Edited by three well-known and respected statisticians, this book is another on exploratory data analysis (EDA), and is part of the prestigious Wiley Series on Probability and Mathematical Statistics. The contributors, in addition to the three editors, seem to be highly competent and write in a way that conveys a healthy exuberance for the work they are doing. The first chapter is by Diaconis and serves as an introductory and motivating piece. This chapter talks about problems with drawing correct conclusions from data, and it relies heavily on the work summarized in Nisbett and Ross (1980). Anyone interested in the psychological research on decisionmaking should refer to that book, although without a glossary of the jargon, it is difficult to read. The impression created is that this first chapter will build toward an advocacy of EDA and this is largely correct. Diaconis gives us seven remedies for the all too common "magical thinking" (his code words for irrational thinking). Five of them are complex mixtures of general methodological or statistical techniques such as cross-validation and bootstrapping. One remedy is negative in that it advocates not reporting p-values, apparently giving up on them being interpreted correctly. The seventh remedy is titled "remedies to come" and this is where EDA is classified. The focus of EDA is clarified when Diaconis says "Yet, none of the classical theories of statistics comes close to capturing what a real scientist does when exploring new data in a real scientific problem" (p. 22). The key is the focus on analysis of new data, but the implication is that the problem area is also new. In other words, EDA appears to be most useful when there is no theory (probabilistic or otherwise). More will be said about this later. In summary, this chapter, unlike many of the others, is a little unfocused, but intellectually very interesting and stimulating. It deals more with ideas than with algorithms.