Date of Award
Doctor of Philosophy (PhD)
The U.S. decennial census is an invaluable source to guide demographic analysis. It enumerates demographic characteristics within different levels of geography to protect privacy. Small statistical units such as census tracts and blocks in different points in time are indispensable to analyze regional and local trends of demographic characteristics. However, the linkage between census demography and those geographies mandates that their boundaries change from one census year to another to reflect underlying population changes. This inconsistency complicates studies of micro-scale nuanced demographic processes. Previous research efforts have aggregated inconsistent census geographies to larger comparable units or used areal interpolation to transfer demographic attributes from geographies of one census year (source zones) to geographies of another (target zones). The former disrupts the required resolution for micro-scale analysis while the latter is susceptible to errors.
This dissertation contributes analytical solutions to the above-mentioned persistent problem in enumerated data, typically used in demography, health sciences or economics. It combines spatial (dasymetric) refinement with areal interpolation methods to increase their accuracy in population estimation over time. This combination leads to a more precise allocation of population, which results in more reliable modeling for different configurations of target zones. The research conducts comprehensive analyses involving various ancillary variables, namely the National Land-Cover Database (NLCD), the Global Human Settlement Layer (GHSL), parcels, buildings and ZTRAX®, to transfer different demographic attributes, namely total population, population by race and age structure and urban population from census tracts in 1990 and 2000 within census tract boundaries in 2010 across different geographic scales (county/state) and under various demographic settings (urban/rural). This constructs demographic estimates within temporally consistent small units over 10- and 20-year periods.
The outcomes of the research affirm the effectiveness of combining spatial refinement with areal interpolation for accurate multi-temporal demographic analysis. The application domain of the methodological advancements of this dissertation includes demography, risk assessments, resource allocation planning, crime analysis and economics, to name a few.
Zoraghein, Hamidreza, "Creating Temporally Consistent Small Area Census Units Using Advanced Combinations of Areal Interpolation and Spatial Refinement: Method Development and Assessment" (2017). Geography Graduate Theses & Dissertations. 137.