Overconfidence effect
The overconfidence effect is the gap between how accurate people think their judgements are and how accurate they actually are. When people say they are "99% sure," they turn out to be wrong far more than 1% of the time.
Why it happens
We have rich access to our reasoning but none to its blind spots, so confidence is built from the evidence we can see, not the evidence we are missing. We also rarely get clean feedback on predictions, so miscalibration is never corrected — and feeling certain is more comfortable than admitting doubt.
Examples
- Studies where people’s "90% confidence intervals" contain the true answer only about half the time.
- The vast majority of drivers, investors, and founders rating themselves above average.
- Experts forecasting timelines and outcomes with a confidence their track records don’t support.
How to counter it
- Calibrate with feedback: record predictions with confidence levels and score them later.
- Widen your confidence intervals — then widen them again; they are almost always too narrow.
- Ask "what would have to be true for me to be wrong?" before committing.
The deeper point
The danger isn’t confidence itself — it’s that confidence and competence feel identical from the inside but correlate weakly from the outside. The most useful question to ask an expert (or yourself) isn’t "are you sure?" but "how often are you this sure and still wrong?"
Frequently asked
- What is the overconfidence effect?
- It is the tendency for people’s confidence in their judgements to exceed their actual accuracy — being "99% sure" while being wrong far more than 1% of the time. Confidence outruns correctness.
- How is overconfidence different from the Dunning–Kruger effect?
- Dunning–Kruger is specifically that the least skilled overestimate the most. The overconfidence effect is broader: nearly everyone, including experts, is more confident than accurate, especially on hard questions.
- How do you reduce overconfidence?
- Calibrate: log predictions with confidence levels and score them against outcomes. Deliberately widen your confidence intervals and ask what would make you wrong. Feedback is the only reliable cure for miscalibration.
Related
Editorial synthesis © ReadGlobe 2026, drawing on Kahneman’s Thinking, Fast and Slow, the Tversky–Kahneman research program, and the primary cognitive-science literature. · Last reviewed 2026-05-29.