See whose days are numbered

Making historians numerate, a necessary precondition for making history count, has always been more of a challenge this side of the Atlantic than in the US. Over there, a combination of differences in school mathematical education, undergraduate history degrees and, above all, the higher training content of US PhD programmes inculcates somewhat more positive attitudes towards quantification among aspirant professional historians. There have been attempts to address the failure of British historians to engage as equally with the quantitative as with the qualitative historical evidence. Nonetheless, I suspect that the stock of historians who can genuinely count is no higher now than it was a generation or so ago when the first of such volumes (Roderick Floud's An Introduction to Quantitative Methods for Historians ) appeared in 1973.

The social science tradition in British history, exemplified perhaps by the rise and institutional decline of economic history, is now but a minority pursuit. And it is not hard to see why an essentially literary-trained historical profession plumps for the delights of gender and postmodernism in preference to the hard graft of making history count.

Moreover, it should be remembered that these preferences have occurred despite the concurrent revolution in personal computing hardware and, above all, software, especially statistical and econometrics packages. The mainframe pioneers of the late 1950s and early 1960s who attempted a cliometric revolution had to write their own software; today's practitioners do not face that formidable barrier to entry.

Charles Feinstein and Mark Thomas are well aware that in Britain there are market failures that account for why the median historian does not count.

In their superb text, they set out a schema, based on teaching experience at the universities of Oxford and Virginia with undergraduates and postgraduates who possess no mathematical education beyond simple arithmetic, that introduces quantitative methods intuitively and through their applications rather on the basis of first learning statistical theory. Given that others have had such ambitions and such teaching experience, why should this text be any more successful?

First, the reputation of the authors as leading economic historians of Britain ensures an audience. Second, no competitor text is, to my knowledge, as effective in taking the student from the basics of descriptive statistics through to the intricacies of multiple linear regression. This is a testament to the quality of exposition - with, in particular, notation kept to the very minimum - and the importance of the historical questions taken as case studies.

Third, there are significant innovations of exposition here. For example, there is a brilliant use of Ballantine charts as an aid to understanding regression analysis, being a device for the visual representation of the respective variances and the extent of overlap between variables (being equal to r2 the coefficient of determination) and showing how, in multiple regression, the addition of further independent variables may increase/decrease the computed statistical significance and thus increase/lessen the explanation.

Fourth, although now somewhat less novel, in constructing the exposition around a number of case studies, and in making those datasets employed available on the web (http://uk.cambridge.org/resources/0521001374/), the authors greatly increase the attractiveness and utility of the text. There are four main datasets provided (in broadly temporal order) to explore statistically the characteristics of the unreformed pre-1834 English Poor Law; the scale and timing of post-famine emigration from Ireland; the influences on children's decision to leave home in mid-19th-century America; and Keynesian versus classical explanations of interwar British unemployment. Each dataset is provided in multiple formats (generic, Excel, SPSS, Stata and Minitab). The datasets are variously cross-sectional, time-series and panel, thereby illustrating the uses and limitations of all the data categories that confront the historian; and the care with which the exercises are set up, paced and developed, demonstrates that they are tried and tested in the classroom.

The authors hope that diligent students working their way through this text would be so equipped and encouraged to read, understand and evaluate much of the leading-edge literature that employs modern quantitative methods to support their analyses of historical questions. Only time will tell whether the absence of model answers hinders their objectives.

I wish this text had been available when I was trying to teach quantitative methods to numerically challenged historians in the 1980s-90s. As for those who feel more attuned to intertextuality than to say the X2 test, it remains to be seen whether any text can breach the historians' fortress, arithmophobia.

Roger Middleton is reader in the history of political economy, University of Bristol.