John Lachin is Professor and Director of the graduate program in biostatistics at George Washington University. The book is intended as a first advanced course for students in that program. The book emphasizes methods for problems in biostatistics. To Lachin this means an emphasis on binary, categorical and survival data that relate to the assessment of risk and relative risk through clinical research. Consequently much of the standard parametric and nonparametric modeling of continuous response data is not considered.
A variety of methods are covered on a number of subjects. The first half of the book deals with classical approaches to single and multiple 2x2 contigency tables used in cross-sectional, prospective and case-control studies. In the second half, the more modern likelihood or model-based approach is presented. Technical mathematical details are covered in the appendix which is referenced throughout the text. The appendix deals with statistical theory (stochastic convergence results and other theory) but does not provide rigorous proofs of the theorems. Real probelms are presented and analyses are illustrated using procedures in SAS.

In the model-based sections, topics include logistic regression, Poisson regression, proportional hazard and multiplicative intensity models. The book is modern, well written, provides a good list of references, has extensive problem sets at the end of the chapters and employs case studies to illustrate the application of the methods. It is not a book for beginners. It is a great reference source for biostatisticians and epidemiologists as well as a fine text for a graduate-level course in biostatistics.

Читать «Biostatistical Methods: The Assessment of Relative Risks»

Купить электронную версию