A.M. Leonowicz

Polish Academy of Sciences, Institute of Geography and Spatial Organization, Warsaw, Poland


Looking for the relationship between geographical phenomena is one of the possible ways to explore spatial data; it is often used in geography as a part of an investigation process. Visualization considerably facilitates that process. Choropleth maps are one of the most popular cartographic visualization tools. However, the most frequently used one-variable choropleth technique is not a perfect method to visualize spatial relationships. It allows depicting only single geographic feature on a separate map. Comparing visually such maps to discover similarities or differences between the distributions was found in many experiments to be a demanding task for the map readers. It can be assumed that information about geographical relationship would be easier to notice if two distributions were combined on a one single map. Such maps (called two-variable choropleth maps) are well known in cartographic literature. They are however not often used in practice, probably because they were thought to be controversial. This opinion was based on few experiments which results evidenced that this mapping method is too difficult for the readers to understand. The aim of my research was to verify this opinion. It was assumed that two-variable choropleth maps can be well understood by their readers, if only the maps are properly designed (by the use of proper graphic and reduction of the number of classes). The experimental test was conducted in order to compare the effectiveness of one-variable and two-variable choropleth maps. Students participating in the test were asked to perform different, typical map-reading tasks, like extracting specific and general information from maps. They were asked questions about the distributions as well as about variables relationship, and after that about their opinions and preferences about maps. The results verify opinion about poor readability of two-variable choropleth maps. Students were more accurate in reading spatial distribution on one-variable maps (especially the general pattern) and in reading spatial relationship on two-variable maps. They found two-variable maps a little bit more difficult to interpret but this form of presentation seemed more unusual and interesting. They found also that one-variable solution is more appropriate to read distribution and two-variable solution – to read variables relationship.