Model Comparison and Model Validation Issues in Empirical Work on Urban Density Functions

摘要

Urban population density functions, as summary descriptions of urban structure, have been based on a variety of theoretical specifications and implemented empirically on a wide scale. In this paper, some issues related to the estimation of density functions and the validation of different specifications are considered me closely. Particular attention is given to the selection of spatial observational units, the use of net or gross density measures, testing for spatial autocorrelation and heteroskedasticity, and the use of different model validation criteria (tests on rum-nested hypotheses versus. measures of fit). An empirical illustration is provided with data for 1980 on the Columbus, Ohio, metropolitan area.