Most managers have heard the term. Many have sat through training on it. A surprising number still cannot give you a clear answer when asked what emotional intelligence actually is. EI has become one of the most discussed constructs in workplace psychology since the 1990s, and also one of the most misrepresented. Books promise it will trump IQ. Consultants market it as the secret to great leadership. Somewhere in all that noise, the actual science gets lost.
This article examines what the peer reviewed research says about what emotional intelligence is, how it works, what it predicts, and what you can reasonably do to improve it. Every claim below is drawn from published peer reviewed sources.
Where the Idea Came From
The term emotional intelligence entered the scientific literature in 1990, when psychologists Peter Salovey and John Mayer published their foundational paper in Imagination, Cognition and Personality. Their definition was precise: emotional intelligence is the ability to monitor one's own and others' feelings and emotions, to discriminate among them, and to use this information to guide thinking and action. They argued that some people were genuinely better than others at reasoning about emotions, perceiving them accurately, understanding how they shift and blend, and using that knowledge to guide thought and behaviour.
Five years later, American Psychologist and journalist Daniel Goleman published Emotional Intelligence: Why It Can Matter More Than IQ. The book transformed a narrow academic construct into a cultural phenomenon. His claim that EI could matter more than IQ in determining life success generated enormous public interest. It also generated enormous confusion. Goleman's version of the concept was far broader than the original, blending emotional abilities with personality traits, motivational tendencies, and social skills that overlap heavily with established personality measures.
As Robinson (2024) summarizes in the Journal of Intelligence, modern interest in EI began with the original Salovey and Mayer definitional effort, but Goleman popularized the construct by arguing, without sufficient evidence, that EI could be more important than general mental ability in determining whether lives were successful. That claim has not held up to scrutiny in the peer reviewed literature.
What Emotional Intelligence Actually Is: The Scientific Definition
The most scientifically rigorous definition comes from the ability model developed by Mayer, Salovey, and Caruso (2004) in Psychological Inquiry. According to this definition, emotional intelligence is a set of four related cognitive abilities: perceiving emotions accurately in faces, voices, and images; using emotions to support and guide thought; understanding how emotions develop and blend over time; and managing emotions in oneself and others to achieve goals.
This definition is testable. It specifies abilities, not personality traits. A 2025 paper in Frontiers in Psychology by Mayer, Caruso, Salovey and colleagues presents updated validation evidence placing EI within the Cattell Horn Carroll (CHC) framework of human intelligence, the same hierarchy that includes verbal, spatial, and mathematical reasoning. The CHC model views intelligence as a hierarchy with general intelligence at the top and broad intelligences nested below it. EI now fits within that hierarchy as a validated broad intelligence.
The original 1990 paper by Salovey and Mayer laid out the theoretical rationale for treating emotion as organized, adaptive information rather than noise that disrupts rational thought. Emotions carry functional signals: fear signals threat; happiness signals that current goals are being met; disgust signals contamination or violation. Emotionally intelligent people read those signals accurately and reason about them effectively. People with lower EI misread them, suppress them, or are overwhelmed by them.
A key distinction in this original definition, as Mayer, Salovey and Caruso (2008) clarify, is that EI is not a motivation, a social attitude, or a personality style. It is a set of cognitive abilities that can be tested, measured, and distinguished from existing personality constructs. That distinction matters enormously for how EI should be measured and applied in organizations.
Three Models, Three Very Different Claims
Understanding what emotional intelligence is requires understanding which model you are working with. As the Noba Psychology Module on EI explains, there are three dominant approaches in the research literature: the ability model, mixed models, and the trait EI model. They are not interchangeable. The claims each model supports differ substantially.
The Ability Model
The ability model, developed by Mayer and Salovey (1997) and formalized in Psychological Inquiry (2004), treats EI as a standard intelligence. It proposes four branches: perceiving emotions accurately, using emotions to support thinking, understanding emotional dynamics, and managing emotions in oneself and others. Ability models approach EI as something that is intercorrelated across its branches, relates to other cognitive abilities, and develops with age and experience.
EI in this model is measured using performance tests, specifically the Mayer-Salovey-Caruso Emotional Intelligence Test (MSCEIT), where responses are scored against expert consensus rather than self-perception. As the Noba EI module explains, performance measures require respondents to actually demonstrate their four emotion skills by solving emotion related problems, rather than self reporting their perceived abilities.
The scientific foundation of this model was significantly strengthened in 2025. A Frontiers in Psychology study by Mayer, Caruso, Salovey and colleagues introduced an updated MSCEIT 2, validated across five studies with a normative sample of 3,000. The revised model confirmed four domains: perceiving emotions, connecting emotions to other sensory information, understanding emotions, and managing emotions.
Robinson's 2024 review in the Journal of Intelligence notes that there is now emerging consensus that ability EI consists of three separable but correlated skill sets covering perception, understanding, and management. However, agreement on precise measurement procedures is still developing.
Mixed Models
Mixed models, most associated with Goleman and Bar On, combine emotional abilities with personality traits, motivational tendencies, and social competencies. As the Noba EI module explains, both mixed and trait models define and measure EI as a set of perceived abilities, skills, and personality traits rather than demonstrated abilities.
Robinson (2024) notes that trait EI, measured through self-report questionnaires, shows substantial overlap with neuroticism, extraversion, and agreeableness from the Big Five personality model. That overlap raises a legitimate question: if mixed EI largely reflects personality, what is it adding that personality science did not already capture? The peer reviewed literature suggests it adds something, but considerably less than the popular accounts claim.
As Mayer, Salovey and Caruso (2008) state explicitly, EI is currently used in too all-inclusive a fashion and in too many different ways. Mixed-model tests that include personality traits such as assertiveness and optimism often measure classically defined personality traits rather than emotional intelligence, even when labeled as EI instruments.
Trait EI
The trait model treats EI as a constellation of emotional self perceptions rather than abilities. Trait EI is measured by self report and is conceptualized as part of personality space. As detailed in the BMC Psychiatry systematic review of EI instruments (2021), this approach is internally consistent and has predictive power, but its incremental validity above established personality measures remains contested in the literature.
The practical implication, as Robinson (2024) concludes, is that the model and measurement method chosen shape what EI predicts. When someone says emotional intelligence predicts success, you need to ask which model they are using and how EI was measured. The claims differ substantially depending on the answer.
What the Meta Analyses Actually Show
A body of meta-analytic evidence now provides a clearer picture of what EI actually predicts in organizational settings. The evidence base is reviewed across three outcomes: job performance, leadership effectiveness, and team dynamics.
Job Performance
The most comprehensive meta analysis on EI and job performance was conducted by O'Boyle, Humphrey, Pollack, Hawver, and Story (2011) in the Journal of Organizational Behavior. Analysing studies across three EI streams covering ability based, self report, and mixed models, the authors found that all three streams of EI correlated positively with job performance. Streams 2 (self report) and 3 (mixed model) incrementally predicted job performance over and above cognitive intelligence and the Five Factor personality model. The study concluded that EI contributes real incremental predictive value beyond what IQ and personality alone explain.
A more recent 2022 meta analysis published in Frontiers in Psychology drew on 253 effect sizes from 287 articles and 118 unpublished studies examining over 30 years of EI research. It found that EI is positively linked to job performance, job satisfaction, organizational commitment, and organizational citizenship behaviour. It also found a meaningful negative relationship between EI and job stress and burnout. Employees with higher EI handle workplace pressure more effectively and report greater satisfaction with their roles.
That burnout finding is significant. The same meta analysis, also available at PubMed Central, identified burnout reduction as one of the most consistent employee outcomes linked to high EI. This has direct implications for healthcare, education, and any sector where emotional labour is a primary job demand. Organizations in these sectors have a measurable return on EI development investments.
Leadership Effectiveness
The evidence for EI in leadership is stronger than for general job performance. A 2024 meta analysis in Management Science Research Archives found a mean correlation of r = 0.511 between EI and leadership effectiveness across a total sample of 4,125 participants. By organizational research standards, a correlation in the range of 0.50 or above represents a strong and consistent relationship, making EI one of the better predictors of leadership quality available to HR practitioners.
A cross cultural meta analysis by Miao, Humphrey and Qian demonstrated that emotionally intelligent leaders generate better task performance and higher organizational citizenship behaviour in their teams. The study showed that leader EI has incremental validity and relative weight in predicting subordinate outcomes after controlling for the Big Five and cognitive ability. These effects were strongest in collectivist, feminine, and high uncertainty avoidance cultures, which describe most of sub Saharan Africa and large parts of Asia. For African organizations, this finding argues strongly for treating EI as a priority in leadership selection and development.
A comprehensive systematic review by Gerhardt, Bauwens, and van Woerkom (2026) in Human Resource Development Review examined 101 empirical studies published between 1990 and 2021. The review concluded that leader EI is consistently and positively associated with transformational leadership behaviour, leader wellbeing, and subordinate performance. The relationship held across industries, cultures, and measurement approaches.
Harms and Crede (2010) in the Journal of Leadership and Organizational Studies conducted a meta analysis on 62 independent samples and found a positive correlation between EI and both transformational and transactional leadership behaviours. This reinforces the finding that EI contributes to leadership effectiveness across styles, not only in contexts that call for empathy or warmth.
Team Dynamics
EI matters beyond individual performance and leadership. A 2023 hybrid literature review in Heliyon by Coronado Maldonado and Benitez Marquez found that EI is directly linked to team effectiveness and team behavioural outcomes. Because teamwork is social by nature, the ability to recognize and manage emotional states in yourself and your colleagues shapes how groups solve problems, handle conflict, and maintain motivation through difficult periods. The review found that recognizing and managing emotions is crucial for both individuals and work teams.
The IQ Versus EQ Debate: What the Evidence Actually Says
One of the most persistent claims in this area is that EI matters more than IQ. Goleman's original argument in his 1995 book drove this narrative. General cognitive ability remains the single strongest predictor of job performance across most roles, a finding confirmed in every major meta analysis that examines both predictors simultaneously. Goleman's specific claim has not held up to scrutiny in the peer reviewed literature.
The more accurate research finding is that IQ and EI predict different outcomes. As Robinson (2024) explains, cognitive ability predicts whether someone can learn the technical requirements of a role. EI predicts how they manage themselves and others once they are in that role, particularly in roles with high interpersonal demand.
Research using the MSCEIT, documented by the EI Consortium, found that EI performance scores correlate modestly with personality traits, at r = 0.25 with Openness and r = 0.28 with Agreeableness. These correlations are low enough to confirm EI is measuring something distinct from personality, but related enough to show conceptual connections. This is precisely the pattern you would expect from a genuine intelligence that overlaps with but is not reducible to existing constructs.
Work by Cote and Miners (2006), referenced by the EI Consortium, found that for employees with lower cognitive ability, higher EI partially compensated for lower job performance and organizational citizenship behaviour. Organizations that rely exclusively on cognitive assessments in selection risk systematically undervaluing candidates who bring strong emotional capability to people facing roles.
Can You Actually Improve Emotional Intelligence?
The research on EI trainability is more encouraging than the debates over definition and measurement suggest. Mattingly and Kraiger (2019) in Human Resource Management Review conducted a meta analysis on multiple studies that attempted to train EI and found meaningful improvement effects. Training works, provided it is designed properly. Programs that combine structured learning with real practice, feedback, and coached reflection show better outcomes than one day workshops that introduce the concept without building actual skills. The review confirmed that lecture, modeling, practice, feedback, and coaching are the training design properties that drive results.
A randomized controlled trial by Gomez Leal, Megias Robles, Gutierrez Cobo and Cabello (2019) in PLOS ONE conducted with senior managers found that a specifically designed EI training program produced significant improvements across key EI dimensions. The program blended face to face sessions with an e learning platform and required participants to apply skills to real workplace problems. Gains were sustained at follow up measurement. The randomized design with an active control group gives this study considerably more evidential weight than the many observational studies that dominate the EI training literature.
A systematic review by Kotsou, Nelis, Gregoire and Mikolajczak examined 25 EI intervention studies and found that specifically EI based programs consistently outperformed generic team building or wellness programs. Programs that were not specifically EI focused, such as ropes courses, peer sessions, and group check in rounds, showed minimal effect on measured EI outcomes. The design and theoretical grounding of the intervention matters enormously.
The neurological basis for this trainability is becoming clearer. As Stoewen (2024) explains in the Canadian Veterinary Journal drawing on neuroscience research, repeatedly engaging in emotionally intelligent behaviours strengthens neural pathways connecting the emotional and rational brain. Each time you deliberately apply an EI skill, whether labelling an emotion accurately, pausing before reacting, or reading another person's emotional state, you reinforce neural connections that make those responses faster and more automatic over time.
Unlike IQ, which typically remains stable across adulthood, Stoewen (2024) notes that although some people are naturally more emotionally intelligent than others, anyone at any stage of life can grow their EI through deliberate practice. This is a significant finding for HR practitioners who design development interventions.
The Gerhardt, Bauwens and van Woerkom (2026) systematic review confirms EI trainability, noting that the most common and effective EI development interventions include lectures, role plays, group discussions, work in tandems, self exploration through diaries, and self directed learning. The review notes that investment in leader EI development benefits not only the leaders themselves but their direct reports, their teams, and the broader organization.
What Emotional Intelligence Is Not
Because the popular understanding has drifted so far from the research base, it is worth being direct about what EI does not mean. Each of the following points is supported by peer reviewed evidence.
EI is not being nice or agreeable. High EI does not make someone conflict averse or easy to manage. As Mayer, Salovey and Caruso (2008) state explicitly, EI is used in too all-inclusive a fashion and in too many different ways. Understanding emotions accurately sometimes means recognizing when a situation calls for directness, disagreement, or confrontation rather than warmth.
EI is not the same as emotional sensitivity or reactivity. Being easily moved by your emotions is not the same as being emotionally intelligent. EI involves effective processing of emotional information. A 2023 study in the Journal of Intelligence on affective decision making found that emotional understanding, the capacity to reason about how emotions work, predicted better decision making even after controlling for general cognitive ability. Emotional reactivity did not produce the same outcome.
EI is not equally relevant across all roles. As O'Boyle et al. (2011) note, the predictive validity of EI is strongest in roles with high interpersonal demand. The research shows consistently stronger effects for managers, healthcare workers, educators, sales professionals, and client facing roles. For technically focused roles with minimal social interaction, the predictive value of EI is considerably lower. Practitioners should size their EI investments accordingly.
EI is not a fixed personality trait you either have or do not have. As the training meta analysis by Mattingly and Kraiger (2019) and the randomized trial by Gomez Leal et al. (2019) both demonstrate, emotional abilities respond meaningfully to well designed development programs. This matters for how organizations structure their leadership pipelines and what they look for in assessments.
Measuring Emotional Intelligence: What HR Practitioners Need to Know
If you are using EI assessments in hiring, promotion, or development decisions, the type of measure you use matters enormously. The BMC Psychiatry systematic review (2021) identified over 60 instruments currently in use, covering ability, self report, 360 degree, and mixed assessment approaches. They are not equally valid for all purposes.
Performance based tests like the MSCEIT measure what a person can actually do with emotional information. They are scored against criterion data and are not susceptible to social desirability inflation. The newly developed MSCEIT 2, published in Frontiers in Psychology in 2025, extends the original across more domains, with a normative sample of 3,000, updated scoring procedures, and expanded item types covering emotion perception, connection, understanding, and management.
Self report measures are easier to administer but are vulnerable to impression management. As Robinson (2024) notes, trait EI measures scored by self report are vulnerable to social desirability biases and limited introspective accuracy. Candidates under selection pressure know what emotional intelligence looks like and can respond accordingly. Self report scores in high stakes selection contexts may systematically overestimate actual EI ability.
The heterogeneity across tools makes comparisons across studies difficult and argues for careful instrument selection based on validated psychometric evidence, as the BMC Psychiatry review (2021) concludes. For development purposes, 360 degree assessments that combine self ratings with peer and manager observations offer a practical middle ground. They capture how a person's emotional behaviour lands on others, which is ultimately what matters in practice.
For selection purposes, ability based tests remain the most defensible choice, as Mayer, Salovey and Caruso (2004) argued in their foundational review of EI theory, findings, and implications. The EI Consortium maintains a comprehensive resource of validated EI measures with psychometric evidence, published reliability and validity data, and guidance on appropriate use contexts.
Where This Leaves HR and Leadership Practice
The research literature supports several conclusions that should shape how organizations think about EI. Each of the following draws on the peer reviewed evidence reviewed above.
First, EI is worth measuring and developing in people facing roles, particularly leadership. The mean correlation of r = 0.511 between EI and leadership effectiveness found by the 2024 management meta analysis is among the stronger relationships in organizational psychology. It is consistent enough across independent studies to justify sustained development investment.
Second, not all EI training is equal. As the Kotsou et al. systematic review concluded, only specifically EI based programs with clear theoretical grounding, practice components, and structured feedback loops consistently improve emotional skills. Generic awareness workshops show minimal effect on measured EI outcomes.
Third, context shapes the magnitude of EI's impact. The cross cultural meta analysis by Miao et al. found that leader EI had stronger effects on subordinate performance in collectivist, high uncertainty avoidance, and long term oriented cultures, characteristics that describe most African and Asian organizational contexts
Fourth, selection decisions should prioritize ability based measures. As documented by the EI Consortium and the peer reviewed work behind the MSCEIT, performance based assessments are more reliable and less susceptible to impression management than self report tools.
Fifth, EI complements rather than replaces technical competence or cognitive ability. As O'Boyle et al. (2011) demonstrated, all three streams of EI added incremental predictive validity above cognitive intelligence and the Big Five personality traits. The manager who combines strong cognitive ability with high emotional awareness is better positioned to retain good people, effectively navigate change , and maintain productive working relationships under pressure.
The Questions That Still Need Answers
The research on EI is substantial, but it is not settled. Robinson (2024) identifies several active debates that practitioners should be aware of when interpreting the evidence.
The overlap between EI and personality remains a live scientific issue. Robinson (2024) notes that several studies have revealed overlap between EI and existing constructs, most notably personality traits, general intelligence, and social intelligence. More research is needed to reach conceptual clarity on how EI is differentiated from related constructs. The field would benefit from greater convergence on both definitions and validated instruments.
Measurement is still fragmented. With over 60 instruments in use, as documented in the BMC Psychiatry systematic review (2021), comparing findings across studies is genuinely difficult. The Frontiers in Psychology MSCEIT 2 paper (2025) represents the most significant recent effort toward a validated, comprehensive ability based instrument with a large normative sample.
The long term development question is not fully resolved. Gerhardt, Bauwens and van Woerkom (2026) note that while EI is generally trainable, research is now needed on the extent to which leader EI can best be developed over time, and what conditions sustain those gains. We know from Mattingly and Kraiger (2019) that training produces measurable improvement in the short to medium term. Less is known about how durable those gains are over the years rather than the months.
These limitations, as Robinson (2024) concludes, do not undermine the case for taking EI seriously. They argue for applying it with precision rather than as a catch all solution to every interpersonal challenge organizations face.
Emotional intelligence is not a soft science complement to hard business metrics. When measured properly with validated ability tests rooted in the Mayer-Salovey-Caruso model, it is a coherent, testable set of abilities with meaningful consequences for how people lead, collaborate, and sustain themselves in demanding roles. The meta analytic evidence from O'Boyle et al. (2011), the 2022 Frontiers meta analysis, and the 2024 leadership meta analysis is rigorous enough to act on. The condition is that you engage with what the research actually says, not what the consulting brochure claims.
