in such a short amount of time remains striking.
In the 2012 election, roughly four-fifths of nonwhite voters, who made up close to 30 percent of the national electorate, voted for Obama, while about three-fifths of white voters, making up about 70 percent of the voters, picked Romney.23 As the country becomes more diverse, it’s not impossible to imagine a scenario where voting becomes even more polarized by race.
Others have noted the increasing political polarization of the American public, such as Alan Abramowitz, who coined along with his colleague Steven Webster the term “negative partisanship,” which describes how voters’ increasingly hostile perceptions of the opposing political party inform their voting. “This has led to sharp increases in party loyalty and straight ticket voting across all categories of party identification,” they write, “and to growing consistency between the results of presidential elections and the results of House and Senate elections.”24
In 1900, just 3.4 percent of US House districts featured split results—that is, only a relative handful of districts supported a different party for president and for US representative. Such low percentages remained common throughout the first half of the 20th century: on average, only 12 percent of districts featured split results in the presidential elections held during this period. (Not all district results are available from this time period, but there’s little reason to think that the missing data would change the results much.)
But throughout the second half of the century and into this century, it was common for congressional districts to support candidates from different parties for president and for the US House. From 1952 to 2008, an average of 28 percent of House districts split their presidential and congressional ballots. However, the number of split districts has been dropping over time, bottoming out at only 26 of 435 districts (6 percent) in 2012.25
Granted, blowout elections will naturally produce more split districts. The two elections with the highest number of split district results were 1972 and 1984, when Republicans Richard Nixon and Ronald Reagan each carried 49 of 50 states in smashing reelections. Also, both parties will draw favorable districts for themselves whenever possible, a process known as gerrymandering. This incentivizes the majority party in a state to draw the minority party’s voters into a small number of districts the majority party cannot win, while drawing a larger number of safe districts for themselves.
But the trend over time is a good measure of polarization—and also of the political trajectory of the South, which throughout the second half of the 20th century often elected Democrats to the House while backing Republicans for president. Indeed, the less polarized second half of the last decade doesn’t exactly have noble roots. Brendan Nyhan has argued that “the less polarized politics of the mid-20th century were driven almost entirely by the issue of race, which created a bloc of conservative southern Democrats who acted as a virtual third party for much of this time.”26
Those conservative Democrats in the South would become Republicans: the Deep South (Alabama, Georgia, Mississippi, Louisiana, and South Carolina) is now largely a one-party preserve. The same can be said of several other southern states, such as Arkansas, Tennessee, and Texas.
The Republican Party used to have a bloc of northeastern liberals/moderates, known as “Rockefeller Republicans” in tribute to the moderate governor of New York, Nelson Rockefeller. One of the last examples of these political anachronisms was Senator Lincoln Chafee of Rhode Island, who lost to a Democrat in the midterm wave of 2006. Nearly a decade later, in 2015, Chafee was not only a Democrat—he was running (hopelessly) for the Democratic presidential nomination before dropping out of the race after a single debate.
The reason to bring all this up is simple: in a divided political world where so few voters and states can be reached, the few states that are near the nation’s political center become even more valuable. When many states are near the average national voting—as was the case in 1960 and 1976, for instance—a truly national electoral strategy is sensible. But when few are, as has been the case in recent elections, much of the country can and should be ignored by any sane presidential campaign.
THE NARROW ELECTORAL BATTLEFIELD
Nowadays, a Nixonian 50-state pledge would be ridiculous. According to an analysis from the Center for Voting and Democracy, the 2012 presidential candidates—President Obama and running mate Joe Biden along with Mitt Romney and running mate Paul Ryan—held public events in only 12 states after the Democratic National Convention, all of which would deviate less than five points from the national average.27 Both parties had an excellent grasp of what the closest states were. If anything, noting that the candidates visited a dozen states makes the number of truly competitive states seem artificially high: the listing includes only one visit apiece in Michigan and Minnesota, and five or fewer in North Carolina and Pennsylvania. The remaining eight states enjoyed—endured?—243 visits from the presidential tickets. Nearly a third of those were in Ohio, whose 73 visits were more than double the total of any other state except for Florida, which had 40.
Additionally, the Center found that the two campaigns spent more than 99 percent of their respective television campaign advertising dollars in just 10 states from mid-April (when Romney effectively clinched the GOP nomination) through the November general election: Colorado, Florida, Iowa, Nevada, New Hampshire, North Carolina, Pennsylvania, Ohio, Virginia, and Wisconsin. Few would be surprised if these 10 states again ate up nearly all the resources spent by the two major-party nominees in the 2016 presidential election.
If campaigns were run 100 years ago the way they are today, with the same technology, extensive candidate travel, and micro-targeting of television ads, a similar dynamic would have prevailed. A small number of electoral votes—36 percent of them in 1916 in 21 states, versus 31 percent of them in 14 states in 2012—would have been in play (with eventual presidential deviations less than five points), and the campaigns in both eras would have been wise to focus only on these states.
The difference would have been in the states targeted. Only five of them appear on both 1916’s list of states closest to the national average and 2012’s: Minnesota, New Hampshire, New Mexico, Oregon, and Ohio. In both eras, Ohio was by far the most valuable prize among these states: in 1916, it had double (24) the number of electoral votes of its next closest rival among this group, Minnesota (12), and in 2012 it had nearly double the electoral votes (18) of the second-ranked state among this group, again Minnesota (10). Additionally, while Minnesota, New Mexico, and Oregon are included here as competitive, they really weren’t in 2012: New Mexico and Oregon did not see a single dollar of ad spending and neither presidential ticket visited them, while Minnesota saw only a pittance of ad spending and a single visit, by Ryan.
Ohio spent the entire post–Civil War era—and, really, before that as well—voting at or near the national political midpoint. Sometimes it had a lot of company in this position, particularly in the middle of the 20th century. Other times—including the elections in the 21st century held while many states were moving further toward the Democrats or the Republicans—it was one of relatively few prizes plausibly available to both parties in a competitive national election. In fact, there’s a lot of evidence that suggests that Ohio has been the best bellwether state in presidential elections for a century or more.
TWO
Ohio at the Head of the Flock
In the late stages of the 2012 presidential campaign, a disconnect emerged between national horse-race polls for the presidential race and those at the state level. Republican Mitt Romney thoroughly outperformed Democratic incumbent Barack Obama in the first presidential debate, held about a month before Election Day, and surged in national polls. Romney took a small lead immediately after the debate and then held within one point of Obama in averages of national polls for the remainder of the campaign, according to the widely cited RealClearPolitics website’s average of national polls.1 The Gallup daily presidential tracker, posted promptly at 1:00 p.m. Eastern Time every day and immediately dissected by political journalists and junkies, proved particularly favorable for Romney, showing him with leads in the mid-single digits for much of the rest of the campaign before narrowing to a final prediction of Romney 49 percent,