Paavo Monkkonen, The University of Hong Kong
Planners and planning scholars frequently portray data about different neighborhoods of cities using a choropleth, a map that displays sections in different colors representing a range of values. However, when the population density is high in some neighborhoods, simple choropleths can be deceptive. Areas that are geographically large seem important as they occupy a large portion of the visual field, but they might not actually contain a large number of people or houses. In this common situation a mapmaker has few options to minimize misinterpretation.
The challenge is prominent in Hong Kong, a city where high-rise buildings cover about a fifth of the land area, with the remainder undeveloped. This yields an average population density in the urbanized areas of over 30,000 people per square kilometer! Thus, roughly half of the city’s population lives in the urban areas of Hong Kong Island and Kowloon, which are circled on the map the map located in Figure A, a choropleth that displays changes in housing prices in different neighborhoods in Hong Kong from 1992 to 1997. Although the main urban area of the city is important in terms of the number of houses, it is hard to see in a map that shows the entire territory. In contrast, the large areas to the north of the city dominate the map though much of them contain relatively few houses or people.
A similar challenge presented itself at a larger scale during the 2008 presidential election in the United States, when maps appeared to show a dominance of states voting for the Republican candidate, when in fact a the Democratic candidate had won. Mark Newman, a professor at the University of Michigan made a series of maps demonstrating the problem and a solution, a distorted map called an area cartogram. In an area cartogram, sub-units are resized according to their values on a different variable, often population. The cartographic technique was made famous by historian Arno Peter’s world map, in which countries boundaries were redrawn based on population; however it is quite uncommon in urban research. Yet it should be, given the vast difference in densities found in different parts of cities.
Area cartograms are now also quite straightforward to implement using add-ons to conventional GIS programs. Figure B represents the same data as that to Figure A, housing price changes in different neighborhoods of Hong Kong; however, the size of neighborhoods reflects the number of housing units rather than the actual geographic size. A circle is drawn around the same neighborhoods in both maps, demonstrating the importance of central urban neighborhoods. The visualization of the data on housing price change is now much more visually accurate. The new map makes it clear that many populous parts of the central urban area did not see large increases in price, whereas those geographically large areas in the north that did experience a price increase had fewer housing units than in the urban area.
One problem with the area cartogram, however, is that it can distort the original map to such an extent that locations become unrecognizable and spatial relationships between neighborhoods inaccurate. Nevertheless, for the purpose of discussing phenomena that affect people, they are a more honest representation of data, and should be used more frequently by urban planners and planning academics. Some discussion of cartograms can be found in the book How to Lie with Maps (Monmonier, 1996), and those seeking an in depth treatment of the topic should consult the publication Area Cartograms: Their Use and Creation (Dorling, 1996).
The author can be reached at firstname.lastname@example.org.