time. For instance, in Republican George W. Bush’s victory in 2000 over Democrat Al Gore, 12 states were decided by less than five points. Four years later, in Bush’s 2.5-point national win over Democrat John Kerry, 11 state margins were less than five points. By 2012, just four states were decided by less than five points in Democrat Barack Obama’s four-point national win over Republican Mitt Romney: Florida, North Carolina, Ohio, and Virginia. So while those elections were slightly more (2000) or slightly less (2004, 2012) competitive than those of 1960 and 1976, the more recent contests featured fewer true swing states.
But that doesn’t tell the whole story, because 1960, 1976, 2000, 2004, and 2012 were all nationally competitive elections. There’s a way to measure how many states vote close to the national average in both close elections and blowouts.
THE TWO-PARTY VOTE AND PRESIDENTIAL DEVIATION
There are two concepts that merit explaining before proceeding. The first is the two-party vote.
This is simple enough. The two-party vote is a way of reporting election results as just the votes cast for the Democratic and Republican candidates in a given race. It subtracts the third-party votes, allowing for comparisons across time without the distorting effects independent and minor party candidacies have on results. Given the longstanding dominance of the two parties, this is a way to cut out the noise that fleeting third-party insurgencies introduce from time to time.
In most modern elections, removing the third-party vote barely makes a difference at all. For instance, in 2004, 2008, and 2012, 99 percent, 98 percent, and 98 percent of all the presidential votes cast, respectively, were for the Democratic and Republican candidates. Removing the third-party votes hardly alters the margins of victory at all. In 2012, Obama beat Romney by 3.86 percentage points in the all-party voting, versus 3.92 points in the two-party vote. In other words, there was no real difference.
However, using the two-party vote creates complications for certain years featuring big third-party votes, like 1912, 1924, 1968, 1992, and 1996, among other years. But only Democratic and Republican candidates have won the presidency since both parties began competing against each other in 1856. Third-party candidates occasionally win electoral votes, but only rarely: the last one to win any state was race-baiting George Wallace in 1968, who won five southern states as an independent candidate.
The main reason to use the two-party vote is to make apples-to-apples comparisons over time. Using two back-to-back elections from the 2000s illustrates why this can be a useful exercise.
In 2000, Gore received 48.4 percent of the total national vote (including all votes cast for all candidates). Four years later, Kerry got 48.3 percent of the all-party total vote. By that metric, it appears that Gore and Kerry performed almost exactly the same.
But in practice, Gore did significantly better, winning the popular vote by about half a percentage point while Kerry lost by about 2.5 points in the national popular vote to Bush. The difference between those years is that close to 4 percent of all voters in 2000 voted for third-party candidates—mostly for Green Party nominee Ralph Nader, who probably cost Gore the election—while just 1 percent of all voters picked a third-party candidate in 2004. The national two-party vote in those years tells the more accurate tale. Gore won 50.3 percent of the two-party tally in 2000, while Kerry captured just 48.8 percent four years later.
Third-party candidacies come and go, but since 1856 the two major parties have remained constant, and tracking the change in the votes for these parties paints a clearer picture about the evolution of the nation’s voting from election to election. While there will be exceptions, most of the results reported throughout the rest of the book will be just the two-party vote.
The second concept is presidential deviation. This is the difference between how a county, state, or other political subdivision votes in a given election compared with, usually, the national results. It’s a way of expressing how reflective a given place is of the national results. This is, again, calculated through the two-party vote (although it can be figured through the all-party vote just as well) and it’s expressed as a rounded number.20
For instance, the national two-party vote in 2012 was 52 percent to 48 percent in favor of Obama over Romney. That same year, Romney won Wyoming 71 percent to 29 percent. Romney’s share of the vote in Wyoming was 23 percentage points larger than his national share (and Obama’s was 23 points lower—again, all numbers are rounded). So Wyoming deviated 23 points from the national average in 2012 in favor of the Republicans. For shorthand, this makes Wyoming an R +23 state.
Measuring this deviation isolates where a state stands in relation to the national voting in elections that are both narrow and lopsided. For instance, in Virginia in 1976, Ford beat Carter by about two points. Four years later, Virginia backed Reagan by about 14 points. That’s a 12-point swing in the two-party vote. But its presidential deviation from the nation was the same in both elections: it was two points more Republican than the nation in 1976 (narrowly backing Ford while Carter won nationally) and then two points more Republican in 1980 (giving Reagan a slightly bigger victory than in his overall national triumph). So, while Virginia’s margin of victory for the Republican presidential candidate changed quite a bit from 1976 to 1980, the Old Dominion didn’t get any more Republican relative to the nation from one year to the next. The deviation separates the swing in the state from the swing nationally.
Presidential deviation is used later in this book to compare county-level results to state and national results, placing the outcome in certain places in both state-level and national-level contexts. For instance, in Ohio in 2008, Obama got 52 percent of the two-party vote, while he got 54 percent nationally. In Athens County, home of Ohio University, he got 68 percent of the vote. So Athens County was D +16 compared to the state versus D +14 compared to the nation.
THE NATION’S SHRINKING MIDDLE
The two-party vote metric, combined with presidential deviation, makes it possible to compare election results over time. It also illustrates which states were close to the national presidential voting average in both blowouts and nail-biters.
The 1960 election, when 20 states were decided by five points or less in the all-party vote, is already noted above. Compare that to 1956, Eisenhower’s reelection victory. That year, only three states were decided by less than five points.
But the two elections are not really comparable: Ike captured 457 electoral votes and 57 percent of the national popular vote against Stevenson in their rematch from four years prior. One wouldn’t expect there to be many close states in such a lopsided election—but looking at the election through presidential deviation tells a far different tale.
A whopping 32 states had deviations of less than five points in 1956, one more than the much closer 1960 election. (Less than five points means any state with a deviation of four or less in the election. Practically speaking, because of rounding, this means any state with a deviation of less than 4.5 points.) This can be less than five points in either direction, which is actually a fairly large range: In an election that was 50–50 nationally, a state that voted 54 percent to 46 percent either way would be included in this definition as a state with a deviation less than five.
The examples of 1956 and 1960 represent a high-water mark for the number of states clustered near the nation’s middle in the 30 presidential elections from 1896 to 2012. They also represent a high mark for the number of electoral votes in states with deviations of four or less. In those two elections, roughly three-quarters of all the available electoral votes were in the states with low deviations. Those are also the two highest in the time frame studied.
So an equal number of states were clustered close to the national average in both elections—it’s just that Eisenhower’s much higher tide of victory meant that the states voting with the middle of the
