Measurement Error and Misclassification in Statistics and Epidemiology: Impact and Bayesian Adjustments

of the mechanics of actual data analysis. I would describe it as a fairly highlevel presentation, long on ideas but short on details, and I doubt the authors would disagree. This is certainly not meant to be a negative characterization, but simply a statement of the book’s overall flavor. By the authors’ own account, this book serves two audiences: (1) advanced undergraduates and graduate students studying transportation issues in academic programs, and (2) researchers and practitioners who need access to information about a broad range of analytical tools and their applications to transportation problems. Wearing my professorial hat, I would say that the second audience is better served than the first. Because no problems or exercises are included and there is very little in the way of detailed worked examples, I would personally find it difficult to use this book as the primary text for a course, simply because the teaching resources are not there. Nonetheless, it would be an excellent academic reference for students and practitioners alike, although individuals engaged in the day-to-day activities of the transportation industry might find it a bit overwhelming. Overall, I would say that the authors have successfully achieved their objective and that the book is fairly well targeted toward the intended audience. The book is divided into three parts. Part I includes two chapters on fundamentals, which encompass elementary descriptive and inferential statistical methods. Embedded in Chapter 1, and sandwiched between the calculation of the correlation coefficient and construction of histograms, is a strange, but short, section on the properties of statistical estimators that one might more typically find in an intermediate text on mathematical statistics (e.g., I seriously doubt the Rao–Cramér lower bound qualifies as an elementary descriptive statistical method!). Otherwise this is a fairly traditional presentation of material contained in a first course on business statistics. Part II includes seven chapters about techniques used to analyze and model continuous response variables, including linear, multiple, and logistic regression; simultaneous equation models; panel data analysis; time series analysis; models for latent variables; and hazard/duration/growth models. Among these topics, regression analysis is presented in the most detail, with the others discussed at a fairly high level but with generally good thematic and conceptual coverage. Part III contains three chapters on techniques used to analyze and model discrete response variables, including ideas about Poisson and negative binomial regression, binary and multinomial probit models, multinomial logit models, and instrumental variables. There are four appendixes encompassing additional information about matrix algebra, special probability distributions, and data transformations, along with copies of the usual tables found in most statistics texts and an excellent glossary. The bibliography is up to date, with a good mix of newer and classical references. Contentwise, there is really nothing new here. The authors’ major contribution is their effective recasting of fairly well-known ideas and notions from statistics and econometrics onto the transportation landscape. Most users, particularly those in academic circles (including students), will find the material familiar and fairly easy to read. The mathematical level is relatively low, although full comprehension does require some previous exposure to matrix algebra, differential equations, and calculus. There are also more econometric terms in this book than ordinarily would be found in books written from a purely statistical perspective, and the selection of topics and associated discussion has a decided econometric flavor (e.g., referring to errors or residuals in a regression model as disturbances). The one significant omission is a discussion of the computational aspects and requirements of many of the techniques that are presented. Like all books, this one is not perfect. It contains the usual editorial gaffes of some misspelled words, inconsistent notation, and nomenclature that is not completely explained. In addition, because we already have such difficulty communicating in the statistics profession, I want to carp a bit about the imprecision with which some statistical terms and concepts are presented (e.g., inconsistent and incorrect interchanging of the words “parameter” and “statistic,” a somewhat sloppy definition and explanation of a confidence interval, attribution of descriptive techniques to statistical inference, and a somewhat nonstandard explanation of how to interpret R2 as a measure of goodness of fit). Perhaps this is just my old-school training showing through! My only other complaint is that, despite the several good transportation examples, I would like to have seen even more; and as noted earlier, I would have preferred some detailed examples worked out step-by-step. I suspect the authors’ response would be that the book would have been much longer than it already is, and that it would have taken much more time to complete. Of course, they would be correct. Despite some of the pickiness of my review, Statistical and Econometric Methods for Transportation Data Analysis really is a pretty good book. I would highly recommend it to anyone engaged in transportation research. I suspect it will be the definitive text on statistics in transportation for some years to come. Simon Washington, Matt Karlaftis, and Fred Mannering are to be commended for their tireless efforts of spreading the gospel of statistics in transportation and for the quality of the product that this particular project has produced. I am pleased to have had the opportunity to read the book and look forward to using it in my work in the future.