Wellyopolis

March 30, 2006

Cool maps

I've always been a little skeptical of the historical geography fascination with maps, as they seem to involve a lot of effort to generate description rather than analysis. Maps summarize averages well, but I've yet to see a good map that shows variation within the geographic units. No doubt it's possible with a little work ...

That said, people love pictures, especially people at conferences, so it seemed to be worth learning how to generate cool maps myself. Apparently there are ways to do maps in Excel, but why use Excel when you can use Stata? Here is the kind of stuff you can produce with the tmap package in Stata.

Black wives in the cotton south worked outside the home more than black wives elsewhere, and white wives anywhere ... something I knew already. But interesting to see in that form nevertheless. I could see the same thing on a table, where I could sort the columns and see more precisely the rate of labor force participation in various areas. But it would be hard on a table—even one where I grouped the states in geographic areas—to see the cotton south like you can see it here.

Anyone who's ever taught American history and done "the map exercise" (getting students to identify the states) knows that the general level of awareness of where the states are is pretty low. You can bemoan that all you like. I know I did at the time, "I'm a freaking foreigner and I know where the states are!!" kind of thinking, but really ... that is not going to get you far. Why should people know exactly where all fifty states are. I'm not sure that that is useful knowledge in and of itself. It's possible that as part of a good liberal education people will end up knowing where all fifty states are, but maybe they won't. It's also quite likely that the ability to visualize something that specific and commit it to memory is not common.

So while I don't see a lot of analytic value in maps communication and presentation is always a part of academic work, and often a very important part. It's easy to begrudge the extra work in making your analysis understandable to an audience, but that is to begrudge part of the job. Good graphs and maps are easier to understand quickly. A good graph or map can convey dramatic differences more clearly. Tables are more precise, to be sure, but will they grab your attention? Not as quickly as the same dramatic trend on a graph. Or a startling geographic trend on a map.

Posted by robe0419 at March 30, 2006 3:59 PM