The Unified Narcissism Scale-Revised: Testing Incremental Validity and Shortening the Measure

Abstract

Narcissism has had a long history of conceptual and measurement confusion. In this paper, we aimed to assess the incremental and external validity of the Unified Narcissism Scale-Revised (UNS-R), and to determine a prototype short form of the measure that is invariant across cultures. In Study 1, we constructed a 15-item short form prototype that was scalar invariant across four countries (United States, China, Sri Lanka, and Australia). Using this short form, we found the Australian sample to be the most different from the other samples. We speculate this is due to Australia having a more horizontal culture, demonstrating resistance to hierarchy and a stronger endorsement of equality. In Study 2, we assessed the incremental and external validity of the UNS-R long and short form and found it to be a superior measure of grandiose narcissism in terms of strength and cogence of external correlations compared to existing measures, but the Five-Factor Narcissism Inventory-Short Form (FFNI-SF) was a better measure of vulnerable narcissism. In conclusion, we have illustrated the robustness of the UNS-R, and its short form, as a measure of narcissism and, in the process, highlighted important cross-cultural differences.

Introduction

Since its inception, the construct of narcissism has had an interesting and complex journey in its conceptualization, theory, and measurement. It has been theorized as a personality trait, a clinical disorder, and a motivational drive (Back et al., 2013; Pincus et al., 2009), and researchers have constructed measures that reflect specific theoretical perspectives on narcissism. For example, the Five-Factor Narcissism Inventory (FFNI) conceptualizes narcissism within a five-factor personality framework (Glover et al., 2012), whereas the Pathological Narcissism Inventory (PNI) captures specific presentations of pathological narcissism (Pincus et al., 2009). As a result, the literature boasts a multitude of measures of narcissism; however, none reflect the contemporary conceptualization of narcissism.

Narcissism is recognized as a multidimensional construct with grandiose and vulnerable dimensions and a shared core (Krizan & Herlache, 2018; Miller et al., 2017). Grandiose narcissism refers to an inflated sense of self, dominance in interpersonal situations, and exhibitionism, whereas vulnerable narcissism refers to low self-esteem, hypersensitivity to criticism and negative feedback, and a contingent sense of self (Krizan & Herlache, 2018; Pincus et al., 2009). Miller et al. (2017), who frame narcissism within the five-factor model, argue that antagonism is the shared core. Antagonism is a broad construct with related but separate dimensions that capture how one relates to another in antisocial ways and includes facets such as callousness, distrust, and arrogance (Crowe et al., 2018). Recent literature supports Krizan and Herlache (2018) theorizing of the specific facet of entitlement being central to narcissism over other facets of antagonism. Entitlement is the belief that one is special and therefore deserving of more than others (Campbell et al., 2004). Network analyses showed entitlement links grandiose with vulnerable narcissism (Dinić et al., 2021). Sivanathan et al. (2023) demonstrated that entitlement is the maintained characteristic of narcissism across cultures. Finally, process-based models of narcissism support entitlement as the underlying belief system driving other antagonistic tendencies in Western samples (Edershile & Wright, 2021).

The Unified Narcissism Scale-Revised (UNS-R; Sivanathan et al., 2023) reflects the current conceptualization of narcissism and further clarifies the theoretical understanding of the construct. The UNS-R captures narcissism with five first-order factors, two second-order factors, and a common third-order Narcissism factor (see Figure 1), and these findings have been replicated across cultures. The UNS-R’s first-order factors are: Leadership (a sense of superiority and overvaluation of oneself), Vanity (a strong focus on physical appearance), Contingent Self-Esteem (a sense of self that requires external validation), Grandiose Fantasy (preoccupation with fantasies of power, achievement, and ideal love), and Hiding One’s Needs (shame and fear of rejection about one’s inner world), with the first two factors representing grandiose narcissism and the latter three vulnerable narcissism. In contradiction to past research, the Grandiose Fantasy factor loaded onto Vulnerable rather than Grandiose Narcissism in the UNS-R (Pincus et al., 2009). In-depth discussions about this discrepancy can be found in Sivanathan et al. (2019, 2021) for interested readers.

Figure 1. Unified narcissism scale-revised factor structure.

Despite not having a separate entitlement factor, Sivanathan et al. (2023) demonstrated that the shared commonality of entitlement is captured throughout the scale and by the common third-order factor, which shows that both grandiose and vulnerable narcissism are located within a broad narcissism construct. Entitlement infuses the five dimensions in the following way: (a) Contingent Self-Esteem through being entitled to constant validation, admiration, and attention from others; (b) Leadership through being entitled to lead and have authority over others; (c) Grandiose Fantasy through being entitled to achieving great things and being admired in the future; (d) Vanity through being entitled to admiration for one’s physical features and appearance; and (e) Hiding One’s Needs through being entitled to acceptance and admiration despite pretending to be someone else and concealing one’s own true individuality. Items reflect how entitlement infuses each factor of narcissism. For example, the item “When others don’t notice me, I start to feel worthless” from Contingent Self-Esteem demonstrates entitlement to others’ validation and attention. The item “I have better leadership skills than other people” from Leadership demonstrates the entitlement to lead and the accompanying self-importance. The item “I often fantasize about being admired and respected” from Grandiose Fantasy demonstrates the entitlement to be admired in the future. The item “I think I turn heads when I walk down the street” from Vanity demonstrates the underlying sense of entitlement to others’ admiration for one’s physical traits. Finally, the item “I would feel so ashamed if someone found out all parts of me” from Hiding One’s Needs demonstrates the underlying entitlement to concealing one’s true self. Furthermore, in previous research, Sivanathan et al. (2023) demonstrated a strong correlation between entitlement and grandiose (r = 0.76, p < .001) and vulnerable narcissism (r = 0.56, p < .001), and correlations with five subscales ranged from 0.43 to 0.71. Taken together, these indicate that despite the lack of a separate factor, entitlement is captured throughout the measure.

The UNS-R, despite being the first scale to theorize and capture contemporary conceptualisations of narcissism is not without its challenges. Sivanathan et al. (2023) assessed measurement invariance between U.S., Chinese, and Sri Lankan samples; however, they were only able to achieve partial scalar invariance, which required a third of the items’ intercepts to be freed. Literature suggests that the more item parameters that are released to achieve partial scalar invariance, the less reliable group mean comparisons become (Putnick & Bornstein, 2016). The UNS-R includes 35 items, with a third of them not showing equivalent performance. Previous research has shown that longer measures result in poorer quality data (Liu & Wronski, 2018). Further, the authors have demonstrated the superior performance of the UNS-R relative to the Narcissistic Personality Inventory (NPI) and the PNI (Sivanathan et al., 2021), but have not made comparisons of the performance of the UNS-R to other existing narcissism measures. Therefore, the main aims of this paper were to determine a prototype short form of the measure that is scalar invariant across cultures and to test the incremental validity of the UNS-R and its short form in relation to existing and widely used measures of narcissism to further expand the external validity of the subscales. We will utilize Chinese, Sri Lankan, and U.S. samples from Sivanathan et al. (2023) and an Australian sample. By constructing a scalar-invariant short form, we can further explore cross-cultural differences in the expression of narcissism and develop a short form of the measure that is more suitable for longer survey designs.

Study 1

Sivanathan et al. (2023) report only achieving partial scalar invariance of the UNS-R between three countries. The UNS-R currently comprises 35 items, with 11 items not working equivalently across cross-cultural samples. Given the current landscape of data collection via online research platforms with multiple self-report measures, having a shorter, well-performing measure of narcissism that works across diverse samples would be beneficial. Therefore, in this study, we aimed to re-analyse the data from Sivanathan et al. (2023) and combine it with an Australian sample to construct a prototype short form of the UNS-R.

Past measures of narcissism have shown mixed cross-cultural invariance and validation. A German translation of the long and short form of the FFNI found a poor model fit of both the three-factor and two-factor structures (Jauk et al., 2022). Similarly, the FFNI short form tested in U.K and Russian samples was only able to replicate the two- and three-factor structures by allowing cross-loadings of items (Papageorgiou et al., 2022). A study exploring the cross-cultural validation of the Pathological Narcissism Inventory (PNI) in a Chinese sample found that the initial factor structure proposed by Pincus et al. (2009) did not replicate, and the authors had to utilize item parceling to achieve adequate fit (You et al., 2013). Similarly, a cross-cultural validation of the PNI utilizing a sample from postwar Croatia found that though the seven-order factor structure of the PNI was replicated, the Exploitativeness factor did not load onto Grandiose Narcissism as it has in Western samples and in China (Jakšić et al., 2014; Wright et al., 2010; You et al., 2013). These findings suggest despite past attempts at constructing or validating measures cross-culturally, there is still a gap in the literature for a robust measure that can perform well across cultures.

In this study, we aimed to re-analyze four samples to construct a cross-culturally invariant short form prototype of the UNS-R. The samples in this study collectively represent vertical individualist (the United States; Nelson & Shavitt, 2002), vertical collectivist (Sri Lanka and China; Li & Benson, 2022), and between vertical and horizontal individualist (Australia; Chen & Li, 2005) cultures. Individualist cultures perceive the self as independent, and one’s goals, values, and pursuits in life reflect this independent self. In collectivist cultures, on the other hand, the self is interdependent and one’s goals, values, and pursuits are often enmeshed with the group one belongs to (Singelis et al., 1995). A further dimension upon which cultures can be understood is vertical and horizontal (Singelis et al., 1995). In vertical cultures, there is an acceptance that individuals are different and that inequality is inherent. In addition, the need for hierarchy, authority ranking, and a strong focus on achievement are pronounced. In horizontal cultures, individuals are perceived to be equal and the same, which then translates to a limited pursuit of achievement (Singelis et al., 1995). Unlike the United States, China, and Sri Lanka, Australia is considered to be between a vertical and horizontal individualist culture, with leanings to be more horizontal than vertical (Chen & Li, 2005). In Australia, there is acceptance of individuality, and a resistance to hierarchy and achievement orientation, manifested in Australia’s culture of “tall poppy syndrome”, where high levels of achievement or fame are actively discouraged.

There has been previous research exploring differences in mean levels of narcissism between cultures. The seminal work by Foster et al. (2003) found that North Americans reported the highest levels of grandiose narcissism, while Asians and Middle Easterners the lowest. In contrast, more recent research by Fatfouta et al. (2021) explored mean differences at the facet level of grandiose narcissism and found that individuals from more collectivistic cultures reported higher levels of Leadership/Authority and Grandiose Exhibitionism than individuals from more individualistic cultures. Neither of these studies considered the vertical/horizontal dimensions of cultural differences. Research exploring changes in levels of narcissism in Chinese samples alluded to the increase in vertical individualism as a driving force behind increasing levels of narcissism amongst younger people in China (Cai et al., 2012). Finally, a recent multi-national study found that countries that value social competition and status differences had higher mean levels of grandiose narcissism (Jonason et al., 2020). Taken together, these findings suggest that cultures where vertical individualism is endorsed are likely to also endorse higher mean levels of narcissism.

Based on the previous limitations of the UNS-R and the lack of culturally-invariant measures of narcissism, this study aimed to construct a prototype short form of the measure that was scalar invariant across all four samples. It also aimed to explore the group differences in the manifestation of narcissism across these cultures utilizing a scalar-invariant short-form measure. It was hypothesized that the short form measure would be scalar invariant across the four groups (Hypothesis 1). Given both the individualist and vertical nature of the United States, we expected this sample would score the highest on Leadership, Vanity, and Grandiose Fantasy, followed by China and Sri Lanka, and finally Australia (Hypothesis 2a). Given the interdependent self that is inherent to collectivist countries, we expected that China and Sri Lanka would score highest on Contingent Self-Esteem, followed by the United States and Australia (Hypothesis 2b). Finally, given the shame and fear of rejection from a group, which are required to sustain a sense of self in collectivist cultures, we also expected that China and Sri Lanka would score highest on Hiding One’s Needs, followed by the United States and Australia (Hypothesis 2c).

Methods

Participants and procedure

For this study, we utilized the Australian sample (labelled Sample 4) collected in Study 2 and the three samples from Sivanathan et al. (2023).

Sample 1 – United States

The sample consisted of 395 participants from the United States recruited via Amazon’s Mechanical Turk, with a mean age of 41.69 (SD = 13.84; 18–78), and 205 participants who identified as men, 187 as women, and three as gender diverse. Participants predominantly identified as White (78.73%). Demographic information about country of origin and ethnicities can be found in Supplemental Materials 1.

Sample 2 – Chinese sample

The sample consisted of 326 Chinese residents recruited via the research participation platform Weidiaocha, with 183 participants identifying as men and 143 as women. The mean age of the sample was 25.52 years (SD = 6.44; 18–53). Except for the missing information of one participant, participants’ registered residencies were distributed across 29 out of the 34 provincial-level administrative divisions of China. No further demographic information about this sample was provided in Sivanathan et al. (2023).

Sample 3 – Sri Lankan sample

The sample consisted of 354 Sri Lankan citizens recruited via social media platforms, with 257 participants identifying as women, 93 as men, and four as non-binary. The mean age of the sample was 28.73 years (SD = 6.97; 18–61). The majority of the participants identified as Sinhalese (63.55%), followed by Tamil (13.84%) and Muslim (11.86%). A complete breakdown of ethnicities can be found in Supplemental Materials 1.

Measures

The samples from Sivanathan et al. (2023) included the UNS-R retained in English for the U.S. and Sri Lankan samples, and a translated Chinese version was used for the Chinese sample. The studies also included the Rosenberg Self-Esteem Scale (RSES) to measure self-esteem (Rosenberg, 1965) and the subscales from the mini-International Personality Item Pool measured extraversion, agreeableness, and neuroticism (Donnellan et al., 2006). Finally, the nine-item Psychological Entitlement Scale was used to capture entitlement (Campbell et al., 2004). As these measures were used in the current study only for purposes of external validity of the short form, please see Supplemental Materials 2 for more detailed information on each of those measures.

Data analytic plan

We initially ran confirmatory factor analysis (CFA) on the 35-item measure separately for each of the four samples using the maximum likelihood estimator with the lavaan package in R (Rosseel, 2012). We also accessed the multi-group CFA results from Sivanathan et al. (2023), which identified the 11 items that did not perform similarly across the three samples. We also ran item response theory (IRT) analyses, using the R package mirt (Chalmers, 2012), and examined individual item functioning with a graded response model (GRM) for polytomous items. Each subscale was run separately due to the assumption of unidimensionality (Reise & Revicki, 2015). We collated all this information to choose three items for each subscale that we hypothesized would perform similarly across cultures. We then ran separate CFAs in each of the four samples to confirm that the hierarchical factor structure was replicated. Following this, we ran further multigroup CFA to determine measurement invariance. The study and planned analyses were not pre-registered. All data and codes are publicly available at: https://osf.io/kb7j9/?view_only=6b08c31b46ad47f0abff2677ddd25ecd.

Results

Measurement invariance and shortening

The CFA, IRT, and initial measurement invariance analyses comparing Samples 1–3 were collated (specific information about these can be found in Sivanathan et al., 2023), along with the CFA output and IRT analyses of items in Sample 4. We did not include the 11 items that were observed to be variant between Sri Lankan, Chinese and U.S. samples. From the remaining 24, we assessed the item loadings from the multigroup CFA, item information curves from item response theory analyses, and the loadings of each item in the CFA models of each sample to choose three items per factor. This resulted in a total of 15 items (see Appendix A) constituting the Unified Narcissism Scale-Revised Short Form (UNS-RSF). We assessed the performance of the short form measure, including its first-order, second-order, and third-order factor structures, in each of the four samples. We ran CFA separately in each sample and found the short form measure to show a good to strong fit for its first-order, second-order, and third-order factor structures. The fit indices along with the item loadings in each sample are provided in Supplemental Materials 3. We ran IRT analyses to assess the performance of the short form measure within each sample and found them to show good information curves across the latent trait (see Supplemental Materials 3 for more information). Finally, we explored the external validity of the short form measure and found it replicated all the expected correlations (see Supplemental Materials 3).

We ran multigroup CFA to assess the measurement invariance of the short form measure across the four samples using the lavaan package for R (Rosseel, 2012). Multigroup CFA sequentially runs increasingly restricted multigroup models in a hierarchical fashion, going from an unrestricted model (named configural) to restricting loadings of items to equality (named metric) and then to intercepts of items (named scalar). Changes in CFI values less than 0.01 and RMSEA values less than 0.015 indicate invariance between the two models and the tested parameters being equal between groups (Chen, 2007). We found the first-order factor structure of the short form measure to show metric (CFI = 0.971; RMSEA = 0.051), configural (CFI = 0.970; RMSEA = 0.050; ΔCFI = 0.001; ΔRMSEA = 0.001), and scalar invariance (CFI = 0.964; RMSEA = 0.052; ΔCFI = 0.006; ΔRMSEA = 0.001). We found restricting the loadings and intercepts of items to equality did not significantly worsen the model fit, providing support for Hypothesis 1. We did not conduct measurement invariance analyses for the hierarchical model as we were most interested in understanding the cultural differences at the first-order level.

We did, however, run subscale intercorrelations to further understand the co-occurrence of grandiosity and vulnerability in the four samples. We found that the correlations between grandiose and vulnerable narcissism for the Australian (r = 0.16, p < .01), Chinese (r = 0.17, p < .01), and Sri Lankan samples (r = 0.15, p < .01) were similar. Nonetheless, the correlation between grandiose and vulnerable narcissism was much stronger in the U.S. sample (r = 0.47 p < .001) than in the other samples. These results allude to a higher co-occurrence of grandiosity and vulnerability in the United States, which may not be the case in the other three countries.

Mean comparisons

Given the scalar invariance of the UNS-RSF, we were justified to use it to conduct mean comparisons between the samples for the five subscales. Each sample’s mean scores on the subscales, along with the standard deviations and alphas, can be found in Supplemental Materials 3. The sample characteristics showed that the U.S. and Australian samples were much older than the Sri Lankan and Chinese samples. As a result, we ran separate regression analyses for the five subscales controlling for age and gender and then conducted post-hoc comparisons of estimated marginal means (Table 1). As expected, we found Australia to score the lowest on Leadership, Vanity, and Grandiose Fantasy (Hypothesis 2a). Grandiose Fantasy, however, showed no significant difference in means for the other three samples. All three samples were similar on Leadership, except the Sri Lankan sample mean was significantly higher than the Chinese sample. The U.S. and Chinese samples were similar on Vanity, and the Sri Lankan sample scored significantly lower than both. Regarding Contingent Self-Esteem, the mean for the Chinese sample was significantly higher than Sri Lanka and Australia, but not significantly different from the U.S. sample. Contrary to Hypothesis 2b, Sri Lanka scored the lowest on Contingent Self-Esteem and was significantly lower than all other samples. Contrary to Hypothesis 2c, we found the U.S. sample to score significantly higher than the Chinese and Sri Lankan samples on Hiding One’s Needs, and the Australian sample to score significantly higher than the Sri Lankan sample.

Table 1. Mean differences for the five subscales of the unified narcissism scale-revised short form.

Subscales	US vs. CHN	US vs. SL	US vs. AU	CHN vs. SL	CHN vs. AU	SL vs. AU
Contingent Self-Esteem	−0.09	0.50***	0.16	0.60***	0.25*	−0.34**
Leadership	0.05	−0.15	0.36***	−0.20*	0.31**	0.51***
Vanity	0.05	0.50***	0.83***	0.46***	0.79***	0.33**
Grandiose Fantasy	0.11	0.15	0.57***	0.04	0.46***	0.42***
Hiding One’s Needs	0.37*	0.48***	0.18	0.11	−0.19	−0.30**

Note. N_US = 395. N_CHN = 326. N_SL = 354. N_AU = 309. US: United States; CHN: China; SL: Sri Lanka; AU: Australia. Estimated marginal mean comparisons in bold represent significant differences between groups.

* p < .05; **p < .01; ***p < .001. The p values are adjusted using false discovery rate method.

Discussion

In this study, we constructed a prototype short form of the UNS-R that was scalar invariant across the four cultural samples and showed similar performance to the long form. The first-order five-factor and higher-order factor structures of the short form measure demonstrated a good fit in the four samples, suggesting the short form replicates the factor structure of its parent measure. The short form also replicated the expected correlations for grandiose and vulnerable narcissism and showed similar strength of correlations in comparison to the long form. This suggests the 15-item measure can capture grandiose and vulnerable narcissism similarly to the longer form measure of UNS-R. Therefore, the UNS-RSF is a good candidate for measuring narcissism across cultures.

We obtained configural, metric, and scalar invariance of the UNS-RSF, suggesting that we could compare the samples’ mean levels of narcissism and addressing an important gap in the literature. The Australian sample scored the lowest on the two factors that represent grandiose narcissism, Leadership and Vanity, whereas the other three samples were comparable. This suggests that grandiose narcissism within the three vertical cultural groups (United States, China, and Sri Lanka) occurs at the same level in comparison to a more horizontal culture, such as Australia, where the occurrence is lower. For the dimensions of vulnerable narcissism, China and the United States scored the highest on Grandiose Fantasy, alluding to the achievement focus that is prolific in both cultures (Salili, 1996). As expected, Australia scored the lowest, reflecting the horizontal nature of the culture where achievement is less celebrated, often captured within the Australian framework of “tall poppy syndrome”. Interestingly, we found United States and China scored similarly on Contingent Self-Esteem and Sri Lanka scored the lowest, suggesting perhaps the recent influence of individualism in Chinese cultures (Cai et al., 2012) promotes a contingent sense of self more so than a collectivist identity. Finally, contrary to our hypothesis, we found the two individualist samples to score significantly higher on Hiding One’s Needs compared to China and Sri Lanka, suggesting individualism promotes greater shame and fear of rejection of one’s inner world.

These findings highlight the importance of vertical versus horizontal dimensions and the need to consider this when hypothesizing sociocultural influences on narcissism. For three out of the five dimensions (Leadership, Vanity, and Grandiose Fantasy), the three vertical cultures were more like each other than Australia (a more horizontal individualist culture) was to any of them. This highlights the importance of understanding narcissism not simply as an expression of elevated individuality but as an expression of elevated entitlement that is embedded within a framework of one being more special than others and requiring special treatment. This explains why a vertical culture—where one accepts the inequality of people—would lend itself to higher levels of narcissism (Singelis et al., 1995). It is important to note that Australian culture does still include elements of hierarchy, and so, to strengthen this finding, it is important to explore the manifestation of narcissism in more horizontal countries, such as Denmark (Nelson & Shavitt, 2002).

Study 2

The UNS-R is a 35-item measure that is a revision of a preexisting measure known as the Unified Narcissism Scale (UNS; Sivanathan et al., 2021). The UNS was initially constructed using items from the NPI and PNI to reflect the contemporary theoretical perspectives on trait narcissism. The UNS showed better psychometric performance in comparison to the NPI and the PNI (Sivanathan et al., 2021), but has not been assessed against other measures. We have chosen to assess the performance of UNS-R and the protype short form from Study 1 against three popular and well-validated measures of narcissism.

Firstly, the FFNI is a measure of grandiose and vulnerable narcissism (Glover et al., 2012). Based on the five-factor model of narcissism, the FFNI describes a three-factor and two-factor model. Its three-factor model captures Antagonism (the perceived core of narcissism), Agentic Extraversion (unique aspects of grandiose narcissism), and Neuroticism (unique aspects of vulnerable narcissism). Its two-factor model captures Grandiose and Vulnerable Narcissism (Miller et al., 2015). Secondly, the Hypersensitive Narcissism Scale (HSNS) is a measure of vulnerable narcissism (Hendin & Cheek, 1997) that has been widely researched and validated across cultures and samples. Finally, the Narcissistic Admiration and Rivalry Questionnaire (NARQ) measures the two pathways via which grandiose narcissists maintain a grandiose sense of self (Back et al., 2013). In comparison to other existing measures of grandiose narcissism, the NARQ is specifically interested in the interpersonal processes of grandiose narcissism and therefore shown to be more predictive of outcomes than the NPI (Back et al., 2013). We have chosen these additional measures to conduct incremental validity analyses against, as they capture unique aspects of grandiose and vulnerable narcissism, and the FFNI especially is an emerging measure of both constructs.

External relationships of grandiose and vulnerable narcissism

To assess incremental validity, we will be exploring the relationship of narcissism to outcome variables and how these relationships are captured by the various measures. Grandiose narcissism is characterized by an elevated sense of self and entitlement that leads to positive behavioral and cognitive manifestations, and the externalizing component serves as a protective factor against internalizing psychopathology (Krizan & Herlache, 2018). Research shows that grandiose narcissism is positively related to general mental health and life satisfaction (Rohmann et al., 2019), and is negatively related to anxiety and depression (Sedikides et al., 2004). The focus on physical attractiveness also motivates those higher on grandiose narcissism to engage in certain behaviors to sustain their level of vanity, with research showing grandiose narcissism to be positively related to healthy eating behaviors and exercise (Hill, 2016). These findings were based predominantly on American college students (Hill, 2016; Sedikides et al., 2004) and European participants (Rohmann et al., 2019).

Grandiose narcissism is interpersonally characterized as dominant and calculating (Dickinson & Pincus, 2003). Research has shown that in long-term relationships grandiose narcissists report higher intentions for infidelity, resulting in lower self-reported relationship satisfaction (Altınok & Kılıç, 2020). Therefore, we predicted that grandiose narcissists in long-term relationships with no intention toward infidelity are likely to report higher relationship satisfaction to sustain the inflated sense of self.

Unlike grandiose narcissism, vulnerable narcissism is characterized by entitlement that interplays with a low sense of self and high neuroticism (Krizan & Herlache, 2018; Miller et al., 2017). Research shows that vulnerable narcissism is positively related to internalizing psychopathology (depression and anxiety), and is negatively related to general mental health and life satisfaction (Rohmann et al., 2019). The contingent sense of self that is a feature of vulnerable narcissism also predicts unhealthy eating behaviors and exercise as an attempt to regulate a low sense of self and the shame and fear of rejection by others (Hill, 2016; Sivanathan et al., 2019). In interpersonal relationships, vulnerable narcissism is characterized by being submissive, aloof, and calculating, and bringing a sense of fear, shame, and suspicion to relational contexts (Dickinson & Pincus, 2003). Research shows those higher on vulnerable narcissism are characterized by an avoidant and anxious attachment style, and report lower relationship satisfaction (Casale et al., 2020).

External relationships of narcissism dimensions

The specific dimensions of grandiose narcissism include Vanity and Narcissistic Leadership. We anticipate that Vanity, with its focus on physical attractiveness, would primarily relate to spending more time on social media, spending more money on appearance, and frequently looking at oneself on reflective surfaces. Alternatively, Narcissistic Leadership would primarily relate to believing oneself to be more competent than one is and an inclination toward holding positions of power.

The dimensions of vulnerable narcissism include Contingent Self-Esteem, Grandiose Fantasy, and Hiding One’s Needs. Contingent Self-Esteem should uniquely relate to constant reassurance-seeking behaviors, rumination about social interactions, and overconcern with body, shape, and weight—things that sustain the contingent sense of self. Grandiose Fantasy would be characterized by dreaming about a fantasy life, which in turn is likely to lead to lower life satisfaction. The characteristics of Hiding One’s Needs would manifest in lower intention to engage in help-seeking behaviors for fear of rejection and constantly keeping people at a distance.

Aims and hypotheses

In Study 2, we had three aims related to the validation of the UNS-R, the UNS-RSF, and the subscales of both measures. Firstly, we aimed to assess the functioning of the scale in an Australian sample to determine if the factor structure would be replicated. Based on this aim, we hypothesized:

Hypothesis 3a: The five first-order factor structure of narcissism would be replicated.

Hypothesis 3b: Leadership and Vanity would load onto the higher-order factor Grandiose Narcissism, and Contingent Self-Esteem, Grandiose Fantasy, and Hiding One’s Needs would load onto Vulnerable Narcissism.

Hypothesis 3c: The two second-order factors would load onto a common Narcissism factor.

Secondly, we aimed to expand the external validity of the UNS-R and its prototype short form and assess its incremental validity. Based on this aim, we hypothesized:

Hypothesis 4a: Grandiose narcissism would positively relate to life satisfaction, relationship satisfaction, healthy eating, physical activity, and general mental health, negatively relate to anxiety and depression, and would not relate to anxious and avoidant attachment styles.

Hypothesis 4b: Vulnerable narcissism would negatively relate to relationship satisfaction, life satisfaction, general mental health, healthy eating, and physical activity, and positively relate to anxiety, depression, and anxious and avoidant attachment styles.

Finally, we aimed to expand the external validity of the five first-order factors. Based on this aim, we hypothesized the following:

Hypothesis 5a: Leadership would primarily relate to applying for a position one is not qualified for and having higher intentions for power.

Hypothesis 5b: Vanity would primarily relate to social media use, money spent on looks, and an inclination to look at oneself.

Hypothesis 5c: Contingent Self-Esteem would primarily relate to anxious attachment, reassurance seeking, rumination about social interactions, and over concern about the body.

Hypothesis 5d: Grandiose Fantasy would primarily relate to lower life satisfaction, preferring fantasy to reality, and rumination about fantasy life.

Hypothesis 5e: Hiding One’s Needs would primarily relate to avoidant attachment, lower help-seeking intention, and lower help-seeking behaviours.

Methods

Participants and procedure

Australian residents were recruited to participate in the study via the online recruitment platform Prolific. Participants were remunerated AUD 3.67, based on Prolific remuneration guidelines, for their participation, and 322 participants completed the survey. Data from 13 participants were excluded due to failing attention checks and requesting to withdraw their data. The final sample consisted of 309 participants (148 identified as men, 152 as women, and five participants identified as gender diverse). The mean age of the sample was 34.01 (SD = 12.97; 18–79), 80.66% were born in Australia, and 60.32% identified as White/Anglo-Australian. There were 111 participants (35.92%) who reported being single and 179 (57.92%) who reported being in a long-term relationship (greater than six months) or married (complete breakdown of demographic information can be found in Supplemental Materials 1). All ethical aspects of this study were approved by the relevant human research ethics committee and were in line with the Helsinki Declaration on Human Research.

Measures

Unified narcissism scale-revised and prototype short form (UNS-R; Sivanathan et al., 2023)

The UNS-R consists of 35 items scored on a six-point Likert-type scale ranging from 1 (Not at all like me) to 6 (Very much like me). In this sample the five UNS-R subscales—Contingent Self-Esteem (α = 0.93), Leadership (α = 0.93), Vanity (α = 0.94), Grandiose Fantasy (α = 0.93), and Hiding One’s Needs (α = 0.94) were reliable, as were the measures of the higher-order constructs of overall Narcissism (α = 0.92), Grandiose Narcissism (α = 0.92) and Vulnerable Narcissism (α = 0.93). The UNS-R prototype short form consists of 15 out of the 35 items.

Hypersensitive narcissism scale (HSNS; Hendin & Cheek, 1997)

The HSNS (α = 0.80) is a 10-item measure of vulnerable narcissism. The 10 items are scored on a six-point Likert-type scale, ranging from Strongly Disagree (1) to Strongly Agree (6).

Five-Factor Narcissism Inventory-Short Form (FFNI-SF; Sherman et al., 2015)

The FFNI-SF is a validated short form measure of the FFNI and has been shown to perform comparably to the long form (Sherman et al., 2015). The FFNI-SF consists of 60 items that can be scored using the two- or three-factor models. For the purposes of this study, we used the two-factor model, including Grandiose (α = 0.93) and Vulnerable Narcissism (α = 0.85) subscales. The items are scored on a five-point Likert-type scale ranging from Strongly Disagree (1) to Strongly Agree (5).

Narcissistic admiration and rivalry questionnaire-short form (NARQ-S; Leckelt et al., 2018)

The NARQ-S is a six-item short form measure of the NARQ that has been empirically validated and performs similarly to its parent scale (Leckelt et al., 2018). NARQ-S distinguishes between agentic (Admiration subscale) and antagonistic (Rivalry subscale) parts of grandiose narcissism. The items are scored on a six-point Likert-type scale ranging from Not agree at all (1) to Agree completely (6). The Rivalry subscale had acceptable internal consistency (α = 0.77), though the reliability of the Admiration subscale was lower (α = 0.66).

Life satisfaction

We included a five-item measure of life satisfaction known as the Satisfaction with Life Scale (α = 0.91; Diener et al., 1985). The items are scored on a seven-point Likert scale ranging from Strongly Disagree (1) to Strongly Agree (7).

Health-related behaviors

We also included items assessing eating and exercise behaviors. We included single items that have been used in previous research to capture healthy eating behaviors. Participants were asked “During a week how often do you eat at least five servings of fruits and vegetables?” to capture eating the daily required serves of fruit and vegetables, scored from 1 (Not at all) to 5 (Everyday). Participants were asked “Is your daily eating pattern well-balanced and healthy? to capture maintaining a balanced diet scored from 1 (Never) to 5 (Always; Hill, 2016). To capture physical activity, we used the Godin Leisure-Time Exercise Questionnaire (Godin & Shephard, 1985). The questionnaire asks about the rate of mild, moderate, and strenuous exercise completed for over 15 min during a 7-day period and participants responding noting down the frequency. The total score is obtained using the formula: (9 × Strenuous) + (5 × Moderate) + (3 × Light).

Adult attachment styles

To capture anxious and avoidant attachment styles, we used the 12-item Experiences in Close Relationship short form scale (ECR–12; Tasca et al., 2018), with two subscales, Avoidance and Anxiety (both αs were 0.89). It is scored on a seven-point Likert scale ranging from Strongly Disagree (1) to Strongly Agree (7).

Mental health

To capture overall mental health, we used a single-item self-rated measure of mental health (“How would you rate your overall mental health”) scored on a range from 1 (Poor) to 5 (Excellent; Ahmad et al., 2014). To capture depression, we included the two-item Patient Health Questionnaire (PHQ-2; Mitchell et al., 2016), and to measure anxiety, we used the two-item Generalized Anxiety Disorder Scale, both were scored on a Likert-type scale ranging from Not at all (0) to Nearly Every Day (3; GAD-2; Plummer et al., 2016). We assessed the intercorrelations between the two-items and found them to be strongly correlated for the PHQ-2 (r = 0.71, p < .001) and GAD-2 (r = 0.81, p < .001).

Factor-specific items

We also created several single items to capture the constructs that we anticipated would be strongly related to each of the five factors. We present these items along with their scoring in Supplemental Materials 4. We included one item from the Body Image Disturbance Questionnaire (Cash et al., 2004), that we expected to be primarily related to Contingent Self-Esteem, but the remaining items were created by the authors.

Data analytic plan

We ran confirmatory factor analysis (CFA) utilizing maximum likelihood estimation to test the hypothesized factor structure in the Australian sample using lavaan package in R (Rosseel, 2012). To assess external validity, we ran Pearson’s r correlations in JASP (Version 0.17.3). To assess incremental validity we ran multiple hierarchical multiple regression analyses with established measures of narcissism in step 1 and UNS-R subscales added in step 2 in JASP (JASP Team, 2023).

Results

Confirmatory factor analysis (CFA)

We initially ran CFA using maximum likelihood estimation to assess the model fit of the UNS-R to the Australian sample. We tested the first-order, second-order, and third-order factor structures. Given the sample size of > 300, we used alternative fit indices to determine adequate/good fit and adopted widely utilized recommendations for alternative model fit estimates (Hu & Bentler, 1999): adequate model fit estimates defined as comparative fit index (CFI) > 0.90, standardized root mean residual (SRMR) < 0.08, and root mean square error of approximation (RMSEA) < 0.08; and good model fit estimates defined as CFI > 0.95, SRMR < 0.06, and RMSEA < 0.06. When we initially ran the five-factor first-order model, we found the fit to be just acceptable based on CFI and SRMR (CFI = 0.90; RMSEA = 0.07 [0.07 − 0.08]; SRMR = 0.06), with the 95% confidence interval for RMSEA. When we explored the modification indices, we noticed a high correlation between two items of the Grandiose Fantasy factor that had similar wording. We allowed the error variances of the two items to covary and found this improved the fit. In this model, although chi-square was significant χ²(549) = 1,221.6, p < .001, the model fitted the data acceptably based on all three other alternative fit indices (CFI = 0.92; RMSEA = 0.06 [0.06, 0.07]; SRMR = 0.06) providing support for Hypothesis 3a.

Similarly, we also found chi-square was significant and SRMR was just above cutoff for the second-order model, χ² (554) = 1,283.9, p < .001, but the model fitted the data acceptably based on CFI and RMSEA fit indices (CFI = 0.92; RMSEA = 0.07 [0.06, 0.07]; SRMR = 0.09), providing support for Hypothesis 3b. Grandiose and Vulnerable Narcissism were significantly correlated with each other (r = 0.40, p < .001). We emphasize that the third-order factor model is mathematically equivalent to the second-order model and is utilized predominantly to re-express the shared correlation between Grandiose and Vulnerable Narcissism. However, we do note the strong loadings would suggest that narcissism is a unitary construct at the highest level of generalization. Figure 2 demonstrates the significant loadings of Grandiose and Vulnerable Narcissism onto a common Narcissism factor, providing support for Hypothesis 3c.

Figure 2. Third-order factor structure showing loadings of latent factors onto higher-order factors. N = 309. All loadings observed are significant at p < .001.

External and incremental validity

Prior to testing our hypotheses, we computed Pearson’s correlations between the measures of narcissism, and they were all correlated with each other (see Supplemental Materials 4), supporting the convergent validity of the UNS-R and its short form. To test for external validity, we ran correlational analyses exploring the relationships of grandiose and vulnerable narcissism with external validity measures.

Table 2 outlines the correlations between various measures of grandiose and vulnerable narcissism with mental health outcomes, attachment styles, and health-related behaviors. In an early revision of the paper we included the Prisoner’s Dilemma task, but found that none of the narcissism measures predicted the participant outcomes on the task. We anticipate this was due to measuring a one-off behavioral outcome rather than aggregate performance, for which personality traits are more likely to be predictive. For interested readers, please see Supplemental Materials 5 for further information. As shown in Table 2, in line with Hypothesis 4a, grandiose narcissism, captured by UNS-R and UNS-RSF, was positively related to life satisfaction, relationship satisfaction, general mental health, healthy eating, and physical activity, negatively related to depression, and unrelated to anxious attachment style. Despite the long form being negatively related to anxiety and unrelated to avoidant attachment style, we found the UNS-RSF was unrelated to anxiety and negatively related to avoidant attachment style. Similarly, in line with Hypothesis 4b, vulnerable narcissism, captured by the long and short form, was positively related to anxiety, depression, and attachment styles, and negatively related to life satisfaction, relationship satisfaction, general mental health, and healthy eating. Contrary to Hypothesis 4b, vulnerable narcissism was unrelated to physical activity.

Table 2. Zero-order correlations between narcissism measures and outcome variables.

Outcome variables	Grandiose Narcissism					Vulnerable Narcissism
	UNS-R	UNS-R SF	FFNI	NARQ-A	NARQ-R	UNS-R	UNS-R SF	FFNI	HSNS
Life Satisfaction	0.40***	0.39***	0.20***	−0.03	0.09	−0.21***	−0.20***	−0.27***	−0.22***
Relationship Satisfaction	0.20***	0.21***	0.14*	0.02	0.01	−0.12*	−0.12*	−0.14*	−0.14*
Mental Health	0.30***	0.29***	0.21***	0.00	−0.08	−0.36***	−0.36***	−0.43***	−0.36***
Anxiety	−0.12*	−0.11	−0.10	0.07	0.01	0.43***	0.42***	0.47***	0.38***
Depression	−0.15***	−0.13*	−0.06	0.15*	0.00	0.40***	0.37***	0.49***	0.40***
Anxious Attachment	0.07	−0.12*	0.15*	0.25***	0.26***	0.59***	0.58**	0.66***	0.53***
Avoidant Attachment	−0.11	0.08	−0.02	0.06	−0.06	0.17***	0.56***	0.16**	0.25***
Five Serves of Fruits and Vegetables	0.14*	0.11	0.06	−0.08	0.02	−0.12*	−0.10	−0.21***	−0.09
Balanced Diet	0.25***	0.25***	0.07	−0.12*	0.02	−0.18**	−0.16**	−0.25***	−0.19***
Exercise	0.30***	0.30***	0.22***	0.07	0.14*	−0.00	0.01	−0.08	−0.08

Note. N = 309. UNS-R: Unified Narcissism Scale-Revised; UNS-R SF: Unified Narcissism Scale-Revised Short Form; FFNI: Five-Factor Narcissism Inventory; NARQ-A: Narcissistic Admiration and Rivalry Questionnaire – Admiration; NARQ-R: Narcissistic Admiration and Rivalry Questionnaire – Rivalry; HSNS: Hypersensitive Narcissism Scale.

* p < .05; **p < .01; ***p < .001.

To test for incremental validity, we ran hierarchical multiple regression analyses (HRMA) separately for each of the outcome variables, with existing measures of grandiose or vulnerable narcissism included in the first step and the UNS-R or its short form subscales in the second step¹. We assessed if the UNS-R and its short form subscales would account for additional variance in the external validity measures once the other existing measures were accounted for. Table 3 outlines the R² change (the additional variance explained by the UNS-R Grandiose and Vulnerable Narcissism subscales) and the standardized regression coefficients of the measures of grandiose and vulnerable narcissism with outcome variables. A more detailed output of each step of the HMRA is provided in Supplemental Materials 5, along with R² change (the additional variance explained by the UNS-RSF Grandiose and Vulnerable Narcissism subscales) and the standardized regression coefficients of the measures of grandiose and vulnerable narcissism with outcome variables.

Table 3. Standardized regression coefficients and R² change outputs from hierarchical multiple regression analyses assessing incremental validity of UNS-R.

	Grandiose Narcissism					Vulnerable Narcissism
Outcome Variable		B					β
	ΔR^2	UNS-R GN	FFNI-GN	NARQ-A	NARQ-R	ΔR^2	UNS-R VN	HSNS	FFNI-VN
Life Satisfaction	0.14***	0.58***	−0.07	−0.15	−0.11	0.00	−0.01	−0.06	−0.22*
Relationship Satisfaction	0.03***	0.26**	0.01	−0.02	−0.17*	0.00	−0.02	−0.08	−0.07
General Mental Health	0.05***	0.34***	0.11	−0.14	−0.11	0.00	−0.08	−0.09	−0.31***
Anxiety	0.01	−0.10	−0.18	0.16*	0.08	0.01*	0.16*	0.07	0.31***
Depression	0.02**	−0.23**	−0.03	0.25***	0.01	0.00	0.08	0.10	0.36***
Avoidant Attachment Style	0.01*	−0.18*	0.11	0.13	−0.11	0.00	0.05	0.25**	−0.05
Anxious Attachment Style	0.01	−0.14	−0.02	0.16*	0.26**	0.02**	0.21**	0.11	0.43**
Five Serves of Fruit and Veg	0.02*	0.21*	0.00	−0.15*	−0.02	0.00	0.05	0.09	−0.30***
Balanced Diet	0.08***	0.44***	−0.11	−0.22**	−0.03	0.00	0.02	−0.04	−0.24***
Regular Physical Activity	0.04***	0.31***	0.05	−0.07	−0.04	0.01	0.13	−0.08	−0.11

Note. N = 309. Existing measures of narcissism were included in the first step of the hierarchical multiple regression analyses and UNS-R subscales in the second step. ΔR²: R² change; β: standardized regression coefficient; UNS-R GN: Unified Narcissism Scale-Revised Grandiose Narcissism; FFNI-GN: Five-Factor Narcissism Inventory- Grandiose Narcissism; NARQ-A: Narcissistic Admiration and Rivalry Questionnaire – Admiration; NARQ-R: Narcissistic Admiration and Rivalry Questionnaire – Rivalry. UNS-R VN: Unified Narcissism Scale-Revised-Vulnerable Narcissism; FFNI-VN: Five-Factor Narcissism Inventory-Vulnerable Narcissism; subscale; HSNS: Hypersensitive Narcissism Scale.

* p < .05; **p < .01; ***p < .001.

Based on the R² change, we found that the UNS-R Grandiose Narcissism subscale (both the long and short form) explained more unique variance in the outcome variables than the NARQ-S and FFNI-SF. In seven out of the 10 models, only the UNS-R, in both its long and short forms, Grandiose Narcissism subscale was the significant predictor of the outcome variables in the hypothesized direction. For example, NARQ subscales were negatively related to healthy eating and positively related to depression and anxiety, contrary to existing literature (Hill, 2016; Sedikides et al., 2004). Additionally, for these relationships, the UNS-R showed the strongest semipartial correlations, indicating that it accounted for more unique variance in the outcome variables.

The FFNI Vulnerable Narcissism subscale explained more unique variance than the HSNS and the UNS-R Vulnerable Narcissism subscales as in most HMRAs it was the only significant predictor of the outcome variables. The UNS-R Vulnerable Narcissism subscale was still a significant predictor of anxiety and anxious attachment style once the shared variance was accounted for.

To test Hypotheses 5a–5e, we computed semipartial correlations, shown in Table 4, between all the subscales, both long and short form, and the external validity measures. We computed semipartial correlations to account for the intercorrelations between the five subscales. In line with Hypothesis 5a, we found Leadership was primarily related to overvaluing competence and higher intentions for power. In line with Hypothesis 5b, we found Vanity was primarily related to money spent on looks and looking at oneself. However, contrary to Hypothesis 5b, Vanity was unrelated to social media use. In line with Hypothesis 5c, we found Contingent Self-Esteem was primarily related to anxious attachment, reassurance seeking, social rumination, and overconcern about the body. We hypothesized that Grandiose Fantasy would primarily relate to lower life and achievement satisfaction, preferring fantasy to reality, and rumination about fantasy life (Hypothesis 5d). Though Grandiose Fantasy was positively related to preferring fantasy over reality and being distracted by fantasy, it was unrelated to achievement and life satisfaction, contrary to our hypothesis. In contrast, Hiding One’s Needs yielded the strongest correlations with these four variables. In line with Hypothesis 5e, Hiding One’s Needs was primarily related to avoidant attachment and lower help-seeking intention but not lower help-seeking behaviors. We observed the same pattern of correlations for both the long and short form versions alluding to the external validity of both.

Table 4. Semipartial correlations of the five dimensions of narcissism with outcome variables.

Variable	LED	LED-SF	VAN	VAN-SF	CSE	CSE-SF	GF	GF-SF	HN	HN-SF
Overvaluing competence	0.31***	0.30***	0.15**	0.18***	−0.00	−0.03	0.16**	0.17**	−0.07	−0.06
Intention for power	0.50***	0.46***	0.03	0.09*	0.01	−0.03	0.15**	0.20***	−0.06	−0.07
Social media use	0.07	0.06	0.03	0.02	0.02	0.02	−0.02	−0.03	0.02	0.04
Money spent on looks	−0.04	−0.06	0.25***	0.26***	0.06	0.04	0.16**	0.19***	−0.14**	−0.16**
Looking at oneself	−0.07	−0.04	0.32***	0.28***	0.26***	0.23***	0.06	0.06	−0.07	−0.06
Anxious attachment	−0.00	0.01	0.04	0.03	0.38***	0.36***	0.01	0.00	0.23***	0.22
Reassurance seeking	0.05	0.03	0.04	0.09	0.35***	0.30***	0.02	0.05	0.06	0.05
Social rumination	−0.07	−0.04	−0.03	−0.03	0.31***	0.32***	0.11*	0.09	0.16**	0.14**
Overconcern about body	−0.10*	−0.09	−0.21***	−0.21***	0.26***	0.25***	0.02	−0.01	0.16**	0.17***
Life satisfaction	0.28***	0.30***	0.17***	0.16**	0.04	0.01	−0.08	−0.06	−0.26***	−0.22***
Fantasy over reality	−0.05	−0.02	−0.11*	−0.11*	0.12***	0.16*	0.17***	0.11*	0.24***	0.23***
Achievement satisfaction	0.23***	0.24***	0.12*	0.13**	0.01	−0.02	−0.02	0.02	−0.31***	−0.28***
Distracted by fantasy	0.14**	0.13**	0.20***	0.23***	0.10*	0.11*	0.11*	0.11*	0.23***	0.18***
Avoidant attachment	−0.07	−0.11*	0.03	0.01	−0.21***	−0.14**	−0.00	−0.02	0.44***	0.37***
Help-seeking from professional services	0.01	0.01	0.09	0.10	0.02	0.02	0.04	0.06	−0.17**	−0.18**
Help-seeking behavior	0.06	0.09	0.08	0.06	−0.18***	−0.18**	0.08	0.05	−0.10	−0.06
Help-seeking from close ones	0.06	0.10	−0.01	−0.01	0.23***	0.19***	0.00	0.02	−0.26***	−0.21***

Note. N = 309. LED: Leadership; VAN: Vanity; CSE: Contingent Self-Esteem; GF: Grandiose Fantasy; HN: Hiding One’s Needs; SF: Short Form. Correlations in bold are in the hypothesized direction.

* p < .05; **p < .01; ***p < .001.

Discussion

In this study, we assessed the functioning of the UNS-R and its short form in an Australian sample and assessed its incremental and external validity. As expected, UNS-R showed an acceptable fit for the first-order five-factor, higher-order two-factor and one-factor structures, in line with past research (Sivanathan et al., 2021, 2023), indicating that narcissism is a hierarchical construct in an Australian sample with dimensions of grandiose and vulnerable narcissism. In line with our hypotheses, we replicated most of the expected external correlations for grandiose and vulnerable narcissism, as captured by both the long and short forms of the measure. In line with the theoretical understanding of grandiose narcissism consisting of an inflated sense of self, with a strong focus on vanity and physical appearance (Krizan & Herlache, 2018; Sivanathan et al., 2021), we found it was positively associated with self-reported measures of mental health, life satisfaction, relationship satisfaction, and healthy eating and activity behaviors, and negatively related to internalizing pathology. Similarly, in line with our theoretical perspective of vulnerable narcissism consisting of a contingent sense of self, an anxious-avoidant approach to interpersonal relationships, and higher levels of neuroticism (Krizan & Herlache, 2018; Sivanathan et al., 2021), we found it to be positively associated with depression, anxiety, anxious and avoidant attachment styles, and negatively related to life satisfaction, relationship satisfaction, mental health, and healthy eating behaviors.

We also found that the five subscales replicated most of the expected external correlations, providing support for the robustness and conceptual clarity of the five-factor structure. Nevertheless, there were a few findings that are worth considering further. We found that the Grandiose Narcissism subscale exhibited good convergent and discriminant validity. It was positively correlated to other measures of grandiosity and not significantly correlated with measures of vulnerability, whilst NARQ and FFNI were positively correlated to all measures. This indicates that the UNS-R Grandiose Narcissism subscale has strong convergent and discriminant validity. On the other hand, the Vulnerable Narcissism subscale was positively correlated to all measures of grandiosity and vulnerability. This was the case for the HSNS and FFNI as well. These findings suggest that either measures of vulnerable narcissism have poorer discriminant validity or that there is a higher co-occurrence of grandiosity in those who score higher on vulnerability, whereas the opposite may not be the case.

We found that the UNS-RSF Grandiose Narcissism subscale was not significantly correlated with anxiety, contrary to our predictions. Therefore, it would be helpful to further assess the external validity of the short form using alternative measures of anxiety. Similarly, vulnerable narcissism was not significantly correlated with exercise across all measures of vulnerable narcissism. These findings could suggest that vulnerable narcissism is more related to a higher occurrence of disordered eating behaviors than to a lower occurrence of healthy eating and exercise behaviors. When we consider the specific five factors, we found interesting relationships for Grandiose Fantasy. Contrary to our expectations, it did not yield a significant negative correlation with life and achievement satisfaction. We expected this finding because we thought the underlying Grandiose Fantasy was a dissatisfaction with reality; however, the non-significant relationship may suggest Grandiose Fantasy as a feature of vulnerable narcissism is less driven by the low sense of self, but perhaps more by the sense of entitlement.

We found that the UNS-R Grandiose Narcissism subscale explained the additional unique variance of the outcome variables over the influence of the FFNI and the NARQ. We also found that the UNS-R Grandiose Narcissism subscale showed the most cogent pattern of external correlations in comparison to FFNI and NARQ, providing support for its incremental validity. We found the Vulnerable Narcissism subscale of FFNI to explain more unique variance in outcome variables alluding to the incremental validity of the FFNI. Despite having only 15-items, the UNS-R short form measure is a brief yet effective measure of grandiose and vulnerable narcissism. We do, however, note that the outcome variables in the study are only a starting point, and there is scope for future research to expand this nomological network.

As a recommendation for future researchers, we recommend the use of the UNS-R/UNS-RSF to capture grandiose and vulnerable narcissism over the FFNI for multiple reasons. Firstly, both our measures are shorter than the shortest FFNI, allowing for a time-efficient way of capturing narcissism with minimal participant burden, especially in a research landscape that utilizes survey designs. Secondly, our measure of grandiose narcissism incremented on the FFNI, and though our measure of vulnerable narcissism did not, it still comprehensively captured the construct of vulnerable narcissism. Finally, choosing a measure for a study is not only based on incremental validity but also on content and alignment with contemporary conceptualisations. Our measure and conceptualization of narcissism are more closely tied to the entitled core of narcissism as described in contemporary literature (Dinić et al., 2021; Edershile & Wright, 2021; Krizan & Herlache, 2018), whereas the FFNI appears to incorporate facets we would consider as correlates of narcissism. For example, distrust is more clearly a core aspect of Machiavellianism according to many conceptualisations of Machiavellianism than of narcissism (see Monaghan & Bizumic, 2023). Furthermore, there seems to be a strong overrepresentation of neuroticism within the FFNI vulnerable narcissism, questioning whether it captures the other entitled features of vulnerable narcissism sufficiently. We do, however, note these findings are based on one study and one sample and recommend further replications.

General discussion

In this paper, we report on two studies, where we constructed a culturally-invariant prototype short form and assessed the incremental validity of UNS-R and its short form, and expanded its nomological network. We have illustrated the ability of the UNS-R Grandiose Narcissism subscale to increment existing measures of grandiose narcissism and shown the same for the UNS-R Vulnerable Narcissism subscale incrementing the HSNS (Hendin & Cheek, 1997). We have also shown the robustness of our factor structure in multiple samples and the external validity of the dimensions of narcissism. Therefore, the UNS-R and its short form are strong measures of both grandiose and vulnerable narcissism with its short form being especially suited for cross-cultural research. Our findings have advanced the field, as we now have a measure of narcissism that can be readily used in cross-cultural research to further our understanding of narcissism across cultures.

Beyond the advancements to the measurement of narcissism, our findings have also strengthened our theoretical perspectives on narcissism. We have confirmed the dimensional structure of narcissism as a multidimensional trait consisting of grandiose and vulnerable subtypes, in line with past research (Krizan & Herlache, 2018). We have also shown that the underlying dimensions of grandiose and vulnerable narcissism are conceptually separate and relate differently to external variables. We have illustrated that Leadership captures a superiority and overvaluation of one’s abilities and competence, that Vanity captures a strong focus on physical attributes, and that these two dimensions represent grandiose narcissism. We have illustrated that Contingent Self-Esteem captures the sense of self that needs external validation, whereas Hiding One’s Needs captures the shame and fear experienced about one’s inner world. We also found that the Grandiose Fantasy dimension of vulnerable narcissism is perhaps not as maladaptive as the other dimensions. Collectively, we found Contingent Self-Esteem, Grandiose Fantasy, and Hiding One’s Needs to represent the underlying dimensions of vulnerable narcissism. Finally, in our cross-cultural comparisons, we also allude to the vertical versus horizontal dimension of cross-cultural comparisons as an important dimension upon which we can understand group differences in narcissism (Singelis et al., 1995), noting that these are preliminary findings and require further exploration.

It is important to consider the findings of these studies within the context of their limitations. Firstly, despite constructing a prototype short form, our assessment of its broader validity in other samples is limited. The non-Australian samples did not include expansive external validity criterion variables. Therefore, it is crucial that future research expands the nomological network of the short form measure in other cultures. One of the major limitations of both studies is the samples used. The Sri Lankan sample was predominantly female and recruited via social media platforms, and all other samples were recruited via online research participation platforms that likely represent the educated, urban, and middle-income parts of the population, limiting the generalizability of findings to other cross-sections of the population. Further, the differences in recruitment methods employed for each sample could have impacted our findings. Therefore, the conclusions about cross-cultural comparisons should be considered tentative, and future research should replicate these findings in more representative samples. Future research should also aim to collect data using more diverse methods of data collection, to allow for a more representative and diverse sample.

Both studies primarily relied on self-report measures and measures of behavioral intentions, with limited inclusion of behavioral data or informant measures. Including only self-report data limits the predictive validity of the measure, and future research should integrate self-reported personality traits with behavioral patterns collected over time to determine if narcissism can predict certain behavioral patterns. Along this vein, it would be useful to integrate longitudinal data to assess the test-retest reliability of the measure. To assess the construct specificity of our five subscales in Study 1, we used several single item measures that have not been validated in previous research. Future research should consider expanding the nomological network with the inclusion of validated measures. The short-form measure has not been validated in an independent sample, and therefore, future research should validate this in a separate sample and, in that process, also compare its performance to other short-form measures of narcissism.

In conclusion, we have demonstrated the utility of the UNS-R as a measure of grandiose and vulnerable narcissism that reflects our contemporary understanding of the construct. We have also identified a culturally-invariant 15-item prototype short form that can be readily used to conduct cross-cultural research, allowing for future expansion of our cross-cultural understandings of narcissism. We further highlighted the relevance of verticality to understanding cross-cultural manifestations of narcissism in moving forward our theoretical perspectives on the construct. Across these two studies, we have furthered our understanding of narcissism and provided a measure of narcissism that can be used further to expand our universal understanding of narcissism, moving away from reliance on only Western samples.

Open scholarship

This article has earned the Center for Open Science badges for Open Data and Open Materials through Open Practices Disclosure. The data and materials are openly accessible at https://osf.io/kb7j9/?view_only=6b08c31b46ad47f0abff2677ddd25ecd. To obtain the author’s disclosure form, please contact the Editor.

Declaration of interest

No potential conflict of interest was reported by the author(s).

Data availability statement

All data have been made publicly available at the Open Science Framework and can be accessed at https://osf.io/kb7j9/?view_only=6b08c31b46ad47f0abff2677ddd25ecd.

Notes

1 Following the suggestion of an anonymous reviewer we conducted the same analyses utilising the three-factor model of FFNI. We found the three-factor model of FFNI to show confusing external relationships alluding to construct creep. These findings can be found in the Supplemental Materials 6 for interested readers. These analyses did not change any of the findings for the UNS-R measure.

Appendix A:

Unified Narcissism Scale-Revised Short Form Items

Contingent Self-Esteem

When others don’t notice me, I start to feel worthless.

When people don’t notice me, I start to feel bad about myself.

I spend a lot of time thinking about what other people think of me.

Leadership

As a leader, I know what is best for my team.

If I became a leader, I would be the best.

I have better leadership skills than other people.

Grandiose Fantasy

I want to amount to something in the eyes of the world.

I often fantasize about being admired and respected.

I fantasize about being a hero.

Vanity

I think I turn heads when I walk down the street.

I am exceptionally good looking.

I enjoy taking photos of myself because I look so good.

Hiding One’s Needs

It’s hard to show others the weaknesses I feel inside.

I would feel so ashamed if someone found out all parts of me.

I become a different person when I am with others for fear of ­disapproval.