How To Lie With Statistics, by Darrel Huff
One of the more telling and unfortunate aspects of life is that those who crusade against something are usually guilty of what they are crusading against. Those who crusade against injustice are often spectacularly unjust in their own worldview and behavior towards others. To truly rage against a problem often requires a certain amount of self-loathing that can only come from suppressed self-knowledge. Such is the case here. The author rages against how one can lie with statistics, and though this book does not reveal it, the author happens to have been a part of a larger campaign to lie with statistics relating to public health and smoking. Given the ways that one can twist and deceive with statistics, it is little question that data that is often used to bolster various positions and stands can easily come into question given the poor levels of data expertise among the general population and even among others. This book is written with a tone that leads the reader to be entertained but also probably more than a little bit irritated and frustrated at the misuse that can befall us when we seek to use statistics as a way of bolstering what we want to be the case.
This book is a short one at between 100 and 150 pages and makes for an interesting read throughout. The book begins with acknowledgements and an introduction. After that the author talks about how one can construct a sample with a built-in bias, a classic way of lying with statistics (1). This is followed by a look at how one makes a well-chosen average to make the point one wants to make (2), and ignoring different insights that can come about from other means of central tendency like mode and median. The author then talks about little figures that are not there (3), as well as the way that people make much ado about what is practically nothing (4), including graphs done in such a way as to skew differences. This is followed by the gee-whiz graph, which is very common in, for example, arguments like climate change (5). After that come chapters on the one-dimensional picture (6), the semi-attached figure (7), and a look at the problem of post hoc (8). The book then ends with chapters on how to statisculate (9) and how to talk back to a bad statistic (10), which is a very necessary skill to have.
One of the more interesting aspects of the book is the way that it leads the reader to be skeptical about the way that data is used and misused so often in the public sphere. Given that the motivation for people to lie is so great, it is little surprise that people end up being so deceptive so often with regards to statistics. The author does provide many ways that statistics can reliably lie, and these are conditions that exist in general over a large part of contemporary society. A great many aspects of contemporary political culture deal with areas where statistics are used and misused on a regular basis. There is a lot about this book that remains relevant even though the book was published in the 1950’s. It was timely at the time and remains timely to the present day, and likely will remain so as long as people lie with statistics, which has been going on for a long time, and is not likely going to stop anytime soon. The only thing one would have preferred to see is an update by someone else on how statistics continue to be used deceptively in the contemporary period by cultural, economic, and political elites.