This first edition seeks to aid applied practitioners or analysts to explore wider classes of models for their data. It explains the processes of data examination, model selection and diagnostics in an accessible style, but which includes dense sections with lots of reliable advice with some brief technical explanations. The methods are illustrated on a small number of largish datasets, demonstrating progressive improvements chapter-by-chapter by repeatedly modelling the same datasets (although without typical textbook chapter-end example problems). The output and B&W graphics are based on R whose code is provided (with some minor typos). As of writing, the parallel SAS code was yet to be made available. This is neither a study textbook nor a technical manual, but rather a distillation of model-building vignettes with an American and astronomy slant in places. Even as an experienced analyst, I found the elucidated analysis considerations and the range of diagnostic approaches and interpretations extremely useful, while learning of a number of more current diagnostic statistics and plots (e.g. logistic histogram, QICu for GEE). While avoiding too much theory, this slim 236 page book packs in a lot of information on fitting a wide range of models. However this consequently means that some terms are used but not explained (e.g. loess, Wald); the index could be more detailed; certain model types are ignored (e.g. Bayesian, GAM); and the analysis examples are not comprehensive but instead suddenly lead to the “most appropriate” model (e.g. a ZIPIG in Chapter 9). One key weakness is not showing all the model-building process: this is referenced rather than explicitly shown. The authors also seem to avoid defining their threshold for significance. Their choice of specialised examples also means they focus on the problems of large datasets (i.e. overly conservative tests) rather than problems of small datasets (i.e. over-fitting). Arguably they imply simpler models are not useful, but most of the modelling issues addressed in these example datasets are clearly and expertly examined. This book is excellent for i) helping match model types to data, ii) showing the most typically useful model classes, iii) explaining which diagnostics are relevant, and iv) linking to references. It easily succeeds in its ambition as an introductory book for such models, encouraging users to adopt more advanced approaches through a few, key, worked examples, whilst tantalisingly acting as a gateway to more comprehensive references.