Applied Bayesian Modeling and Causal Inference from Incomplete Data Perspectives
Isaac Dialsingh · Journal of the Royal Statistical Society Series A (Statistics in Society) · 2005
The authors’ intent is to show how complex statistical theory and methods can be applied to real world problems, when data of different types have missing values. The book is basically a showcase for Donald Rubin's methods, and the input that he has made in the lives of his students that enables them to follow his footsteps. It is divided into four parts and has a well-equipped index. The main areas covered are causal inference and observational studies, missing data modelling, statistical modelling with computation and applied Bayesian inference. Methods that are used include propensity measures, instrumental variables and Bayesian inference. These make it an attractive reference book for those who are interested in pursuing data analyses through some of the more up-to-date methods available in an easily accessible format, via the use of real world data selected from a variety of fields ranging from medi-cine to census surveys. The book is edited by two of the best-known statisticians in the field. Its concise reference section and the ease with which each author describes the data set and the problem, and then gently guides the reader into the analysis, are its strongest points. This is really appealing and will be useful to applied statisticians and practitioners, as well as to academics who teach courses related to multiple imputation at an advanced undergraduate and post-graduate level. However, those who do not have the mathematical background should consider going through a good multivariate statistics text like Applied Multivariate Statistical Analysis by Johnson and Wichern before even attempting to read it. I strongly recommend that libraries have a copy of this book in their reference section.