Birmingham-Southern College1, Computer Science, Birmingham, AL 35254 Samford2, Computer Science, Birmingham, AL 35229
Data which is made available is not necessarily made accessible. In this talk, we present a web application allowing non-experts to explore heterogeneous spatial, or georeferenced, data using an intensity mapping visualization of nation level statistics. A number of statistical visualization tools are available on the web, notably GapMinder’s interactive plotting tool [www.gapminder.org/]. Ours is similar to Geoclip’s interactive cartographic visualization [www.geoclip.fr/an/] but additionally provides a temporal component and simple data reduction functionality. Data reduction is performed through elimination and aggregation of areal units or rolling up values. The datasets available reflect a variety of classes of statistical measures (continuous, ordinal, categorical), and units (square kilometers of land, percentage of people with a certain attribute). Spatial presentation provides an intuitive interface and takes advantage of users’ recognition of geographical entities to relate visualized data to prior knowledge. It also inherently shows information about spatial variation and spatial autocorrelation, important aspects of the data which are often overlooked. This makes spatial data accessible to users with limited, if any, knowledge of conventional statistics, and no knowledge of spatial statistics. The user is able to aggregate areal units freely, increasing the user’s awareness of the modifiable areal unit problem despite lacking formal knowledge of spatial statistics. In this talk, we examine the system’s responsiveness and scalability along with perceptual effects in the interface relating to shape, size, and color. We also compare the context suggested by cartographic visualization with the benefits of higher dimensionality in interactive plotting. In general, we emphasize the utility of visualization and data reduction techniques to non-experts.