Positional Errors in Small-Area Demographic Analysis: Estimating Population for an Oregon School District

Irina V. Sharkova, Portland State University
Kenneth Radin, Portland State University

Using a recent study to develop enrollment forecasts for the Medford School District (Oregon) as an example, this paper discusses common yet frequently overlooked sources of error in small area demographic analysis: positional errors. It identifies their sources and types and focuses on boundary mismatch errors arising from different spatial representations of the same study area. These errors are especially challenging when conflicting data are provided by trusted sources, such as GIS departments with the school district, city, or county. Using spatially referenced tax assessors’ inventories, fine-resolution GIS imagery, and expert judgment, we corrected positional errors in the data and created "true" boundaries for the study area. Next, we developed population estimates for the school district and its attendance areas from "true" and conflicting boundary configurations. The paper compares gains in spatial accuracy with improvements in estimates' accuracy and discusses the results in light of efforts required to achieve them.