Cognitive Assessment of Geospatial Survey Data
Roe, D., Currivan, D. B., Ghneim, G. S., & Holt, J. (2006, May). Cognitive Assessment of Geospatial Survey Data. Presented at American Association for Public Opinion Research 61st Annual Conference, Montréal, Canada.
Mapping survey data has become an increasingly important means of displaying survey information and highlighting differences across geographic areas. In public health, geospatial data are used by policy-makers, researchers, and other health professionals to identify and address issues of public health concern. Traditionally, geospatial data are presented in choropleth maps, where defined geographic units (i.e., political boundaries) are filled with a uniform color or pattern. Such maps are appropriate for data that have been scaled or normalized. For example, high school graduation rates displayed by county are typically represented using choropleth maps. Alternatively, geospatial data can be displayed using isopleth maps, in which the data are not aggregated to pre-defined geographic units, but rather are typically smoothed across adjacent geographic boundaries. There is little empirical research, however, on the differential effectiveness of choropleth versus isopleth maps. In particular, how do these two different mapping techniques affect the users ability to extract information from the map? Using maps of health data collected from the Behavioral Risk Factor Surveillance System (BRFSS), a widely used source of public health information, we present the results from a series of focus groups and cognitive testing sessions which were used to evaluate the way in which public health professionals and epidemiologists commonly use mapped data. Among the issues explored are usability, general preferences, ease of pattern recognition, and rate retrieval for both choropleth and isopleth maps. A clear majority of participants preferred the isopleth format, however, their assessment varied somewhat by the data retrieval task. The color scheme used also had an impact, regardless of the type of map used. Implications for the assessment of geospatial presentation of survey data are also discussed.