Written to be easily absorbed by its audience, many majors may profitably use this book

It is always a pleasure to review a text that is easy to read and understand, when targeted to a novice audience. This book was written for business majors at the junior undergraduate level, and not statistics majors. However, it is recommended that readers have a course in introductory statistics before using this book. As such, it contains mostly business datasets (more on that later); it is well-suited for its target audience. For the rest of us, it is still an excellent introduction to the understanding, calculation, uses, and pitfalls of the most common forms of regression.

The style is not unusual for some (read very few) undergraduate textbooks, especially in math and physics. The author deliberately keeps mathematical details to a minimum, avoiding all calculus, while concentrating on understanding and interpreting results. This is done by avoiding tedious hand calculation through the use of statistical software. Mercifully, while using “R” to generate the book’s graphics, the standard commercial statistical packages (SPSS, Minitab, SAS, JMP, Stata, Statistica, and S-Plus) are recommended. Some data is also provided for “R”, DataDesk and EView (I admit the last two mentioned are new to me).

The seven chapters cover the following topics: Foundations, Simple Linear Regression, Multiple Linear Regression, Regression Model Building I & II, Case Studies, and Extensions. The last two chapters involve three case studies (two business, one pharmaceutical) and advanced topics, such as Generalized Linear Models, Variance Component Analysis and Bayesian inference.

The five appendices include software, a t-table with critical values, notations and formulas, a math refresher (natural logs, exponentials, rounding and accuracy), and the inevitable “answers to selected problems.”

The chapter problems are fairly easy to do with the suggested software. So, after the software learning curve is met, the students can concentrate on concepts and understanding.

Written to be easily absorbed by its audience, many other majors may profitably use this book to master regression. It is very well-written for its level, definitions are set off by italics and each chapter ends with a concise and clear summary. On his attempts to facilitate learning, the author succeeds admirably.

Explanations of concepts are clear, the flow of material logical, the examples simple-to-grasp and the Web site for the book contains all the data sets in most of the formats for the recommended software. As an added bonus in each section, after the student learns how to build a model, the author suggests how these models contain assumptions that need to be examined and results validated. One feature students will particularly appreciate is the fact that X- and Y-variables are given different names in different texts (and software!) and the author lists all of the common ones. Same goes for various specialized modes of analysis, such as Variance Component Analysis.

The book’s negatives are few or not of paramount importance. The Web site indicates some errors, and the examples overwhelmingly favor business. Yes, the book was originally written for business majors but could be more profitably used by other students with examples from their own fields to increase relevance and motivation.
I found it rather strange that the author’s students were confused by the standard Greek letters, e.g., µ for the mean and σ for the standard deviation, as well as familiar nomenclature, such as Ho for the null hypothesis. In my experience, I haven’t found that this confuses most undergraduates. He also claims that Bayesian analysis is more common than standard mixed-models analysis (frequentist, employing reduced maximal likelihood estimation), although the latter is extensively used in industry.
The only major complaint is that, in several mathematical derivations, important steps are omitted and students may have difficulty completing the algebra.

Availability
Applied Regression Modeling, by Iain Pardoe (2nd Ed.) John Wiley & Sons, Inc. NJ. Hardcover (346 + xx). $110

John Wass is a statistician based in Chicago, IL. He may be reached at editor@ScientificComputing.com.