ABSTRACT
The hypotheses outlined in the previous chapter suggest tests using the conventional regression framework, with the dependent variables referring to aspects of the geography of voting. However, most of the hypotheses suggest relationships that cannot be identified using the aggregate data available from the census plus the voting returns. For example hypothesis 1 suggests spatial variations in the percentage of members of a particular class group voting Conservative, but data with which to investigate such variations are not available. How, then, can the hypotheses be tested?
