This book is primarily intended as a text for a one or two semester course on statistical research methods for graduate students in biology, agriculture and related life sciences. The book contains a variety of examples and exercises drawn from these areas. Many examples are drawn from the consulting experience of the author.
The book supplemented with Appendix~B can be used as the text for a one or two semester applied statistics course for first-year graduate students majoring in statistics. Appendix~B collects together a number of theoretical results that a masters level statistician should be familiar with. Instructors can selectively use Appendix~B as a source of reference for theoretical results that form the basis for the statistical methodology covered in the text. It should be emphasized that Appendix~B is not intended as a substitute for books on statistical distribution theory and the theory of least squares.
The book also comes with a companion text, authored by Mary Sue Younger (1997). This recommended companion text shows how to use the statistical computing software SAS to perform the calculations described in this book.
The book, with the exception of Appendix~B, is written at a mathematical level typical among first year graduate students in life sciences. The ability to manipulate simple mathematical formulas with symbols and interpret graphs of simple functions is expected, but knowledge of calculus or a background in statistics is not necessary. Some familiarity with upper division linear algebra and calculus will be needed when reading Appendix~B.
The book introduces statistical models early and uses a model based approach in the development of the statistical methods. The main focus is on planning and analyzing designed experiments. Point and interval estimation take priority over statistical hypothesis testing. Particular attention is paid to methods of constructing and interpreting one-at-a-time and simultaneous confidence and prediction intervals (one and two sided). Methods and guidelines for determining sample sizes get more emphasis in this text than is normally found in books written for similar audiences. The important difference between hypothesis testing and confidence interval approaches to sample size calculation is emphasized and guidelines are given on how to use SAS to determine sample sizes.
Even though the book emphasizes interpretation of results over computational details, computational details are given when it is felt that such details help in the interpretation of the computed quantities. Printouts from the popular statistical software, SAS and StatXact, are frequently used to display the results in the worked examples in the text.
The book can be divided into seven parts. The first part, consisting of the first three chapters, is devoted to a discussion of some basic concepts and definitions focal in the study of statistics. Throughout this initial part, the focus is on the theme that statistics deals with methods of collecting and using information obtained in samples to draw conclusions about populations. The notion of population, sample, sampling distribution, estimation, hypothesis testing and prediction are introduced in this part. In the second part, consisting of Chapters 4, 5, and 6, the most commonly occurring situations in which the researcher is interested in making inferences about one or two populations, are used to describe statistical methods of estimation, hypothesis testing and prediction. The third part, consisting of Chapters 7, 8, and 9, discusses some general issues pertaining to designing research studies and presents methods based on one-way analysis of variance (ANOVA) for the design and analysis of comparative experiments; that is, experiments in which the objective is to compare several treatments with each other. The fourth part consists of Chapters 10 and 11, and is devoted to a discussion of regression methods. A detailed discussion of simple and multiple linear regression is included in these chapters. Chapter 12 is the fifth part in which ANOVA models and regression models are treated as special cases of the general linear models. Analysis of covariance is treated as an example of the use of general linear model. Chapters 13, 14 and 15 constitute the sixth part, in which the one-way ANOVA is extended to cover experiments involving multiple factors with fixed and random effects. Finally, the seventh part, consisting of Chapter 16, provides an introduction to analysis of repeated measures designs, an important application of the methods based on general linear models involving random and fixed effects. In this text, split-plot experiments are treated as special cases of repeated measures studies.
Suggestions for course use.
Most of the material in the book has been class tested in graduate level applied statistics courses taught by the author to graduate students in the life sciences and statistics. The book contains sufficient material to cover a two semester sequence of courses. Depending on the emphasis of the course, some material may be omitted or used as extra reading assignments. For example, for students with some statistics background, Chapters 1, 2 and 3 may be covered very quickly spending class time on only the main definitions and using the remaining material as reading assignments. In some courses, one or more of the chapters and sections dealing with ordinal data, categorical data and sample sizes may be skipped. One possible division of the text material into two semesters will cover the ANOVA and regression topics in the first semester with second semester devoted to topics on ordinal data, categorical data, and factorial (fixed and mixed models) analyses. Such a division will correspond to the following coverage of the topics in the text.