Abstract
This is a comprehensive book covering many aspects of statistical analysis appropriate for researchers and students in the animal sciences, agriculture, biological and veterinary sciences. Perhaps the biggest praise I can give the book is that despite being originally trained as a statistician in a mathematics department and thus not perhaps its main target audience, I found this book very readable and it refreshed my memory on many aspects of experimental design and even taught me a little! The book starts from the basics and works through material appropriate for a first few lecture on statistics all the way to more complex random effect models and repeated measures designs. At over 500 pages its coverage is impressive and detailed. Most topics are followed up with numerical examples to illustrate the methods described using data-sets primarily from farm animal experiments (although the statistical topics covered are equally important in laboratory animal studies). The same examples are then generally also illustrated using the SAS software package so that the researcher can replicate the work on their own examples (if they possess SAS).
The book begins with six chapters giving basic building blocks to statistics and probability and covering summarizing and presenting data; probability; random variables and probability distributions; the concepts of samples and populations; parameter estimation and hypothesis testing. There are next four chapters on correlation and regression including curvilinear relationships and multiple predictor variables. The material then switches topic to categorical predictors with chapters on one-way analysis of variance (ANOVA) (both fixed and random ANOVA) and mixed models.
There follow 10 chapters building on the ANOVA material for specific experimental designs and covering topics such as general design, blocking, crossover designs, factorial designs, split plots, repeated measures and analysis of covariance. The book finishes with a long chapter dealing with non-normal responses and in particular covering logistic regression models and diagnostic tests.
There are several approaches to writing statistics books for a specific applied audience, in particular a decision has to be made into how much detail one gives on the underlying mathematics. In this respect, Kaps and Lamberson pull no punches exposing the reader not only to the sums of squares tables in a variety of ANOVAs but also to the matrix algebra approach of representing the many statistical models covered. This may scare off the more maths phobic potential reader for whom I am sure there are possible alternative statistics books. I however particularly like the approach that the authors take in working through the matrix algebra via numeric examples, which demystifies the selecting of statistics methods via a click of a button as in modern computer packages and shows what is really going on behind the button pressing. In this respect, the book would also work well as a text for mathematics students for their early statistics courses.
The book works best if read through sequentially, at least on first viewing, rather than dipping into it as many of the chapters build on earlier material. It would however then make a good reference guide as one of the most impressive aspects is the level of coverage of material. Some of the more advanced material and less common tests are quoted rather than derived but this is to be expected in an applied statistics book. There are unfortunately several typographic mistakes but this should not take away from what is a nice book.
One suggestion to the authors with regard to the level of mathematics is that it might be a worthwhile investment to introduce a star system where more advanced mathematical material is indicated with stars, so that the less mathematical reader is not put off. Here starred material could be described as optional material for more mathematical readers. For example, some of the sections on matrix algebra, although interesting and well presented, are not vital for later material and could be starred. In particular, the short mixed models chapter does not add much, given its lack of illustrative examples.
So summing up I would definitely recommend this book to any researcher/student looking for a comprehensive text on statistical methods for the animal sciences and who is familiar with some matrix algebra.