Visualizing Political Change

I am mulling over a post on the events in Tunisia, but I've been distracted. I may get around to it eventually, but right now I've been busy trying to make some animated maps for use in my "dictatorships and revolutions" class. Here's the first result:

I tried something like this a few months ago, but this is an improvement over the previous map. First, it uses the update to the Alvarez, Cheibub, Limongi, and Przeworski dataset of political regimes by Cheibub, Gandhi, and Vreeland (the "DD" dataset) that covers the entire period 1946-2008 (not just to 2002). And second, it uses a dataset of historical maps of state borders by Nils Weidmann, Doreen Kuse, and Kristian Skrede Gleditsch that allows you to visualize such things as the breakup of the Soviet Union or the reunification of Germany.

The DD definition of democracy is very minimalist: a country counts as a democracy if both the government and the legislature are elected in competitive elections. Thus, some countries are classified as democracies which seem to have all sorts of political problems. Moreover, by "competitive elections" CGV mean that a) the opposition can contest the election, and b) the government actually relinquishes power if the opposition wins. Since in some countries the current regime has never lost an election (e.g., Botswana), it is not always possible to unambiguously code the country as a democracy or a dictatorship given their coding rules. In such cases, they err on the side of classifying the country as a dictatorship (this is their "type II error" rule), which leads to some curious outcomes: for example, South Africa never turns into a democracy (look at the video at around the year 1994), and Botswana is always classified as a dictatorship. But they identify these cases with their "type II" variable, so it is possible to see which countries might be democracies but are classified as dictatorships: these are the "ambiguous" cases in the video.

One thing that comes out very clearly in the animation is that regime types tend to cluster temporally and spatially. There are waves of civilian dictatorships and of military dictatorships (see, for example, Africa in the late 1970s and 1980s), as well as of democracies. Most communist regimes clustered around the Soviet Union, and most absolute monarchies are in the Middle East/North Africa. There seem to be strong "regional" influences on regime change, which suggests that the events in Tunisia are unlikely to remain isolated.

Some people criticize the DD data for conceptualizing the democracy/dictatorship distinction as a categorical rather than a gradual distinction (most of my students, for example, really dislike this categorical distinction when I assign a reading from Gandhi in my class). So most political regime datasets (like Freedom House or Polity IV) have some kind of scale from most autocratic to most democratic. The choice is, to some, extent, pragmatic, but I think there is something to the idea that regimes come in types, not just gradations of a single underlying dimension. So I like CGV's effort to identify different regime types, and I am largely in agreement with many of their criticisms of "graduated" indexes of democracy like Polity IV here. Nevertheless, I've also made a similar animation using Polity IV data:

There is less to note here, except the march of democracy. You miss some of the geographic and temporal patterns visible in the DD data.

I'm thinking of making an animated map that shows coups as they happen (using the Coups d'Etat dataset by Marshall and Marshall) now that I've mastered the process of making these maps (it took a while: R and ArcGIS are not the most easy to use pieces of software). Other ideas?

Update, 1/21/2011: I went ahead and did the animated map showing coups d'etat.