How to Lie with Statistics: 8 key ideas
The core ideas of How to Lie with Statistics, each in under a minute, written in our own words. Want the full depth? Get the book.
- IDEA 1 OF 8
Statistics can mislead easily
Huff's short, witty book shows how numbers can be used to mislead people, whether deliberately or by accident.
He argues that understanding these tricks helps readers protect themselves from false impressions.
- IDEA 2 OF 8
Samples can be biased
Huff explains that conclusions from surveys depend on who is asked.
A biased sample, such as only asking certain people, gives misleading results. So a result from a small or hand-picked group may not be true for everyone.
- IDEA 3 OF 8
Averages can be chosen to suit
Huff shows that the word average can mean mean, median or mode, which can give very different numbers.
People may choose whichever suits their argument. So whenever someone quotes an average, it is wise to ask which kind they mean.
- IDEA 4 OF 8
Missing figures matter
Huff warns that statistics often leave out important information, such as sample size or margins of error.
Without these, numbers can mislead. For example, a result from just a handful of people may look impressive but could be pure chance.
- IDEA 5 OF 8
Graphs can exaggerate
Huff describes the gee-whiz graph, which cuts off part of the scale to make small changes look dramatic.
Visuals can distort reality. A line that looks like a steep climb may actually show only a tiny rise.
- IDEA 6 OF 8
Pictures can mislead
Huff shows how pictures scaled in two or three dimensions can exaggerate differences.
A picture twice as tall may look much more than twice as big. So the picture can impress the eye even though the real numbers differ only a little.
- IDEA 7 OF 8
Unrelated figures are attached
Huff describes the semi-attached figure, when a statistic about one thing is used to prove another.
The connection is weak or false. A claim may sound scientific because of the figures, but they may not really prove what is being said.
- IDEA 8 OF 8
After does not mean because
Huff warns against assuming that because one thing followed another, it was caused by it.
This post hoc reasoning is common. Two events happening one after another may be only a coincidence, or both may come from some other cause.
Get one idea like this every daySwipe through ideas, go deeper on any of them, and save the ones that matter. Free.
Read How to Lie with Statistics in the app