How to Read Numbers: A Guide to Statistics in the News (and Knowing When to Trust Them)

I hadn’t heard of this book until recently. Glad I found it (thanks to goodreads!). The news is SO misleading SO much of the time. Especially with data. Tom and David Chivers help sort all this out.As they highlight, context with data matters a ton. Take this simple statement: Nearly ten percent of the world’s people live in extreme poverty.This is true. And standing alone, without context, this statement would lead many to think the world has a huge poverty problem. They'd be right. But they'd also be missing a big part of the equation: Just twenty-five years ago nearly 30 percent of people lived in extreme poverty.So the world’s extreme poverty problem is actually getting much better.The prospects for the future would be very different if instead of consistently improving extreme poverty was stagnating at nearly ten percent. Or, worse, if extreme poverty was increasing. But there's no way of knowing which of these distinct scenarios is true from the simple statement alone.You need context. And the news so often doesn’t provide any.Terrorism is another good example. The events of September 11, 2001 highlighted the great dangers of terrorism. It obviously has received a lot of press. But what about other threats? While terrorists have killed about 3,000 Americans since 2000, over 20,000 Americans die by homicide or murder annually. And more than 38,000 die from traffic accidents each year.Is terrorism, in context, as threatening as the media can make it seem?Hardly.The book explains in great detail how people embrace facts presented by the press without context all the time. They disregard trends. They ignore the net impact of something and focus on discrete effects. They gloss over comparables and instead focus on isolated samples. They confuse outliers and aberrations with the mean or the median. And—a big recurring challenge—many have trouble with scale, hardly distinguishing 100 million units of something from a billion units.Anecdotal thinking, moreover, is particularly widespread, as the authors delve into. The press often tells stories without considering the big picture. Global warming deniers, for example, describe dramatic blizzards to assert that the world is not warming. But to know whether there's global warming you need to know, well, whether the globe is warming—not how much snow there was in a single place on a single day. And the data clearly establish that Earth, on the whole, is getting hotter.It’s often the case, like with global warming, that context with data is not just important—it is a necessary precondition to understanding something at all.This short little book isn’t just enjoyable; it’s super useful. All consumers of the news should read it. And—perhaps more importantly—so should every journalist.

I would love it! But $11.91 for 172 pages?Nope!

Even though I am not a total beginner in the field of statistics and data analysis I found this book entertaining, amusing, and above all, informative. There are many examples of what to watch out for when reading or hearing reports on numerical data, with topical cases and explanations taken from real life. Even more impressive, my wife, who claims to know nothing about maths let alone statistics, has delved in and enjoyed reading it too. The book will be a great help to anyone who wants help to see through misleading or incorrect data reports.

This is one of my favourite kinds of book - it takes on the way statistics are presented to us, points out flaws and pitfalls, and gives clear guidance on how to do it better. The Chivers brothers' book isn't particularly new in doing this - for example, Michael Blastland and Andrew Dilnot did something similar in the excellent 2007 title The Tiger that Isn't - but it's good to have an up-to-date take on the subject, and How to Read Numbers gives us both some excellent new examples and highlights errors that are more common now.The relatively slim title (and that's a good thing) takes the reader through a whole host of things that can go wrong. So, for example, they explore the dangers of anecdotal evidence, tell of study samples that are too small or badly selected, explore the easily misunderstood meaning of 'statistical significance', consider confounders, effect size, absolute versus relative risk, rankings, cherry picking and more.This is all done in a light, approachable style that makes the book a delight to read. Just occasionally the jokes are a little heavy-handed (as when they gave examples based on people buying their book), but it's all nicely balanced, informative and laser-accurate in pinpointing the errors we see day after day from the media. The book finishes with a 'statistical style guide' for journalists, which should be printed out and placed on every desk in a media outlet.Off the top of my head, the only major issue they don't address is spurious accuracy - though this is indirectly covered in asking for statistics to be given with confidence intervals. Overall, an excellent addition to the armoury of good use of statistics. I suspect a self-selecting readership will result in the Chivers brothers largely preaching to the converted - but if even a few media sources take the hint it will be well worth it, and even if they don't, it gives readers the tools to recognise misuse of statistics in many cases.

実はこの本を読んでいて、先日の山行で降車駅を間違えるという失態を起こしてしまった。格別面白いというわけではないのだが、どうもある部分で熱中してしまい、われを忘れてしまったようだ。中身はというと、統計学の超入門書。大学生の入門書レベルだろう。統計学の初歩中の初歩を、メディアの報道で使われる数字を反面教師として、一般人にもわかり易く解説したもの。実はこの統計学には昔からかなりのコンプレックスがあり、ときどき入門書なるものを手に取ってみる。ただ頭の元からの悪さとこの種の領域でのセンスの無さは年を重ねても変わらないようで、いつも途中で投げ出してしまう。今回は、例のウイルス騒動で目にしたあまりものメディアの報道のひどさとウイルス専門学者のお粗末さに、さすがのワテも呆れて、買うか買わないかだいぶ迷ったが、値段が非常に安くなったため購入。サンプリングとバイアス、統計的な有意の意味、因果関係と相関関係、ランキングの無意味さ、相対的なリスクと絶対的なリスクなどの基本的な概念がわかり易く解説されている。のだが、これほどの入門書でも理解不能な部分がまだあることに愕然とする。とはいえ、一言でまとめると、いうまでもなく、メディアの報道はどうしても新奇性（novelty)を求めることが第一の優先順位となるため、そこでとりあげられる数字については相当に割り引いて受け止めなければいけないということだ。つまり僕らが目にするheadlineのほとんどが眉唾もの。さらにそこにはメディアの否定しがたい先入観が混入されている。つまり文脈を無視した結論ありきの下での数字の報道ということだ。さらには、発表された科学論文なるもの自体が相当な問題性を含んでいるということだ。新奇性の追求は、書かれる論文の発表にも必然的に影響を及ぼしている。さらには、本書でも指摘された「goodhartの法則」ーある指標が目的となると、もはやその指標は良い指標とはなりえないーの意味合いは深い。指標はあくまでも複雑な現実の限定的な代用品であり、それ以上ではない。それが一人歩きしてしまう。

I bought this because I'm studying statistics and wanted some of the concepts explained in plain English with relatable examples. This book fits the bill, it's brilliant. I read ten chapters in one go on a train journey and was absorbed throughout. A really accessible read, the 'technical' explanations are in boxes so you can ignore them if you want to, but even they are so clearly explained. I feel much more confident in conversation about numbers and how they can be used/ misused. I'd definitely recommend this book.

A concise and readable account of statistics that everyone should read.