March 2015 – The past quarter century has seen impressive growth of emotional intelligence (EI) as a topic of interest in the fields of psychology and management (Grandey, 2000; Law, Wong, & Song, 2004; Matthews, Zeidner, & Roberts, 2002; Mayer, Roberts, & Barsade, 2008), likely fueled in part by claims that EI predicts job performance better than general intelligence does (Goleman, 1995). Claims regarding the strong relationship between EI and work performance have also stimulated interest among consultants and practitioners, who have made EI a widely used tool for personnel hiring and training (Fineman, 2004).
Despite the impressive commercial success of emotional intelligence (EI), scholars have levied the following criticisms against the construct: (a) EI has definitional ambiguities, (b) there is considerable overlap between EI and related constructs of personality and general mental ability/intelligence (Landy, 2005; Locke, 2005; Murphy, 2006; Zeidner, Matthews, & Roberts, 2004), and (c) many EI measures are proprietary and expensive to administer, which makes it unnecessarily difficult for researchers to build a strong scientific foundation for the EI construct. Below we discuss each of these issues in the hope of clarifying recent progress in EI research, while pointing out a few remaining difficulties associated with the concept.
Definitional Ambiguity: Ability EI vs. Mixed EI
The definitional ambiguity surrounding EI stems from a jingle fallacy (i.e., an assumption that two concepts are the same because they share the same name; Kelley, 1927), wherein the label “emotional intelligence” has been used to refer to two separate constructs: ability EI and mixed EI. Ability EI models propose that EI is a form of intelligence and should therefore overlap with measures of general mental ability/intelligence, whereas mixed EI models do not classify EI as a form of intelligence but rather as a composite or mix of personality traits and abilities (Petrides & Furnham, 2001).
Ability EI is generally considered to be the “gold standard” model of the construct (Daus & Ashkanasy, 2005; Murphy, 2006) because: (a) it has a relatively clear theoretical definition and (b) it is typically measured with performance-based (i.e., multiple-choice, correct/incorrect) assessments, which are arguably less susceptible to social desirability biases than are self-reports of EI. Recent evidence supports the construct validity (i.e., the extent to which a measure captures its intended construct) of ability EI measures, indicating that ability EI is indeed a facet of intelligence, as the construct label emotional intelligence implies (MacCann, Joseph, Newman, & Roberts, 2014; also see Mayer, Caruso, & Salovey, 1999). Specifically, modern conceptualizations of intelligence such as the Cattell-Horn-Carroll theory of mental abilities (McGrew, 2009) suggest that intelligence consists of various specific abilities including verbal ability, quantitative reasoning, visual processing, and retrieval ability.
MacCann et al. (2014) showed that the Mayer-Salovey-Caruso Emotional Intelligence Test (MSCEIT; Mayer, Salovey, Caruso, & Sitarenios, 2003), a popular measure of ability EI, can be represented in the Cattell-Horn-Carroll theory of intelligence as an additional specific ability, which formalizes ability EI as an aspect of intelligence characterized by the ability to process information in the emotion domain. It should be noted, however, that the MSCEIT has been criticized for its potential underrepresentation of the entire domain of EI and for construct-irrelevant variance. For example, although the MSCEIT was designed to assess all dimensions of Mayer and Salovey’s (1997) model of EI, some have noticed that the MSCEIT fails to assess various dimensions of this model such as the ability to express emotion, the ability to perceive emotions in oneself, and the ability regulate emotion in oneself (Maul 2012).
The MSCEIT may also unintentionally assess abilities other than EI (i.e., the MSCEIT exhibits construct-irrelevant variance) such as the ability to match emotions to colors and tastes (i.e., the Sensations measure of the MSCEIT; Maul, 2012). Therefore, although evidence suggests the MSCEIT can be represented in the Cattell-Horn-Carroll theory of intelligence (MacCann et al., 2014), this type of construct validity evidence should be replicated on additional EI measures beyond the MSCEIT.
Ability EI: Does it Predict Job Performance and Other Work Behaviors?
Meta-analytic evidence indicates there is little-to-no incremental validity of ability EI in predicting job performance, above and beyond Big Five personality and general mental ability/intelligence (ΔR2= .002 = 0.2%; Joseph & Newman, 2010; ΔR2= .004 = 0.4%; O’Boyle, Humphrey, Pollack, Hawver, & Story, 2011). In other words, ability EI does almost nothing to predict overall job performance, after one has accounted for personality and general intelligence.
Although the incremental validity of ability EI is slightly larger in jobs that require emotional labor (ΔR2= .015 = 1.5%; Joseph & Newman, 2010; that is, emotional intelligence can predict emotional job performance), and meta-analytic evidence has identified emotion regulation ability as the key driver of the relationship between ability EI and emotional job performance (i.e., Cascading Model of EI; Joseph & Newman, 2010), it appears that empirical evidence does indeed support the long-standing criticism of EI as having questionable incremental validity for predicting overall job performance beyond well-established general intelligence and personality constructs.
It follows that the claims about the predictive power of EI with respect to job performance that are common in the popular press are substantially overstated; we currently have no consistent scientific evidence that scoring high on ability EI tests predicts how strong your job performance will be, beyond smarts and personality traits (in particular, Conscientiousness, Emotional Stability, Agreeableness, and Extraversion).
Despite this criticism, it should be noted that the incremental validity of ability EI for predicting other work-related outcomes, besides job performance, is an open question. As suggested by Landy (2005), ability EI might end up predicting job satisfaction, leader emergence, or the size of one’s social network much better than it predicts job performance. We await further research on these topics.
Ability EI: How Many Facets (or Branches) are There?
In addition to the limited incremental validity of ability EI, recent theoretical and measurement-related developments involving ability EI have raised new questions about the facet structure of ability EI. Specifically, although the original four-branch Mayer and Salovey (1997) model of ability EI included four dimensions (i.e., the ability to perceive, understand, use, and regulate emotion), recent work has called for the use of three, rather than four, dimensions of ability EI (i.e., by removing the ability to use emotion facet, or by combining it with the ability to perceive emotion facet; Fan, Jackson, Yang, Tang, & Zhang, 2010; Joseph & Newman, 2010).
Moreover, other work has proposed the addition of a new EI dimension (i.e., the ability to use emotional displays to influence others; Côté & Hideg, 2011), and the specification of sub-facets of existing ability EI dimensions (Côté, 2014). In order to establish a consensus definition of ability EI and its facets, we will need a multidimensional measure of ability EI that consistently supports the theorized facet structure. Our own viewpoint follows from Fan et al.’s (2010) meta-analysis of ability EI dimensions and suggests there is consensus on three of the four facets of the Mayer-Salovey model: ability to perceive, ability to understand, and ability to regulate emotion.
Mixed EI: What is it, and Why Does it Predict Job Performance?
In comparison to ability EI, mixed EI is defined as a constellation of personality traits, emotional skills and abilities, and self-evaluations such as general self-efficacy (Goleman, 1995; Petrides & Furnham, 2001). Whereas the ability EI literature is largely based on Mayer and Salovey’s (1997) model, mixed EI has not been based on a single, consensus model, and as such has been referred to as an umbrella construct that is a “grab bag of everything that is not cognitive ability,” (Joseph & Newman, 2010), and consists of a “list of healthy individual differences” (Côté, 2014, p. 462).
Unfortunately, to this point, the constructs that occupy this list of healthy individual differences have been unclear, leaving us to ask, “Which constructs are mixed EI measures really assessing?”. Research has noted that a majority of mixed EI items can be classified as Big Five personality items (de Raad, 2005), and others have suggested that mixed EI likely borrows content from achievement motivation (i.e., Conscientiousness), impulse control (e.g., Conscientiousness), gregariousness/assertiveness (e.g., Extraversion), and self-evaluations—such as general self-efficacy (Mayer, Roberts, & Barsade, 2008) and self-rated performance (Newman, Joseph, & MacCann, 2010).
Given that a majority of items from mixed EI measures seem to borrow from the content domains of other, well-established constructs, it appears that mixed EI measures might have been developed via a process that we call heterogeneous domain sampling (Joseph, Jin, Newman, & O’Boyle, 2015). That is, the developers of mixed EI measures have sampled items from a broad set of well-established content domains, and then simply pooled these items under the umbrella construct label that is mixed EI.
If mixed EI measures do indeed borrow content from a laundry list of other well-known constructs (e.g., neuroticism, extraversion, conscientiousness, self-rated performance, self-efficacy), then one important question is, “How much of mixed EI represents emotional skills and abilities?” If the construct of mixed EI has only limited content involving emotional skills and abilities, we question the extent to which the label “mixed emotional intelligence” is appropriate.
Interestingly, compared to ability EI, we note that mixed EI appears to display more incremental validity in the prediction of job performance (ΔR2 = .142 = 14%; Joseph & Newman, 2010; ΔR2 = .068 = 7%; O’Boyle et al., 2011). That is, mixed EI strongly and uniquely predicts job performance. This incremental prediction of job performance by mixed EI is perhaps due to mixed EI measures’ having been developed via domain sampling of self-efficacy and self-rated performance constructs, in addition to Big Five personality constructs and general mental ability. Due to its sizable incremental validity, the construct of mixed EI may be a useful practitioner tool that would benefit from future investigations into its construct validity. [We note that although mixed EI measures are typically self-report personality measures, relatively little of their content directly assesses the degree to which individuals perceive themselves to be emotionally competent.]
Do EI Scores Vary Across Sex? Is There a “Dark Side” of EI?
Well-established stereotypes of women as more emotional than men (Timmers, Fischer, & Manstead, 2003) have captured the attention of EI researchers who wish to understand which sex is more adept at expressing, understanding, and/or regulating emotion. Some evidence suggests biological sex differences in EI (e.g., areas of the brain dedicated to the processing of emotion are larger in females; Baron-Cohen, 2002, 2003) as well as situational/socialization factors that can explain sex differences in EI (e.g., girls are taught to label their emotional feelings more often than boys; Cervantes & Callanan, 1998).
Meta-analytic evidence suggests that sex differences in EI are only found for measures of ability EI (e.g., MSCEIT; Joseph & Newman, 2010), upon which females score notably higher than males (d = .52). In contrast, self-report measures of EI exhibit near-zero sex differences (self-report ability EI: d = .01; self-report mixed EI: d = .02), suggesting that although females may have higher ability EI than males, males and females perceive themselves to be equally emotionally competent/skilled. Unfortunately, research in this area has yet to examine the extent to which sex differences in EI might vary across discrete emotions; e.g., to discern the extent to which females are more proficient in perceiving, understanding, and regulating happiness, sadness, anger, etc. The issue of gender differences in EI across specific emotions should be addressed in future research.
Another area of recent interest to EI researchers involves the extent to which emotional skills and abilities can be used to manipulate, deceive, or engage in other socially abusive behavior. Despite the EI construct’s typically being described as a positive trait, preliminary evidence suggests that EI has both a positive side and a “dark side” (i.e., emotion understanding facilitates interpersonally deviant behavior; Côté, DeCelles, McCarthy, Van Kleef, & Hideg, 2011). A thorough examination of both the positive side and the dark side of EI may help us better understand clinical conditions that are characterized by lack of emotional skills (e.g., autism) or the use of emotional skills for manipulation and personal gain (e.g., psychopathy). This also points to the potential advantages of EI research that borrows theory and findings from the domain of clinical psychology.
When Will Scientists be Able to Measure EI Freely, for Research Purposes?
We appreciate that instruments designed to measure ability EI and mixed EI can constitute intellectual property, and thus charging fees for their administration is a way to recoup the costs of measurement research and development. We also understand the need for test security in the arena of high-stakes testing, where personnel decisions are being made on the basis of test scores. However, when the developers of a proprietary EI instrument insist on charging prohibitively high fees to researchers who simply want to advance scientific knowledge on the topic—after the R&D costs for developing the EI measure have already been recouped—then test developers are condoning the appearance that they do not want scientific research to be conducted on their measures and constructs.
This is our one major gripe. Enough is enough. We think the time has come for the proponents of the dominant measures and models of EI to make their instruments (or some versions of their instruments) either freely available, or available at only nominal cost, to researchers seeking to advance EI science. As it currently stands, it is quite difficult for an interested graduate student to conduct a simple study of EI without first paying prohibitively large test administration fees (e.g., an EI study with N = 200 participants would cost well over $1,000 in test administration fees). Most mainstream psychological constructs can be studied for free—EI is an exception.
In sum, existing measures of ability EI have been demonstrated to represent the expression of intelligence in the emotion domain, but to also suffer from a lack of incremental validity for predicting job performance. In contrast, mixed EI appears to have strong incremental validity for predicting job performance, but suffers from construct validity problems including unknown content (i.e., the content appears to be recycled from other well-established constructs in psychology), questions involving the appropriateness of the label “emotional” intelligence, and lack of a clear theoretical explanation for why mixed EI predicts job performance so well. These and other questions could be better answered if test publishers would make at least some versions of their EI instruments less cost-prohibitive to researchers.
Baron-Cohen, S. (2002). The extreme male brain theory of autism. Trends in Cognitive Science, 6, 248-254.
Baron-Cohen, S. (2003). The essential difference: the truth about the male and female brain. New York: Basic Books.
Cervantes, C. A., & Callanan, M. A. (1998). Labels and explanations in mother–child emotion talk: Age and gender differentiation. Developmental Psychology, 34, 88–98.
Côté S. (2014). Emotional intelligence in organizations. Annual Review of Organizational Psychology and Organizational Behavior, 1, 459-488.
Côté, S, DeCelles, K. A., McCarthy, J. M., Van Kleef, G. A., & Hideg, I. (2011). The Jekyll and Hyde of emotional intelligence: Emotion-regulation knowledge facilitates prosocial and interpersonally deviant behavior. Psychological Science, 22, 1073–80
Côté S., Hideg I. (2011). The ability to influence others via emotion displays: A new dimension of emotional intelligence. Organizational Psychology Review, 1, 53–71.
Daus, C. S., & Ashkanasy, N. M. (2005). The case for an ability-based model of emotional intelligence in organizational behavior. Journal of Organizational Behavior, 26, 453-466.
de Raad, B. (2005). The trait-coverage of emotional intelligence. Personality and Individual Differences, 38, 673-687.
Fan, H., Jackson, T., Yang, X., Tang, W., & Zhang, J. (2010). The factor structure of the Mayer-Salovey-Caruso Emotional Intelligence Test V 2.0 (MSCEIT): A meta-analytic structural equation modeling approach. Personality and Individual Differences, 48, 781-785.
Goleman, D. (1995). Emotional intelligence. New York: Bantam.
Joseph, D. L., & Newman, D. A. (2010). Emotional intelligence: An integrative meta-analysis and cascading model. Journal of Applied Psychology, 95, 54-78.
Kelley, T. L. (1927). Interpretation of Educational Measurements. Yonkers-on-Hudson, N.Y.: World Book Company.
Landy, F. J. (2005). Some historical and scientific issues related to research on emotional intelligence. Journal of Organizational Behavior, 26, 411-424.
Locke, E. A. (2005). Why emotional intelligence is an invalid concept. Journal of Organizational Behavior, 26, 425-431.
MacCann, C., Joseph, D. L., Newman, D. A., & Roberts, R. D. (2014). Emotional intelligence is a second-stratum factor of intelligence: Evidence from hierarchical and bifactor models. Emotion, 14, 358-374.
Mayer, J. D., Caruso, D. R., & Salovey, P. (1999). Emotional intelligence meets traditional standards for emotional intelligence. Intelligence, 27, 267-298.
Mayer, J. D., Roberts, R. D., & Barsade, S. G. (2008). Human abilities: Emotional intelligence. Annual Review of Psychology, 59, 507-536.
Mayer, J. D., & Salovey, P. (1997). What is emotional intelligence? In P. Salovey & D. Sluyter (Eds.), Emotional development and emotional intelligence: Implications for educators (pp. 3-34). New York: Basic Books.
Mayer, J. D., Salovey, P., Caruso, D. R., & Sitarenios, G. (2003). Measuring emotional intelligence with the MSCEIT V2.0. Emotion, 3, 97-105.
McGrew, K. S. (2009). Editorial: CHC theory and the human cognitive abilities project: Standing on the shoulders of the giants of psychometric intelligence research. Intelligence, 37, 1–10.
Murphy, K. R. (2006). Four conclusions about emotional intelligence. In K. R. Murphy (Ed.), A critique of emotional intelligence (pp. 345-354). Mahwah, NJ: Erlbaum.
Newman, D. A., Joseph, D. L., & MacCann, C. (2010). Emotional intelligence and job performance: The importance of emotion regulation and emotional labor context. Industrial and Organizational Psychology, 3, 159-164.
O’Boyle, E. H., Humphrey, R. H., Pollack, J. M., Hawver, T. H., & Story, P. A. (2011). The relation between emotional intelligence and job performance: A meta-analysis. Journal of Organizational Behavior, 32, 788-818.
Petrides, K. V., & Furnham, A. (2001). Trait emotional intelligence: Psychometric investigation with reference to established trait taxonomies. European Journal of Personality, 15, 425-448.
Timmers, M., Fischer, A. H., & Manstead, A. S. R. (2003). Ability versus vulnerability: Beliefs about men’s and women’s emotional behavior. Cognition and Emotion, 17, 41–63.
Zeidner, M., Matthews, G., & Roberts, R. D. (2004). Emotional intelligence in the workplace: A critical review. Applied Psychology: An International Review, 53, 371-399.