Demystifies the Use of Advanced Statistcal Methods
Unlike other texts, Primer of Applied Regression & Analysis of Variance teaches both how to understand more advanced multivariate statistical methods, as well as how to use statistical software to get the correct results. This new edition offers the modern, intuitive approaches that won the first edition a wide following, while adding traditional methods for complete coverage of applied statistical methods.
*Reader-friendly style that makes complicated material approachable and usable
*Practical guidelines for the correct application of statistical software
*Examples from biological and health sciences research that clarify key points
*End-of-chapter study problems that quickly test mastery of the material
NEW IN THIS EDITION
*Expanded coverage of traditional ANOVA (analysis of variance)
*Expanded coverage of ANOVA extensions, assumptions, and workarounds for "problem" data
*Cox proportional hazard models
*Expanded coverage of repeated measures
*New examples from biological and health sciences research
*Expanded and revised coverage of statistical software
*Web site (http:www.vetmed.wsu.edu/AppliedRegression/) to support statistics instruction and facilitate use of example and problem data sets
This book provides fairly standard coverage of both regression and the analysis of variance. Most books that cover both topics do so in a cursory way and those that provide more complete coverage do so for one topic or the other, but not both. Therefore, this is a BIG book. Examples are numerous and focus on medical applications. This will appeal to certain audiences but also limits the market to those audiences. I don't know that I was able to divine any real statement of purpose in the preface to the book. "We wanted to popularize the methods we present" is the stated objective. The methods presented could hardly get any more popular then they are now. It appears to me that they wanted to write a "second course in statistics" book. That is a perfectly reasonable thing to do and there is surely some market for such an effort. This book is written for students, although exactly which is a debatable point. In my opinion, it probably is most appropriate for graduate students in biostatistics/epidemiology and in Masters of Public Health (MPH) programs with a strong emphasis in those same areas. The authors are entirely credible in the subject matter. This book begins with regression analysis, which is covered thoroughly in what I would call the standard order. Basic ideas are treated in the linear regression chapter and those are followed by the inevitably more complex issues surrounding multiple regression. Multicollinearity is important and it is appropriately given substantial coverage in the text. I think there is too much emphasis on assumptions, particularly normality, but that is not a major point. In fact, much of what appears in the chapter on assumptions isn't really aboutassumptions. The material on the analysis of variance is, again, pretty standard. Coverage of unbalanced design analyses and "missing cell" problems is unusual in biostatistics books, however. The coverage of multiple comparison methods is more complete than I have seen elsewhere, too. The author's treatment of hierarchical methods is completely couched in terms of "repeated measures" analyses. That seems a bit limiting to me because cluster randomization is quite common, especially in epidemiology. The chapter on nonlinear methods is one that I read carefully and it provides a nice overview of the topic. However, I don't think many instructors are going to cover that chapter and I don't see it being used in medical research very much. The book is uniformly well written and the coverage of several statistical packages is laudable. There isn't much that is truly innovative. There are some 3-D displays that one doesn't see in texts very often. I don't see any major shortcomings. This is a solid reference work. Biostatisticians and epidemiologists will find it useful. There are myriad other books that cover the same general ideas. This particular effort covers ANOVA and REG in some detail but does not cover other topics usually found in introductory books. Therefore, it is aimed at the "second course" niche. I have my doubts about how many schools might adopt it, but the best predictor of that is how well it has done in the past and I am not privy to that information.