I always enjoy the maps and other data features at the Texas Tribune, but sometimes they inspire me to try to improve them.
A case in point is the Tribune’s recent map of party “drift” or Democratic-Republican change between the 2008 and 2012 presidential elections.
The Tribune map strikingly shows that Obama got fewer votes in 2012 than in 2008 in most Texas counties, but it made me wonder about the significance of the change, both as a percentage of votes cast and in terms of the population of each county.
A couple of approaches to adding more information to the map come to mind. The most obvious would be to use more shades to show percentage change. Another would be to use a cartogram, or a map where political units are resized according to population, like the famous election results cartograms of Mark Newman.
What can we see in this map? First, it supports the Tribune’s thesis that most Texas counties had Republican gains in 2012, at least in terms of presidential counts. However, it calls into question how significant the change was, for both Republicans and Democrats. None of Texas’ most populous counties experienced over 3% change, and in the three largest the change was less than 1%. Only a few tiny counties have the bright red indicating Republican gains over 6%. On the other hand, the Democrats have no bright blue counties, and most of the pockets where the Tribune showed Democratic gains turned out to be either under 1% change or in counties so small that they can barely be seen. The only notable blue on the map is in the Valley, and the only county of a significant size that broke 3% growth for the Democrats was Webb County (Laredo).
My map has plenty of room for improvement: the convention of the red-blue political scale makes light blue hard to distinguish from neutral gray, and keeping the circles small enough that Harris County doesn’t swallow the whole Houston area turns most West Texas counties into barely perceptible dots. Using Excel rather than a proper GIS means the map is missing some cartographic niceties: the projection is distorted, and it lacks an outline of the state that would make it a lot easier to recognize as Texas. Finally, while my data ranges and scales seem reasonable to me, it would be better to base them on some measure of statistical significance.
Nevertheless it’s a start. If you’d like to see my data sources or check my work, here’s the spreadsheet: texas-drift-2008-2012.xlsx.