Statistics Critical Thinking Big Ideas

How to Lie with Statistics (Summary)

by Darrell Huff

Imagine a chart showing a company's profits skyrocketing. The line on the graph shoots up from the bottom left to the top right. But look closer. The chart's vertical axis doesn't start at zero—it starts at $10 million and ends at $10.5 million. That dramatic 'skyrocketing' growth is actually a tiny, almost negligible increase. This is the 'Gee-Whiz Graph,' just one of the many simple tricks used every day to manipulate your perception of the world through data.

The Sample with the Built-in Bias

A conclusion is only as good as the sample it's based on. Many impressive-sounding statistics are worthless because they are drawn from a sample that doesn't accurately represent the whole population.

In 1936, the Literary Digest magazine polled its readers and confidently predicted a landslide victory for presidential candidate Alf Landon. They were spectacularly wrong. Why? Their sample was drawn from telephone directories and car registration lists—which in 1936 heavily overrepresented the wealthy, who favored Landon, while completely missing the majority of voters who would elect Franklin D. Roosevelt.

The Semi-Attached Figure

This trick involves using a true but irrelevant statistic to support an unrelated claim. It creates a powerful but false connection in the reader's mind, proving one thing while making you think it proves another.

An ad for a cold remedy proudly states that it killed over 31,000 germs in a laboratory test tube in eleven seconds. While factually true, this says nothing about how the product works in the human body against the viruses that actually cause the common cold. The impressive number is 'semi-attached' to the claim of curing your cold.

Post Hoc, Ergo Propter Hoc

One of the most common logical fallacies is assuming that if B comes after A, then A must have caused B. This confusion of correlation with causation is the source of endless misinformation and superstition.

A study might find that college graduates earn more money than non-graduates. The easy conclusion is that a college education causes higher income. However, it's also true that people who go to college tend to come from wealthier families and may have more ambition and intelligence to begin with, factors that also lead to higher earnings. The diploma isn't the sole cause.

The Little Figure That Isn't There

Often, the most crucial information is what's left out. A statistic can be technically true but deeply misleading if it's presented without its context, such as the margin of error or the sample size.

A toothpaste ad claims that '27% of a group of dentists' recommend their brand. This sounds impressive until you realize the question might have allowed dentists to recommend multiple brands. More importantly, we don't know the sample size. The 'group of dentists' could have been just a handful of people, making the percentage statistically meaningless.

Go deeper into these insights in the full book:
Buy on Amazon
Listen to the full audio book with an Audible Free Trial.
As an Amazon Associate, qualifying purchases help support this site.