Simplifying complexity

Chapter 2 is devoted to applied numismatic data: whether from site finds or coin hoards, they consist of tables of counts of coins comparing the distribution of coins—grouped by date, mint, design etc.—by find-spot. As ever larger collections of data are created, such as that being gathered by the Oxford Coin Hoards of the Roman Empire Project, the task of teasing out the underlying patterns in that data becomes ever more difficult. Numismatists have, however, generally eschewed using statistical techniques designed to do just that in favour of home-grown, and generally graphically based, methods. This chapter explores the application of statistical methods in numismatics over the last 50 years and suggests reasons why this is the case. It then presents an analysis of a small, simple data set using Correspondence Analysis that shows how a mathematically complex technique is, in fact, simple to apply and interpret. Finally, a more complex data set is analysed to demonstrate the efficacy of the approach