What went into deciding the 10 regions?
By 
Thanks again to Michael Barone for linking to the 10 Regions of American Politics. But Barone did say that some of the regions "don’t make much sense to me." That’s understandable; I haven’t had the time or resources to post all the data that led me to draw the boundaries on the 10 regions. I will post more as we get closer to the general election, but I want to point out that the main factor was the voting history of each county in presidential elections going back to 1948. In particular, I looked at changes from one election to the next, as opposed to margins of victory by one party or another in individual elections. That is, I wanted to give a sense of where shifts in voting patterns led to shifts in party control of the White House — where John F. Kennedy got new votes in 1960, where the Republicans built an Electoral College advantage in the 1980s, and how we got to the red vs. blue stalemate of 2000.
The graphs below (drawn using the widely hated Powerpoint program) give an idea of how votes have shifted in each of the presidential elections beginning in 1948 (but skipping the landslides of 1952, 1956, 1964, and 1972, for which I haven’t had the stomach to tabulate all the data). They apply to Republican margins (or losses) vs. Democratic candidates in each case, since I needed to use a constant measure. Strong third-party candidates in 1968, 1980, and 1992 obviously skewed things, but not enough to change what is, at heart, a two-party system of presidential politics.
In the case of Mega-Chicago vs. Chippewa (which Barone singles out), the former region stayed pretty close to the national pattern and even shifted toward the Republicans in the Ford vs. Carter race of 1976, but has lately become more and more Democratic. In contrast, the more rural Chippewa was noticably more Democratic than the nation as a whole in the 1980s but has recently trended a bit toward the Republicans, perhaps because of social issues.
