── ── Cognitive bias

Dunning-Kruger Effect

The Dunning-Kruger effect is the systematic self-assessment asymmetry demonstrated by Kruger & Dunning (1999): bottom-quartile performers overestimate rank by ~50 percentile points; top-quartile performers underestimate by ~5 points. Mechanism: the cognitive skills needed to perform a task are the same ones needed to evaluate performance — so novices lack the metacognition to see their own gap. The corrective is external…

How it works

Step 1 — State the claim: Domain · verbatim claim · self-assessed percentile · basis for assessment · stakes.

Step 2 — Test metacognitive prerequisite: Can the person evaluate others' performance in this domain? Have they received external feedback? Can they articulate what failure looks like?

Step 3 — Get external data: Objective metrics · blind peer comparison · expert evaluation · track record · calibration test (predict score → take test → compare).

When to use it

  • someone says 'how hard could it be' about an unfamiliar domain
  • a novice dismisses expert opinion or says 'I could do that'
  • someone's self-assessed confidence seems disconnected from their actual track record
  • a hiring or promotion decision is driven by self-presentation
  • feedback loops are absent and someone is operating on gut confidence alone

When not to use it

the person is a known expert with an established external track record; the confidence claim is in a domain with tight, recent feedback loops that already calibrate performance.

Worked example

Kruger and Dunning's 1999 Cornell Studies

The empirical foundation is Kruger and Dunning's four-study sequence at Cornell University in 1999, which together demonstrated the effect across multiple domains and tested its mechanism.

Install this skill (free, MIT)

$npx skills add deciqAI/knowledge-skills
View Dunning-Kruger Effect source on GitHub →

Start free. Pay when it pays off.

These skills are open source. deciqAI is the operator team that runs them — autonomously, on your company.

Start free