|Also available as an Acrobat File|
Maps of the Census: a rough guide
In the scientific visualization community the motivations are primarily technical, related to the types of graphic that technology allows us to generate. Only seldom do users consider the fundamental design aspects of their work and then usually this involves a passing mention of the well known works of Tufte. The result is often a graphic of either terrifying complexity or else one that is a poor representation of the data it is supposed to reveal. Yet almost all of the graphics produced would benefit from some attention to the various principles of map design that have been assembled by map makers (cartographers).
In this Case Study, Dykes and Unwin take what at first sight seems to be a simple graphic, the so-called choropleth map for area aggregated data, and show how different practive in the data to be visualized and symbolism used can lead to totally different maps and, hence, conclusions about the underlying geography. One solution would be to counsel perfection, that there is only one monosemic graphic for any one data set. The other is the approach they adopt, which is to use a series of graphic devices, most notably multiple linked representations and the ability to brush subsets, to present a multiplicity of views of the same data. A critical notion here is that of map stability, it being argued that any patterns that seem to be the same, irrespective of the display techniques used are almost certainly real effects. This seems to us to be an idea that could equally well be used in any visualization work.
Graphics Multimedia Virtual Environments Visualisation Contents