Notes on the Inconsistency of Subcounty Scale Demographic Data: A Comparison of Five Data Vendors

Abstract

Business geographers spend a large portion of their time developing new models to answer complex spatial questions. Although these models of economic systems are frequently innovative, they often fail to produce useful results (Colander et al. 2009). Inconsistent and inaccurate input data are a primary source of error in these models and the location analysis process as a whole (Rando 2014). The willingness of professional and academic geographers to take the accuracy of subcounty demographic data for granted exacerbates the problem of model inaccuracy. The sluggish update cycle of U.S. Census Bureau data forced professional geographers to rely on spatial demographic data provided by third-party vendors for model construction and application. Frequently these data are packaged with analytical software such as ESRI Business Analyst or Alteryx or accessed via an online subscription model. These vendors generate their small-area data by applying proprietary algorithms to base-year census data to extrapolate contemporary data (ESRI 2014). These data are expensive—the high price simultaneously serving as a signal of quality as well as a barrier to evaluative comparisons between vendors. Accuracy assessments are further impeded by the lack of significant metadata or historical data. The result of pricing, opaque sourcing, and the common genealogy of third-party demographic data is users must blindly accept whatever data source is purchased as the best available (Graves and Chabot 2015). The purpose of this article is to directly compare five sources of basic spatial demographic data to reveal the variation that exists among them. The goal of this evaluation is to illuminate sources of error in the location analysis process and discuss the problems that data inaccuracy poses for the discipline of applied geography.