Emotional intelligence is measured two fundamentally different ways, and the two approaches don't measure the same thing, don't predict the same outcomes, and have been in genuine scientific dispute for two decades. The ability model — developed by Peter Salovey, John Mayer, and David Caruso — treats emotional intelligence as a set of cognitive abilities, measurable with right-and-wrong-answer tests the same way IQ is measured. The trait model — associated with Konstantinos Petrides and others — treats emotional intelligence as a constellation of emotional self-perceptions and dispositions, measured through self-report questionnaires. Understanding which model you're looking at when you encounter an EQ test or EQ-based claim matters considerably, because they make different predictions and have different relationships to other well-established constructs.
The Ability Model: EI as Cognitive Capacity
Mayer, Salovey, and Caruso's four-branch model, operationalised in the MSCEIT (Mayer-Salovey-Caruso Emotional Intelligence Test), defines emotional intelligence as the ability to:
- Perceive emotions accurately: Recognise emotions in faces, voices, images, and designs with accuracy (scored against consensus responses of how most people interpret the emotional content)
- Use emotions to facilitate thought: Generate and use emotional states to improve thinking — recognising, for example, that mild sadness facilitates careful attention to detail while happiness facilitates creative association
- Understand emotions: Know how emotions combine, develop, and transition — understanding that contempt involves both anger and disgust, that grief evolves through specific stages
- Manage emotions: Regulate emotional states in oneself and others effectively, calibrating emotional responses to situations
Critically, ability EI is scored against performance criteria — you're either more or less accurate at reading emotion in a face, or more or less correct about how a specific emotional blend is typically understood. This makes it directly analogous to other cognitive ability tests and allows it to be correlated with IQ (the correlation is positive but modest, suggesting EI is partly distinct from general intelligence).
The Trait Model: EI as Self-Perceived Disposition
Trait EI, as defined by Petrides and operationalised in the TEIQue (Trait Emotional Intelligence Questionnaire), is a different construct entirely. It measures how people perceive their own emotional tendencies, capabilities, and behavioural dispositions through self-report. Trait EI encompasses items like:
- Perceived ability to regulate emotions
- Self-assessed empathy
- Perceived social skill and relationship quality
- Subjective wellbeing and optimism
- Emotion expression comfort
Since these are self-perceptions rather than performance measurements, trait EI occupies the conceptual space of personality — specifically, it shows substantial overlap with the Big Five personality dimensions of extraversion, agreeableness, conscientiousness, and emotional stability (low neuroticism). Petrides explicitly locates trait EI in the personality domain and argues it should be interpreted and studied as a personality construct rather than as an intelligence.
Why the Distinction Matters
Different predictive validity
Ability EI and trait EI predict different outcomes. Ability EI (particularly the emotion perception and understanding branches) shows stronger relationships with cognitive outcomes — academic performance, real-world reasoning quality, the ability to accurately read social situations. Trait EI shows stronger relationships with wellbeing, life satisfaction, relationship quality (as reported by the respondent), and some aspects of occupational success, particularly in social-demand roles.
The distinction becomes practically important when organisations use EI measures for selection or development: a self-report "EQ" assessment may be measuring something closer to emotional stability and extraversion than the cognitive emotional abilities the organisation thinks it's assessing.
The self-report problem
A fundamental methodological issue with trait EI is that people are notoriously poor at accurately assessing their own emotional abilities. The correlation between self-assessed emotional intelligence and measured ability EI is consistently modest — often around 0.15-0.30 — meaning that how smart about emotions you think you are is only weakly related to how accurately you perform on emotional reasoning tasks. Dunning-Kruger effects are well-documented in this domain: people with the lowest ability EI scores tend to significantly overestimate their emotional abilities, while high performers are more calibrated.
The overlap with existing personality measures
Critics of trait EI (particularly in the personality psychology community) argue that when the construct's overlap with Big Five dimensions is controlled statistically, trait EI adds little unique predictive validity beyond what personality measures already provide. This is a live debate, but it suggests that much of what makes trait EI useful is also captured by well-established personality instruments — raising questions about whether trait EI is a genuinely new construct or a repackaging of existing ones.
Mixed Models: The Popular Version
The EI frameworks most widely known in popular and business contexts — particularly Goleman's model, which formed the basis of the widespread "EQ" popularisation in the 1990s — are mixed models that combine elements of ability, trait, and motivational variables into a broader framework. Goleman's original book and the associated assessments include self-awareness, self-regulation, empathy, motivation, and social skill — a combination that spans ability EI, trait EI, and personality characteristics.
Mixed models have faced consistent criticism from ability EI researchers for lacking the theoretical precision and measurement validity of the pure ability model, and from personality researchers for poorly distinguishing EI from existing personality constructs. They remain widely used in training and coaching contexts partly because their broader, more actionable framing is easier to work with practically. For a structured assessment of emotional intelligence in the ability-adjacent sense — across emotional perception, understanding, and regulation — our free EQ test maps your profile with interpretation across these dimensions.
Frequently Asked Questions
Which EI model is more scientifically legitimate?
The ability model has more rigorous theoretical grounding and more consistent empirical support for the distinctiveness of the construct from existing measures. It also has better measurement properties because it uses performance-based rather than self-report assessment. However, "more scientifically legitimate" doesn't mean more practically useful for every purpose. Trait EI has strong evidence for predicting wellbeing, relationship satisfaction, and certain occupational outcomes — it's just a different (and less precise) construct than ability EI claims to be.
Can you improve ability EI?
Modest improvability through specific training — particularly for emotional perception (facial expression reading can be improved through deliberate feedback training) and emotional regulation (which overlaps with skills taught in CBT, DBT, and mindfulness-based interventions). The improvement evidence is stronger for specific components than for overall EI scores, and the transfer from training to real-world performance is variable. Ability EI is more stable than mood states but less fixed than crystallised intelligence.
Why does Goleman's model dominate corporate EI discourse if ability EI is more rigorous?
Several reasons: Goleman's 1995 book was a popular science bestseller that preceded most of the scientific debate about measurement; the mixed model's broader scope is easier to teach in workshops; the self-report format of most commercial EI tools is cheaper and faster to administer than performance-based assessment; and the corporate training industry has significant investment in frameworks that produce actionable development plans, which mixed models support more easily than the more theoretically precise ability model.
Does high ability EI correlate with high IQ?
Positively but modestly — correlations typically in the range of 0.15-0.35. This suggests that ability EI captures something related to but partially distinct from general cognitive intelligence. The ability to perceive emotion accurately and understand complex emotional blends is a real cognitive skill, but it's not simply general intelligence applied to emotional content.
Is emotional intelligence heritable?
Twin studies on both ability and trait EI show moderate heritability estimates, broadly similar to other cognitive and personality traits. Ability EI heritability estimates are typically 30-50%; trait EI estimates are somewhat higher, consistent with its heavier personality component. Environmental factors (particularly quality of emotional education in childhood and adult attachment experiences) account for the remaining variance.