The Manager's Guide to Statistics

Author Erol A. Peköz, on the faculty of the Boston University School of Management, has been, for at least a decade, improving and expanding his introduction to statistics for would-be managers and decision makers desiring to move past low-level details of statistical theory. With no mathematical expertise beyond that obtained by an MBA or the typical undergraduate, one can here gain confidence in developing practical applications and making sense of typical statistical summaries and diagrams.

Topics include probability, hypothesis testing, correlation, multiple regression, standard error, normal distribution, and more. There are many examples using Excel, including rich implementations of histograms and stepwise regression. There are a few examples using R. The reader can download several data sets and spreadsheets from the book’s site. Histograms and standard error are some of the topics delved into here with a depth and clarity that would add to any of the similar introductory statistics texts I have seen.

Much like a textbook for an introductory statistics course, key terms are in bold and core concepts are repeated in callouts. There are plenty of well-composed exercises. Although solutions are not provided, I feel the book is self-contained enough to make this not a serious absence. There is no mathematical notation until page 27 and the first “technical notation” is the formula for sigma on page 84. The focus throughout is on general concepts over mathematical formulas. Often, these sections cover at length and in depth the counterintuitive phenomena and unexpected sources of error that may arise in doing statistics. This includes biases such as survivor bias, the skewing effect of non-respondents in surveys, size biasing, collinearity, and the inspection paradox

Among the many areas of applied statistics touched on here, there is stock valuation (such as the beta), and property valuation, with house values returned to a few times in moving from simple linear regression to multiple regression. Multiple regression culminates this book and is held up with convincing arguments and examples as a technique comparable to more advanced, i.e. complex, approaches for making predictions on an unknown population based on several variables. This includes several pages covering the careful implementation of dummy variables to handle data categories effectively.

This book is also superior, I find, to other non-mathematical introductions in presenting binomial and hypergeometric distributions. Along with the t-test, comparable texts typically dispense with these topics with less detail and lucidity than this effective guide for those aspiring to employ and understand basic statistics in the context of real-world business problems.

There is a high likelihood that Tom Schulte is biking the St. Tammany Trace rails-to-trails bike path, given that it is sunny and dry.