in book

How to Lie with Statistics Summary

Here’s the How to Lie with Statistics summary. Learn the errors that people make in the interpretation and presentation of data. Also learn how to become vigilant in interpreting graphs and numbers.

Overview

How to Lie with Statistics author Darrell Huff explained how people use statistics to deceive people. He enumerated several examples that cover businesses, universities, and government agencies.

He also explained how pros use data to confuse people. He talked about errors in sampling and causation vs correlation. If you’re aware of the common tricks and errors, you’ve got a good defense.

About the Author

Darrell Huff wrote How to Lie with Statistics that became one of the best-selling statistics books of all time. He also wrote hundreds of “How to” articles over 15 books.

He shared practical statistics information that readers can understand. He presented lessons in a simple, practical, and meaningful way. Bill Gates even recommended the book in TED 2015 Conference.

Main takeaways: How to Lie with Statistics Summary

Let’s discuss the key takeaways from his book:

  1. Numbers can fool you
  2. Correlation is different from causation
  3. Look at the sampling and samples

Numbers can fool you

Which statement has more impact?

  • Many people believe in life after death.
  • 82% of adult male respondents believe in life after death.

It’s likely you’ll choose the second statement. Why? Because it uses numbers. That ‘82%’ catches your attention. It also adds credibility.

Here’s the thing though. It’s just a made up number. I could have used 75%, 23%, or 64%. It doesn’t matter. The numbers add credibility to any statement.

Numbers can fool you. If an article mentions a number or any statistic, check its source. Doubt the whole article if the writer didn’t provide the study or reference.

Remember that one of the purposes of using numbers in an article is to make it credible. This way you’ll always be aware what those numbers really mean.

Correlation is different from causation

Two unrelated events can lead to correlation. But that doesn’t mean one thing caused the other.

For example you dropped a spoon on the floor. For superstitious folks it means a visitor is coming. Minutes later it actually happened. Did the dropping of the spoon caused the visitor to come? Or did the visitor caused the dropping of the spoon?

Let’s see another example. In your town the birth rate is increasing. In another nearby town the same thing is happening. Does the birth rate of one town affect the others?

Confusion between correlation and causation is common. Some people use that by using two unrelated data. The goal is to change the readers’ perceptions. It could be they want to show that sudden increase in one thing “caused” the decrease in the other.

Keep in mind the difference between correlation and causation whenever you look at a set of data. That will help you derive accurate conclusions.

Look at the sampling and samples

“8 out of 10 mommies prefer using this product.” Or other variants of that. People use that in TV commercials and ads. But what is the story behind that?

It could be the sampling is repeated until a favorable result emerged. There’s also a chance that the samples chosen are already predisposed to using that product.

It could also be the sample size is very small. Maybe it’s only 10 samples. Or when a favorable result appeared, further sampling stopped.

The basic sampling method is “random.” It means each unit has an equal chance of belonging into a particular sample. But this is expensive. To obtain a totally random sample, it should be large enough.

That’s why people use stratified random sampling. It’s less expensive. It’s also subject to easy manipulation. The reason is people can choose the categories or “strata.” Bias can come in with that sampling method.

Whenever you look at data, take a minute to look at the actual samples and sampling method used. That will give you more insights on the results. You’ll also learn the thought processes of statisticians in gathering and interpreting the data.

Aside from how numbers can mislead people, you can also learn the following from the How to Lie with Statistics book:

  • How people use graphs to distort reality
  • Why averages don’t always reflect the truth
  • Why numbers hide other numbers
  • How to figure out if the data makes sense
  • Why visuals in infographics sometimes exaggerate the truth

My personal takeaways

How to Lie with Statistics was first published in 1954. But its contents are more timely now than ever. Why?

We see in articles and Facebook posts the deliberate use of numbers and graphs to convince people. Just add a certain statistic and the article becomes more credible overnight.

But those numbers and graphs can be misleading. They can be exaggerated to shift people’s perspectives. It can be used in political and business propaganda.

I believe the purpose of statistics is to highlight the truth. The purpose of data is to make a correct decision. If the numbers are wrong in the first place, we should not expect much on the conclusions and decisions.

Buy this book through my Amazon affiliate link

You might also like these summaries:

Made to Stick Summary

The Power of Less Summary

The Effective Executive Summary