A spatial analysis of county-level outcomes in US Presidential elections: 1988–2000

This paper identifies spatial patterns of county-level presidential election outcomes from 1988 to 2000, and tests the retrospective (reward–punishment) and issue–priority models of voting behavior within the context of county-level geographical clusters. Based on our spatial analyses, we find that: the geographical concentration of the partisan vote has increased at both the global and regional scales. Globally, counties have become more likely to be clustered with similar counties in terms of their partisan support. Regionally, Democrats have increasingly received more votes from the East and the urban areas than Republican candidates while the opposite is true in the West and the rural areas. The regression analyses also support aspects of the issue–priority model of voting behavior, while the retrospective theory is confirmed only for 1996.