---
title: "The Dunning–Kruger effect, as it's usually misquoted"
url: https://erroneo.us/daily/dunning-kruger-effect/
description: "The 1999 study found low scorers overestimated their rank, not that 'dumb people think they're geniuses.' Part of the famous pattern may be statistics."
type: daily-entry
section: "Thinking"
date_published: 2026-11-10
author: erroneo.us Editors
date_modified: 2026-11-10
publisher: erroneo.us
license: All rights reserved. Quote with attribution and a link.
---

# The Dunning–Kruger effect, as it's usually misquoted

> It's the most-cited bias on the internet and the one most often used as an insult. The real finding is narrower, more interesting and still argued over.

*Error of the Day for November 10, 2026 · Thinking · by erroneo.us Editors*

- **Wrong:** Dunning–Kruger proves that dumb people think they're smarter than experts.
- **Right:** In Kruger and Dunning's studies, the lowest scorers overestimated how they ranked, and part of that pattern may be a statistical artifact.
- **Memory trick:** It's about misjudging your rank on a test, not a law that the ignorant feel like geniuses.

## What people get wrong

Online, "Dunning–Kruger" usually means "you're too stupid to know you're stupid." It's often accompanied by a chart of a confidence curve that shoots up to a peak labeled "Mount Stupid," crashes into a "Valley of Despair," then climbs slowly toward expertise.

That chart isn't from the research. The original 1999 paper, ["Unskilled and unaware of it"](https://pubmed.ncbi.nlm.nih.gov/10626367/) by Justin Kruger and David Dunning, reported four studies in which participants took tests of humor, grammar and logic and then estimated how well they'd done compared with their peers. Nobody's confidence was tracked across a learning journey. There's no mountain in it.

## What the study actually found

People who scored in the bottom quarter "grossly overestimated their test performance and ability." On average their scores put them around the 12th percentile, but they guessed they were around the 62nd. The authors argued that the skills needed to do a task well are often the same skills needed to judge whether you did it well. The abstract calls this a "dual burden."

Two details rarely make it into the meme. First, the claim is about estimating your *rank* on a specific test, not about general intelligence. Second, the paper reported that improving participants' skills helped them recognize the limitations of their earlier performance. That's a hopeful finding, not an insult.

## Why the effect may be partly a mirage

Here's the nuance the pop version leaves out: some researchers argue that much of the classic pattern would appear even if nobody had any bias at all.

The reason is statistical. Self-assessments are noisy, and people tend to place themselves somewhat above average. If you sort people by their actual score, the bottom group has far more room to err upward than downward, and the top group the reverse. So the bottom quartile will tend to appear to overestimate and the top quartile to underestimate, even when everyone's guesses are equally inaccurate. Part of the gap is regression to the mean, not a lack of self-awareness.

[Nuhfer and colleagues (2016)](https://digitalcommons.usf.edu/numeracy/vol9/iss1/art4/) ran random-number simulations and showed that common ways of graphing self-assessment data, including line charts of quartile averages like the classic figure, "introduce artifacts that invite misinterpretation." [Gignac and Zajenkowski (2020)](https://doi.org/10.1016/j.intell.2020.101449) put their conclusion in the title of their paper: the Dunning–Kruger effect is "(mostly) a statistical artefact."

"Mostly" is doing honest work there. The debate is about how big the real effect is, not whether people ever overrate themselves. They clearly do. What's in doubt is the dramatic, uniquely clueless bottom quartile.

## The fix

Wrong: He doesn't get it, ~~classic Dunning–Kruger: the dumber you are, the smarter you think you are~~.

Right: He may be overrating himself. Low performers in Kruger and Dunning's studies overestimated their rank, though some of that pattern is statistical.

Why: The finding concerns self-assessed rank on specific tasks, and its size is disputed.

- **Don't use it as an insult.** Everyone is in the bottom quartile at something, and the study's point applies to you as much as to the person you're arguing with.
- **Drop the mountain chart**, or at least label it as a cartoon about learning, not data from the paper.
- **Check your own calibration.** Before you see a result, write down a guess (your score, your estimate, your deadline) and compare. Feedback works better than self-reflection alone, which is close to what the original paper found.

For more biases that are real but routinely overstated, see our guide to [cognitive biases](https://erroneo.us/thinking/cognitive-biases/).

## Sources

1. [PubMed: Kruger & Dunning, Unskilled and unaware of it (J Pers Soc Psychol, 1999)](https://pubmed.ncbi.nlm.nih.gov/10626367/)
2. [Numeracy: Nuhfer et al., Random Number Simulations Reveal How Random Noise Affects the Measurements and Graphical Portrayals of Self-Assessed Competency (2016)](https://digitalcommons.usf.edu/numeracy/vol9/iss1/art4/)
3. [Intelligence: Gignac & Zajenkowski, The Dunning-Kruger effect is (mostly) a statistical artefact (2020)](https://doi.org/10.1016/j.intell.2020.101449)

## Go deeper

- https://erroneo.us/thinking/cognitive-biases/

---
Source page: https://erroneo.us/daily/dunning-kruger-effect/ · More: https://erroneo.us/llms.txt
