Data Mining Book Review: How to Lie with Statistics

howtolieThis short and well written book by Darrell Huff is a journey in the world of misleading statistics. The book is full of funny drawings from Irving Geis. Although not technical, the book covers several important topics. Bias, for example in polls, is detailed through several examples. The misleading word “average” is compared to terms such as mean, median and mode.

Throughout his book, Huff warns the reader about statistics traps. He shows plenty of cases which, although dating from the 50’s, are still more than valid now. When you read an average, you should ask yourself: average of what? who is included? Huff also shows how to play with visualization. According to what you want to show, you can “zoom” in a plot to change the effect it has on people.

Huff gives nice examples of group comparisons where, in fact, groups are not comparable. The often discussed topic of correlation vs causation was also well known at that time, as can be read in chapter 8. Finally, the last chapter presents five questions to ask yourself when confronted to statistics. A must-have book for anyone involved in statistics!

How to Lie with Statistics

Share

Recommended Reading