Nonparametric analysis of longitudinal data in factorial experiments (Q2769921)

Longitudinal data (data collected from the same individuals over time) arise in numerous diverse fields such as biological sciences, clinical trials, agriculture, forestry, sociology, psychology, and ecology. Data collected from different subjects (individuals, experimental units) are generally assumed to be independent whereas those from the same experimental unit have dependence to a certain degree. Thus the techniques dealing with the analysis and interpretation of longitudinal data are much more difficult and complex. This book contains several such procedures for the analysis of data collected under different conditions. The authors restrict themselves for the collection of the data to factorial designs -- dividing the subjects into several groups and subjecting the groups and different times of taking observations to a factorial structure. The goal here is to investigate models and develop techniques which require very few assumptions for applications. The methods presented are robust with regard to outliers, applicable to arbitrary data types, and the results obtained using these methods are invariant under arbitrary monotone transformations of the data. Furthermore, these techniques permit the analysis of designs with at random missing data, are approximated for small sample sizes, and permit the absence of variability in some trial groups.NEWLINENEWLINENEWLINEThere are 11 chapters in all, of which Chapter 1 includes all the examples, illustrating the ideas concerning longitudinal data to be used for analysis in subsequent chapters. After presenting nonparametric models in Chapter 2, the next chapter is on the formulation of nonparametric hypotheses and their interpretation, describing the influence of time on the measurements, and defining the nonparametric treatments in these models. Chapter 4 deals with point and interval estimators for the nonparametric treatment effects computed from the midranks of the observations. The general form of various test statistics for nonparametric hypotheses and their asymptotic distributions are presented in Chapter 5.NEWLINENEWLINENEWLINEThe next chapter is an overview of macros written in SAS-IML, separate macros for each experimental design, and naming these macros according to the models given in Chapter 2. Statements needed to perform the computations are also given. The next two chapters include designs involving a single homogeneous group of experimental units, and designs involving several groups of objects. Chapter 9 deals with dependent and repeated measurements, while Chapter 10 is set aside for techniques dealing with higher factorial designs. The final chapter is concerned with suggesting simple solutions for problems of several time points. Other features of the book are: a list of references, an index set, a glossary of numerous symbols used in the book, an appendix A (subdivided into A1-A15) containing original data sets, an appendix B containing results from the SAS macros (web site to download: www.wiley.com/statistics), and problem sets at the end of Chapters 4, 7, 8, 9, 10, and 11.NEWLINENEWLINENEWLINEIt is an interesting and a very book on nonparametric techniques for the analysis of longitudinal data (could be count data, metric data, ordered categorical data, or dichotomous data) in factorial designs. The subject matter is carefully planned, well-organized, and presented with great clarity. It is heavily oriented towards applications, and provides a unified approach for analyzing longitudinal data. For those interested in the theoretical aspects of the techniques, the authors present the basic mathematical principles for the techniques at a level which is consistent with the modest prerequisite required by the authors. A salient feature of the book is use of numerous examples to introduce basic and fundamental concepts, to construct models, to evaluate, to interpret, and to represent the results and observed effects. An excellent book on a topic of great importance, and researchers from different disciplines will find it very valuable for analyzing longitudinal data from their fields of study.