Date of Publication


Document Type


Degree Name

Master of Science (M.S.) in Geography




Elections -- Oregon -- Multnomah County, Voting -- Oregon -- Multnomah County, Political geography



Physical Description

1 online resource (3, x, 123 p.)


A significant and challenging topic for contemporary geography is an understanding of how the social construction of space both reflects social forces and at the same time structures those forces. This thesis is a case study of how political and social attitudes, measured as an outcome of votes for different issues, reflect the social organization of space in Multnomah County, Oregon. It employs an arealstructural (ecological) approach. Using eleven different issues voted upon in the general elections of November, 1990, it analyzes relationships between ballot items and socioeconomic characteristics of the electorate for small geographic areas in the county. The 1990 election was selected to permit a minimum possible temporal gap in the comparative analysis data with returns from the 1990 census of population. Using a technique from the field of Geographic Information Systems (GIS), "block group" census divisions were "overlayed" with voting precinct boundaries. This permitted identification of clusters or proportions of block groups falling within each precinct. Factor analysis, correlation analysis, and cluster analysis were used to identify relationships among the measures themselves, to establish associations between the measures and socioeconomic data from the census, and to characterize spatial patterns of voting. The following conclusions emerge: (1) Factor analysis confirms that voting patterns for the eleven issues can be aggregated into two basic trends: "westside vs. eastside" and "inner city vs. periphery". (2) Cluster analysis shows that neighboring precincts have common voting patterns that create distinctive geographic regions. A new GIS method was developed to permit quantification of the geographic component of cluster analysis. (3) Correlation analysis of census and electoral data at the precinct level indicates high correlations of voting patterns with two socioeconomic dimensions: status (education, income, occupation) and position in the life cycle (age, marital status, family size). Such analysis was permitted by new GIS capabilities that allow the use of less aggregated block group data. ( 4) Cluster analysis of residuals shows a significant geographic patterning that suggests the existence of a "neighborhood effect" in Multnomah County, although confirmation requires further analysis.


If you are the rightful copyright holder of this dissertation or thesis and wish to have it removed from the Open Access Collection, please submit a request to and include clear identification of the work, preferably with URL

Persistent Identifier

Included in

Geography Commons