This is an introductory book. It also fits in introductory level of Mathematical Statistics. The prerequisites are introductory calculus and linear algebra. Most theorems are proved in calculus style but there are some gIt can be shownsh that are not proved. So some readers may not be satisfied with the book, especially Math majors.
Logical steps are shown in detail; else logical gaps are contained within a level such that a first time reader can fill in the gap with a pencil and paper. Occasional mix with Bayesian perspective is also a feature. Answers to odd-numbered exercises are provided except ones that ask derivations and proofs. Exercises that require some tricks are provided with hints. In these respects, this textbook is suitable for self-study.
Upon completion of the entire material, I feel concepts are developed well up to Hypothesis testing Chapter 8 where the presentation of material reaches climax and its level of exposition is somewhat higher than other chapters. Thereafter, simple linear regression is treated in detail, but coverage and detail of materials seem to deteriorate from the following general regression section, nonparametrics and thereafter. Kolmogorov-Smirnov Tests section is treated nicely though. Anova section lacks in coverage. The new simulation chapter is presented more like a demonstration rather than an introduction.
I have never seen the previous 2nd edition (unfortunately Dr. Degroot is no longer with us), but according to the preface of this 3rd edition, Dr. Schervish describes 8 major changes from the previous edition. Notable are some material removed from the previous (likelihood principle, Gauss-Markov theorem, and stepwise regression), some added (lognormal distribution, quantiles, prediction and prediction intervals, improper priors, Bayes test, power functions, M-estimators, residual plots in linear models and Bayesian analysis of simple linear regression), more exercises and examples, special notes, introduction and summary to each section, and so on. I find the last in the list is somewhat disturbing, especially introduction parts that are often redundant with the very next paragraph. On the other hand, I find that special notes provide good insights.
I wish they included introduction to Statistical Decision theory, full coverage of regression analysis to be usable such as diagnosis, transformation and variable selection, coverage of Multivariate Normal distribution, more coverage and depth in nonparametrics and simulation, and lists of recommended readings for further study at the end of each section with comments.
There are a noticeable number of typos as of this first printing I have. I sent suggestions for typos and was impressed that Dr. Schervish updated errata list within a few days at his homepage. I wish all authors were like him being responsible.