About

Unstatistical is for anyone who reads a headline with a percentage in it and wonders whether to believe it. Each page covers one classic way numbers mislead and lets you trigger the effect yourself, because it is much easier to remember a trap you have fallen into than one you have read about.

How we treat evidence

Every page names the people and publications behind its main example and links to them. Where a demo uses real data, such as the Berkeley admissions counts, it says so. Where it uses simulated or invented numbers, it says that too, and nothing simulated is ever presented as a finding. If you find a mistake, tell us and we will correct it.

The guide leans on the work of Darrell Huff, Daniel Kahneman, Amos Tversky, Gerd Gigerenzer, David Spiegelhalter, Jordan Ellenberg and others cited on the individual pages. It covers: Simpson's paradox, Base-rate neglect, Survivorship bias, Misleading axes, p-hacking, Regression to the mean, Correlation is not causation, Relative vs absolute risk, The law of small numbers, Mean vs median.

Get in touch

Corrections, suggestions for a trap we should add, or a number you would like checked: james@unstatistical.com.