One always hesitates to get way out in front of the crowd or in this case the other political analysts. So far none of the other analysts have been willing to go above a 57 seat gain for the Republicans or a 235 to 200 House in favor of the Republicans. So where do we get off projecting an 89 net seat gain for them?
Unlike other analysts, we have been using the results of the likely voter generic congressional ballots as developed by Gallup over the last 60 years for midterm elections. When you have 15 data points like this, you can apply a regression analysis to develop some mathematical equations that can allow you to predict how many seats the Republicans will wind up taking from the Democrats.
Never before in the last 60 years have the Republicans been able to get a spread of more than 7 points in their favor and that was in 1994. Right now the Real Clear Politics average of the generic ballot is 7.3 points in their favor and the Republicans picked up a net 51 seats when it was 7. This already says that the Republicans should do better than a 51 seat gain.
Remember that Gallup is the pollster that has data going back 60 years so they have all that experience in developing a screen to pick out the likely voters from a sample of registered voters. This year Gallup is offering two types of likely voter screens depending upon how many voters actually do turn out to vote. The spread in favor of the Republicans is a whopping 9 to 14 points.
If a 7 point spread got the Republicans a net gain of 51 seats, what would 9 points get them let alone 14? A simple proportional analysis says that 9 is 1.29 times greater than 7 so 51 X 1.57 = 65 seats. On the other hand, 14 is 2 times greater than 7 so 51 X 2 = 102 seats. While this kind of progression is way too simple and way outside any existing data points, it does tell you that our projection of an 89 net seat pickup for the Republicans does not seem so outrageous.
So why do we pick 89 seats instead of 65 or 102? We use results from previous equivalent elections as a starting point and then add in an adjustment for the Obamacare votes and poll results. Actually we project that the Republicans will pick up 91 seats from the Democrats but will lose 3 of their own (DE-AL, IL-10 and LA-3). On the other hand they will gain back their now vacant seat (IN-3) so 178 + 91 - 3 + 1 = 267. This gain means the final generic ballot spread will be 12.5 points in favor of the Republicans.
We go further and try to project which seats will be among the 91 seats the Democrats will lose at Who's Winning the House? As you can see from that table, the 91st seat is Carolyn McCarthy in NY-4. An October poll from McLaughlin and Associates has her up 1 point over the Republican Becker, 46 to 45. That 1 point can be made up in the 8 days remaining before the election. The 92nd seat is Ben Chandler in KY-6 and he is ahead by 4 points in an October poll from Mason-Dixon, 48 to 44. While it is possible for the Republicans to pick up this seat since Chandler is polling under 50%, it is not an odds on bet.
Likewise, Tierney in MA-6 is up 6 and Braley is up 11 in IA-1. On the other hand, in the 95th seat, there is a tie in an August poll from Action Solutions, but that is an old poll so who knows where that race is today. At best, it could make up for one of the ones listed in the first 91 that the Republicans end up losing. The point is that the pickings are pretty slim below that 91st seat so that is why we stopped there.