Making History Count: A Primer in Quantitative Methods for Historians (review)

This is the best textbook on statistical methods ever written for a historical audience. Realizing that computers can relieve the burden of most calculations, and that equations often frighten many historians whose work and comprehension of others' work would benefit from more than a passing acquaintance with statistics, Feinstein and Thomas require that their readers have only a general familiarity with high-school algebra, and they include no mathematical proofs. They offer clear, well- illustrated, and exceptionally well-presented intuitive explanations of the basic methods employed to analyze much historical data, particularly economic data, along with well-designed exercises using real or invented, but plausible, historical information. They return again and again to four case studies downloadable without charge from the Cambridge University Press website for further analysis and practice. Well-written and jargon-free, neither simplistic nor patronizing, offering acute advice on research design, as well as on statistical techniques, this work is equally appropriate for the classroom or self-study.

Authors of introductory statistics books that do not use either calculus or linear algebra often produce "cookbooks"—collections of formulas and facts about calculations, without much explanation of general concepts. Or they may spend so many pages trying to convince math-phobic students that statistics might be useful even to them that little space remains to discuss actual methods and the assumptions behind them. Sometimes such books concentrate entirely on a limited range of topics, such as cross-tabulation or linear regression, avoiding more general approaches. They may even key discussions to particular computer software, which might have already changed or disappeared by the publication date. Feinstein and Thomas avoid these difficulties, assuming that readers are actually interested in their subject, and that basic understanding and general references to statistical software will be more useful than calculating trivia. Thus, they introduce correlation and regression, hypothesis testing and elementary probability theory, cross-classification (the only simplistically treated topic in the book), and a considerable number of variations on the general linear model, including logit and probit analysis.

The discussions of each topic are sufficiently detailed and clear to launch analyses using the techniques, as well as to permit the understanding, and even critiquing of, many statistical historical studies. Readers will also have a basis for studying more advanced techniques, such as the application of loglinear models to cross-classified data, and more specialized topics, such as the use of regression and regression-like models to scrutinize aggregate voting data.

In the last two chapters, which concentrate on the authors' four historical data sets and will be especially useful to those who read this book on their own, Feinstein and Thomas proceed through the statistical analyses step by step, examining the assumptions and technical choices in each case, bringing research design, methodology, and a few details of calculation together to sum up the book's substance in concrete examples. They succeed superbly. Every historian who teaches a methodology course or who wants to begin learning about statistics central to the analysis of historical data should own this book.