Morning Jay: Special “Ultimate Predictions” Edition!
6:30 AM, Nov 2, 2010 • By JAY COST
1. Intro. I have to say that I am of two minds about this midterm election. On the one hand, it is great to see the Republican party in resurgence. And not just in terms of raw politics: The Tea Party movement has given the GOP a sense of meaning and purpose in the domestic political debate that it has not really had since 1996 when Bill Clinton outfoxed the Republicans in the budget battle.
On the other hand, as somebody who makes a living discerning trendlines before they develop, this has been frustrating. I’m not going to beat around the bush, my friends: we are in uncharted water here. Consider:
(a) The generic ballot average at RealClearPolitics is predicting a Republican margin of 9.4 points. If that were to happen, it would be the largest Republican margin since 1946.
(b) The Gallup poll, historically the most accurate, is an outlier this cycle. What to make of this? If it is correct, and the margin is 15 points today, that would be the largest margin since 1928, when Herbert Hoover did this to Al Smith.
(c) The political landscape has changed so much since the 1920s. Liberal pundits like to talk about how narrow the Republican Party of today is, but actually it has never been geographically broader. The party of Lincoln, McKinley, Teddy Roosevelt, Calvin Coolidge, and Dwight Eisenhower was a regional party, limited mostly to the North. The party of today is a national one. So it’s fair to ask: how far back in time can we go to get a reasonable sense of what happens next? And if the Republican Party’s appeal now is geographically broader, what’s to say a 1928-style election isn't possible?
I mention this up front to tell you that the estimates that follow are my best guesses, but there is a great deal of uncertainty. I’ve done my best to quantify matters, but this is still about 40 percent gut instinct. Keep that in mind.
2. Macro Perspective. For starters, we have to deal with the Gallup generic ballot. It’s showing R +15 right now, higher than any other pollster. But Gallup is also the most accurate historically, while the other polls have historically tended to underestimate Republican strength, often by substantial margins. How to proceed?
I’ve put together a model that predicts the number of seats based on the number of votes. It goes back to 1942, and predicts about 92 percent of the variation in seats every cycle. Models like this have been used in other places, but mine takes into account the annual redistricting process, and it finds the GOP has enjoyed a systematic advantage during the 2000s.
Plug a 15-point Republican victory into this model, and it spits out a Republican gain of about 90 seats. On the other hand, if we take the RealClearPolitics average and plug it into the model, we arrive at a gain of about 70 seats.
I am well aware that these numbers are substantially higher than what just about every other pundit is predicting. And as I noted at the beginning, I am not pleased about this! But look: this is not the “Price Is Right.” The goal here is not to get closest without going over. The goal is to get the closest period. With the caveat that there remains a great deal of uncertainty, here is my take.
Frankly, my gut tells me that Gallup’s R +15 is too high, but then again I think the ABC News/Washington Post poll (R+4) is wider off the mark than Gallup. Mark Blumenthal has noted that there is no real middle in the final generic ballot numbers. They are all either high or low. I’m going high because of the metric’s historic tendency to undercount Republican strength. And while Gallup may be higher than normal this year, it is still the smart bet that Gallup is on the correct side of the ledger.
Recent Blog Posts