The Influence of Narcissistic Vulnerability and Grandiosity on Momentary Hostility Leading up to and Following Interpersonal Rejection

Abstract

Wide empirical support exists for two aspects of narcissism — grandiosity and vulnerability. Hostility is a form of interpersonal antagonism, which is considered central to narcissism broadly. Though it has often been subsumed by the concept of narcissistic grandiosity, interpersonal antagonism is associated with vulnerability as well. Rejection represents an interpersonal stressor that evokes hostility to a greater degree in those high in narcissism, with mixed evidence regarding whether it stems from threat to one’s egotism (grandiosity) or low self-esteem (vulnerability). Therefore, investigating the associations between narcissistic dimensions and individuals’ trajectories of hostility leading up to and following rejection may provide a basis for a more unified conceptualization. In this study, we leverage the wide range of narcissistic expression displayed in a combined sample of borderline personality disorder (N=56) and community (N=60) individuals who completed ambulatory assessments approximately six times per day for 21 consecutive days. We examine whether narcissistic vulnerability and grandiosity, as measured by NEO Personality Inventory facet combinations constructed based on the Five Factor Narcissism Inventory, moderate trajectories and overall levels of hostility surrounding self-reported interpersonal rejections. Grandiosity and vulnerability were independently positively associated with a faster rise in hostility leading up to rejection; however, greater grandiosity was uniquely associated with a greater spike in hostility at the occasion of rejection and subsequent faster recovery. These results are consistent with both the idea that grandiosity is proportionately more central to interpersonal antagonism and that antagonism serves as a bridge, connecting and reinforcing both narcissistic grandiosity and vulnerability.

Much of the research into Narcissistic Personality Disorder (NPD) has found grandiose and vulnerable dimensions underlying narcissism (Miller et al., 2011) or, relatedly, an “overt” dimension encompassing such Diagnostic and Statistical Manual of Mental Disorders (DSM; APA, 2013) criteria as grandiose self-importance, belief in one’s own uniqueness, entitlement, exploitativeness, arrogance, and lack of empathy, and a “covert” dimension encompassing fantasies, envy, and need for admiration, respectively (Fossati et al., 2005).¹ However, it is not the case that overt and covert expressions necessarily map directly onto narcissistic grandiosity and vulnerability (Pincus, Cain, & Wright, 2014). Nevertheless, entitlement and low agreeableness (e.g., antagonism) seem to be two major common aspects of these two proposed dimensions, and there is evidence of substantial consensus among clinicians, researchers, and laypeople about the prototypical grandiose and antagonistic narcissist (Miller, Lynam, Hyatt, & Campbell, 2017). The concept of narcissistic vulnerability and its proposed components (e.g. neuroticism and negative affectivity), however, pose a greater challenge to reaching a consensus about the existence of a unified concept of narcissism and about its nature, particularly in terms of the relationship between vulnerability and grandiosity. For example, the extent to which certain features should be considered “central” to narcissism (i.e., would be expected to exist “uniformly” across highly narcissistic individuals) or “peripheral” to narcissism (i.e. could be used as “diagnostic specifiers”) is still under debate. Evidence exists, however, for a trifurcated model, where NPD, grandiosity, and vulnerability represent combinations of two or all of three basic personality dimensions — agentic extraversion, antagonism, and neuroticism. While all three dimensions are thought to underlie NPD, grandiosity reflects the combination of antagonism and agentic extraversion, and vulnerability stems from the combination of antagonism and neuroticism (Miller et al., 2017). Such a model suggests that interpersonal antagonism is an aspect more central to NPD while neuroticism and agentic extraversion are specifiers for vulnerable and grandiose manifestations, respectively (Miller et al., 2017).

It has been argued that these two facets of narcissism do not represent distinct subtypes, but rather coexist within many individuals to varying degrees. Individual differences in the relative levels of grandiosity and vulnerability within people high in total narcissism not only can be linked to differential behavioral and affective expression of core narcissism, but also may be a prime metric for distinguishing narcissistic patients for clinical purposes (Pincus & Lukowitsky, 2010; Pincus, Cain, & Wright, 2014). Narcissistic patients with high vulnerability are also more likely to seek treatment; therefore, clinicians are more likely to deal with high levels of this aspect of narcissism than with grandiosity (Pincus & Lukowitsky, 2010); however, the association of both narcissistic facets with interpersonal antagonism means that high levels of grandiosity, vulnerability, or both can lead to social impairments (Pincus, Cain, & Wright, 2014; Czarna, Zajenkowski, Maciantowicz, & Szymaniak, 2019; Edershile & Wright, 2020).

Although narcissism is prototypically associated with interpersonal and social dysfunction, it also impacts areas such as emotional functioning. Narcissism in general has been linked to maladaptive emotional responses, such as shame, to stressors (Pincus & Lukowitsky, 2010). Grandiose narcissism has been linked to strategies such as using self-enhancing humor and grandiose fantasies to regulate self-esteem and affect (John & Robins, 1994; Pincus, Cain, & Wright, 2014). Vulnerability, by contrast, has been linked to depression, hyper-sensitivity, suicidality, and shame in the face of narcissistic injury (Pincus & Lukowitsky, 2010; Pincus, Cain, & Wright, 2014).

However, refining approaches to the diagnosis and treatment of narcissism therefore relies on resolving the question of whether a unified narcissistic construct exists. A major hurdle in reaching such a resolution is to find out which, if any, fundamental narcissistic components are shared between the vulnerable and grandiose manifestations of narcissism. Evidence is mounting that suggests interpersonal antagonism as a bridge between narcissistic vulnerability and grandiosity (Miller et al., 2017; Vize et al., 2019). Thus, narcissistic vulnerability and grandiosity may drive similar patterns of affect in certain situations in which interpersonal antagonism is a particularly salient factor.

Hostility in response to rejection is one of the prime candidates for overlap, as it has been long associated with both grandiose and vulnerable narcissism (Baumeister & Bushman, 1998; Baumeister, Bushman, & Campbell, 2000; Pincus, Cain, & Wright, 2014; Krizan & Johar, 2015). Individuals with high self-esteem tend to cluster into relatively homogeneous groups on the high and low ends of the spectrum of hostile tendencies according to the stability of their self-esteem (Baumeister et al., 2000). Individuals with high but unstable self-esteem, a profile tightly linked to trait narcissism, comprise the group in the high range of trait hostility. Indeed, a key unifying feature of the narcissistic construct may be the pattern in which dysregulation of one’s self-concept accompanies changes in one’s affect, social cognition, and interpersonal behaviors (Wright et al., 2017; Krizan & Johar, 2015).

Samples biased towards narcissistic vulnerability may lead to an over-emphasis on its role in the construct of narcissism (Miller et al., 2017) and may skew structural models and analysis of its outcomes (e.g. interpersonal and professional functioning) and correlates (e.g. basic personality domains such as neuroticism). Vulnerability is found at much higher rates in clinical settings, which are common sources of research samples that study narcissism, and results of studies (and resultant theories) in such samples may conflict with those conducted in community samples (e.g., Bushman & Baumeister, 1998; Cale & Lilienfeld, 2006). Moreover, the high co-occurrence of other personality disorders in clinical samples may jeopardize generalizability to more grandiose manifestations of narcissism, which are more commonly found in community samples (Miller et al., 2017). Conversely, other sampling approaches may lead to inadequate representation of vulnerable narcissism (e.g., recruiting on the basis of DSM-5 NPD diagnosis or other instruments that emphasize grandiosity) and/or of the range of narcissistic severity (e.g. categorical classifications that do not reflect the dimensionality of narcissism; c.f. Aslinger et al., 2018). ² In this study, we leverage the wide range of narcissistic severity and manifestation expected in a combined clinical-community sample to conduct an exploratory investigation of the trajectories of hostility leading up to and following rejection and their relationship to Five Factor Narcissism Inventory (FFNI)-based facets of narcissistic vulnerability and grandiosity (Glover, Miller, Lynam, Crego, & Widiger, 2012), as measured by the Revised NEO Personality Inventory (Costa & McCrae, 1992).

A shared antagonistic component of narcissistic grandiosity and vulnerability would be expected to drive similar trajectories of hostility leading up to and/or in response to rejection. Results from several cross-sectional studies have suggested there may be distinct pathways by which vulnerable and grandiose traits relate to hostility. For instance, one study found that entitlement explained the association between grandiosity and reactive aggression while angry rumination accounted for vulnerability’s relationship to the same outcome (Krizan & Johar, 2015). In another study, neuroticism explained much of narcissistic vulnerability’s relationship to hostility while grandiosity exhibited a protective effect; however, the authors also found that grandiosity positively associated with “dispositional” hostility after adjusting for emotional stability and posited that some unifying feature may exist between the narcissistic expressions with respect to hostility (Czarna, Zajenkowski, Maciantowicz, & Szymaniak, 2019).

Antagonism has been suggested as the “common core” of vulnerable and grandiose narcissism (Miller et al., 2017), and empirical studies have found evidence of the key role of antagonism in links between narcissistic facets and aggression (Vize et al., 2019). Nevertheless, while certain aspects (e.g., antagonism, entitlement) shared by vulnerability and grandiosity may drive a relationship to hostility for both facets, personality features specific to grandiosity may enable grandiose individuals to cope with their emotional reactions to ego threats while interpersonal styles and traits (e.g. tendency to social withdrawal, neuroticism) typically found among vulnerably narcissistic individuals contribute to persistence of negative emotionality (John & Robins, 1994; Pincus, Cain, & Wright, 2014; Czarna, Zajenkowski, Maciantowicz, & Szymaniak, 2019; Edershile & Wright, 2020).

Thus, both theory and past empirical results give reason to expect that grandiosity and vulnerability may associate with different trajectories of hostility before and after rejection, and that such patterns could contribute to the taxonomic discussions surrounding narcissistic traits. If antagonism is subsumed by grandiosity rather than being an area of overlap between grandiosity and vulnerability, then vulnerability would not be expected to predict the same trajectories of hostility as would grandiosity. Unique components of grandiosity and vulnerability would be indicated by each explaining unique variance in hostility profiles, either in the same or opposite directions. Any divergent relationships between grandiosity and vulnerability with momentary hostility could be demonstrated by a difference in their predictions of pre- and/or post-rejection trajectories. For instance, the agentic extraversion associated with narcissistic grandiosity may contribute toward a faster return to baseline while the neuroticism associated with narcissistic vulnerability may contribute to a slower recovery from rejection hostility. Post-rejection emotion regulation strategies that mitigate hostility, such as seeking a different person to provide validation, may be more available to agentically extraverted individuals than to neurotic individuals high in narcissism (John & Robins, 1994; Pincus, Cain, & Wright, 2014; Czarna, Zajenkowski, Maciantowicz, & Szymaniak, 2019; Edershile & Wright, 2020). In general, to the extent that interpersonal antagonism is largely represented by grandiose narcissism, then vulnerability should account for relatively less unique variance in hostility profiles surrounding rejection events.

Hypotheses

1. Grandiosity and vulnerability will be associated with higher levels of hostility upon the occasion of rejection (i.e. the intercept of the model, with time centered on rejection).

2. Grandiosity will be associated with a faster return to baseline hostility following rejection.

3. Vulnerability will be associated with a slower return to baseline hostility following rejection.

Exploratory Questions

Do vulnerability and grandiosity interact in predicting mean levels and trajectories of post-rejection hostility? We did not have specific a priori hypotheses regarding the impact of vulnerability and grandiosity on hostility trajectories leading up to rejection reports.

Methods

Sample

The current sample combined two archival datasets, which together included 116 participants, 56 of whom were individuals who met criteria for borderline personality disorder (BPD) diagnosis and who specifically endorsed the affective instability criterion (APA, 2000). As the data were originally collected to investigate emotion dysregulation and alcohol use among drinkers with and without BPD, we required this feature in order to ensure those in the BPD group were characterized by emotion dysregulation (Trull et al., 2008), a phenomenon also strongly linked to the nomological connections between vulnerable narcissism and BPD (Miller et al., 2010). The remaining 60 were a group of community controls (COM) who did not meet the syndromal requirements for BPD nor endorse affective instability. BPD participants were recruited from local psychiatric outpatient clinics and the general community. In clinics, potential participants were made aware of the study through flyers in the waiting rooms and from their assigned therapists or doctors. In addition, advertisements were placed in a local weekly advertisement circular, on bulletin boards of local businesses, and online via Craigslist and a weekly university-wide email announcement. Advertisements targeted individuals experiencing symptoms associated with BPD (i.e., intense mood swings, impulsive behavior, unstable relationships, and intense anger).

Community controls (COM) were recruited through advertisements which included no language regarding BPD symptoms. No restrictions were placed on controls regarding treatment or presence of psychiatric diagnoses, other than those regarding absence of BPD, affective instability, and general exclusion criteria noted below. Potential participants who were recruited from outpatient clinics or through BPD targeted advertisements could not be considered for the control group, even if they proved to be ineligible for the BPD group.

After either completing an initial participation form, which was returned to research staff, or contacting research staff directly, participants were briefly screened over the phone for presence of BPD features and absence of exclusionary criteria. If participants appeared eligible, they were next brought in for a face-to-face screening diagnostic interview. Based on research questions informing the collection of the original sample (Carpenter et al., 2017; Lane, Carpenter, Sher, & Trull, 2016), to be eligible, participants were required to be between the ages 18 and 45 and report alcohol consumption on average at least once a week (or four times over the previous month). They were excluded if they were currently seeking treatment or interested in seeking treatment for alcohol use and/or problems, or if they reported significant unsuccessful efforts to cut down or stop using alcohol or physiological withdrawal symptoms when not using alcohol over the past year. Participants were also excluded if they reported current psychosis, intellectual disability, severe neurological dysfunction, or history of head trauma that affected mood or concentration. If female, participants had to report not being pregnant or not planning to become pregnant. Eligible participants were scheduled for an orientation session. All potential participants who completed the screening interview were paid $20.

The overall sample was predominantly women (78.4%), of Caucasian ethnicity (84.5%), with an average age of 26.4 years (SD = 7.1). A majority of participants were single/never married (68.7%) or currently married (20.0%), did not have any children (74.1%), were employed (78.5%), and had an annual income less than $50,000 (76.7%). All of the BPD participants were currently in treatment and most (73.2%) were taking psychotropic medication (e.g. antipsychotics, mood stabilizers, stimulants, anxiolytics, depressants, hypnotics, anticonvulsants). In contrast, only 2 (3.3%) COM participants were currently receiving treatment for conditions other than BPD and 5 (8.3%) were taking psychotropic medication. BPD and COM participants did not differ significantly in terms of gender, age, ethnicity, relationship status, number of children, or employment status (all ps > .109). The BPD group was more likely than the COM group to have an income less than $25,000 (χ²(4) = 16.72, p = .002). The BPD group also had a higher proportion of individuals who met diagnostic criteria for current anxiety, mood, eating, substance use (alcohol, drug), and personality disorders.

A dataset with approximately equal clinical and community groups can be expected to capture a fuller range of the general narcissism spectrum than would be found in a community or clinical sample alone. BPD and NPD co-occur nationally in men at a rate of 47% and in women at a rate of 32% (Grant et al., 2008). Additionally, because DSM-5 NPD criteria capture grandiose narcissism, mainly, while BPD overlaps more with vulnerable narcissism, a sample including individuals with a BPD diagnosis could be expected to capture manifestations (i.e., vulnerable narcissism) that an NPD sample would likely under-represent. Indeed, BPD has been posited as a closer taxonomic relative of vulnerable narcissism than is NPD: BPD and vulnerable narcissism have been found to demonstrate strongly similar personality profiles (r = 0.94; Miller et al., 2010) and are both associated with externalizing responses to emotionally dysregulated reactions to interpersonal situations (Miller et al., 2011).

Procedure

Diagnostic information was obtained from two semi-structured interviews, Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I; First, Spitzer, Gibbons, & Williams, 1995) and Structured Interview for DSM-IV Personality (SIDP-IV; Pfohl, Blum, & Zimmerman, 1994), to assess Axis I and II disorders, respectively. Inter-rater reliability was computed for 20 participants in the larger sample by having a second trained interviewer rate videotapes of interviews; reliability was excellent for the diagnosis of BPD (κ = 1.00) and for presence of affective instability (κ = .89).

At the initial orientation session, eligible participants first completed a battery of self-report questionnaires. Participants were then issued an electronic diary (ED; Palm Tungsten E2© handheld computer) programmed with customized software that they carried for approximately 21 days (M = 21.6; SD = 2.1, Range = 7-25). Two participants dropped out of the study early, but completed at least 7 days of the ED protocol, and so are included in the present analyses. The other 114 participants each completed at least 18 days. Participants returned to the lab weekly for data downloads and to receive payment. They were paid $50 at each visit for adequate compliance (i.e. at least 80%) of the previous week’s random prompts; however, payment was reduced by $10 for every 10% below 80%. During the last visit, participants also completed a follow-up self-report battery for which they received $10. Thus, the maximum payment per participant was $180 (initial screening, three weeks of data collection, and follow-up).

Electronic Diary Protocol

The current protocol was a modified version of protocols from previous studies using different samples (see Piasecki et al., 2012; Trull et al., 2008). Participants completed seven different types of reports over the course of the study (morning reports, random prompts, initial drink reports, drinking follow-ups, initial cigarette reports, initial self-harm reports, self-harm follow-ups; for more details see, Lane et al., 2016). However, only random prompts, which were randomly interspersed across six approximately two-hour intervals within a given day, starting after an initial morning report or 12:00PM, whichever came first, contained items pertaining to whether individuals experienced interpersonal rejection. These prompts were used to center the daily trajectories of affect, based on whether a rejection was reported or not. Reports of affect, which included hostility items, were collected at all prompt types.

The compliance in this sample was high, with an average completion rate of 90.26% of random prompts, 92.95% of follow-up prompts, and 93.28% of morning reports. On average participants completed 137.0 reports per person. A majority of the prompts were random (60.2%), followed by self-initiated or follow-ups (25.7%) and morning (14.1%). A majority of all self-initiated and follow-up reports were alcohol reports (78.7%), followed by cigarette reports (19.7%) and self-harm (1.6%). The total number of prompts included in the analyses was 15,889 over 2,508 person-days.

Measures

Narcissistic Grandiosity and Vulnerability.

At the between-person level, vulnerability and grandiosity were estimated using the framework of the Five Factor Narcissism Inventory (FFNI), which conceptualizes narcissism within the Five Factor Model (FFM) of personality (Glover et al., 2012). FFM facets were measured using the Revised NEO Personality Inventory (Costa & McCrae, 1992; see supplemental materials, Table S4 for zero-order correlations among the facets). Vulnerability was measured by the mean of the “self-consciousness,” “angry hostility,” and “stress vulnerability” facets of NEO neuroticism. Grandiosity was measured using facets of NEO extraversion (i.e. “gregariousness,” “assertiveness,” and “excitement-seeking”), the “fantasy” facet of NEO openness, the “achievement striving” facet of NEO conscientiousness, and reverse-coded “trust,” “straight-forwardness,” “altruism,” “modesty,” and “tendermindedness” facets of NEO agreeableness (i.e. antagonism). Both subscales were centered and standardized based on the mean of person-means (i.e. the grand mean). Table 1 contains descriptive statistics for group for narcissistic grandiosity and vulnerability. BPD and community participants differed significantly on grandiosity and vulnerability. Grandiosity scores of BPD participants (M = 15.73, SD = 2.09) were significantly higher than those of community participants (M = 14.90, SD = 1.51) (b_Group = 0.83, p < 0.05; Cohen’s D = 0.44). Vulnerability scores of BPD participants (M = 20.26, SD = 4.26) were significantly higher than those of community participants (M = 12.14, SD = 3.79) (b_Group = 8.12, p < 0.0001; Cohen’s D = 2.02). Moreover, vulnerability and grandiosity did not correlate significantly in the overall sample (r = 0.14, p = 0.12), COM (r = 0.08, p = 0.52), or BPD (r = −0.09, p = 0.53) groups.

Table 1:

Descriptive statistics of narcissistic grandiosity and vulnerability differences by group

Vulnerability	N	Mean	SD	SE	Minimum	Maximum	df	t-value	p
COM	60	12.14	3.79	0.49	4.33	25.33
BPD	56	20.26	4.26	0.57	10.67	30.33
Difference		−8.12	4.03	0.75			114	−10.81	<.0001
Grandiosity	N	Mean	SD	SE	Minimum	Maximum	df	t-value	p
COM	60	14.90	1.51	0.19	12.10	18.90
BPD	56	15.73	2.10	0.28	10.89	20.20
Difference		−0.83	1.82	0.34			114	−2.43	0.017

Note. BPD=Borderline Personality Disorder. COM=Community Sample.

Hostility.

The emotion used as the outcome variable was “hostility,” a composite measure of six specific emotion items (angry, irritable, hostile, loathing, scornful, disgusted) out of the 21 emotions contained in the PANAS-X (Watson & Clark, 1999). The PANAS was administered at all prompts such that individuals were asked to report on how they were feeling “in the past 15 minutes”. The average of the hostility person-means of BPD participants (M = 1.41, SD = 0.46) was significantly higher than the average of community participants’ (M = 1.09, SD = 0.14) person-means (t = 5.117, p < 0.001, df = 64.418; Cohen’s d = 0.64). The overall sample had a mean hostility of 1.24 (SD = 0.37). The reliabilities for the hostility composites for each group ranged from acceptable to excellent for a) person averages across the assessment period (Rkfs > .99), b) ratings for any single time point (Rifs > .84), and most importantly, c) changes in moment to moment ratings (Rcs > .72; Shrout & Lane, 2012).

Rejection.

At each random prompt (~every 2 hours), participants reported whether they had, “since [their] last prompt, felt rejected.” On average, participants experienced 4.04 rejections over the course of the study (SD = 5.68), with 71.6% of participants and an average of 13.28% of participant-days having at least one reported rejection (18.6% among those with at least one rejection day). For participants in the BPD group, an average of 6.41 rejections were reported (SD = 6.93), and for those in the community group, an average of 1.83 rejections were reported (SD = 2.78). In the BPD group, rejections occurred on 20.99% of days on average and in the community group rejections occurred on 6.73% of days on average. BPD and community groups differed in the average number of rejections reported over the course of the study (t = 4.58, p < 0.001, df = 76.263).

Data Processing and Quality Checks

Because the NEO was administered at both the beginning and end of the study, one community participant’s missing pre-study NEO score was replaced with the participant’s corresponding post-study NEO score. The rate of missingness was low (0.9%) and these scores would not be expected to have changed substantially in less than a month’s time, so we were confident that these replacements would not introduce bias to the composite vulnerability and grandiosity scoring. Moreover, for participants with data available for both, the correlations between the pre- and post-study grandiosity and vulnerability scores were 0.84 and 0.92, respectively.

Analyses

Daily trajectories of PANAS hostility before and after experiencing a rejection were modeled in multilevel linear spline regressions using the lme4 package in R (Bates, Mächler, Bolker, & Walker, 2014) and visualizations were created in Python 3.7 using the Matplotlib and Seaborn packages. Unlike purely piecewise functions, this method allows the initial and spline trajectories to correlate. A linear model was chosen because the shape of the graph, which uses a nonparametric LOESS fit of the data, suggests that a linear term would reasonably model the data given our research question.³

On days in which rejection occurred, the continuous time variable was centered on the first within-day occasion of that event.⁴ For days without rejections, time was centered on the person-specific mean time of day at which the participant reported being around other people over the course of the study — their “typical interaction times.” These non-event centered time points closely resembled those of the event days. As they tended to fall near the middle of the day, this centering scheme offered the additional benefit of maximizing “pre” (ascending limb) and “post” (descending limb) data. On average, 3.29 measurement occasions fell in the ascending limb and 3.05 measurements in the descending limb, and participants provided an average of 6.34 total daily measurements.

The time variables combined rejection and non-rejection days. A daily rejection/non-rejection dummy-coded variable, with rejection occurrence as the reference category, served to distinguish between rejection and non-rejection days. We compare these effects to parallel profiles on days where rejection does not occur to test if they represent more situationally dependent, temporally acute profiles or chronic profiles. Spline variables had zeros entered in for pre-rejection measurement occasions but had identical data to their corresponding time variables for post-rejection measurement occasions. These splines reflected time elapsed since rejection (or on non-rejection days, since the person’s typical interaction time), and their coefficients therefore represented the post-rejection deviation from (i.e. moderation of) the initial trajectory. The spline trajectory therefore represents, in essence, a moderation of the initial trajectory, (i.e. a time by “post” phase interaction.)

Model Specification

Predictors in the full model included a “no rejection” dummy coded variable, NEO vulnerability and grandiosity variables (γ₀₂, γ₀₃), and time variables for “expected trajectory” in the ascending limb and the spline (deviation from the ascending limb after entering the descending limb). The “no rejection” coefficients indicated the expected difference in intercepts (or the expected difference in estimated trajectories or effects, in the case of interactions) on days in which rejection did not occur.⁵ The ascending limb represented the trajectory of hostility across time up until the occurrence of rejection. The spline represented the deviation from that trajectory after the rejection. These predictors were included as main fixed effects (γ₁₀, γ₂₀). Also included in the model were the two through four-way interactions among them.

To adjust for mean level differences between the BPD and community groups, a dummy coded “group” variable, with the community group as the reference category, was included as a main effect (γ₀₁) in the model. A LOESS smooth showed a vertical shift in trajectories in the BPD versus the community sample for both hostility and positive affect. A simple vertical shift would indicate that slopes should not differ substantially as a function of BPD/community group. Most interactions by group were therefore excluded from the model to preserve power; however, like for the other effects in the model, a Group X No Rejection interaction was included to account for the possibility that groups might differ more or less on rejection days compared to on non-rejection days.

Random effects included random intercepts for person (u_0i) and person-day (u_0i(j)) as well as random slopes for the time (i.e. trajectory) and spline variables (β_1i and β_2i). Random slopes and intercepts were constrained to be uncorrelated in the presented model, as initial analyses allowing for correlated slopes did not improve model fit or result in differential interpretations/inferences of model results.

The equations below represent the overall model, excluding effects and interactions with the person-day level “No Rejection” variable for simplicity.

Level 1

$$y_{ijk}={\beta }_{0i(j)}+{\beta }_{1i}trajector{y}_{ij}+{\beta }_{2i}splin{e}_{ij}+{ϵ}_{ijk}$$

Level 2

$${\beta }_{0i(j)}={\gamma }_{00j}+u_{0i(j)}$$

Level 3

$$\begin{gathered} {\beta }_{0i}={\gamma }_{00}+{\gamma }_{01}Grou{p}_i+{\gamma }_{02}Vulnerabilit{y}_i+{\gamma }_{03}Grandiosit{y}_i+{\gamma }_{04}(Vulnerabilit{y}_i)(Grandiosit{y}_i)+u_{0i} \\ {\beta }_{1i}={\gamma }_{10}+{\gamma }_{11}Vulnerabilit{y}_i+{\gamma }_{12}Grandiosit{y}_i+{\gamma }_{13}(Vulnerabilit{y}_i)Grandiosit{y}_i+u_{1i} \\ {\beta }_{2i}={\gamma }_{20}+{\gamma }_{21}Vulnerabilit{y}_i+{\gamma }_{22}Grandiosit{y}_i+{\gamma }_{23}(Vulnerbilit{y}_i)(Grandiosit{y}_i)+u_{2i} \end{gathered}$$

Results

On rejection days, hostility upon rejection was 1.35 on the 5-point Likert scale on average for the community group and 1.65 for the BPD group, a statistically significant difference (b = 0.31, p = 0.001; Tables 2 and S1). On non-rejection days, hostility at “typical interaction time” was 1.16 on average, which was statistically lower than hostility at the time of rejection on rejection days (b = −0.192, p < 0.001; Table S1).

Table 2:

Summary of Results for Hostility Before and After Rejection

	Estimate	SE	t-value	p-value
Intercept	1.347	0.06	22.90	0.000***
BPD	0.306	0.09	3.276	0.001**
Grandiosity	0.086	0.04	2.35	0.020*
Vulnerability	0.055	0.05	1.15	0.252
Time	0.037	0.00	12.19	0.000***
Spline	−0.056	0.01	−10.94	0.000***
Grandiosity X Vulnerability	−0.056	0.03	−1.68	0.095
Grandiosity X Time	0.006	0.00	2.16	0.031*
Grandiosity X Spline	−0.019	0.01	−3.83	0.000***
Vulnerability X Time	0.010	0.00	3.15	0.002**
Vulnerability X Spline	−0.002	0.01	−0.48	0.633
Grandiosity X Vulnerability X Time	−0.001	0.00	−0.35	0.725
Grandiosity X Vulnerability X Spline	0.002	0.00	0.52	0.601

Note. Significance:

^***p < 0.001,

^**p < 0.01,

^*p < 0.05.

BPD = Borderline Personality Disorder. Terms in bold obtained statistical significance after a familywise adjustment using the Benjamini and Hochberg False Discovery Rate adjustment (1995).

Hostility increased (b = 0.037, p < 0.001; Table 2) leading up to rejection. The rejection-day positive slope for hostility was overcome by the effect of the spline (b = −0.056, p < 0.001; Table 2), indicating that hostility decreased after rejection. This pattern of escalation and decline did not occur on non-rejection days (refer to Supplementary Materials) and was not the result of a peak discontinuity at the occasion of a rejection report, as indicated in our initial modeling fitting of the overall trajectories.

Vulnerability did not significantly predict hostility upon rejection (i.e. the intercept).⁶ It did, however, predict a steeper positive trajectory of hostility leading up to rejection (b = 0.01, p = 0.002; Table 2). Grandiosity predicted higher levels of hostility at rejection (b = 0.086, p = 0.02; Table 2), as well as a steeper increase in hostility (b = 0.006, p = 0.031; Table 2) leading up to rejection. Grandiosity also predicted a faster decline in hostility after rejection (b = −0.019, p < 0.001; Table 2). For example, on a rejection day, a typical participant’s hostility would be expected to increase at a rate of approximately 0.037 (points per hour) leading up to rejection and would be expected to decrease at a rate of approximately −0.056 following rejection (Table 2). By contrast, the hostility of a participant one standard deviation above mean grandiosity would be expected to increase at a rate of approximately 0.043 leading up to rejection and would be expected to decrease at a rate of approximately −0.118 following rejection (Table 2).

On days without a rejection, grandiosity very slightly attenuated the increase in hostility leading up to the “typical interaction” time (b = −0.007, p = 0.015; Table 2); however, it also positively moderated the spline (b = 0.023, p < 0.001; Table 2). On days without a rejection, more grandiose participants were less hostile at the “typical interaction” time (b = −0.061, p < 0.01). On a non-rejection day, both typical and grandiose (+ 1 SD) participants would be expected to have a nearly flat trajectory (see Figure S1). Grandiosity and vulnerability did not display a significant interaction, and their interaction did not significantly moderate pre- or post-rejection trajectories.

Discussion

Grandiosity and vulnerability were associated with similar trajectories of hostility leading up to rejection, but their effects differed at the time of rejection and following rejection. Grandiosity, but not vulnerability, predicted a higher “spike” in hostility at the occasion of rejection. Both aspects of narcissism were associated with a steeper trajectory of pre-rejection hostility. Vulnerability, however, did not affect the spline (i.e. the post-rejection deviation from the initial trajectory). Nevertheless, although vulnerable individuals may experience the same deviation from their initial trajectory, their initial trajectory would be steeper to start. Thus, vulnerability would predict a slower return to baseline following rejection. Grandiosity, by contrast, predicted a faster decline in hostility following rejection, suggesting that more grandiose individuals may be better able to recover from the accumulation of hostile feelings they experienced before and upon rejection.

Grandiosity may lead people to react with hostility when they anticipate an impending rejection or experience the types of interpersonal situations which precipitate rejection. Likewise, high grandiosity individuals’ antagonistic behaviors may themselves lead to such rejections. However, higher grandiose experience/expression may buffer against the kind of self-consciousness that leads others to ruminate and become more hostile after a rejection. Moreover, several adaptive traits were included in our measure of grandiosity. Individuals high in traits such as “assertiveness,” “gregariousness,” or “achievement striving”, characteristic of extraversion, may have more social or emotional ability or motivation to overcome rejection or to seek approval elsewhere. By contrast, higher levels of vulnerability may lead narcissistic individuals to withdraw from social interaction following rejection.

Antagonistic individuals (whatever their levels of vulnerability and grandiosity) may not be as invested in the opinions of others, may tend to discount them, or may be more easily able to divest from relationships. Narcissistic grandiosity and vulnerability thus may share an association with a hostile dismissiveness of the opinions of others. Opinions of others may be ignored as a defense against a vulnerable person’s adverse affective responses. Even in the case of certain traits buffering against those affective responses, the rejection inherently conflicts with the narcissistic self-concept, which is typified by an inflated sense of self-worth and unrealistic sense of entitlement. Though others may reject these unrealistic views and expectations, their reactions may not be able to penetrate a narcissistic individual’s grandiose delusions. Emotions that lead to antagonistic behaviors can precede and even precipitate the experience of rejection, driven by anticipation of rejection and self-enhancement motives among grandiose individuals (Gore & Widiger, 2016; Miller et al., 2011) as well as hostile attribution bias and emotional lability among narcissistically vulnerable individuals (Dickinson & Pincus, 2003; Miller et al., 2011). Whatever the source or purpose, this sequestration from the thoughts and feelings of others may underlie narcissistic interpersonal impairment. Affective dysregulation and deficits in empathy could jointly hinder the incorporation of important social information into behavioral planning.

To explore this possibility, we included explicit antagonism as a covariate, and also as an additional focal variable of interest that accounts for possible shared patterns in vulnerability and grandiosity effects. For the latter, we included antagonism, measured using the reverse-coded NEO-PI “agreeableness” dimension score, as a variable that could moderate typical trajectories both leading up to and following rejection. Results provided suggestive evidence for antagonism as an explanatory bridge such that certain effects of vulnerability and grandiosity were reduced to non-significance while antagonism was statistically significant. However, these constructs were all highly correlated with one another, creating a high probability of a partialling problem regarding interpretability and power (Lane & Hennes, 2019; Lynam, Hoyle, & Newman, 2006). While consistent with our conceptual framework, more controlled (experimental) or dynamic (personality) designs are necessary to more definitively support an antagonism bridge mechanism between vulnerable and grandiose narcissism.⁷

Thus, the affective antecedents of rejection and related interpersonal impairments may be common to narcissistic vulnerability and grandiosity. Narcissism has been associated with responses to rejection such as denigrating other people (Morf & Rhodewalt, 1993) or evaluating sources of negative feedback as less competent than sources of positive feedback (Kernis & Sun, 1994); however, further investigation is needed to explore whether the buffering effect of grandiosity contributes to different behavioral responses to rejection than does narcissistic vulnerability. For instance, grandiosity is associated with approach behaviors, such as seeking out validation using self-enhancing humor (John & Robins, 1994), while vulnerability is more strongly associated with withdrawal (Miller et al., 2017; Dickinson, K. A., & Pincus, 2003). The impairments in emotion regulation and resultant social isolation of more vulnerable individuals may contribute to the higher rates of narcissistically vulnerable patients seeking out voluntary treatment.

Individuals higher in grandiosity, by contrast, may cope after the experience of interpersonal stressors (however maladaptively) and not see the need for treatment. Like vulnerability, however, grandiosity may be associated with maladaptive affective patterns which lead up to those rejections. Moreover, the effects of grandiosity on post-rejection decline in hostility may not fully overcome the consequences of the initial rise in hostility. Thus, even narcissistic individuals low in vulnerability could be expected to experience substantial social and emotional impairment. An acute rise in a person’s hostility also carries the potential to lead to harm to others, whatever the person’s own affective response. Indeed, the seeking out of social contact (namely, praise and admiration), which is associated with narcissistic grandiosity, may enhance the risk of harm to others compared to a withdrawal response.

Limitations

Nevertheless, our data do not allow a rigorous investigation of causal processes. Future research should employ experimental methods to investigate whether grandiosity and vulnerability differentially influence behavioral responses or is associated with behavioral or affective antecedents that make rejection more likely. Another limitation of our study was that the sample size limited our ability to include interactions by group for all terms. It is possible that more adaptive forms of grandiosity exist in the community sample than in the BPD sample, which could lead to different sizes or directions of effects in each group; however, groups’ trajectories of raw hostility did not qualitatively differ upon visualization using LOESS smooth (see Figure S1 in the Supplementary materials).

While grandiosity scores in the community group could have been driven by adaptive traits included in the measure, scores constructed without potentially adaptive items correlated with the original grandiosity scores at 0.76, 0.78, and 0.75 in the full, BPD, and community samples, respectively.⁸ Thus, potentially adaptive items did not show signs of systematically affecting the results. Moreover, we found evidence that even with the adaptive items, the scales used in the present analyses captured maladaptive grandiosity in both groups. In both sub-samples, grandiosity and antagonism correlated at 0.55 and had non-significant negative correlations with conscientiousness (a domain from which potentially adaptive traits were drawn). Thus, while the grandiosity measure includes some traits that can be expressed adaptively, the composite grandiosity scores’ correlational patterns show signs that it captures maladaptive characteristics associated with narcissism (i.e. antagonism) in both samples. These results are consistent with past evidence and theory of narcissism as arising from a combination of traits, some of which can be adaptive in isolation or in certain contexts (Holtzman & Strub, 2011). Nevertheless, more research is needed into the nature and measurement of adaptive and non-adaptive narcissism.

Generalizability could be limited by the high proportion of women to men. Some studies have found gender differences in comorbidity patterns, with women more likely to have comorbid internalizing disorders such as Post-Traumatic Stress Disorder (PTSD) and eating pathology and men more likely to have comorbid externalizing disorders and personality pathology, such as substance use disorders, intermittent explosive disorder, antisocial PD, and narcissistic PD (Johnson et al., 2003). Women may be more likely to internalize and/or manifest their narcissism vulnerably while men may be more likely to externalize and/or manifest their narcissism grandiosely or with more overt hostility.

The inclusion of a main effect for BPD diagnosis influenced whether a significant main effect for vulnerability was present — when the BPD and BPD × No Rejection terms were removed from the model, vulnerability was associated with higher hostility at the average daily interaction time (i.e., t = 0; b = 0.151, p < 0.001); however, unlike for grandiosity, this effect did not significantly differ on rejection versus non-rejection days (b = 0.013, p = 0.602), suggesting a similar conclusion to the original results (i.e., that grandiosity uniquely accounts for acute reaction to rejection). The influence of high vulnerability and grandiosity on trajectories for participants with and without BPD strongly resembled each other (see supplemental materials, Figure S2). Some place BPD and vulnerable narcissism together under a vulnerably-antagonistic umbrella: the personality profile similarities between BPD and vulnerable narcissism (r = 0.94; Miller et al., 2010) have been specifically linked to hostile attribution bias, a key mechanism by which interpersonal hostility is thought to occur among vulnerably-antagonistic individuals (Miller et al., 2010; Miller et al., 2011). Taken together, while the inclusion of participants with BPD remains an important consideration in the interpretation of our results, there is reason to believe that our results were not systematically influenced in a way that would have led to different conclusions.

A final consideration is that our study only had an average of six measurements per day, which leaves few data points on either side of the zero time-point. Moreover, rejections tended to occur later in the day, around 3 to 5 pm, and post-rejection was estimated solely from data after those points; therefore, pre-rejection hostility may have been more reliably measured than post-rejection hostility.

Future Directions

Our study only examined trait-level measures of vulnerability and grandiosity; however, some have called into question this static conceptualization of narcissism as divisible into static, between-person differences in trait grandiosity and vulnerability, and more dynamic within-person changes. One recent study found a daily process component, where grandiosely narcissistic individuals exhibited lower state narcissism on stressful days and higher state narcissism on days in which they experienced more agentic events, such as holding power over someone or having an important role in a group, as well as on days in which they experienced more communal events, such as caring for someone or giving a gift (Giacomin & Jordan, 2016). Looking at within-person interpersonal processes and fluctuations in vulnerability and grandiosity across time may lend insights into how the two relate and differ.

Investigating all the PANAS subscales such as fear, sadness, and positive and negative affect more generally in a multivariate framework across different interpersonal interactions such as receiving compliments or disagreeing may flesh out the patterns and processes (Hepp et al., 2017, 2018). For instance, people high in vulnerable narcissism, rather than non-antagonistic vulnerability, might be expected to experience more specific negative affect (e.g., depression, anxiety) than more grandiose individuals but also to experience hostility at similar levels to grandiose individuals. People high in both grandiose and vulnerable narcissism might manifest more of their grandiosity in interpersonally secure situations but behave more vulnerably or obsessively seek out compliments if facing a strong ego-threat. Investigating these processes in both a multivariate and vulnerable-grandiose fluctuation framework could be a particularly promising direction to explore the overlap and differences between narcissistic vulnerability and grandiosity as they unfold within-person and to distinguish between them.

Conclusion

Grandiosity and vulnerability independently predicted similar profiles in hostility leading up to rejection. One would expect grandiosity to provide the sole contribution to rejection hostility if it subsumed interpersonal antagonism. Thus, the conceptualization of interpersonal antagonism as a core component of narcissism unifying its vulnerable and grandiose aspects would be supported. Moreover, elevated hostility also often precedes rejection, suggesting a potentially reinforcing process between the two. The two facets differed, however, in that grandiosity uniquely predicted a greater rise in hostility at the time of rejection and a faster decline afterwards. Thus, while grandiosity may aggravate acute reactions, grandiosely narcissistic individuals may also engage in approach behaviors to regulate the narcissistic need for approval, thereby promoting faster affective recovery from rejection.

Figure 1.

Loess smooth of predicted hostility trajectories of participants with and without high (+1 standard deviation) scores on grandiosity (left) and vulnerability (right) before and after rejection