## The Elusive Convention-Site Effect

by

When you read an article about the 2016 Republican convention in Cleveland, it almost always mentions that the last five states in which the Republicans held their convention voted Democratic in that year’s election. Yet as NRO’s Nick Frankovich (a loyal follower of Chief Wahoo since the days of Duke Sims) has pointed out, it’s still possible that the state you hold your convention in does tend to give you extra votes, even if you don’t win the state (by the way, the Democrats have won the electoral votes of five of their last six convention states) — and Ohio should be ripe for Republican appeals in 2016.

To test this hypothesis, for each election from 1988 through 2012, I calculated (with help from NRO’s Spencer Case) the margin between the two parties’ popular-vote percentages in that election and in the previous one, then calculated the difference between these two figures to get the swing. For example, looking at 2012 vs. 2008, the national popular-vote percentages were:

2012 D 51.0, R 47.2, difference = 3.8
2008 D 52.9, R 45.6, difference = 7.3

So while the Democrats won both elections, the Republicans did better in 2012 than in 2008. The swing between the two elections was 3.5 points in favor of the Republicans (R +3.5), or 3.5 points against the Democrats (D –3.5).

Having done this, I then calculated the swing between elections for the state in which each party’s convention was held, and compared it with the national swing that same year. For example, the 2012 Democratic convention was held in Charlotte, N.C. Here are the results for the 2012 and 2008 elections in North Carolina:

2012 D 48.4, R 50.4. diff = –2.0
2008 D 49.7, R 49.4. diff = 0.3

So while the national swing in 2012 was D –3.5, the North Carolina swing was only D –2.3. This means the Democrats did 1.2 points better in North Carolina than they did nationally.

Whether this was because of the convention or some other reason, or just a random statistical fluctuation, is unclear. But for what it’s worth, here are the swings for both parties’ convention states from 1988 to 2012, compared with the national swing:

Republican National Convention

2012 – Tampa, Fla.
National swing R +3.5 / Fla. swing R +1.9
Republican gain/loss: –1.6

2008 – St. Paul, Minn.
National swing R –9.8 / Minn. swing R –6.8
Republican gain/loss: +3.0

2004 – New York, N.Y.
National swing R +3.0 / N.Y. swing R +6.7
Republican gain/loss: +3.7

2000 – Philadelphia, Pa.
National swing R +8.0 / Pa. swing R +5.2
Republican gain/loss: –2.8

1996 – San Diego, Calif.
National swing R –2.7 / Calif. swing R +0.6
Republican gain/loss: +3.3

1992 – Houston, Texas
National swing R –13.4 / Texas swing R –9.1
Republican gain/loss: +4.3

1988 – New Orleans, La.
National swing R –10.5 / La. swing R –12.7
Republican gain/loss: –2.2

Democratic National Convention

2012 – Charlotte, N.C.
National swing D –3.5 / N.C. swing D –2.3
Democratic gain/loss: +1.2

2008 – Denver, Colo.
National swing D +9.8 / Colo. swing D +13.7
Democratic gain/loss: +3.9

2004 – Boston, Mass.
National swing D –3.0 / Mass. swing D –2.2
Democratic gain/loss: +0.8

2000 – Los Angeles, Calif.
National swing D –8.0 / Calif. swing D –1.1
Democratic gain/loss: +6.9

1996 – Chicago, Illinois
National swing D +2.7 / Ill. swing D +3.2
Democratic gain/loss: +0.5

1992 – New York, N.Y.
National swing D +13.4 / N.Y. swing D +11.7
Democratic gain/loss: –​1.7

1988 – Atlanta, Ga.
National swing D +10.5 / Ga. swing D +0.1
Democratic gain/loss: –10.4

My first reaction is that I wish I’d started with 1992 instead of 1988, to get rid of that Georgia outlier. I could still cut the survey off at 1992 and come up with some retroactive rationale for why that’s the natural starting point, but I’ll leave that sort of thing to the climate scientists. Anyway, if you include 1988, the Democrats average a tiny 0.2 percentage-point gain in their convention states; exclude 1988 and it’s an average gain of 1.8 points. For the Republicans, it’s an average gain of 1.2 points (or 1.6 if you exclude 1988).

Looking at the data impressionistically (since this is hardly the sort of large, randomized trial from which one can draw firm statistical conclusions), it seems plausible to suggest that the state you hold your convention in will probably give you a boost of about 1 percentage point in the popular-vote margin. Since Ohio has usually been within 3 or 4 points in recent elections, and since it has gone with the winner in every election from 1964 to 2012, the convention site might have more influence in 2016 than it usually does. And this would be a great time for the GOP to break its convention-site losing streak.