How to Lie with Statistics
by Darrell Huff
Statistics mislead less by fabrication than by selection and presentation; learning the standard manipulations is the practical defence, because the numbers themselves are usually true.
What it teaches
Huff's method is to teach deception in order to inoculate against it, and he works through the manoeuvres in turn. The sample that was not random, so the result describes whoever happened to answer. The word average, which can mean the mean, the median or the mode, chosen after the fact to suit the claim. The axis that does not start at zero, or the graph where a doubled quantity is drawn as a shape with four times the area. The precise-sounding figure with no error attached. The correlation offered as a cause, when the arrow may run the other way or a third factor may drive both — and the related failure of reading an extreme result as a trend when it is a fluctuation that will settle back toward the ordinary. He is equally sharp on the difference between a real pattern and the clusters random data reliably produces, on the survivors studied while the failures go uncounted, and on the missing base rate that makes a scary percentage meaningless. Written in 1954 with cartoons, it remains the fastest available training in reading a number sceptically.
The ideas this book explains
Read the idea in two minutes, then read the book behind it.
Read or listen to How to Lie with Statistics
Hear the whole book free: start an Audible trial and your first audiobook — this one, if you like — is on the house.
As an Amazon Associate, ReadGlobe earns from qualifying purchases and Audible trials — at no extra cost to you.
