The Attribution of Variance in Electoral Returns: An Alternative Measurement Technique

Swings in district vote for Congress are conditioned by many factors. An attempt is made here to apportion the variance in the partisan distribution of votes for U. S. representative among three levels of influence—national, state, and district—measuring the degree of nationalization, regionalization, and localization of voting. Previous attempts have defined “nationalization†of voting as the degree to which district interelection differences are numerically identical. Here this concept is defined in a two-stage regression model as the degree to which districts behave as if these differences were caused by the same factors. In contrast to previous research, national factors are found to be responsible for more than 50 per cent of the variance in local vote, with state and district forces accounting for 19 per cent and 26 per cent, respectively. Several analytic uses for this measure are suggested and illustrated.