The local context and the spatial diffusion of multiparty competition in Urban Mexico, 1994–2000

Some researchers believe that voting behavior is spatially clustered and varies across places because of the different local political and economic conditions where voters reside (local contextual effects). Others have also argued that the spread of support for political parties is affected by geographical proximity of one place to another (contagious diffusion). This study tests these two propositions. This study analyzes the results of the 1994 and 2000 federal representative elections in the 89 largest cities in Mexico. Aggregate data about voting behavior were analyzed, using Moran spatial autocorrelation coefficients and a spatial lag regression model with diagnostics for spatial effects. Dependent variables were the percentage of the vote for the Partido Revolucionario Institucional (PRI), Partido Accion Nacional (PAN), and Partido de la Revolucion Democratica (PRD). To test for local contextual effects, regional dummy variables were used in combination with socioeconomic covariates of voting behavior. To test for contagious diffusion, this study used a temporal–spatial lag. Moran coefficients show a pattern of spatial clustering of votes for the PRI and dispersion for the PAN and the PRD. Results from the regression analysis show evidence of significant local contextual effects operating in several regions of the country. Results also show the temporal–spatial lag variable to be non-significant for all political parties. This finding suggests that the spread of the PAN and the PRD and the retrenchment of the PRI in the period of study were independent of physical proximity to areas of previous electoral support. In summary, the findings in this study suggest (1) the acceptance of the local contextual effects proposition for explaining spatial clustering in voting behavior and (2) the rejection of the contagious diffusion proposition.