mainstream.
The District of Columbia, with its heavy Democratic lean, has never been within five points of the national voting (it first voted for president in 1964). The formerly Democratic Solid South, which is now staunchly Republican, is almost a perfect anti-bellwether: Alabama and South Carolina have had deviations less than five points in only two of the last 30 elections, and Mississippi only did twice.
Table 2.3 shows the average presidential deviation for each state over the past 30 elections. For each election, it doesn’t matter whether a state deviated in a Democratic or a Republican direction—the average deviation in either direction over the 30 elections is what’s expressed here.
Table 2.3. States’ average deviation from two-party presidential vote, 1896–2012
The results of these calculations offer another strong argument for Ohio as the most accurate bellwether over the last 30 elections. On average, Ohio’s presidential vote deviated just 2.2 points from the national results. New Mexico, noted above as a historic bellwether, was second at 2.8 points.
The deviations calculated above track fairly well with the presidential win totals discussed above, with one major exception. While states like Ohio, New Mexico, Illinois, and Missouri lead this list—just as they do above—bellwether Nevada’s vote has historically deviated much more strikingly over the last 30 elections, placing it in the middle of the pack of both tables 2.2 and 2.3.
Again, Nevada’s strong support of silver bug Bryan, particularly in 1896 and 1900 at the start of this analysis, skews the numbers. Based on two-party vote, the Silver State was 33 points more Democratic than the nation in 1896, and 15 points more Democratic four years later. More recently, though, Nevada had presidential deviations of R +15, R +8, and R +7 in 1980, 1984, and 1988, respectively, before settling into a deviation near the national average moving into the 21st century.
If we narrow the time frame, Ohio’s position as the leading bellwether becomes even more striking. Since 1964, Ohio has been at most just two points from the middle of the country either way, voting with the winner every time. Meanwhile, Illinois, the one-time battleground turned Democratic stronghold, has been at least five points more Democratic than the nation in every election since 1992. Missouri was four and seven points more Republican than the nation in 2008 and 2012, respectively, and was slightly more Republican in the three elections prior to that.
Using two-party presidential deviations can skew some state results. For instance, this method shows Alabama with a whopping R +52 deviation in 1948. Why? Because Truman was not even on the ballot there: conservative Democrat Strom Thurmond was on the ballot instead, so Truman got zero votes in Alabama. Thurmond won the all-party vote with nearly 80 percent, but Republican nominee Thomas Dewey got 100 percent of the two-party votes, or 52 points better than his national share of roughly 48 percent of the two-party tally. As is obvious, the two-party vote is not perfect, but when it performs poorly it is generally in the states of the South, which supported third-party candidates such as Thurmond in 1948 and Wallace in 1968. Nobody would consider those places bellwether states.
The District of Columbia, meanwhile, has never been less than 24 points more Democratic than the nation in its entire electoral history, which began in 1964. That gives it the highest average deviation from the nation’s results of any place with electoral votes, by a significant margin.
Of all the other states (and DC), only two states besides Ohio have never been more than 10 points from the national voting since 1896: Ohio’s neighbor to the west, Indiana, which is currently the most Republican state in the Midwest (and has been more Republican than the nation in every election since 1928), and historic bellwether Illinois, which is now the Midwest’s consistently most Democratic state.
Averaging the presidential deviation over the last 30 elections helps confirm what Ohio’s nearly flawless record in picking presidential winners suggests: Ohio is almost always reflective of the national popular vote. Its place in the middle of the national voting leads to a third argument for its importance as a presidential bellwether state. More often than any other state, Ohio puts the winning candidate over the finish line in the Electoral College.
OHIO: THE DECIDER
Technically, in 27 of the 30 elections from 1896 to 2012, the winning candidate still would have won without Ohio’s electoral votes. The exceptions are 1916, 2000, and 2004. However, that statistic doesn’t tell the full tale of how often Ohio votes close to the national average and how often it casts the decisive vote for the winner in both competitive and uncompetitive elections.
While Nevada and New Mexico have similar batting averages to Ohio’s when it comes to voting with the winning presidential candidate, the Buckeye State has an obvious but nonetheless important advantage: it’s always been much more populous than these far more sparsely populated western states. Ohio cast 18 electoral votes in 2012: while Ohio’s number of electoral votes has been declining because of slow population growth, it was still the seventh highest in the nation, and Ohio has always ranked among the Electoral College’s biggest prizes.
Nevada and New Mexico cast just six and five electoral votes, respectively, in 2012, and they’ve never cast more in any election. The small size of those two states means that they are not nearly as valuable to the candidates in terms of assembling a winning electoral coalition. They are too small, practically, to make the difference between winning and losing in all but the closest elections. Indeed, neither state has actually been decisive in the past 30 presidential elections.
Meanwhile, Ohio has produced the winning electoral vote for the victorious presidential candidate more times over the last 30 elections than any other state. In five of those elections, Ohio’s electoral vote put the winner over the finish line.
Here’s what that means: One can take the states that voted for the presidential winner and put them in order, from biggest margin to smallest, rather like Trende’s hypothetical pool cited in chapter 1. Under this model, the state that produced the biggest percentage-point margin for the winner casts the first votes. For instance, in 2012, the District of Columbia voted for President Obama by an 83.6 percentage-point margin, by far the biggest margin won by either candidate that year in any place that cast electoral votes. So Obama got his “first” three electoral votes from DC. Obama’s next four votes came from his birth state of Hawaii, which he won by 42.7 points. That put him at seven electoral votes. In that same election, Romney got his first six electoral votes from Utah, which he won by 47.9 points thanks to overwhelming support from his fellow Mormons (although Utah is frequently among the most Republican states in presidential elections). And so on. Once the winner gets to 270 electoral votes, the rest is gravy.
Table 2.4 lists the 30 elections from 1896 through 2012 and the “decisive” state in each election. In recent years, this means the 270th electoral vote in the current Electoral College, which has 538 electoral votes, a total reached in 1964 with the addition of the District of Columbia and its three electoral votes.
Table 2.4. States providing decisive electoral vote, 1896–2012
Ohio cast this decisive vote five times: 1896 for Ohioan William McKinley, 1936 for Franklin Roosevelt, 1968 and 1972 for Richard Nixon, and
