Regression methods in biostatistics. Linear, logistic, survival, and repeated measures models. (Q5916154)

This book is, as the title indicates, about regression methods, with examples and terminology from the biostatistics field. It should, however, also be useful for practitioners from other disciplines where regression methods can be applied. A prerequisite for the book is a first course in statistics, including simple linear regression, one-way ANOVA, and the chi-square test for independence in contingency tables. After a motivating introductory example in Chapter 1, the authors continue in Chapter 2 with a presentation of descriptive and graphical methods for different kinds of variables, e.g., the histogram, the boxplot, and the \(qq\)-plot, along with ordinary numerical summary statistics for numerical variables, tables for categorical variables, and scatterplots, correlation coefficients, and crosstabulation for the relation between two variables. In Chapter 3, ``Basic Statistical Methods'', the above mentioned prerequisites are reviewed, i.e., the \(t\)-test, analysis of variance, correlation, simple linear regression, methods for binary outcomes, survival analysis, and bootstrap confidence intervals. In Chapter 4, multipredictor methods are introduced in the context of the basic linear regression model. This chapter also covers the general topics confounding, mediation, interaction, and model checking. Chapter 5 deals with predictor selection, Chapter 6 with binary outcomes and logistic regression, and Chapter 7 with survival analysis. In Chapter 8, dependence is introduced in the form of repeated measurements, and its various forms are discussed in sections about hierarcical data, longitudinal data, generalized estimating equations, and random effects models. Chapter 9 returns to independence and introduces the generalized linear models. Finally, Chapter 10 introduces briefly the concept of analysis of complex surveys. The book concludes with a summary in Chapter 11, where the authors try to give practical advises about which of the methods covered in this book should be used in a particular situation, and also under what circumstances none is suitable. Most chapters end with a Problems section, and a section of further notes and references, making the book suitable as a text for a course on regression methods for Ph. D. students in medicine, but also in other appplied fields, e.g., in the social sciences. Many of the analyses in the book are illustrated with output from the statistical package Stata.