In the 1990s, a concept entered the popular imagination with the force of revelation: intelligence as we had understood it — the kind measured by IQ tests — was only part of the story. There was another kind of intelligence, and it might matter even more for success in work and life. This was emotional intelligence, and Daniel Goleman's 1995 book of that name made the concept a fixture of management culture, business school curricula, and corporate training programs worldwide.
Thirty years later, the scientific picture is considerably more complicated than the popular accounts suggested. Emotional intelligence is real, measurable, and genuinely predictive of certain outcomes. But the magnitude of its effects, its relationship to personality, and whether it can be meaningfully trained are all more nuanced than the self-help framing implied.
Two Models, Very Different Claims
The first important distinction in understanding emotional intelligence is that there are two substantially different models that share the same name.
The Mayer-Salovey Ability Model
John Mayer of the University of New Hampshire and Peter Salovey of Yale developed the original scientific model of emotional intelligence in a 1990 paper. Their framework defines EQ as a genuine cognitive ability — a specific type of intelligence applied to the domain of emotional information, analogous to how verbal intelligence involves reasoning about linguistic information.
The Mayer-Salovey model has four components organized in a hierarchy:
- Perceiving emotions — accurately reading emotional expressions in faces, voices, images, and situations
- Using emotions — employing emotional states to facilitate cognitive tasks like creativity and focused attention
- Understanding emotions — knowing how emotions develop, blend, and shift over time
- Managing emotions — regulating one's own emotions and influencing the emotions of others
This model is measured through performance-based tests — the Mayer-Salovey-Caruso Emotional Intelligence Test (MSCEIT) — where participants demonstrate emotional ability by, for example, accurately matching emotional expressions to faces or correctly identifying which emotion best facilitates a particular type of thinking. Performance is scored against expert consensus.
Goleman's Mixed Model
Daniel Goleman's popular model, introduced in his 1995 book Emotional Intelligence and developed further in subsequent work for Harvard Business Review, is substantially broader. His framework encompasses:
- Self-awareness — knowing your own emotions, strengths, weaknesses
- Self-regulation — managing disruptive emotions and impulses
- Motivation — being driven by internal standards rather than external rewards
- Empathy — understanding others' emotions and perspectives
- Social skills — managing relationships and building networks
Goleman claimed in his 1998 HBR article that emotional intelligence accounted for 67 percent of the abilities required for excellent performance in leadership — more important than technical skills or IQ. This claim, based on research at the time, was widely cited and drove the enormous commercial success of EQ training programs.
The scientific community has been consistently critical of this broader model. Critics, including psychologist Edwin Locke in a 2005 paper, argued that Goleman's model is too expansive to be meaningfully defined, substantially overlaps with established Big Five personality traits (particularly conscientiousness, agreeableness, and neuroticism), and conflates personality with intelligence in ways that muddy both constructs.
| Feature | Mayer-Salovey (Ability) | Goleman (Mixed/Trait) |
|---|---|---|
| What is measured | Performance on emotional tasks | Self-reported traits and behaviors |
| Overlap with Big Five | Low | High |
| Scientific rigor | Higher | Debated |
| Correlation with IQ | Low but positive | Near zero |
| Popular recognition | Low | Very high |
What the Research Actually Shows
EQ and Job Performance
The most comprehensive meta-analysis of EQ and job performance, conducted by Dana Joseph and Daniel Newman and published in the Journal of Applied Psychology in 2010, analyzed 43 studies with a combined sample of over 5,000 workers.
The headline finding: ability-based EQ (Mayer-Salovey model) correlates with job performance at r = 0.24 after controlling for cognitive ability and Big Five personality traits. This is a meaningful effect — comparable in magnitude to the predictive validity of structured behavioral interviews and substantially larger than unstructured interviews.
Critically, the relationship was moderated by emotional labor demands. Jobs requiring significant interpersonal interaction, emotional display management, and customer service showed stronger EQ-performance correlations than technical or independent roles. For sales representatives, customer service managers, and healthcare workers, EQ was a more useful predictor than for engineers, programmers, or researchers working largely independently.
Trait-based EQ measures (closer to Goleman's model) showed larger raw correlations with performance, but much of this predictive validity was shared with Big Five personality measures. After removing the variance explained by personality, the incremental validity of trait EQ was smaller.
EQ and Leadership
Leadership is the domain where EQ evidence is strongest. A 2002 meta-analysis by Lopes and colleagues found that higher MSCEIT scores predicted higher supervisor ratings of leadership effectiveness and positive peer relationships. A separate analysis of 44 studies by Walter, Cole, and Humphrey found that EQ accounted for significant variance in transformational leadership — the leadership style associated with inspiring followers, building commitment, and driving change — even after controlling for personality.
The mechanism appears to involve several capabilities that are genuinely useful for leaders:
Accurate emotion perception allows leaders to read the room — to sense when a team is anxious, when an individual is struggling, or when apparent consensus masks unspoken disagreement. Leaders who miss these signals make worse interpersonal decisions.
Emotion regulation allows leaders to manage their own emotional reactivity under pressure. Leaders who lose composure in crises, display unpredictable emotional states, or project anxiety to their teams create conditions that impair team performance.
Empathic accuracy — understanding what others feel and why — improves the quality of feedback conversations, conflict resolution, and coalition building. The ability to take the emotional perspective of a resistant stakeholder improves the quality of change management.
"The most effective leaders are all alike in one crucial way: they all have a high degree of what has come to be known as emotional intelligence. It's not that IQ and technical skills are irrelevant. They do matter, but mainly as threshold capabilities; that is, they are the entry-level requirements for executive positions." — Daniel Goleman, Harvard Business Review, 1998
The legitimate criticism of this passage — and Goleman's broader claims — is about degree rather than direction. EQ does appear to contribute something to leadership effectiveness. The claim that it contributes more than IQ is much less well-established, and depends heavily on how EQ is defined and measured.
EQ vs. IQ: A False Competition?
The popular framing of EQ vs. IQ as competing predictors of success is misleading in several ways.
First, the two are largely independent. The correlation between ability-based EQ and IQ is typically around r = 0.10 to 0.20 — knowing someone's IQ tells you almost nothing about their EQ, and vice versa. They are different things, not competing versions of the same thing.
Second, what they predict differs by context. IQ is a consistently strong predictor of job performance across virtually all job types, with correlations averaging around r = 0.51 in comprehensive meta-analyses. EQ adds incrementally to this prediction, especially in roles high in emotional labor demands. They are complementary, not competing.
Third, the EQ-versus-IQ framing sometimes functions rhetorically to devalue cognitive ability in contexts where it is genuinely important. A strong case can be made that society undervalues interpersonal intelligence; a much weaker case can be made that it overvalues analytical intelligence, particularly in technical fields where analytical ability genuinely drives performance.
The Big Five Overlap Problem
One of the most persistent methodological criticisms of trait-based EQ measures is substantial overlap with established Big Five personality dimensions — particularly:
- Neuroticism (emotional stability): People low in neuroticism experience less intense negative emotions, regulate them more easily, and are less emotionally reactive — all of which look like high trait EQ
- Agreeableness: People high in agreeableness are more attuned to others' feelings and more motivated to maintain harmony — also resembling high EQ
- Conscientiousness: Self-regulatory capacities central to Goleman's EQ model correlate with conscientiousness
If trait EQ measures largely recapture variance already explained by the Big Five, the incremental scientific value of the EQ construct is reduced. The ability-based EQ model is less susceptible to this criticism because performance-based emotional ability tests show much less overlap with personality dimensions.
Can Emotional Intelligence Be Trained?
This question has substantial practical importance because billions of dollars are spent annually on EQ training programs in organizational settings.
A 2011 meta-analysis by Mattingly and Kraiger synthesized 58 studies on EQ training interventions and found that programs produced significant improvements in measured EQ, with a mean effect size of d = 0.53 — a moderate effect. Programs that used active practice, feedback, and behavioral modeling outperformed those based primarily on knowledge transmission.
However, the evidence base has important limitations:
Short follow-up periods. Most studies measured EQ changes immediately after training or within weeks. Long-term retention studies beyond six months are rare.
Measurement ambiguity. Many studies used self-report EQ measures, which may reflect changes in how participants describe themselves rather than changes in actual emotional ability.
Transfer to behavior. Demonstrating that training changes measured EQ scores does not establish that it changes on-the-job behavior. Studies tracking behavioral outcomes in workplaces — actual conflict resolution, leadership effectiveness, team climate — are fewer and show more modest effects.
Placebo and demand effects. Participants in EQ training who expect to improve may rate themselves higher post-training regardless of actual change.
Social and emotional learning (SEL) programs delivered in schools over sustained periods show stronger evidence for lasting behavioral and academic outcomes than brief adult workplace training, which is consistent with the general finding that skill-building interventions work better with longer time horizons and during developmental windows.
Practical Implications for Individuals
Despite the scientific caveats, several practical conclusions hold up.
Know what you are measuring. EQ assessments vary enormously in quality and what they actually measure. Performance-based measures (MSCEIT) are more scientifically defensible than self-report questionnaires, and self-report measures inflate perceived scores because emotional competence is precisely the kind of ability people tend to overestimate.
Focus on specific, coachable behaviors. Rather than trying to "improve EQ" as an abstraction, identify specific behaviors with clear connections to outcomes: listening without interrupting, naming emotions in difficult conversations, asking questions before giving feedback, creating deliberate space for quieter voices in meetings. These are trainable behaviors whose improvement is observable.
Match role requirements. If you are in a role with high emotional labor demands — management, sales, negotiation, healthcare — the research supports investing in interpersonal and emotional competencies. If your role is primarily individual technical work, the incremental return on EQ development is lower.
Use accurate feedback. 360-degree feedback from multiple raters, ideally including subordinates and peers who observe you regularly, is a more reliable signal of interpersonal effectiveness than self-assessment or scores on self-report EQ tests.
Summary
Emotional intelligence is a real and measurable set of capabilities that predicts meaningful outcomes at work, particularly in roles requiring sustained interpersonal engagement, leadership, and emotional labor management. The ability-based Mayer-Salovey model is the most scientifically defensible version of the construct; Goleman's broader trait model captures more variance in outcomes but substantially overlaps with established personality dimensions. EQ and IQ are largely independent and predict different things — they are complementary tools, not competitors. Training interventions show moderate short-term effects on measured EQ, with less certain long-term and behavioral transfer. The practical value of emotional intelligence at work is real; the claims made for it in the popular management literature have sometimes exceeded the evidence.
Frequently Asked Questions
What is emotional intelligence?
Emotional intelligence (EQ or EI) is the ability to perceive, understand, manage, and use emotions effectively in oneself and in interactions with others. The scientific model developed by John Mayer and Peter Salovey defines it as a specific cognitive ability involving the processing of emotional information. Daniel Goleman's popular model broadens it to include personality traits, motivation, and social skills, though this broader model is more contested scientifically.
Does emotional intelligence actually predict workplace performance?
A comprehensive 2010 meta-analysis by Dana Joseph and Daniel Newman found that ability-based EQ (the Mayer-Salovey model) predicts job performance with a correlation of about r = 0.24 after controlling for IQ and personality. The effect is stronger for jobs with high emotional labor demands — management, sales, customer service — and weaker for technical or independent roles. Goleman's broader trait-based EQ model shows larger correlations but faces criticism for overlap with established personality measures.
How is emotional intelligence different from IQ?
IQ measures general cognitive ability — reasoning, problem-solving, and learning speed across abstract domains. EQ measures a specific type of intelligence applied to emotional information — recognizing facial expressions of emotion, understanding how emotions influence judgment, and regulating emotional responses. The two are largely independent, with correlations between them typically around r = 0.10 to 0.20, meaning knowing someone's IQ tells you almost nothing about their EQ.
Can emotional intelligence be trained?
The evidence for training EQ is modest but positive. A 2011 meta-analysis by Mattingly and Kraiger found that EQ training programs produced significant short-term improvements in measured EQ, with effect sizes ranging from small to moderate. However, long-term retention and transfer to workplace behavior are less well established. Social and emotional learning programs in educational settings show stronger evidence for lasting behavioral change than adult workplace training programs.
What is the relationship between emotional intelligence and leadership effectiveness?
A meta-analysis by Danielle Dulewicz and Malcolm Higgs found significant correlations between EQ and transformational leadership ratings. Leaders who score higher on ability-based EQ tend to create more psychologically safe environments, receive higher subordinate satisfaction ratings, and perform better in roles requiring coalition-building and conflict resolution. However, EQ accounts for a modest share of leadership effectiveness variance after personality and general intelligence are controlled.