Preface

Statistical hypothesis testing has been introduced almost one hundred years ago and has become a key tool in statistical inferences. The number of available tests has grown rapidly over the decades. With this book we present an overview of common statistical tests and how to apply them in SAS and R. For each test a general description is provided as well as necessary prerequisites, assumptions and the formal test problem. The test statistic is stated together with annotations on its distribution. Additionally two examples, one in SAS and one in R, are given. Each example contains the code to perform the test using a tiny dataset, along with output and remarks that explain necessary program parameters.

This book is addressed to you, whether you are an undergraduate student who must do course work, a postgraduate student who works on a thesis, an academic or simply a practitioner. We hope that the clear structure of our presentation of tests will enable you to perform statistical tests much faster and more directly, instead of searching through documentation or looking on the World Wide Web. Hence, the book may serve as a reference work for the beginner as well as someone with more advanced knowledge or even a specialist.

The book is organized as follows. In the first part we give a short introduction to the theory of statistical hypothesis testing and describe the programming philosophy of SAS and R. This part also contains an example of how to perform statistical tests in both programming languages and of the way tests are presented throughout the book. The second part deals with tests on normally distributed data and includes well-known tests on the mean and the variance for one and two sample problems. Part three explains tests on proportions as parameters of binomial distributions while the fourth part deals with tests on parameters of Poisson and exponential distributions. The fifth part shows how to conduct tests related to the Pearson's, Spearman's and partial correlation coefficients. With Part six we change to nonparametric tests, which include tests on location and scale differences. Goodness-of-fit tests are handled in Part seven and include tests on normality and tests on other distributions. Part eight deals with tests to assess randomness. Fisher's exact test and further tests on contingency tables are covered in Part nine, followed by tests on outliers in Part ten. The book finished with tests in regression analysis. We provide the used datasets in the appendices together with some tables on critical values of the most common test distributions and a glossary.

Due to the numerous statistical tests available we naturally can only present a selection of them. We hope that our choice meets your needs. However, if you miss some particular tests please send us an e-mail at: [email protected]. We will try to publish these missing tests on our book homepage. No book is free of errors and typos. We hope that the errors follow a Poisson distribution, that is, the error rate is low. In the event that you find an error please send us an e-mail. We will publish corrections on the accompanying website (http:\www.d-taeger.de).

Last but not least we would like to thank Wiley for publishing our book and especially Richard Davies from Wiley for his support and patience. We hope you will not reject the null hypothesis that this book is useful for you.

Dirk Taeger
Sonja Kuhnt
Dortmund