Overprecision
Overconfidence in one’s estimates, Excès de précision
Being too certain of the accuracy of one’s knowledge or estimates: the margins of error one allows are too narrow.
The essentials
Mechanism
We build an estimate from what comes to mind, without accounting for all that we do not know. The sources of error we do not think of are not built into the margin.
In practice
Example
An engineer announces that a prototype will be delivered “between 10 and 15 March”. It arrives on 2 May.
Famous case
In Alpert and Raiffa’s study, students had to give 98% confidence intervals: the true value fell outside in 42% of cases, instead of 2%. For 90% intervals, studies typically find that the true value is included in only 30 to 60% of cases (McKenzie, Liersch and Yaniv, 2008).
Defending yourself
How to protect yourself
Deliberately widen your ranges. Estimate the lower bound (“I’m almost certain it’s at least…”) and the upper bound separately. Write down your forecasts to measure your own calibration.
Don’t confuse
Not to be confused with: Dunning-Kruger effect
Overprecision concerns the width of margins of error, in beginners and experts alike; the Dunning-Kruger effect, the overestimation of their level by the least competent.
Going further
Source
Alpert and Raiffa (1969 report, published in 1982). Moore and Healy (2008) distinguish overprecision from overestimation of one’s abilities and the “better than average” effect.
Practise
In the app, each entry offers ways to practise recognising the concept. No sign-up: your progress stays on your device.