A Cambridge-linked team has just put a very large stack of papers through the mill: 1,279 studies, more than 600,000 people, 76 countries, three decades. The folk belief they were testing is the one that sells leadership books and HR modules: women are the emotionally intelligent sex. The result is not a clean rebuttal and not a clean confirmation. It is a demolition of the single score.
Women, on average, do better at recognising and understanding emotions, and at managing other people's. Men and boys tend to score higher at managing their own. The female edge is bigger on performance tests (do this task, get scored) and almost vanishes on self-report surveys (rate yourself). The overall gap has narrowed over time, especially in children. Differences shift with leadership status, industry, and region. The authors' own punchline is that "who has more EI?" is the wrong question. Which skill, measured how, aimed at whom, in what context?
Hausfeld, Hampel, Bach and Menges treated "emotional intelligence" as a bundle: recognise, understand, regulate; self versus other. Once you cut the bundle, the stereotype splits with it. The office myth that women are naturally better at the whole emotional job is, in their framing, a way of handing them the other-focused work: soothing rooms, reading the boss, absorbing other people's weather, while treating self-regulation as a separate, less gendered talent. Menges's line writes itself for the press release: women are emotionally intelligent; men think they are. The data are more modest. Averages, not destinies. Moderators everywhere. Causes not identified. Socialisation, confidence, biology: all still on the table. Useful, as far as it goes. A stereotype that assigns unpaid emotional labour by sex should have to survive contact with a split skill list. It does not.
Emotional intelligence arrived as a marketing success and a psychometric argument. Ability models (Mayer–Salovey style) try to treat it like a test with right answers. Mixed and trait models fold in personality, optimism, grit, and whatever else looks hireable. Critics noticed the join decades ago.
Edwin Locke's 2005 case still sits on the table: it is not clear EI is intelligence, and the definitions are so wide they stop meaning anything distinct. Intelligence, in the ordinary scientific sense, is a capacity to grasp and manipulate abstractions, measured in ways that converge. Emotions are appraisals and motives. Using reason about emotions is just reason. Calling the package a second IQ invites a cottage industry of tests that either collapse into general ability and the Big Five, or else score "correct" feelings by committee. Consensus scoring on an MSCEIT item is not the same kind of fact as a matrix item. Who decided that face was anger rather than contempt, and why is that the intelligent answer?
The measurement split in the new meta-analysis is the same old crack. Performance tests and self-reports do not behave like two windows on one trait. They behave like two different questionnaires. When the female advantage "almost disappears" on self-report, you have learned something about self-presentation and sex-typical modesty or swagger. You have not necessarily measured a latent mental organ called EQ.
Ability EI also has a habit of correlating with ordinary cognitive ability. Mixed EI has a habit of correlating with personality. Incremental validity over IQ and the Big Five is the test that keeps coming back thin once the models get strict. A construct that is "intelligence" when you want prestige and "personality" when you want workshops is doing little work.
Even if you grant a loose cluster of emotion-related skills, ranking the sexes on a composite was a category error dressed as woke equity science.
Recognition of facial cues is not the same as keeping your temper in a night shift. Managing a meeting's feelings is not the same as not catastrophising. Those skills have different training paths, different costs, and different payoffs. Collapsing them into one league table licensed two bad institutional habits at once: treat women as the designated feelers, and treat men who score low on other-focused items as emotionally stunted rather than differently specialised.
The new paper is honest about the second risk. Persistent "women have more EI" talk dumps the interpersonal regulation onto women and calls it nature. The reverse cartoon, men are robots, is the same error with the signs flipped. Average differences in a noisy, context-soaked cluster do not tell you who should run the team or who should get the promotion. They tell you that a slogan was doing too much work.
You can study emotion recognition. You can study regulation. You can study whether people are accurate about their own competence. Those are researchable. You can notice that other-focused emotional work is unevenly assigned. That is a labour fact.
What you do not need is a master trait with a trademarked name that pretends to sit beside IQ as a second general factor of the mind. The Hausfeld analysis is strongest when it refuses the master trait and reports the parts. It is weakest when the press converts the parts back into a gender Olympics with a more sophisticated medal table. Ask which ability, measured how, directed at whom. Then ask whether "intelligence" was the right word for any of it. It probably isn't.
https://phys.org/news/2026-09-assumption-women-emotional-intelligence-men.html