If therefore, even one or two of these few instances be insufficiently known . . . and therefore not adequately compared with other instances, nothing is more probable than that a wrong empirical law will emerge.
(John Stuart Mill)
This volume is concerned with the comparative analysis of a particular type of long-term historical process. As such, it has drawn on a diversity of theoretical and empirical sources, and has included some extended discussions of key methodological issues (such as the meaning of critical concepts, the role of metaphors in theory building, and the merits of an 'interpretavist' approach to theorizing about historical processes). This chapter aims to provide a more focused discussion of one methodological issue that is particularly central to the approach of this book, namely how to undertake the comparative side of the analysis.
The standard approach to comparative work may be labelled the 'reporting unit' approach. Here the procedure is to identify the universe of units included in the study, and then to assemble standardized objective indicators for each unit. This permits such statistical procedures as sampling the universe, measuring distributions of relevant characteristics within it, and investigating correlations between the indicators, calculating probabilities of outcomes. The 120 competitive electoral regimes listed in Annex Table 2 can clearly be analysed in this way, and a substantial amount of academic energy has been devoted to such efforts, sometimes yielding revealing results. 1 Comparison requires a minimum of shared