Cartograms are evil…

I recently engaged in a pretty heated debate with a senior researcher at my company about the merits and drawbacks of various data visualization techniques.  This particular person used mapping as a great example of visualization techniques, unfortunately started to focus his attention on cartograms (insert image here) as the best example of data visualization out there.  I nearly choked…  There is a fascination with cartograms that is bordering on fanaticism.  As we continued our conversation, I tried to articulate the pros/cons of cartograms (mostly cons).

What are cartograms?

The problems with cartograms (or why they are evil).

There are three problems with cartograms that make them evil and (IMHO) useless and dangerous as a means of communication.

1. Assumption of geographic knowledge:  The most common cartograms deform the underlying geography so that the areas end up representing the proportions in the numbers.  For example, you could resize all states to represent electoral votes they have in the electoral college, and create a new map of the US where Alaska is tiny, California huge, the East coast fairly dominant, and the middle of the country fairly small.  These areas in the middle of the country have a lot of land, but not a lot of people, and thus have (generally) fewer seats.  This, as the example shows, is often used in showing national information in the US, but is also used to show global trends like the share of global GDP that every country has, etc…  The first problem is that most people don’t have a good sense of size and geography.  Do you know (at first reaction) how many times larger Alaska is then California?  If you did, do you get a sense of the impact of the electoral college is?  Most likely you’ll answer that you don’t know the relationship, but seeing California blow up like a balloon is a sign.  However, I’d argue that there this map does not give you anything more than a table that shows bar graphs for each state.
2. Projection issues & education: The second issue is that most cartograms are based on one form or another of cartographic project, and most people have been brought up on the Mercator project which renders areas closer to the pole much larger, and areas around the equator much smaller.  So our basis for comparison is often biased to start out with
3. Arbitrary Algorithms: I realize that this is an oxymoron, however, the algorithms prioritize relative scale, and physical proximity (i.e. New Hampshire, Vermont and Maine need to be side by side, and need to be above Massachusetts, which needs to be above Rhode Island and Connecticut.  As a result you got odd ballooning and squeezing effects which poorly reflect the reality of the underlying information.  For example, if you choose to represent something that is worth 100 units, you could choose to represent as a 10 by 10 square, or a line that .1 of a square wide, and 1000 units long.  This line could be barely perceptible at the same overall scale, and would then create the perception that the 10 by 10 could be larger.

Hope in Cartograms:

The biggest hope in cartograms, I believe, is with those that use a known shape (a 1 by 1 square, or circle of radius of 1), and scale them up.  You can still keep the alignment mostly right, but then ensure that you are a) free of project, b) don’t have to assume that people know geography too well, and c) can be sure that two geographies worth 100 units are equally represented.

So why are they used so frequently:

In short – sex appeal.  They are sexy, and make people pause for a second.

What do you think?

(link for more info on cartograms: )

http://www.ncgia.ucsb.edu/projects/Cartogram_Central/types.html