Authors: Robert Gooljar*, University of North Carolina - Charlotte, William Graves, University of North Carolina - Charlotte
Topics: Business Geography, Location Theory, Quantitative Methods
Keywords: Location analysis, data bias, MAPE, urban development, site selection
Session Type: Paper
Start / End Time: 11:10 AM / 12:25 PM
Room: Plaza Court 6, Sheraton, Concourse Level
Presentation File: No File Uploaded
Graves and Gerney (2018) found significant underestimation bias in current year, vendor-provided demographic data estimates (relative to the American Community Survey (ACS)). Questions remained about the persistence of these estimation biases and their impact on the urban retail landscape. This paper updates Graves and Gerney (2018) by examining the change in estimation errors in the population and median household income data provided by five vendors (Experian, Synergos, Scan/US, ESRI and EASI) between 2015 and 2017. The dynamic aspects of these errors are explored in 80 Census tracts in the 40 fastest-growing US metropolitan areas. Half of the tracts are stable, suburban neighborhoods, half revitalizing urban neighborhoods. The change in mean absolute percent error (MAPE) (benchmarked by the ACS) is evaluated across the sample for population and median household income in each tract. Categories of tracts with decreasing and increasing error are identified, informing a discussion of the source of this estimation bias. The role of persistent estimation bias on the under-provision of retail in urban settings will be discussed. Finally, we will touch on strategies for minimizing the impacts of these biases in the location analysis process.