical Psychol-
ogy, 76, 22–27.
Klonsky, E. D., Oltmanns, T. F., & Turkheimer, E.
(2003). Deliberate self-harm in a nonclini-
cal population: Prevalence and psychologi-
cal correlates. American Journal of Psychia-
try, 160, 1501–1508.
Knorr, A. C., Jenkins, A. L., & Conner, B. T.
(2013). The role of sensation seeking in
non-suicidal self-injury. Cognitive Therapy
and Research, 37, 1276–1284.
Kroenke, K., Spitzer, R. L., & Williams, J. B.
(2001). The PHQ‐9. Journal of General In-
ternal Medicine, 16(9), 606–613.
Lieb, K., Zanarini, M. C., Schmahl, C. G., Line-
han, M. M., & Bohus, M. (2004). Bor-
derline personality disorder. Lancet, 364,
Linehan, M. M., Comtois, K. A., Brown, M. Z.,
Heard, H. L., & Wagner, A. (2006). Sui-
cide Attempt Self-Injury Interview (SASII):
Development, reliability, and validity of a
scale to assess suicide attempts and inten-
tional self-injury. Psychological Assessment,
Links, P. S. (2013). Pathological narcissism and the
risk of suicide. In J. S. Ogrodniczuk (Ed.),
Understanding and treating pathological
narcissism (pp. 167–182). Washington, DC:
American Psychological Association.
Links, P., & Prakash, A. (2014). Strategic issues in
the psychotherapy of patients with narcis-
sistic pathology. Journal of Contemporary
Psychotherapy, 44, 97–107.
Locke, S. D., & Gilbert, B. O. (1995). Method of
psychological assessment, self-disclosure,
and experiential differences: A study of
computer, questionnaire, and interview as-
sessment formats. Journal of Social Behav-
ior and Personality, 10, 255–263.
Loeys, T., Moerkerke, B., De Smet, O., & Buysse,
A. (2012). The analysis of zero‐inﬂated
count data: Beyond zero‐inﬂated Poisson
regression. British Journal of Mathematical
and Statistical Psychology, 65, 163–180.
Long, J. S. (1997). Regression models for categori-
cal and limited dependent variables. Thou-
sand Oaks, CA: Sage.
Loranger, A. W., Sartorius, N., Andreoli, A., Berg-
er, P., Buchheim, P., Channabasavanna, S.
M., ... Jacobsberg, L. B. (1994). The In-
ternational Personality Disorder Exami-
nation: The World Health Organization/
Alcohol, Drug Abuse, and Mental Health
Administration international pilot study of
personality disorders. Archives of General
Psychiatry, 51, 215–224.
Maniaci, M. R., & Rogge, R. D. (2014). Caring
about carelessness: Participant inattention
and its effects on research. Journal of Re-
search in Personality, 48, 61–83.
Mathews, C. A., Waller, J., Glidden, D. V., Lowe,
T. L., Herrera, L. D., Budman, C. L., …
Reus, V.I. (2004). Self injurious behavior
in Tourette syndrome: Correlates with im-
pulsivity and impulse control. Journal of
PATHOLOGICAL NARCISSISM AND SELF-INJURY 107
Neurology, Neurosurgery, & Psychiatry,
Miller, J. D., Dir, A., Gentile, B., Wilson, L., Pryor,
L. R., & Campbell, W. K. (2010). Searching
for a vulnerable dark triad: Comparing fac-
tor 2 psychopathy, vulnerable narcissism,
and borderline personality disorder. Journal
of Personality, 78, 1529–1564.
Morf, C. C., Schürch, E., Küfner, A., Siegrist,
P., Vater, A., Back, M., … Schröder-Abé,
M. (2016). Expanding the nomological
net of the Pathological Narcissism In-
ventory: German validation and exten-
sion in a clinical inpatient sample. As-
sessment. Advance online publication.
Muehlenkamp, J. J. (2005). Self-injurious behavior
as a separate clinical syndrome. American
Journal of Orthopsychiatry, 75, 324–333.
Muthén, L. K., & Muthén, B. O. (2014). Mplus
user’s guide (7th ed.). Los Angeles, CA:
Muthén & Muthén.
Nicolai, K. A., Wielgus, M. D., & Mezulis, A.
(2016). Identifying risk for self‐harm: Ru-
mination and negative affectivity in the pro-
spective prediction of nonsuicidal self‐inju-
ry. Suicide and Life-Threatening Behavior,
46, 223–233. doi:10.1111/sltb.12186
Nock, M. K., Holmberg, E. B., Photos, V. I., & Mi-
chel, B. D. (2007). Self-Injurious Thoughts
and Behaviors Interview: Development, reli-
ability, and validity in an adolescent sample.
Psychological Assessment, 19, 309–317.
Pilkonis, P. A., Yu, L., Colditz, J., Dodds, N., John-
ston, K. L., Mahioefer, C., … McCarty, D.
(2013). Item banks for alcohol use from
the Patient-Reported Outcomes Measure-
ment Information System (PROMIS): Use,
consequences, and expectations. Drug and
Alcohol Dependence, 130, 167–177.
Pincus, A. L. (2013). The Pathological Narcis-
sism Inventory. In J. Ogrudniczuk (Ed.),
Understanding and treating pathological
narcissism (pp. 93–110). Washington, DC:
American Psychological Association.
Pincus, A. L., Ansell, E. B., Pimentel, C. A., Cain,
N. M., Wright, A. G. C., & Levy, K. N.
(2009). Initial construction and validation
of the Pathological Narcissism Inventory.
Psychological Assessment, 21, 365–379.
Pincus, A. L., Cain, N. M., & Wright, A. G.
(2014). Narcissistic grandiosity and narcis-
sistic vulnerability in psychotherapy. Per-
sonality Disorders: Theory, Research, and
Treatment, 5, 439–443.
Pincus, A. L., Dowgwillo, E. A., & Greenberg, L. S.
(2016). Three cases of narcissistic personal-
ity disorder through the lens of the DSM-5
alternative model for personality disorders.
Practice Innovations, 1, 144–177.
Pincus, A. L., Roche, M. J., & Good, E. W.
(2015). Narcissistic personality disor-
der and pathological narcissism. In P. H.
Blaney, R. F. Krueger, & T. Millon (Eds.),
Oxford textbook of psychopathology (3rd
ed.; pp. 791–813). New York, NY: Oxford
Poreh, A. M., Rawlings, D., Claridge, G., Free-
man, J. L., Faulkner, C., & Shelton, C.
(2006). The BPQ: A scale for the assessment
of borderline personality based on DSM-IV
criteria. Journal of Personality Disorders,
Posner, K., Oquendo, M. A., Gould, M., Stanley,
B., & Davies, M. (2007). Columbia Clas-
siﬁcation Algorithm of Suicide Assessment
(C-CASA): Classiﬁcation of suicidal events
in the FDA’s pediatric suicidal risk analysis
of antidepressants. American Journal of
Psychiatry, 164, 1035–1043.
Roche, M. J., Pincus, A. L., Conroy, D. E., Hyde,
A. L., & Ram, N. (2013). Pathological nar-
cissism and interpersonal behavior in daily
life. Personality Disorders: Theory, Re-
search, and Treatment, 4, 315–323.
Roche, M. J., Pincus, A. L., Hyde, A. L., Conroy,
D. E., & Ram, N. (2013). Within-person
covariation of agentic and communal per-
ceptions: Implications for interpersonal
theory and assessment. Journal of Research
in Personality, 47, 445–452.
Ronningstam, E. F., & Maltsberger, J. T. (1998).
Pathological narcissism and sudden suicide‐
related collapse. Suicide and Life-Threaten-
ing Behavior, 28, 261–271.
Ronningstam, E. F., Weinberg, I., & Maltsberger,
J. T. (2008). Eleven deaths of Mr. K.: Con-
tributing factors to suicide in narcissistic
personalities. Psychiatry, 71, 169–182.
Schoenleber, M., & Berenbaum, H. (2012). Aver-
sion and proneness to shame in self- and
informant-reported personality disorder
symptoms. Personality Disorders: Theory,
Research, and Treatment, 3, 294–304.
Simeon, D., & Hollander, E. (Eds.). (2008). Self-
injurious behaviors: Assessment and treat-
ment. Washington, DC: American Psychiat-
Skinner, H. A. (1982). The Drug Abuse Screening
Test. Addictive Behaviors, 7, 363–371.
Soloff, P. H., & Chiappetta, L. (2012). Subtyping
borderline personality disorder by suicidal
behavior. Journal of Personality Disorders,
Stone, M. H. (1989). Long-term follow-up of
narcissistic/borderline patients. Psychiatric
Clinics of North America, 12, 621–641.
Taylor, J., James, L. M., Bobadilla, L., & Reeves,
M. D. (2008). Screening for disinhibited
disorder cases in a college population: Per-
formance of the SMAST, DAST, SCID-II-Q,
and PDQ-4. Psychological Assessment, 20,
108 DAWOOD ET AL.
Thomas, K. M., Wright, A. G . C., Lukowitsky,
M. R., Donnellan, M. B., & Hopwood,
C. J. (2012). Evidence for the criterion va-
lidity and clinical utility of the Pathologi-
cal Narcissism Inventory. Assessment, 19,
Tromp, N. B., & Koot, H. M. (2010). Self-and par-
ent report of adolescent personality pathol-
ogy: Informant agreement and relations to
dysfunction. Journal of Personality Disor-
ders, 24, 151–170.
Vuong, Q. (1989). Likelihood ratio tests for mod-
el selection and non-nested hypotheses.
Econometrica, 57, 307–333.
Weathers, F. W., Litz, B. T., Keane, T. M., Palmieri,
P. A., Marx, B. P., & Schnurr, P. P. (2013).
The PTSD Checklist for DSM-5 (PCL-5).
Boston, MA: National Center for PTSD.
Whitlock, J., Eckenrode, J., & Silverman, D.
(2006). Self-injurious behaviors in a college
population. Pediatrics, 117, 1939–1948.
Whitlock, J., Muehlenkamp, J., Purington, A.,
Eckenrode, J., Barreira, P., Baral Abrams,
G., … Knox, K. (2011). Nonsuicidal self-in-
jury in a college population: General trends
and sex differences. Journal of American
College Health, 59, 691–698.
Wright, A. G., Pincus, A. L., & Lenzenweger, M.
F. (2012). An empirical examination of
distributional assumptions underlying the
relationship between personality disorder
symptoms and personality traits. Journal of
Abnormal Psychology, 121, 699–706.
Wright, A. G., Pincus, A. L., Thomas, K. M., Hop-
wood, C. J., Markon, K. E., & Krueger, R.
F. (2013). Conceptions of narcissism and
the DSM-5 pathological personality traits.
Assessment, 20, 339–352.
Zeileis, A., Kleiber, C., & Jackman, S. (2008). Re-
gression models for count data in R. Jour-
nal of Statistical Software, 27, 1–25.