Statistics for experimenters - George E. P. Box

This fresh approach to statistics focuses on applications in the physical, engineering, biological, and social sciences. From the beginning it source of ideas is the scientific method itself and the need of the investigator to make his research as effective as possible through proper choice and conduct of experiments and appropriate analysis of data. While the book is unusual in the attention given to the scientific philosophy, it requires only elementary mathematics.

Material is presented with the non–statistician in mind. After a problem is stated, appropriate statistical methods of design and analysis are discussed. And frequently, examples are presented for which standard mathematical assumptions are wrong, thus forcing the reader's attention onto the essential precautions necessary in the conduct of the experiment to ensure valid conclusions.

In addition to many worked examples in the text, there are frequent exercises (with answers) throughout the book. There are also questions at the end of each chapter (for purposes of preview and review) and numerous problems at the end of each part of the book.

Box, Hunter, and Hunter–all statistical practitioners themselves–provide scientists and engineers with a lucid introduction to the basic statistical methods needed for research. In the last part of the book, advanced topics, selected for their special interest to scientific investigators, are introduced in readily understood language. These include time series analysis, response surface methods, regression analysis, study of error transmission, and mechanistic model–building.