ABSTRACT
I now turn to changes in dissimilarity indices for PAN and for other parties between the initial conditions of 1964 and ending conditions of 2000 at the three electoral levels. The dissimilarity index can be regarded as a measure of the dispersion of vote shares about the overall vote for the party under study. It is described by Duncan (1957 p.30) as “the sum of the positive differences between the two percentage distributions.” Thus, I computed the dissimilarity index as the sum of positive differences between the percentage of votes for minority party members (PAN or other parties) and the percentage of PRI voters, at both initial and ending conditions. The dependent variable is the change between these indices from 1964 to 2000. A decrease implies more competitive party politics. An increase represents the increasing concentration of support in a state. As in chapter five, ten models are calibrated using the same independent variables. Because the dissimilarity index is an absolute value, the direction of change is not captured. It does not tell us which accounts for any increased concentration that may be discovered. For purpose of analysis I first explore changes in dissimilarity indices for PAN’s presidential elections. The comparison is repeated for the PAN senatorial and representative elections. The discussion then turns to dissimilarity indices for the set of other parties, where the sequence of analyses is repeated.
