For all the time we spend worrying about swing states—who’s polling well in them, how candidates are tailoring their messages to them, and so on—it’s not really that clear what a swing state even is. (There even used to be a blog initially devoted to that very question, and I’m not sure it ever got definitively answered there …) There’s a general sense that it’s the band of states that were the closest in the previous election, and thus will probably be closely contested again in the next election. But there’s also the sense that it’s the “bellwether” states, the ones that are close to the national average and that narrowly break in favor of whoever wins nationwide.
The problem is, those two definitions don’t necessarily match up. The states that were closest in the previous election may not dovetail with the next election; a state’s position in the spectrum of states may vary a lot, depending on how much money and manpower the campaigns pour into the state and whether that state’s unique demographics mesh with the candidate’s appeal. (Take the case of Indiana, for instance; nobody would have guessed that it would be the third-closest state in the 2008 election, based on how dark red it was in the 2004 election, and nobody seems to think that its closeness in 2008 will mean that it’s again in play in 2012.)
And, conversely, just because a state is usually close to the national average doesn’t mean it’s always going to be close. If it’s a close election (like 2000 or 2004), then they will, but if it’s not a close election, the “swing” states don’t wind up being that close, and the tide starts sloshing up into the safer states. Like, for instance, a blue state (Illinois) being the closest in 1988, or a red state (Indiana) one of the closest in 2008.
The constant change in who is and isn’t a swing state is easier to see in graphic form than it is to describe, though, and here’s a handy chart putting the states in the last six presidential elections into rank order. The percentage in each cell is the percentage the Democratic candidate got in a two-way heat. Where it says “Par” in each column, that’s the two-way percentage that the Democratic candidate got nationwide; that helps you see how the candidates fared in each state in relation to the rise or fall of the overall tide each year. As you can see, some states (Iowa, Pennsylvania) usually appear right near the national average regardless of who’s winning. In other places, you can see one-time swing states turning safely blue (New Jersey) or safely red (Louisiana), safely red states gradually turning into swing states (Florida, Virginia), and even safely blue states not even pausing in swing state territory before becoming safely red states (West Virginia).
|D.C. 85.2%||D.C. 90.8%||D.C. 90.1%||D.C. 90.5%||D.C. 90.5%||D.C. 93.4%|
|Rhode Is. 55.9%||Mass. 62.1%||Rhode Is. 69.0%||Rhode Is. 65.6%||Mass. 62.7%||Hawaii 73.0%|
|Iowa 55.1%||Rhode Is. 61.8%||Mass. 68.6%||Mass. 64.8%||Rhode Is. 60.5%||Vermont 68.9%|
|Hawaii 54.8%||Vermont 60.3%||New York 66.0%||New York 63.1%||Vermont 60.3%||Rhode Is. 64.2%|
|Mass. 54.0%||Arkansas 60.0%||Hawaii 64.3%||Hawaii 59.8%||New York 59.3%||New York 63.6%|
|Minnesota 53.6%||New York 59.5%||Vermont 63.2%||Conn. 59.3%||Maryland 56.6%||Mass. 63.2%|
|Oregon 52.4%||Illinois 58.6%||Maine 62.7%||Maryland 58.5%||Conn. 55.3%||Maryland 62.9%|
|W. Virginia 52.4%||California 58.5%||Conn. 60.4%||New Jersey 58.2%||Illinois 55.2%||Illinois 62.7%|
|New York 52.1%||Maryland 58.3%||New Jersey 60.0%||Delaware 56.7%||California 55.0%||Delaware 62.6%|
|Wisconsin 51.8%||W. Virginia 57.8%||Illinois 59.6%||California 56.2%||Maine 54.6%||California 62.3%|
|Wash. 50.8%||Minnesota 57.7%||Minnesota 59.4%||Illinois 56.2%||Hawaii 54.4%||Conn. 61.3%|
|Illinois 49.0%||Wash. 57.6%||Arkansas 59.4%||Vermont 55.4%||Delaware 53.8%||Maine 58.8%|
|Penn. 48.8%||Hawaii 56.7%||Maryland 58.6%||Wash. 52.9%||Wash. 53.6%||Wash. 58.8%|
|Maryland 48.5%||Oregon 56.6%||Delaware 58.6%||Maine 52.7%||New Jersey 53.4%||Oregon 58.4%|
|Vermont 48.2%||Missouri 56.5%||W. Virginia 58.4%||Michigan 52.6%||Oregon 52.1%||Michigan 58.4%|
|California 48.2%||Maine 56.1%||Michigan 57.3%||Penn. 52.1%||Minnesota 51.8%||New Jersey 57.9%|
|Missouri 48.0%||Penn. 55.5%||California 57.2%||Minnesota 51.3%||Michigan 51.7%||New Mex. 57.7%|
|New Mex. 47.5%||Delaware 55.2%||Wash. 57.2%||PAR 50.3%||Penn. 51.3%||Wisconsin 57.1%|
|Conn. 47.4%||New Mex. 55.1%||Louisiana 56.6%||Oregon 50.2%||New Hamp. 50.7%||Nevada 56.4%|
|Montana 47.0%||Michigan 54.6%||Wisconsin 55.9%||Iowa 50.2%||Wisconsin 50.2%||Penn. 55.2%|
|S. Dakota 46.8%||Connecticut 54.1%||Iowa 55.7%||Wisconsin 50.1%||Iowa 49.7%||Minnesota 55.2%|
|PAR 46.1%||Iowa 53.7%||New Hamp. 55.6%||New Mx. 50.003%||New Mex. 49.6%||New Hamp. 54.9%|
|Colorado 46.0%||PAR 53.5%||* Penn. 55.2%||* Florida 49.995%||* Ohio 48.9%||Iowa 54.8%|
|* Michigan 46.0%||* Colorado 52.8%||PAR 54.7%||New Hamp. 49.3%||PAR 48.8%||* Colorado 54.6%|
|Louisiana 44.8%||Wisconsin 52.8%||Oregon 54.7%||Missouri 48.3%||Nevada 48.7%||PAR 53.6%|
|Ohio 44.5%||Louisiana 52.7%||New Mex. 54.0%||Ohio 48.2%||Colorado 47.6%||Virginia 53.2%|
|Maine 44.2%||Tennessee 52.6%||Ohio 53.6%||Nevada 48.1%||Florida 47.5%||Ohio 52.3%|
|Kentucky 44.1%||Kentucky 51.9%||Missouri 53.5%||Tennessee 48.0%||Missouri 46.4%||Florida 51.4%|
|Delaware 43.8%||Nevada 51.8%||Florida 53.2%||Arkansas 47.2%||Virginia 45.9%||Indiana 50.5%|
|Texas 43.7%||Montana 51.7%||Tennessee 51.3%||W. Virginia 46.8%||Arkansas 45.1%||N. Carolina 50.2%|
|N. Dakota 43.4%||New Jersey 51.4%||Arizona 51.2%||Arizona 46.7%||Arizona 44.7%||Missouri 49.9%|
|Kansas 43.3%||Ohio 51.2%||Nevada 50.9%||Louisiana 46.1%||N. Carolina 43.8%||Montana 48.8%|
|New Jersey 43.1%||New Hamp. 50.8%||Kentucky 50.5%||Virginia 45.9%||W. Virginia 43.5%||Georgia 47.4%|
|Arkansas 42.8%||Georgia 50.3%||Georgia 49.4%||Colorado 45.5%||Tennessee 42.8%||S. Dakota 45.7%|
|N. Carolina 41.8%||N. Carolina 49.5%||Colorado 49.2%||Georgia 44.0%||Louisiana 42.7%||Arizona 45.7%|
|Tennessee 41.8%||Florida 48.8%||Virginia 48.9%||N. Carolina 43.5%||Georgia 41.7%||N. Dakota 45.6%|
|Oklahoma 41.6%||Arizona 48.7%||Montana 48.3%||Alabama 42.4%||S. Carolina 41.4%||S. Carolina 45.5%|
|Alabama 40.3%||Texas 47.8%||S. Dakota 48.1%||Kentucky 42.3%||Miss. 40.1%||Texas 44.1%|
|Indiana 39.9%||S. Dakota 47.7%||N. Carolina 47.5%||S. Carolina 41.9%||Kentucky 40.0%||Miss. 43.4%|
|Georgia 39.8%||Virginia 47.4%||Texas 47.3%||Indiana 42.0%||Indiana 39.6%||W. Virginia 43.3%|
|Virginia 39.6%||Kansas 46.5%||Miss. 47.3%||Miss. 41.4%||Montana 39.5%||Nebraska 42.4%|
|Miss. 39.5%||Wyoming 46.2%||Indiana 46.9%||Kansas 39.1%||S. Dakota 39.1%||Kansas 42.4%|
|Nebraska 39.5%||Alabama 46.2%||S. Carolina 46.8%||Texas 39.0%||Texas 38.5%||Tennessee 42.4%|
|Arizona 39.3%||Indiana 46.2%||Alabama 46.3%||Oklahoma 38.9%||Kansas 37.1%||Kentucky 41.8%|
|Nevada 39.2%||S. Carolina 45.4%||N. Dakota 46.1%||S. Dakota 38.4%||Alabama 37.1%||Louisiana 40.5%|
|Florida 38.7%||Miss. 45.1%||Oklahoma 45.6%||Montana 36.3%||Alaska 36.8%||Arkansas 39.8%|
|Wyoming 38.6%||Oklahoma 44.4%||Wyoming 42.5%||N. Dakota 35.3%||N. Dakota 36.1%||Alabama 39.1%|
|S. Carolina 37.9%||Alaska 43.4%||Kansas 39.9%||Nebraska 34.8%||Oklahoma 34.4%||Alaska 38.9%|
|Alaska 37.8%||N. Dakota 42.1%||Alaska 39.6%||Alaska 32.1%||Nebraska 33.2%||Idaho 37.0%|
|New Hamp. 36.8%||Idaho 40.3%||Nebraska 39.4%||Idaho 29.2%||Idaho 30.7%||Utah 35.5%|
|Idaho 36.7%||Nebraska 38.7%||Idaho 39.2%||Wyoming 29.0%||Wyoming 29.7%||Oklahoma 34.4%|
|Utah 32.6%||Utah 36.2%||Utah 38.0%||Utah 28.3%||Utah 26.7%||Wyoming 33.4%|
If you’re wondering why this uses only the Democratic and Republican shares of the overall vote, that’s the best way to smooth out differences between different elections, where the presence of third parties may or may not matter. For instance, that’s the way that Charlie Cook calculates Partisan Voting Index. As you can see, it doesn’t seem to affect the overall results much, particularly in the case of Ross Perot’s 1992 and 1996 runs, where he seemed to draw votes about equally from both sides of the aisle. In fact, it seems a little more pronounced with Ralph Nader’s 1996 and 2000 runs, where some of the states where he performed the strongest (Oregon, Wisconsin) show a noticeable dip into swing state territory thanks to votes being siphoned from only one side of the ledger.
And if you’re wondering what the little asterisks are next to one state each year, that’s the “tipping point” state, a form of analysis that Nate Silver came up with back in 2008. In each year, that’s the state that would have put the Democratic candidate barely over the top in the Electoral College, starting with the first three in the District of Columbia and working your way down the totem pole until getting to 270 votes. As you can see from the chart, the whole hyperventilating over swing states may be a little overdone in the first place; the tipping point is almost always right next to the national average. The one exception, as you can see from the distance between the tipping point state (Florida, naturally) and the national average, seems to be Al Gore’s 2000 campaign, which seems to point to a tactical misfire on their part: They ran up the score in the safely blue states, enough to win the Popular Vote, without allocating the resources to bring home enough swing states.
You’ve probably noticed, if you’ve stared hard enough at the chart without your eyeballs falling out, that some states stay pretty constantly in the narrow band near the national average, while others are on more of a diagonal trajectory, moving from safe to swing or vice versa. (Or a V-shaped trajectory if you’re Arkansas, which went red to blue back to red, thanks to the presence of favorite son Bill Clinton.) That left me asking, which of the states are truly the swingiest? In other words, which ones float constantly with the national tide, instead of going in one direction or the other?
I developed a simple formula for figuring that out, and the result may not surprise you. It’s the state that usually is one of the first to get mentioned when pundits talk about what states the road to the presidency goes through, and it really did hold the key to the 2004 election (in terms of where the pre-election focus was, where the post-election threats of litigation were, and also what state was closest to the national mean): Ohio.
This, of course, doesn’t mean that any Barack Obama electoral strategy should be based, at all costs, on winning Ohio, as if it magically holds the key to the entire country. As you can see by looking at the overall totem pole of states from top to bottom each year, and where the “tipping point” always falls, it’s always a national election, just in a nation where Ohio falls right in the middle. If you’re losing the country badly, Ohio is long gone, and if you’re winning easily, then Ohio is already in the bag.
Much more discussion over the flip …
The math used to arrive at Ohio as the swingiest may not be immediately apparent; here’s what it means. The D+ and R+ figures are probably familiar to those of you who use PVIs to talk about electoral propensities; it’s how far that state, that particular year, deviated from the overall national norm. In Ohio, for instance, Mike Dukakis got 44.5 percent in Ohio in 1988, which was 1.6 percent less than his national average of 46.1 percent; in 1992, Bill Clinton got 51.2 percent in Ohio, which was 2.3 percent less than his national average of 53.5 percent, and so on. As you can see, Ohio lagged the national average every year except 2004 (when John Kerry heavily contested it at the expense of, say, Colorado), but consistently by only a small margin. Add up the six deviations, and the total is 8.6, the smallest total of any state.
A fair question might be why I’m adding up the absolute value of each deviation, rather than, say, averaging them out by adding D+ figures and subtracting R+ figures. I tried it that way too, but the results don’t seem to be as “swingy” that way. If you use that approach, Missouri winds up as the swingiest state (1.9 + 3.1 – 1.2 – 2.0 – 2.4 – 3.8 = -4.4). That’s fitting, given its historical “bellwether” status, but not too fitting, as it’s slowly been receding into the red column lately. This alternate method manages to capture Missouri balanced on a knife point, averaging the years (1988, 1992) when it was slightly Democratic-friendly against the years (2004, 2008) when it started to become Republican-friendly instead. The Missouri result instead would be an artifact of what year I started counting. (If I went further back, to the 50s or 60s, I could probably get an even stranger result, like Illinois or New Jersey being the swingiest state.) Of course, there’s always going to be an artificialness to the result, depending on the start date; even using absolute values, if I’d started in 1992 instead of 1988, Iowa would be the swingiest state (by lopping off Mike Dukakis’s lopsided victory there in ’88, thanks to the now-long-forgotten mid-80s farm crisis) and Florida would be right behind it (as it was redder than most of the Deep South back in the 80s, when it was a much whiter state).
There’s also one other potential approach to this question, which, instead of using the deviation from the floating national average (which is kind of a pointy-headed political science idea), uses an idea that’s more resonant to the layperson: how close the election in each state was over the six elections. In other words, not asking how far the state was from “par,” but rather from 50 percent in a two-way race, or the point where you actually win. Interestingly, the swingiest state is still Ohio, though this shows traditional bellwether Missouri coming close too. (Here, the deviation from 50 percent in Ohio in 1988 was 5.5, the deviation in 1992 was 1.2 percent, and so on; add it up for a total of 15.5.)
If at this point you’re still haven’t succumbed to numbers overload, this kind of analysis can be applied to smaller units, too. Been wondering what the swingiest county is? Wonder no more! Switching back to the original method, based on absolute deviation from the national “par,” the nation’s swingiest county is: Shiawassee County, Michigan! This rural/exurban county of about 70,000, located in the gap between Lansing and Flint, is, out of more than 3,000 contenders, the most average place in America.
Or maybe you’re wondering what is the swingiest county of the swingiest state? That honor falls to Ottawa County, Ohio, a rural lakefront county of 40,000 near Toledo. Right behind it is Stark County, the most populous of the top 10 swingiest counties and location of Canton (and, apparently, a whole lot of swing voters).