The risks of using choropleth maps

Choropleth maps use existing spatial units (such as census blocks, cities, countries) to map statistical data. They are commonly used to map census data, which is where I was introduced to them in the 1980s.

One of the risks of using them is that a few geographically large units can swamp the visual communication. This is evident in an interesting visualization produced by J. Samson, D. Berteaux, B. J. McGill, and M. M. Humphries, researchers at McGill University, and reported by John Cook in the Huffington Post.  They produced the following sets of maps comparing the countries contributing the most to climate change and those likely to be most negatively affected by it.

The map shows clearly the major contibutors, although there could be an argument for using total emissions rather than per capita emissions.  Australia in particular has high per capita emissions but low total emissions because of its relatively small population, something not evident from the map.

But it is in the map of impacts that the limits of choropleth mapping are really evident.  Small island states which are particularly vulnerable to the changes wrought by rises in sea levels are literally invisible in this map.

Wikipedia has a good discussion of the risks of using choropleth maps.

2 comments to The risks of using choropleth maps

  • Agreed! I was at a training last week where we discussed the shortcomings of choropleth maps. So, how do we solve the problem? What would you suggest for reporting in this instance that would bring to light to disparate impact on small nations? What do you recommend as a solution (I love to learn from you!)? Thanks for shining a light on poor mapping.

  • Thanks for your comment, Susan, and I’d be very interested to learn what solutions were suggested at the session last week. Wikipedia suggests transforming variables (for example using people per unit of area) , which works for some variables. In the case of the Pacific Islands, I think some sort of infographic is needed, perhaps showing a graph of the percentage of the population likely to be seriously negatively impacted, (and another one showing the numbers of people – both are important stories) for each region or for specific countries, and use the world map simply as a spatial background for the graphs.
    The other change I’d suggest is to use a world map where the equator is actually in the middle of the map (yes, the southern hemisphere IS the same size as the northern, although you’d never know this from the maps that are commonly used) or maybe even a Peters projection map, which shows the land mass of countries accurately. (Check it out http://www.petersmap.com/ for a different view of the world)