Author: Stephen M. Stigler Rating: 7/10 Buy from Amazon.com

 

This year I decided to improve my knowledge of statistics. Along with technical books and online courses (to be reviewed in separate articles), I felt curious about the history of statistics.

This book was the first one I came across that promised a detailed coverage of the history of statistics, at least the history before the 20th century. The book didn't disappoint on that objective. Stephen Stigler comes across as very knowledge on the subject matter, and explains in some detail the main developments in the area of statistics. The book starts with Bernoulli, de Moivre, Laplace and Gauss, and progresses all the way through the great late 19th century British statisticians: Galton, Edgeworth, Pearson and Yule. There's a particular focus on the technique of "least squares" leading up to the development of "regression". The importance of regression as a statistical method is implicit in this book, but you don't realize its importance until you read other texts such as Super Crunchers. For example, most of the personalized predictions and recommendations websites provide these days use some form of regression. I wish the book had tied these things together to provide some motivation behind the coverage of regression. The book is quite technical and assumes from the reader some relatively advanced prior knowledge of the matter. I was reading some technical material just before reading this book, and even so, I found some of the material difficult to follow. Therefore, this is not a "popular science" book that I would recommend to someone without a good foundation on statistical theory. The book takes at times an overly academic and rigid tone which isn't always enjoyable. The author's style is not to everyone's liking, and you might find more enjoyment out of authors like William Dunham when writing about Mathematics.

The period of history covered by the book is also somewhat limiting. On the one end, the book excludes the very early developments in the field and you won't find much discussion of probability theory for example. On the other end, because of the "1900" upper limit, important modern statisticians like Fisher are excluded. This is however not an entirely fair criticism of the book since, as stated at the outset, the author clearly set out to write a book focused on this limited time period. However, I believe I would have enjoyed a broader coverage even if sacrificing some of the detail (which one can get in the technical books). In summary, this is a well researched and thorough walk through the developments in statistic in this period for the more mathematically inclined. For those looking for a more entertaining and broader coverage, there are better choices out there.

Buy from Amazon.com