Delimiting Residential Structure in Knoxville, Tennessee: A Geo-Demographic Approach

Geo-demographic database systems consist of standardized procedures to classify small geographic entities within cities according to the similarity of the demographic and socioeconomic characteristics of an entity's population. The theoretical development of these systems represents a new application of more traditional urban structure models. Geo-demographic systems employ the same factoring and clustering techniques used in past intraurban studies, but typically employ a wider range of variables. Some uses of these systems include target marketing, demographic studies, and sales territory definition. It is unlikely that users of these proprietary commercial geo-demographic systems have any knowledge of their theoretical development or potential limitations. In this research, we employ the geo-demographic database system developed by the United Way of America to classify entire cities according to their expected level of charitable philanthropy. Using similar data employed in the United Way's interurban applications, we develop an intraurban geo-demographic database system for the Knoxville, Tennessee SMSA at the census tract level. We also use the United Way's factor analytic technique and rotational schema in order to derive a solution which simultaneously explains a large percentage of the variance, minimizes the number of interpretable factors, and seems to make intuitive sense. The standardized factor loadings produced by this factor analytic procedure were classified using the cluster technique employed by the United Way. The results indicate that geo-demographic database systems do provide some general information about the spatial structure of Knoxville. The census tract level is, however, too large and diverse to produce the desired degree of residential lifestyle homogeneity. Subjectivity is introduced when using both factor analysis and cluster analysis. In the end, it is the researcher's responsibility to determine the number of relevant factors and clusters that best characterize the target population.