Abstract
The study of emotional intelligence (EI) and its relationship with the dark triad has emerged as a popular research area. However, the complex nature of the dark triad and EI, including multiple measures for assessment, has led to inconsistent findings. A systematic review was conducted to focus on the multifaceted nature of the dark triad traits. Included studies must have been conducted with adult samples using standardized EI and dark triad measures. Forty-eight studies were identified; all bar one reported overall negative associations between the dark triad and EI. These associations were more complicated than expected. Further examination found these relationships significantly differed when examined at the facet level. Our results highlight that future research requires more than a one-size-fits-all approach.
There has been a recent surge of interest in the relationship between emotional intelligence (EI) and the dark triad (Muris et al., 2017). EI rose to popularity in the research (Mayer & Salovey, 1990) and lay fields (Goleman, 1995) in the 1990s, while the dark triad was introduced in 2002 (Paulhus & Williams, 2002). Dark triad traits have been linked to success in the workplace, sexual pursuits, and attaining leadership positions (Jonason & Kavanagh, 2010; Jonason et al., 2009; Lilienfeld et al., 2012; Spurk et al., 2016). Considering the apparent success of those high in dark triad traits coupled with the reported role of EI in successful life outcomes, it is not surprising there has been growing interest in exploring the emotional competencies of individuals with high levels of dark triad traits. That is, individuals with high EI know how to act to achieve social, personal, and professional goals. Interest has grown over the past decade into how positive life outcomes and interpersonal success in individuals with high EI interact with the darker personality traits (i.e., the dark triad). The combination of high dark triad trait levels combined with high EI levels has given rise to the concept of a dark EI (Austin & O’Donnell, 2013). Systematically reviewing the association between EI and the dark triad enhances the understanding of the complexities of manipulation and interpersonal exploitation.
To date, there have been two recent meta-analyses looking at these relationships. The first (Megías et al., 2018) reviewed the relationship between total psychopathy and ability EI, finding a negative association between psychopathy and EI. The second meta-analysis (Miao et al., 2019) reviewed the associations between narcissism, Machiavellianism, psychopathy, and both ability and trait EI. Miao et al. found a negative relationship between trait and ability EI with both psychopathy and Machiavellianism. Narcissism was unrelated to either ability or trait EI. Notably, neither of these meta-analyses considered the multifaceted nature of the dark triad traits. Understanding the relationship between the dark triad traits and other constructs requires examining the potentially divergent association patterns at the facet level of these traits. As such, this systematic review aims to provide a comprehensive investigation of the relationship between the dark triad and EI at the facet level. An examination of the existing literature in the following sections describes key elements and theories of the dark triad and EI to justify an expected pattern of results between the dark triad and EI. Further, this review provides a timely overview of a rapidly growing area of research interest.
Emotional Intelligence
EI was popularized in 1995 with Daniel Goleman’s book, Emotional Intelligence: Why It Can Matter More than IQ. This sudden rise in popularity in research and industry led to researchers developing a myriad of theories and assessments. Following a call for clarity in this rapidly expanding field, a distinction was made between ability and self-report measurement of EI (Mayer et al., 2000). EI has been linked to various outcomes, including academic success (Ashkanasy & Dasborough, 2003; MacCann & Roberts, 2008). A recent meta-analysis found that higher EI students tend to have better academic performance, with a greater relationship for ability EI than trait EI (MacCann et al., 2020). EI has also been linked to improved mental health and well-being (Schutte et al., 2007). Further, individuals high in EI are more successful in social situations than their less emotionally intelligent peers (Mavroveli et al., 2009). A 2014 meta-analysis also found higher romantic and marital relationship success for couples with higher EI levels (Furnham et al., 2012; Malouff et al., 2014).
Ability Emotional Intelligence
Ability EI (AEI) is defined as a set of abilities measured with tests of maximum performance and positively correlated with intelligence (Mayer et al., 2004). The most commonly used AEI measure is the Mayer–Salovey–Caruso Emotional Intelligence Test (MSCEIT; Mayer et al., 2004). AEI is based on a hierarchical four-branch model introduced by Mayer and Salovey (1997), comprising (a) perceiving emotions, (b) using emotions, (c) understanding emotions, and (d) managing emotions. Perceiving emotions is understood as the ability to identify emotionally salient visual and auditory information in others and oneself while accurately perceiving one’s own emotions. Using emotions relates to the ability to use the information one has identified to guide one’s decision-making and problem-solving approach. Understanding emotions is understood as integrating knowledge about emotions such as emotional vocabulary, the evolution of emotions over time, or understanding the impact of one’s emotions on others. Finally, managing emotions is understood as the ability to manage the emotions of others and oneself to achieve a desired outcome (Mayer et al., 2011). While the four-branch model is the currently accepted model of AEI, there have been concerns raised about the theoretical model structure as measured by the MSCEIT (Maul, 2012). These concerns largely relate to the perceived lack of causal explanations of item responses (Maul, 2012) and the inability to discriminate EI at the upper end of the trait (Fiori et al., 2014). Regardless, the MSCEIT remains the foremost measure of AEI.
Trait Emotional Intelligence
Trait EI (TEI) is operationalized by self-report measures, positively correlated with personality traits (Petrides et al., 2004), and often understood as an emotion-relevant personality trait (Bucich & MacCann, 2018). TEI is considered a collection of self-perceptions about one’s behavior, dispositions, and self-perceived emotional abilities (Petrides et al., 2004). Furnham and Petrides (2003) argue that emotional experience is inherently subjective, with TEI measures thought to capture this subjective experience more thoroughly than AEI. TEI has been positively linked to substantial happiness (Furnham & Petrides, 2003) and the size of one’s social group, but not to the quality of social group interactions, life satisfaction, or health (Austin et al., 2005).
The Dark Triad
The dark triad is an umbrella term referring to a constellation of three socially aversive personality traits comprised of narcissism, Machiavellianism, and psychopathy (Paulhus & Williams, 2002). These traits are associated with ethical, moral, and socially deviant behavior. There are overlapping characteristics shared among these traits, though debate remains about the presence of a common core. Despite this ongoing debate, it is becoming more widely accepted that antagonism is the common core connecting narcissism, Machiavellianism, and psychopathy (Jones & Figueredo, 2013; Jones & Neria, 2015; Paulhus & Williams, 2002). Furthermore, there is a shared exploitative demeanor across the three traits enabling the goal-focused manipulation of others’ emotions.
Jonason et al. (2012) found the dark triad traits predicted a preference for short-term sexual relationships over long-term commitments. Further, Jonason et al. (2009) found the dark triad facilitated exploitative behavior in men’s short-term mating behavior. These studies support the notion that exploitative behaviors link the dark triad traits. In contrast, workplace relationships appear more complex. A recent meta-analysis of workplace behavior found individuals higher in dark triad traits engaged in counter-productive workplace behaviors, such as bullying and stealing, more frequently than those with lower dark triad scores (O’Boyle et al., 2012).
However, the dark triad traits have also influenced career success. Individuals with higher narcissism scores reported receiving higher salaries, while those with higher Machiavellianism scores attained more influential leadership positions than individuals with lower levels of narcissism and Machiavellianism (Spurk et al., 2016). Additionally, Volmer et al. (2016) found that leaders high in narcissism provided employees with a pathway to achieving subjectively positive, short-term career goals and advancement opportunities.
Narcissism
Traditionally recognized as a personality disorder, narcissism’s subclinical characteristics have been found in the broader population on a spectrum of mild to extreme (Raskin & Hall, 1979; Samuel & Widiger, 2008). As such, narcissism is commonly regarded and investigated as a personality trait (Miller & Campbell, 2008). Narcissism is widely understood as patterns of behaviors and attitudes aligned with grandiosity, self-importance, and desire for admiration (Krizan & Herlache, 2017). More succinctly, Krizan and Herlache (2017) defined narcissism as possessing entitled self-importance. Additionally, narcissism has been characterized by a desire to appear socially desirable and interpersonally manipulative (Dickinson & Pincus, 2003).
Narcissism comprises two distinct but related factors: grandiose and vulnerable narcissism (Besser & Priel, 2010; Grijalva et al., 2015; Krizan & Herlache, 2018; Miller et al., 2011; Rohmann et al., 2012; Wink, 1991). Though grandiose and vulnerable narcissism share several core characteristics such as an aggressive interpersonal style, self-importance, entitlement, and hypersensitivity to criticism (Dickinson & Pincus, 2003; Krizan & Herlache, 2018; Rauthmann & Kolar, 2012; Weiss et al., 2019), the diversity between these traits is evident in the manifestation of these characteristics (Krizan & Herlache, 2017; Miller et al., 2018). For example, individuals high in grandiose (or overt) narcissism are characterized by grandiosity, self-confidence, exploitation of others, and reliance on self/internal validation. When threatened, individuals high in grandiose narcissism tend to blame and devalue others while denying their own weaknesses (Dickinson & Pincus, 2003; Zhang et al., 2017). Individuals high in grandiose narcissism have been found to reappraise negative situations to maintain self-confidence (Walker, Gorodezki et al., 2021). Prior research indicates that reappraisal requires an individual to perceive, use, understand, and regulate emotions adequately (Megías-Robles et al., 2019; Smieja et al., 2011). Therefore, it is expected that grandiose narcissism would be positively associated with EI. Comparatively, vulnerable narcissism tends to rely on external validation. It is also characterized by grandiose fantasies that tend to oscillate between self-love and self-loathing, thereby exhibiting fragile self-esteem, defensiveness, and resentment (Weiss et al., 2019). Individuals high in vulnerable narcissism are hypersensitive to negative feedback and tend to act aggressively when their sense of self is threatened (Wink, 1991). Additionally, it has been argued that vulnerable narcissism is more a disorder of neuroticism characterized by distinct differences in the experience of distress between vulnerable and grandiose narcissism (Miller et al., 2018). Based on difficulties with emotional expression and management, it is expected that vulnerable narcissism will be negatively associated with EI.
Machiavellianism
Unlike narcissism and psychopathy, Machiavellianism is most commonly understood as a unidimensional construct characterized by goal-focused manipulative and callous social interactions (Christie & Geis, 1970). Individuals high in Machiavellianism tend to be viewed as strategic, capable of delaying gratification for bigger and better rewards in the future, possessing low moral commitment, and engaging in long-term strategic planning with a cold and cynical world view (Christie & Geis, 1970; Furnham et al., 2013). Machiavellianism was derived by Christie and Geis (1970) from the writing of Niccolo Machiavelli, a political advisor to the Medici family in 16th-century Firenze. Machiavelli’s ideas were captured in his seminal work Il Principe (1513). Il Principe contains advice to rulers about successful governance, mainly through others’ fortune and power (Machiavelli, 1891). Specifically, Machiavelli asserts that the use of manipulation and cunning is more successful than truthfulness.
Very little is understood about the underlying mechanisms driving the development of Machiavellianism as a personality trait. There is substantially less empirical work in the Machiavellianism literature than either in the literature on narcissism or psychopathy. Additionally, theoretical development has been minimal since Christie and Geis’s theoretical and empirical work in the 1970s. The limited empirical work may be due, in part, to the problematically high intercorrelations between psychopathy and Machiavellianism (Miller et al., 2017). A recent meta-analysis found the corrected Machiavellianism–psychopathy correlation to be .59 (O’Boyle et al., 2012). These high intercorrelations make it challenging to ascertain how much the relationship between Machiavellianism and other constructs is influenced by psychopathy. Current Machiavellianism and psychopathy scales appear to be measuring the same construct despite theoretical distinctions (Miller et al., 2017). Recent findings suggest the intercorrelations are reduced by revising existing measurement tools (Collison et al., 2018). Collison et al.’s (2018) development of the Five-Factor Machiavellianism Inventory has provided a timely reevaluation of Machiavellianism’s theory and measurement.
The current debate surrounding accurate measurement and existing empirical evidence makes it difficult to effectively predict the type of relationship expected between Machiavellianism and EI. From a theoretical perspective, we might expect to see a strong positive relationship between Machiavellianism and EI. However, based on the empirical work in the literature to date and the aforementioned measurement issues, we expect a negative association between Machiavellianism and EI.
Psychopathy
Psychopathy has been typified by interpersonal, affective, and behavioral characteristics, including superficial charm, pathological lying, and lack of empathy/conscience/remorse (Cleckley, 1941; Hare, 1991, 2003). Often confused with antisocial personality disorder, the subclinical characteristics of psychopathy, like narcissism, have been found in the broader population on a continuum (Berg et al., 2013). Based on Karpman’s (1941) work, enduring classifications of psychopathy distinguish between two subtypes differing on their etiology and symptomology. Hare (1991, 2003) expanded on earlier work, developing a model of psychopathy distinguishing primary and secondary psychopathy. Both facets of psychopathy involve affective elements suggesting some indifference to their own and others’ emotions, with each underpinned by an antagonistic interpersonal style (Miller & Lynam, 2012). It is generally understood psychopathy lies on a continuum. However, individuals may manifest both primary and secondary psychopathy, with one dominating facet providing a better description of an individual’s pattern of psychopathic personality (Karpman, 1941).
Primary psychopathy is characterized by (a) a lack of guilt and remorse, with elevated levels of callousness, manipulation, and socially desirable behavior (Hare, 1991, 2003), (b) having superficial affect (Casey et al., 2013), and (c) deficits in affective empathy (Wai & Tiliopoulos, 2012). Primary psychopathy is also associated with lower levels of fear (Patrick et al., 1993) and lower indications of repentance (Hare, 2003; Lee & Salekin, 2010) than seen in nonpsychopaths. In contrast, secondary psychopathy is characterized by higher levels of antisocial behaviors such as aggressiveness, impulsivity, and anxiety (Levenson et al., 1995). These characteristics are potentially a result of experiencing strong emotions, which individuals cannot regulate effectively (Hare, 2003). Additionally, individuals high on secondary psychopathy have been shown to possess guilt and fear responses not typically observed in individuals high in primary psychopathy (Lykken, 1995).
Primary psychopathy is considered affective and interpersonal (Cleckley, 1941) with goal-directed interest in maintaining social desirability (Hare, 1991, 2003). Therefore, it seems reasonable to assume that an individual high on primary psychopathy should possess the ability to identify and understand others’ emotions to manage them. As such, it is expected that there will be a positive relationship between primary psychopathy and EI. In contrast, secondary psychopathy comprises antisocial and impulsive behaviors that tend toward aggressive outbursts (Hare, 1990, 2003). That is, those with high levels of secondary psychopathy exhibit a lack of emotional awareness coupled with excessive emotional reactions. Therefore, it is expected that there will be a negative association between secondary psychopathy and EI.
Method
Literature Search
A search was conducted in August 2018 and updated in August 2020. The search terms (“dark triad” or “narciss*” or “psychopathy” or “Machiavellian”) AND (“emotional intelligence” or “EI”) yielded 1,249 results from the databases PsychInfo, Web of Science, ProQuest Dissertations & Theses, and Scopus. The search was limited to English language studies. Additionally, reference searches were conducted to identify potential studies that may have been missed in the initial search. The database and reference searches yielded a total of 122 studies that progressed to full-text review to check for specific inclusion and exclusion criteria. The overall literature search resulted in 42 studies containing 48 samples fitting our inclusion criteria (see Figure 1).

Quorum chart of articles included in the review.
Coding
The coding procedure was developed based on Cochrane collaboration standards (Higgins & Green, 2006). Study characteristics comprising author, date, sample size percentage of female participants, mean age, EI instrument, dark triad instrument, EI type (trait vs. ability), dark triad trait, Cronbach’s alpha, and Pearson correlations were extracted and coded into a worksheet. The method and results are presented according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA; Moher et al., 2015). Before beginning this review, a protocol was developed to guide the search data extraction.
Inclusion and Exclusion Criteria
The following inclusion and exclusion criteria were defined for eligibility of studies to ensure only studies that adequately addressed the research questions were included: (a) a dark triad measure and EI measure, both of which must have either a publication manual, a previous journal article, or an academic book chapter providing validity information; (b) effect size information relating to the association between the dark triad trait and EI; (c) only adult samples (18 years and over) were included; and (d) only English language articles were considered. Theoretical articles, systematic reviews, and meta-analyses were excluded.
Results and Discussion
Our search identified 48 studies in total (42 articles). See Table 1 for a complete list of all studies included. Fifteen studies assessed AEI, 37 assessed TEI, and four used both AEI and TEI measures. Seven studies used a forensic sample, 17 used a community sample, and 24 used undergraduate samples, for a total of 14,029 participants. The mean age range of participants was between 18.59 and 58.08 years. Reviewed studies were grouped into two sections: AEI and self-report TEI. A summary of the main results obtained with each measure can be found in Table 1.
Included studies and study characteristics for the systematic review of the relationship between emotional intelligence and the dark triad of personality.
Note.
Perceive, Facilitate, Manage, and Understand = Branches 1–4 of the MSCEIT; Interpersonal, Stress mgt = facets from the EQi; Attention and Repair = attention, and repair facets from the TMMS; Mach = Machiavellianism; TP = total psychopathy; PP = primary psychopathy; SP = secondary psychopathy; GN = grandiose narcissism; VN = vulnerable narcissism; TN = total narcissism; EI = emotional intelligence; np = not provided.
p < .05. **p <.01. ***p < .001.
The results are presented alongside the discussion separately for each of the dark triad traits. A General Discussion section will follow this combined Results and Discussion section. The purpose of presenting the results and discussion for each of the dark triad traits together is to help the reader interpret each construct’s results.
Narcissism
Eighteen studies measured the relationship between narcissism and EI. Of these studies, nine used the Narcissistic Personality Inventory (NPI; Austin et al., 2014; Link & Bailey, 2010; Petrides et al., 2011; Stead & Fekken, 2014; Szabó & Bereczkei, 2017; Yocum, 2007; Zacher et al., 2014; Zajenkowski et al., 2018), two used the Pathological Narcissism Inventory (PNI; Vonk et al., 2013, 2015), two used the Short Dark Triad (Plouffe et al., 2017), five used Hendin and Cheek’s (1997) Hypersensitive Narcissism Scale (Austin et al., 2014; Casale et al., 2019; Srivastava et al., 2020; Zajenkowski et al., 2018), one used Foster et al.,’s (2015) Grandiose Narcissism Scale (Casale et al., 2019), and three used the Dirty Dozen (Hyde et al., 2020; Jauk et al., 2016; Saklofske et al., 2016). Of the 18 studies, four studies used AEI measures (Jauk et al., 2016; Vonk et al., 2015; Yocum, 2007; Zajenkowski et al., 2018), with correlations ranging from −.26 to .10. Seventeen studies (Austin et al., 2014; Casale et al., 2019; Hyde et al., 2020; Jauk et al., 2016; Link & Bailey, 2010; Petrides et al., 2011; Plouffe et al., 2017; Saklofske et al., 2016; Srivastava et al., 2020; Stead & Fekken, 2014; Szabó & Bereczkei, 2017; Vonk et al., 2013; Zacher et al., 2014; Zajenkowski et al., 2018) used TEI measures, with correlations ranging from −.55 to .49.
The dual nature of narcissism is well established in the literature (Pincus et al., 2009; Rohmann et al., 2012). As such, it is not surprising there were divergent relationships at the facet level found in this review. As expected, grandiose narcissism was positively related to AEI and TEI, while vulnerable narcissism was negatively related to AEI and TEI, though there were three exceptions. The first exception, Zajenkowski et al. (2018), measured the relationship between narcissism facets using both AEI and TEI scales. While Zajenkowski et al. reported a positive correlation of .49 between TEI and grandiose narcissism, and a negative correlation of −.55 between TEI and vulnerable narcissism, these differed from the AEI relationships. A correlation of −.11 was reported for AEI and grandiose narcissism, and a correlation of −.12 was reported for AEI and vulnerable narcissism, though these did not reach significance. Closer examination revealed that vulnerable narcissism was positively related to the understanding branch of AEI, correlating at .10. In contrast, grandiose narcissism was negatively related to the understanding branch of AEI, correlating at −.15.
The second exception was Vonk et al. (2015), who used the Situational Test of Emotion Understanding (STEU; MacCann & Roberts, 2008) to measure AEI. Similarly, they found a small negative correlation of −.05 between understanding and grandiose narcissism. However, in contrast to Zajenkowski et al. (2018), Vonk et al. found a negative correlation of −.05 between understanding and vulnerable narcissism. Similarly, Zajenkowski et al. found grandiose and vulnerable narcissism were negatively related to the managing branch of the MSCEIT, whereas Vonk et al. found an inverse relationship between the two using the Situational Test of Emotion Management (STEM; MacCann & Roberts, 2008). That is, Vonk et al. found that vulnerable narcissism was negatively related to management, while grandiose narcissism was positively related to management. While only two studies presented results across both AEI and TEI measures, it is unclear whether these differences are due to something systematically different between the measures of AEI or narcissism, or whether there are specific emotional competency differences found at the facet level of narcissism and the branch level of AEI. The final exception was Yocum (2007), who used the MSCEIT to measure the four-branch model of AEI and grandiose narcissism. The only significant association reported was a correlation of −.28 between the understanding branch of AEI and grandiose narcissism.
EI results range from −.55 to .49, with grandiose narcissism consistently positively related to TEI, and vulnerable narcissism consistently negatively related to TEI. These differences between AEI and TEI echo the idea that TEI measures reflect self-knowledge and significantly correlate with personality (Petrides et al., 2007). However, the extent to which self-knowledge is accurate for individuals high in narcissism is unclear. Carlson et al. (2011) found that individuals high in narcissism were somewhat aware of how others perceived them. Furthermore, John and Robins (1994) found that individuals high in narcissism tended to evaluate themselves more positively than others perceive them. Petrides (2010) hypothesized that the positive association between grandiose narcissism and TEI may be no more than hubris (Petrides, 2010), which warrants questioning the prevalence of socially desirable responding among those high in self-reported narcissism. Neither Carlson et al. nor John and Robbins examined their findings at the facet level to determine how self-knowledge differed between grandiose and vulnerable narcissism.
Measurement concerns
There are important differences between the scales used to measure narcissism. For example, studies examining the Dirty Dozen’s psychometric properties show the scale captures a mix of vulnerable and grandiose narcissism items, though it is not possible to differentiate between these facets (Maples et al., 2014; Savard et al., 2017). Two of the three studies using the Dirty Dozen found trivial, positive correlations between narcissism and EI (Hyde et al., 2020; Saklofske et al., 2016), which is possibly due to the Dirty Dozen total scores consisting of grandiose and vulnerable items. In contrast, the Short Dark Triad predominately measures grandiose/entitlement traits (Jones & Paulhus, 2014; Maples et al., 2014). As a result, the two studies using the Short Dark Triad reported a significant, moderate positive correlation between narcissism and EI (Plouffe et al., 2017). These findings provide further evidence that the practice of collapsing into a total narcissism score can provide a misleading account of the relationship between narcissism and other constructs.
Further, we identified several general issues within the empirical literature. First, 60% of participants across all studies of narcissism were female, and 11 of the 18 total reviewed studies used university samples. A meta-analysis by Grijalva et al. (2015) found significant evidence of gender differences, with males scoring higher on grandiose narcissism than females. Additionally, they found that narcissism tends to decline after young adulthood, which is an important consideration for future studies of narcissism and EI when using university samples. In this review, only Jauk et al. (2016) reported gender differences but found no differences between male and female university students.
Further, researchers commonly collapse grandiose and vulnerable narcissism into a total narcissism score. This assumes the two facets measure the same thing. However, while only a few studies measured both grandiose and vulnerable narcissism, grandiose narcissism tends to have a small positive relationship with EI, and vulnerable narcissism tends to have a substantial negative relationship with EI. These results provide additional evidence that grandiose and vulnerable narcissism should be considered separately rather than combined into a total score.
Finally, researchers sometimes measure only one facet of narcissism yet report their findings on narcissism instead of reporting their findings on either grandiose or vulnerable narcissism specifically. This tendency goes against the predominant theory of narcissism, which includes both facets as sharing underlying characteristics but being distinct from each other (Wink, 1991). Furthermore, collapsing grandiose and vulnerable narcissism scores into a total score does not accurately represent the relationships with other constructs. Doing so removes the opportunity to investigate the important outcome differences of each facet. It appears that the complex relationship between EI and narcissism requires further investigation along with careful methodological consideration coupled with a strong theoretical foundation.
Machiavellianism
Eighteen studies measured the relationship between Machiavellianism and EI. Of these 18 studies, five used AEI measures (Austin et al., 2007; Baytalskaya, 2008; Côté et al., 2011; Jauk et al., 2016; Vonk et al., 2015), with correlations ranging from −.32 to .03. Fifteen studies (Ali et al., 2009; Austin et al., 2007, 2014; Hyde et al., 2020; Jauk et al., 2016; O’Connor & Athota, 2013; Petrides et al., 2011; Plouffe et al., 2017; Saklofske et al., 2016; Stead & Fekken, 2014; Szabo & Bereczkei, 2017) used TEI measures, with correlations ranging from −.60 to .04. Two studies used both AEI and TEI measures. Thirteen of 18 studies measured Machiavellianism using the Mach-IV (Christie & Geis, 1970), three used the Dirty Dozen (Jonason & Webster, 2010), and two used the Short Dark Triad (Jones & Paulhus, 2014). As expected, Machiavellianism was negatively associated with EI across 14 of the 18 studies, with correlations ranging from −.60 to −.03. Jauk et al. presented a positive but not significant association of .04 between Machiavellianism and TEI. Baytalskaya also presented a positive but not significant association of .03 between Machiavellianism and AEI.
Unfortunately, only two of the included studies reported their findings on sex differences (Austin et al., 2007; Jauk et al., 2016), and one on age differences (Austin et al., 2007). Their results were consistent with prior research. Females had higher levels of AEI, males had higher levels of Machiavellianism and nonsignificantly higher TEI levels, and students had higher levels of Machiavellianism compared to a community sample. However, given there were so few studies, the degree to which this review can provide a cohesive overview of sex and age differences in Machiavellianism is limited.
An area in which Machiavellianism research is growing is within the dark EI literature. The dark EI concept posits that EI has a dark side, whereby individuals with high levels of emotional manipulation can successfully achieve personal goals through the emotional manipulation of others (Austin et al., 2007). This concept moves away from the typical positive conceptualization of EI, instead offering an additional perspective on the interpersonal impact of high EI, particularly when combined with morally dubious attitudes and behavior.
A core characteristic of Machiavellianism is the manipulation of others for personal gain. Indeed, it has been suggested that in order for manipulation to be successful, a certain level of EI is required (Austin et al., 2007; Côté et al., 2011). The pattern of results in this review presents a convincing strong negative association between EI and Machiavellianism. Austin et al. suggest this may be due to the inherently positive elements of existing EI scales, which do not tap into some of the darker attitudes and behaviors associated with EI (i.e., interpersonal manipulation). Presently, it is unclear what underlies the degree to which interpersonal emotionally manipulative behavior is successful.
Measurement concerns
Like for narcissism, there are distinct differences between AEI and TEI measures and their association with Machiavellianism. The relationship between total AEI and Machiavellianism sits around −.20, whereas total TEI and Machiavellianism have a much wider range of correlations between −.60 and .04. This variability seems to revolve around the use of short dark triad combination measures. Specifically, studies that used Jonason and Webster’s (2010) Dirty Dozen (Hyde et al., 2020; Jauk et al., 2016; Saklofske et al., 2016) and Jones and Paulhus’s (2014) Short Dark Triad (Plouffe et al., 2017) revealed nonsignificant, small negative correlations between Machiavellianism and TEI. By comparison, studies using the Mach-IV (Christie & Geis, 1970) show less variability among studies, with correlations ranging from −.23 to −.42. The only exception to this was Szabo and Bereczkei (2017), which reported a −.13 association between Machiavellianism and TEI using the Self-Report Emotional Intelligence Scale (SREIS; Schutte et al., 1998). This anomaly may be due to the limited set of TEI facets measured by the SREIS, which is based on an earlier model of AEI (Gardner & Qualter, 2010; Mayer & Salovey, 1990). Without the full coverage of TEI facets, the SREIS may not be measuring the association with Machiavellianism as precisely as the other TEI measures.
Psychopathy
Thirty-six studies measured the relationship between psychopathy and EI. Of these 36 studies, 10 used AEI measures (Copestake et al., 2013; Curci et al., 2017; Ermer et al., 2012; Howe et al., 2014; Jauk et al., 2016; Ling et al., 2018; Lishner et al., 2011; Vidal et al., 2010; Visser and Bereczkei, 2010; Vonk et al., 2015), with correlations ranging from −.52 to .33. Twenty-nine studies used TEI measures, with correlations ranging from −.66 to .56 (Ali et al., 2009; Austin et al., 2007, 2014; Copestake et al., 2013; Fix et al., 2015; Garofalo et al., 2020; Grieve & Mahar, 2010; Grieve & Panebianco, 2013; Grieve et al., 2014; Hyde & Grieve, 2014; Hyde et al., 2020; Jauk et al., 2016; Malterer et al., 2008; Owens et al., 2018; Petrides et al., 2011; Pham et al., 2010; Plouffe et al., 2017; Porter et al., 2011; Ray et al., 2011; Sacco et al., 2016; Saklofske et al., 2016; Stead & Fekken, 2014; Szabo and Bereczkei, 2017; Watts et al., 2016).
Eleven of the 36 studies measuring the relationship between psychopathy and EI used Levenson et al.’s (1995) Levenson Self-Report Psychopathy Scale (LSRP; Ali et al., 2009; Austin et al., 2014; Grieve & Mahar, 2010; Grieve & Panebianco, 2013; Grieve et al., 2014; Hyde and Grieve, 2014; Szabo and Bereczkei, 2017; Vonk et al., 2015), nine used Lilienfeld and Andrews’s (1996) Psychopathy Personality Inventory-Revised or short form (PPI-R; Copestake et al., 2013; Fix et al., 2015; Garofalo et al., 2020; Howe et al., 2014; Ray et al., 2011; Sacco et al., 2016; Vidal et al., 2010; Watts et al., 2016), five used Paulhus et al.’s (2012) Self-Report Psychopathy Scale-III (SRP-III; Lishner et al., 2011; Owens et al., 2018; Petrides et al., 2011; Stead & Fekken, 2014; Visser et al., 2010), seven used Hare’s (2003) Psychopathy Checklist-Revised (PCL-R; Copestake et al., 2013; Curci et al., 2017; Ermer et al., 2012; Garofalo et al., 2020; Ling et al., 2018; Malterer et al., 2008; Pham et al., 2010), two used Jones and Paulhus’s (2014) Short Dark Triad (SDT; Plouffe et al., 2017), one study used Paulhus et al.’s (2016) Self-Report Psychopathy Scale-IV (SRP-IV; Porter et al., 2011), and three used Jonason and Webster’s (2010) Dirty Dozen (DD; Hyde et al., 2020; Jauk et al., 2016; Saklofske et al., 2016).
Eighteen studies reported negative correlations ranging from −.41 to −.03 between total psychopathy and EI (Copestake et al., 2013; Ermer et al., 2012; Fix et al., 2015; Garofalo et al., 2020; Howe et al., 2014; Hyde et al., 2020; Ling et al., 2018; Lishner et al., 2011; Malterer et al., 2008; Petrides et al., 2011; Plouffe et al., 2017; Porter et al., 2011; Ray et al., 2011; Saklofske et al., 2016; Stead & Fekken, 2014; Vidal et al., 2010; Visser et al., 2010). These negative correlations were spread across both AEI and TEI, with total AEI displaying slightly larger negative correlations with total psychopathy than total TEI. This may be due to individuals overstating their emotional competencies on self-report measures of EI to make themselves appear more emotionally intelligent (Petrides, 2011). It does not appear that sample characteristics drove these results.
A more complex pattern begins to emerge when examining psychopathy at the facet level. As expected, there were positive correlations ranging from .05 to .33 between total AEI and primary psychopathy, but only when the PPI-R was used to measure psychopathy. However, when the SRP, LSRP, or PCL-R were used, there were negative correlations with AEI ranging from −.47 to −.01. Also, as expected, secondary psychopathy was consistently negatively related to total AEI across all measures of psychopathy, ranging from −.52 to −.10.
Examining the branch level of AEI, primary psychopathy was positively related to perceiving emotions when psychopathy was measured using the PPI-R or the PCL-R, but negatively related when using the SRP-III. The relationship between secondary psychopathy and perceiving emotions was more straightforward, with negative correlations across the studies. A similar pattern emerged for understanding emotions, which was positively associated with primary psychopathy when measured using the PPI-R and PCL-R, but negatively associated with primary psychopathy when measured using either the SRP-III or LSRP. This same pattern emerged for both the facilitation and managing branches of AEI, and their relationship with primary psychopathy. Secondary psychopathy was negatively related to the facilitation and managing branches of AEI.
While levels of secondary psychopathy found in male and female samples were negatively related to AEI, there was a stronger negative relationship found for males than females. In contrast, the negative relationship between primary psychopathy and AEI was substantially larger for females than males (Ermer et al., 2012; Grieve et al., 2014; Grieve & Mahar, 2010; Jauk et al., 2016), suggesting females with higher levels of primary psychopathy exhibited lower levels of EI than males. These findings highlight the importance of considering gender differences in the EI–psychopathy relationship to enhance our understanding of what underlies interpersonal manipulation.
The relationship between TEI and the facets of psychopathy is significantly more variable than with AEI. Across the reviewed studies, there are both positive and negative correlations between total TEI and primary psychopathy, ranging from −.31 to .56. Six studies report a positive correlation between primary psychopathy and total TEI across undergraduate, community, and forensic samples (Copestake et al., 2013; Grieve & Mahar, 2010; Malterer et al., 2008; Pham et al., 2010, Study 1; Sacco et al., 2016, Studies 1 and 2). A similar pattern emerges for secondary psychopathy and total TEI, whereby secondary psychopathy is generally negatively related to total TEI across the board, ranging from −.66 to −.19. However, Pham et al. (2010) and Copestake et al. (2013) used forensic male-only samples, reporting that secondary psychopathy was positively related to total TEI, with correlations of .21 and .39, respectively. By comparison, Howe et al. (2014) using a mixed-gender community sample and Copestake et al. found equally significant moderate negative correlations between secondary psychopathy and EI. These results are consistent with the impulsive and aggressive behavior associated with secondary psychopathy, which is more likely found in forensic than community samples (Falkenbach et al., 2008). Additionally, Copestake et al. used both AEI and TEI measures, reporting opposite results between the two. AEI was negatively related to secondary psychopathy and TEI was positively related to secondary psychopathy. These findings are consistent with prior research demonstrating AEI and TEI are distinct constructs.
With such large variation between and within studies, it is vital to investigate whether results are driven by deceptive responding or whether individuals high in primary and/or secondary psychopathic traits are inhibited by impaired self-reflection. Carlson et al. (2011) found that individuals high in narcissistic traits were more likely to be at least somewhat self-aware of their narcissistic tendencies. Although Lilienfeld and Fowler (2006) discuss the lack of insight and limited realistic self-evaluation of individuals high in psychopathic traits, there does not appear to be an equivalent study to that of Carlson et al. for psychopathy.
Additionally, these results appear to be consistent across samples and genders, although results appear to be influenced by age. Undergraduate samples (mean age of 19.78, collapsed across studies) presented moderate negative correlations (range −.47 to .10) between the managing emotions branch of AEI and primary psychopathy, compared to community/forensic samples reporting small nonsignificant correlations (range −.05 to .16). Differences in EI across the lifespan may explain these findings. Atkins and Stough (2005) found a positive relationship between age and both TEI and AEI, while Kafetsios (2004) found age particularly influenced scores on the managing emotions branch of AEI.
Prior research found that theory of mind was not impaired in individuals exhibiting high levels of psychopathic traits (Richell et al., 2003). This suggests that the capacity to extrapolate another individual’s goals and intentions is intact in high psychopathic individuals. Additionally, a recent meta-analysis reported that primary psychopathy was positively related to cognitive empathy, and secondary psychopathy was negatively related to both cognitive and affective empathy (Wai & Tiliopoulos, 2012). Considering these prior findings relating to theory of mind and empathy, the somewhat inconsistent findings between the relationship of EI with primary and secondary psychopathy are unsurprising. While EI and empathy are separate constructs, they remain somewhat related. As such, these findings similarly lend credence to the importance of exploring the relationship of psychopathy and EI at the facet level of psychopathy rather than collapsing across facets into a total psychopathy score.
The implications of these findings for the study of psychopathy within a forensic setting merit comment. Seven of the included studies used forensic samples incorporating a mix of AEI and TEI measures, and either the PCL-R (Hare, 1991, 2003) or Lilienfeld and Andrews’ (1996) PPI-R (Copestake et al., 2013; Curci et al., 2017; Ermer et al., 2012; Garofalo et al., 2020; Malterer et al., 2008; Pham et al., 2010; Ray et al., 2011). The pattern of results demonstrates a negative association between secondary psychopathy and EI when AEI measures are used, and generally a positive association when TEI measures are used. On the other hand, primary psychopathy is in general consistently positively associated with both AEI and TEI. This association seems consistent whether the psychopathy measure used is the PCL-R or the PPI-R, suggesting that TEI measures may be particularly vulnerable to socially desirable responding. Importantly, these seven studies predominantly tested males, with only Ray et al. (2011) including females in their study, thus limiting generalizability within a forensic context. Taken together, these findings indicate that the use of these measures in a forensic setting should be balanced with a consideration of the impact of socially desirable responding on the validity of results. Researchers might consider using additional measures, such as lie scales or specific statistical approaches, to mitigate the impact of socially desirable responding and improve the validity of their findings.
Measurement concerns
Discrepancies found across the psychopathy measures may reflect the nature of measurement combined with inherent differences between scales. For example, the PPI-R (Lilienfeld & Andrews, 1996) measures central characteristics of psychopathy rather than representing highly discriminative items at each end of the scale (Tsang et al., 2017). In contrast, the SRP (Paulhus et al., 2012) and LSRP (Levenson et al., 1995) were developed to more closely parallel the two-factor model measured by the PCL (Hare, 1991). Examination of the psychometric properties of the SRP and LSRP shows support for a two-factor model capable of discriminating at extreme ends of the scale. When considering the pattern of results presented in this review, it may be that when collapsing the facets of the PPI-R, SRP, and LSRP into a total score, the meaning of those total scores is slightly different between scales.
Additionally, the pattern of results in this review shows that primary psychopathy relates both positively and negatively to EI, while secondary psychopathy is consistently negatively related to EI. As psychopathy theory continues to develop, so too will scales that more adequately capture the entire domain. Previous comparison studies of psychopathy scales suggest that it is necessary to use a multimeasure approach when measuring psychopathy to ensure coverage of the full domain (Tsang et al., 2017).
General Discussion
This systematic review shows that dark triad traits are negatively associated with EI at the domain level. In other words, people with high scores on narcissism, Machiavellianism, and psychopathy have lower EI scores. These patterns were consistent across AEI and TEI. More importantly, these findings provide a unique perspective by exploring the facet-level associations between dark triad traits and EI. That is, on the surface, the pattern of results in this review reinforces prior meta-analytic findings that psychopathy (Megías et al., 2018; Miao et al., 2019), narcissism, and Machiavellianism are negatively related to EI (Miao et al., 2019). However, on closer examination of the studies, the relationship between the dark triad and EI is more complicated than typically portrayed. While each of the dark traits appears to be negatively related to EI, results at the facet level of narcissism and psychopathy tell a different story. Grandiose and vulnerable narcissism were differentially related, such that grandiose narcissism was generally positively related to EI, while vulnerable narcissism was negatively related. The total narcissism score was also negatively related to EI, suggesting this negative association is driven by vulnerable narcissism.
Similarly, primary and secondary psychopathy were differentially related. Unlike narcissism, the positive association between primary psychopathy and EI seems to be a function of the scale used rather than the construct. Again, the total psychopathy score seems largely driven by the negative association between secondary psychopathy and EI. This complexity appears to be due to the limited reporting at the facet level of both the dark triad and EI indices, suggesting these relationships are neither accurately nor adequately explained by collapsing the facet scores into a total score.
Perhaps more concerning is that by collapsing the facets into a total score, it is impossible to discern the differences between the facets and their relationships with other constructs. For example, vulnerable narcissism is considered to have properties relating to poor self-esteem and positively correlates with neuroticism (Clarke et al., 2015; Pincus et al., 2009; Wink, 1991). Both poor self-esteem and neuroticism have been shown to negatively relate to performance on EI tests (Hjalmarsson & Daderman, 2020; Schutte et al., 2002). Therefore, it is reasonable to expect to see important differences between dark triad traits and other constructs, in this case with EI, when exploring the data at the facet level.
While this review includes several studies investigating the dark triad at the facet level, existing studies exploring the relationships between facet-level dark triad traits and other constructs are limited (Tran et al., 2018). Instead, facet-level investigations seem to be tagged onto the end of domain-level studies rather than presenting theoretically grounded interpretations for the differences between dark triad facets. Understandably, this can give the impression that facet-level interpretations have not been well considered (Dane et al., 2018). However, with a theoretical approach underlying interpretation of these findings, it is becoming clear that collapsing results across the dark triad facets presents a misleading analysis of the relationship between the dark triad and EI.
Future Research Directions
Given that the relationship between the dark triad and EI is complex, relying on simple correlations without considering mediators and moderators may be misleading. Though zero-order correlations are often reported as indicators of the association between the dark triad and EI, they capture only a simple linear relationship. They do not, by definition, well represent the nuanced nature of both constructs. Mediation and moderation begin to address these issues. The faceted approach to the conceptualization and measurement of the dark triad and EI may help to uncover this nuance.
Additionally, the development of personality, dark or otherwise (Roberts et al., 2006), and EI (Kafetsios, 2004) over the lifespan may mean that the prevalence of undergraduate samples that have rather narrow age diversity in the studies reviewed makes generalization difficult. This homogeneity is particularly important because prior research has found that young adulthood is a time of significant personality development, coupled with a decline of negative emotionality (Roberts et al., 2006). As such, this may further adversely impact the generalizability of these findings to a wider population.
Additionally, the use of self-report measures of dark triad traits warrants consideration. Though this review did not examine response distortion among the included studies, the high TEI scores in forensic samples suggest there may be an element of impression management worth further investigation. Although contentious, it is well supported that individuals may underreport negative traits and behaviors to appear socially desirable (Muris et al., 2017; Paulhus & Vazire, 2007; Paulhus & Williams, 2002). Response distortion is especially true for persons high in dark triad traits, which is problematic as self-report measures are developed to rely on the respondent’s accurate self-reflection and honest self-assessment (Muris et al., 2017). In fact, a recent meta-analysis found measures of the dark triad were susceptible to faking when participants were instructed to present themselves in either a favorable or unfavorable way (Walker, Double, Birney & MacCann, 2021). These findings were consistent with similar meta-analytic findings using Big Five measures (Viswesvaran & Ones, 1999).
Furthermore, as the core characteristics of the dark triad traits involve impression management and manipulation, self-report measures alone may not be sufficient. A multimethod approach may provide less opportunity for deception. In conjunction with self-report measures, the use of observer ratings, the development and use of measures less susceptible to socially desirable responding such as situational judgement tests, and experience sampling methods provide a few examples of how researchers may choose to gather information beyond the use of traditional self-report measures. Finally, scales such as the Paulhus Deception Scale (Paulhus, 2012) should be used to capture the faceted nature of impression management and self-deception, providing additional checks for response bias when measuring the dark triad.
Importantly, our review shows that there are currently a limited number of studies reporting facet-level relationships between the dark triad and EI. Thus, while a meta-analysis would have been preferable in order to provide quantitative analyses at the facet level, the small sample of studies and inconsistencies in approaches used would limit the reliability of conclusions that might be drawn. Until further empirical research is conducted focusing on a facet-level analysis of the dark triad–EI relationship, a systematic review of the literature is the most appropriate method. Despite this limitation, our work highlights emerging trends in the existing body of literature and the value of pursuing a facet-level approach.
Conclusion
The relationship between EI and the dark triad is complicated and requires more than a one-size-fits-all approach. Our work highlights key differences between facets of dark triad traits and their relationships with facets of EI. Furthermore, it highlights the importance of having a firm theoretical foundation on which to base facet-level research. The state of the field could be clarified by substantial additional research including factors that moderate EI–dark triad facet-level relationships, followed by a meta-analysis. It is hoped that this review’s overall findings will provide the impetus for such research.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
