Abstract:
The United States decennial census gathers population information for the entire country at the block, block group, tract, county, and state levels. The population survey data are accordingly aggregated (or summed) to areal units of these levels. This causes problems because (1) individual persons and households are more "real" objects than artificially defined areal units for the representation of population but the spatial distribution of these real objects is lost in the process of aggregation, and (2) the actual population may not be distributed evenly within census polygons. In GIS representations and analyses, the ecological fallacy and modifiable area unit problems may ensue when such data are used. In this research, we used household (building) location data to map population distribution. We also disaggregated census population into locations where households are located and re-defined population boundaries. This was done by counting households (buildings) of local, small areas in GIS to identify population clusters and then expand the size of the local areas to delineate and classify the clusters at multiple spatial scales. A new population classification map was produced to better represent urban and rural populations. The study area is in McDonough County, Illinois. Household location data was based on digital othrophoto quarter quadrangles.
