**In his book “A Field Guide to Lies and Statistics”, psychologist Daniel Levitin elaborates on some commonly made mistakes when it comes to interpreting data. Although a lot of the topics are closely related to the chapters from the best-selling 1960’s booklet “How to lies with Statistics” by Darrell Huff, it brings some new stuff to the table.** **Although it was published in 2016, it’s never too late to review a book about a timeless topic.**

Through six chapters on numbers, four chapters on words and five chapters on the world, Levitin wants to help us think more critically about the world. Some examples: don’t trust averages, cumulative numbers can be deceiving and significant is not the same as noteworthy. It’s great that Levitin doesn’t stick to just numbers. Because lies don’t come in statistics and graphs only. In three chapters, Levitin explains what expertise is, that you need a control group to draw conclusions and that your brain is wired to do some cherry-picking.

Personally, my favourite chapter wasn’t even about numbers, it was titled “Logical Fallacies”. Here’s a perfect example of how risk can get wrongly framed and how you can identify it through some very easy sanity checks:

“Misframing is often used by salespeople to persuade you to buy their products:

Ninety percent of home robberies are solved with video provided by the homeowner. It sounds so empirical. So scientific.”

Levitin encourages us to do a sanity check: does it seem plausible that ninety percent of home robberies get solved? No. What the company means is that ninety percent of solved robberies are from video provided by the homeowner. That’s not the same. It doesn’t say that ninety percent of all home robberies are solved. It simply means that if *and only if* a home robbery gets solved, it’s ninety percent likely that it was solved because the provider offered video tapes. Ten percent gets solved through other proof.

This brings him to the final chapter: bayesian reasoning, which tackles conditional probability: the probability that some theory is correct given that some other event has happened. Here’s an example: if you are at a party of 1000 people, and a body is found in the room with someone else’s blood on it. What’s the probability that you are guilty if the lab guys explain that there’s an 85% chance that your blood matches the sample of the body? That’s what bayesian reasoning is all about.

Surprisingly, the book only spends five pages on bayesian reasoning. If you’re curious about more, I can recommend “The Signal and the Noise” by FiveThirtyEight’s Nate Silver.

This book isn’t an encyclopedia. It’s not something that you have sitting in your book case, waiting to be referenced. It’s a field guide, a list of *some* things you should be warry of when interpreting information matters. If this topic is new to you, it’s a really good start to take your first steps towars becoming more resilient to bullshit.