Comparative Cross-Regional Analysis

The purpose of this chapter is twofold. In the first half of the chapter, I introduce the method of analysis that I use to assess the empirical support for the models and hypotheses presented in the previous chapter. Although no specific component of the analysis represents a new analytic technique or statistical tool, the manner in which they are combined here – and applied to the substantive questions at hand – is novel, and thus warrants a systematic explication of the method. In this vein, the second half of the chapter considers in greater detail the advantages and disadvantages of the method of analysis. The aims of the method are fourfold. First, it is intended to facilitate a comparative analysis of the effect of economic conditions on election results across multiple cases; in other words, it is explicitly designed for comparative study. Second, the method of comparison is intended to be as transparent and easily replicable as possible; faced with the same data and statistical results, others should be able to come to similar conclusions about the relative empirical support for different hypotheses, or at the very least understand exactly why I have drawn the conclusions that I did. Third, the method is explicitly designed to treat seriously issues of comparability of data, an especially important issue in new democracies. Finally, the method is intended to link the empirical tests of the data – and the conclusions drawn on the basis of these tests – as closely as possible to the actual hypotheses proposed in the previous chapter. The method of comparison involves the following components. Separate data sets of regional level data are utilized to conduct statistical analyses of the effect of cross-regional variation in economic conditions on crossregional variation in election results for each of the twenty elections in the dataset. Using these separate datasets, separate estimates for the effect of