Cartographic techniques for communicating class separability: enhanced choropleth maps of median household income, Iowa

In mapping population characteristics, data are usually portrayed as accurate without error. However, many population datasets provide estimates derived from surveys or samples, and a certain level of uncertainty is associated with each estimate. Ignoring estimated uncertainty information in mapping may produce misleading maps and generate spurious spatial patterns. In this paper, we introduce a measure of separability to indicate the likelihood that units assigned to different classes are truly different statistically. A series of map symbolization techniques is proposed to communicate class separability to the cartographer or map reader, and presented in four series of maps of American Community Survey data on median household income for Iowa counties. These map series illustrate several different techniques: a legend designed to communicate separability between classes, graduated line symbols to communicate separability between individual map units, and a color scheme in which perceptual color differences are related to class separability. Each map series presents three alternative classifications to illustrating how the proposed symbolization techniques could assist a cartographer in choosing the more preferable classification scheme.