Skip to content
All apps

Overconfidence test

About this app

Give a 90 % interval for each of ten estimation questions. Calibrated would be nine hits — most manage far fewer. Interval bars with the true answer expose the overconfidence.

What overconfidence means

Overconfidence is the tendency to believe one’s knowledge is more precise than it really is. It can be measured with confidence intervals: giving a 90 % interval for an estimation question claims that the truth lies inside it with 90 % probability. Across ten such questions, well-calibrated people should hit about nine on average. In studies many people score well below that because they choose intervals that are too narrow. This is a familiar problem in forecasting, planning and risk assessment.

What the app shows

The test panel holds ten estimation questions from different areas, for example the length of the Nile or the gestation period of elephants. For each question you enter a lower and an upper bound that should contain the right answer with 90 % confidence. After evaluation the app shows how many of the ten intervals contain the truth and draws all intervals as bars. A green dot lies inside the interval, a red one outside. A verdict ranges from “Overconfident” to “Very cautious”.

Things to try

Answer all ten questions honestly and without looking anything up. Set the bounds so that you are genuinely 90 % sure. After evaluating, look at the red bars: was the truth just outside or off by a large factor? Then start a new round and deliberately widen your bounds. Notice how your hit count changes when you take your uncertainty more seriously.

Frequently asked questions

What is overconfidence bias?
Overconfidence is a cognitive bias in which people overestimate the accuracy of their knowledge or judgement. A typical sign is estimation intervals that are too narrow. It affects laypeople and experts alike.
What is a 90 % confidence interval for an estimate?
A 90 % interval is a range with a lower and upper bound that, in your own judgement, contains the true answer with 90 % probability. Someone who is well calibrated captures the truth in about nine out of ten cases. Far fewer hits point to intervals that are too narrow.
How can you reduce overconfidence in estimates?
One simple remedy is to widen your intervals deliberately when unsure. It also helps to look for reasons why your estimate might be wrong. Regular feedback on your own hit rate improves calibration.
How does the test in the app work?
You enter a lower and upper bound for each of ten estimation questions and click “Evaluate”. The app shows your number of hits, a verdict and, for every question, a bar with your interval and the true value.

Subject: Data & decisions

More from Data & decisions

A VisuApp by heyprof: interactive, free in your browser, no sign-up. What is a VisuApp? · All apps