Several pundits have claimed that the second major victor in yesterday’s U.S. presidential election was statistician Nate Silver, who correctly picked the winner in every state, thereby seemingly demonstrating the power of Bayesian analysis—when done correctly. In scrutinizing Silver’s final pre-election map, I can find only a few minor instances in which was not fully on-target (Iowa, for example, was not as close as he had depicted it). In a Slate column, however, Daniel Engber claims that the real credit should go to the pollsters who generated the date that Silver used. Engber notes that Silver, unlike most pollsters, missed the Democratic victory in the Montana senatorial contest.
The New York Times website features some excellent cartographic work on the election. One innovative map shows the shift in voting patterns from the 2008 election at the country level. As can be clearly seen, in the majority of U.S. counties, Mitt Romney gained a larger share of the vote than Republican candidate John McCain had received in the previous election. The exceptions to this pattern are intriguing. Across much of the Deep South, overall a Republican stronghold, Barack Obama gained votes in 2012 over his 2008 showing. Many of these “blue-shifted” counties are heavily African-American, which may indicate a greater voter turnout among Blacks in this election; if this is indeed the case, such a change runs counter to most of the predictions made prior to this election. An alternative thesis is that a considerable number of evangelical Whites in these counties declined to vote, not wanting to endorse a Mormon candidate. Yet in most other parts of the country dominated by conservative Protestants, Romney outpolled McCain. Other areas that moved in the Democratic direction include much of New Jersey and New York, which may in part reflect the aftermath of Hurricane Sandy. Central Ohio, perhaps the most crucial battleground area in this election, also shows a distinct shift in the direction of the Democratic Party.
At the state level, the map of the 2012 election looks very much like that of 2008, with only Indiana and North Carolina switching back to the Republican candidate (provided that Florida stays within the Obama camp). More significant is the fact that this map is also strikingly similar the maps of the 2004 and 2000 elections. The only state-level difference between yesterday’s election and that of 2000 was the movement of a few closely contested swing states from the Republican to the Democratic candidate: Nevada, Colorado, Ohio, Florida, Virginia, and New Hampshire. I suspect that migration patterns are pushing a few of these states, especially Colorado and Virginia, into the Democratic camp. But otherwise, the basic electoral geography of the U.S. has shown little change over the past twelve years. Even at the county level, the differences are relatively modest. The coal-mining region of Appalachian has definitely turned to the Republicans Party over this period, as have a number of counties located elsewhere in the Upper South. At the same time, the Democratic Party has solidified its advantage in the coastal West and in the Northeast. In 2000, George W. Bush took thirteen coastal counties on the West, whereas in 2012, Romney won only six. And whereas Bush was the victor in fifteen counties in northern New England, in this election Romney took only four.
Although the geographical changes in U.S. presidential voting since 2000 have been minor, the situation is quite different if we look back to the 1996 election, as well as those preceding it. In 1996, Bill Clinton took the interior states of the Upper South as well as Louisiana. In the early twenty-first century, it would be highly unlikely that such states as Kentucky, Tennessee, and Arkansas would vote for a Democratic presidential candidate. In yesterday’s election, Obama received less than 40 percent of the vote in all three states, and in West Virginia, which was recently a Democratic stronghold, he barely got 35 percent. Obama did significantly better in such Deep South states as South Carolina, Georgia, and Mississippi, where the African-American population is much larger.
Only two states gave more than 70 percent of their votes to one candidate: Utah, where Romney got roughly 73 percent, and Hawaii, which went for Obama by 70.6 percent. Although the overall trend in U.S. politics is clearly one of increasing regional differentiation, most states are still more “purple” than “red” or “blue.” At the county level, however, it is a different story, as many localities in the Great Plains and the Inter-Mountain West went for Romney by well over 80 percent. In contrast, it is difficult to find any county that gave more than 80 percent of its votes to Obama. Holmes County in Mississippi, however, did go for Obama by 83.9 percent. Holmes County, not surprisingly, is mostly African-American, with only 20 percent of its population classified as White.