The evidence for emotional intelligence at work isn't thin. It's built on people grading themselves.

You're two candidates into a final round for a plant manager job. One has the numbers. The other doesn't have quite the resume, but she read the room in the interview — calmed down an anxious reference call, picked up on a tension between two of your team leads before you mentioned it. You find yourself leaning toward her because she seems, in the word everyone reaches for, emotionally intelligent. That instinct has more academic support behind it than almost any other soft-skill hunch you could name. It also has a specific, built-in blind spot, and the same research that backs the instinct is the research that names the blind spot.

The paper is "Robust associations of emotional intelligence with human flourishing: A second-order meta-analysis," by Tyler J. Robinson and Ethan Zell, both in the Department of Psychology at the University of North Carolina at Greensboro, published in the Proceedings of the National Academy of Sciences in May 2026. A second-order meta-analysis doesn't collect new data — it pools existing meta-analyses. This one pooled 62 of them, covering more than 3,000 individual studies and over a million participants, across 77 different life outcomes: mental health, cognitive ability, relationship quality, and — the category that matters here — workplace performance and leadership. PNAS kept the full paper behind a wall I couldn't get past, so what follows comes from the study's own published abstract and from reporting on it, not from a page-by-page read of the method and results sections.

The headline number is real and it's not small: emotional intelligence correlated with human flourishing at r = 0.28 (95% CI, 0.25 to 0.31), holding up across ability-based, trait-based, and mixed measures of EI, and across age groups and cultures. Workplace outcomes were among the strongest of the 77. If you've ever bet on the more emotionally attuned hire, this is real support for that bet.

Here's the qualifier the authors built into their own abstract, not one I'm layering on top of it:

"Nevertheless, effect sizes were most pronounced for self-report EI measures, especially when flourishing was likewise measured subjectively." — from the study's abstract, Proceedings of the National Academy of Sciences

In plain terms: when the emotional-intelligence score and the flourishing score both came from the same person describing themselves, the relationship was strongest. When one side was measured by something other than the subject's own account of themselves, it shrank. The abstract also flags "substantial heterogeneity" in the effect sizes across studies and calls for more research into what's driving that spread — that's their language, not mine, and it means the 0.28 is an average sitting on top of a lot of variation, not a number that applies evenly everywhere.

Here's why that lands differently for you than for a researcher reading the abstract in an office. You are not a survey instrument. But when you sit across from that plant manager candidate and come away thinking "she gets people," you are doing something structurally close to what the weakest studies in this pool did: forming an impression of someone's emotional skill based on how they present themselves, and then checking that impression against your own read of how well things go with her — which is also, in the end, an impression. Two subjective reads confirming each other will feel like strong evidence. It is exactly the pattern the authors found producing their most inflated numbers.

The part the research doesn't say, but that I'll say plainly: a person who scores well on self-described emotional intelligence, and who also makes you feel good about her in the room, may be genuinely skilled at reading and managing people. Or she may simply be good at knowing what a hiring manager wants to see and comfortable telling you she has it. The meta-analysis can't and doesn't distinguish those two people from each other — that distinction is mine, drawn from the measurement problem the authors describe, not a finding they report themselves.

One thing to do with this before your next promotion or hire: don't drop the read on emotional skill, but stop letting it stand alone. Pair it with at least one data point that isn't self-description and isn't your own gut — a 360 from people who've actually reported to the candidate, a specific behavioral account from a reference ("tell me about a time she had to deliver bad news to her team" and press for what actually happened), or a work-sample situation where you watch her handle a real disagreement instead of describing how she'd handle one. The self-report and the vibe check are each cheap. The two of them agreeing with each other is not the same thing as either one being true.

If you want a structured version of that second data point before your next hire or promotion decision, the Leadership Profile at /wiring-map is built to give you a read that doesn't depend on someone's account of themselves.