How to Lie with Statistics Reissue Edition

In 2016, all the best money was on Hillary Clinton to defeat Donald Trump in the U.S. Presidential election. Massive polls analyzed by highly-credentialed experts said so. The election went the other way. How could the analysts have gotten it so wrong? A potential answer is found in the pages of this short book written more than sixty years earlier by Darrell Huff and charmingly (if whimsically) illustrated by Irving Geis.This book was recommended to me in passing by one of my professors when I was completing the capstone course for my BS in mathematics. Largely because of who suggested it, I expected a book about mathematical statistics. Instead, this is a book about understanding how statistical analysis can be abused (by journalists, politicians, advertisers, etc., etc.). It does not denigrate the practice of statistical analysis itself (though you will not learn even a single technique from statistical theory in its 144 pages), but rather serves as a lighthearted cautionary tale about how easy it is to become convinced that statistics carry all the weight of science even though statistical analysis is both science and art.The reader already well-versed in statistics will not find any new information but will still be pleased by the book's artful presentation of known ideas. Readers who are not so well-versed in statistics should consider this book required reading because it succinctly explains how the information we all consume every day may have been manipulated--intentionally or otherwise--to give us false impressions.In fact, I would argue the value of this book has only increased in the decades since its initial publication. While the reader picking up this book upon its publication in 1954 would surely encounter plenty of statistics and graphs throughout the week, our modern 24-hour news cycle and constant immersion in a multimedia world has magnified the opportunity for statistical deception. Of course, you'll find that the book's examples are outdated (references to an exorbitant $25,000 salary for Yale graduates might seem at first more quaint than informative). However, despite the dated examples, the statistical phenomena described are as relevant as ever. Indeed, an argument could be made that the examples from yesteryear might even aid the book's pedagogical value by avoiding the contemporary issues that might cause the reader to don partisan intellectual blinders.If I were to criticize, I would say that the book fails as an introduction to statistical thinking. For example, it rightly cautions the reader to beware of the difference between median and mean when interpreting reported "averages," but fails to provide much insight regarding when each of these measures of central tendency might be superior to the other. As such, the reader looking for insight regarding the practice of statistics, even from a non-technical perspective, may be disappointed. However, the reader interested in the consumption of statistical information will find a wealth of information packed into a charming little book.

Great little book. Catchy, slightly misleading title. Thed author does not show you how you can lie with statistics. This book illustrates how statistics can be manipulated to conform with someone's preconceived conclusions or messages. Advertising, of course is full of this sort of shenanigans, and the world of advertising provides numerous examples of misleading statistics.Because of the sorts of abuse outlined in this book, statistics fare often suspect, mistrusted in our day-to-day existence. This is unfortunate, as the basic quantification of various phenomena is essential to our understanding of the world around us. All the more reason to read this book carefully: you'll be better equipped to spot the intentional (and to be fair, sometimes unintentional) misuse of statistics and the misuse of analysis and logic.Highly recommended, and truly a classic "caveat emptor."

This is a useful book for those who have no familiarity with probability and statistics. The level of treatment is basic - no "equations" but concepts well explained. Inquisitive 9th graders should be able to understand it with no trouble.My complaint is the book is too simple. Less verbose with more insights on problematic applications would be good. For example, quoted probabilities for medical maladies frequently call for Bayesian concepts that are wrongly interpreted, often poor understanding of false positives/negatives. The abstract Bayesian concept would be a good one to explain in simple language (graphical explanation is elegant). Then there is also the issue of correlation and causation ... could be treated better. It's bad enough that rigorous experimental designs are difficult. Not understanding the impact of (independent, dependent) variables (or that a multivariate analysis correlation with a huge number of variables can be meaningless) is deadly. Granted, the book was written in 1954 however, probability and statistics has long been part of the human equation, and there is a clear need for everyone to have a better understanding (well stated by other reviewers). I suggest this book falls short, hence 4* (4* only because it is well written, my bias lower but wish to be objective).I read editorial book reviews and 5* and 1-2* star reviews before buying. In this case should have paid more attention to the low side comments that on rereading had clues. Buried were sentences I wish I'd seen before buying.

This book is 70yrs old and is as relevant as ever and thoroughly enjoyable to boot, far exceeding my expectations. Not only does the reader gain the benefit of learning statistical fallacies he may already be aware of, but sees that they still apply 70yrs later. This book has aged like a fine whisky and serves as confirmation bias to my long held belief that the every day person is easily tripped up by simple percentages. Highly recommend for fellow stats nerds

Is more storytelling and text than graphics or charts