By Joe Williams and Anthony Salvanto
It's one of the great paradoxes in American politics. Why do Americans say they hate Congress but keep re-electing most of it? Part of that answer, at least this year, is that nationally Americans are split nearly even on their preference for their House candidates; part of it is the relative lack of competitive districts. And a big part of it is redistricting, which really just keep the number of competitive districts about the same as it was before - and that isn't too many.
First, here's a look at the findings of our 2012 House model estimate:
Republicans are widely expected to keep control of the House this year, and our new model simulation confirms that finding as well. We analyzed the partisan makeup of all the districts and ran a computer simulation to see how the nation's votes would fall across all of them, and which seats might tip because of it. The answer: not many.
Short of a massive wave materializing in the final week, Democrats aren't likely to tilt the 25 seats they'd need to retake the majority. The absence of that wave, along with a relatively small number of competitive seats after redistricting, will keep the House reasonably close to its current breakdown. This table shows why:
If the national House vote is 50-50, we expect the new House to be about where it is. Democrats could lose seats, or gain one.
But if the national House vote is 51 to 49 in favor of the Democrats, they still won't gain enough seats to retake the House, they will get perhaps 205 seats, and somewhere between a four seat pickup and - less likely - up as high as 205, with a range midpoint of 201, but that is well short of the 218 needed to take control.
Here's how it works: We start with the expectation - bourn out by past history - that the competitive races are by definition more likely to swing and contribute a lot to whatever national movement we see. We know that the national House vote is just the assemblage of all the votes in all the districts - the only question is how they all move and contribute to that national total.
We next computed the partisan makeup of the new districts from county- and precinct-level past vote as it applies now within the boundaries of all the new districts. Each district is known to have a certain number of swing votes, based on that past data.
Then we use a simple probability algorithm, based on the underlying partisanship and current race competitive ratings, to allocate swing votes between the Democratic and Republican candidates in the competitive districts until we reach the total expected national vote. So the more the national vote swings toward one party or the other, the more swing votes are expected to move within the districts. We then simulate this allocation process thousands of times and report the mean - the most likely outcome - in the table.