November sees not just a presidential race but a general election. So can the Democrats recover from the 2010 debacle?
House Speaker John Boehner caused a stirwhen he declared on Sunday that Republicans only have a two-thirds chance of maintaining control of the House of Representatives. Some took the speaker's words verbatim, while others thought Boehner was just working to ensure donations.
My review of the evidence suggests that the speaker has no real reason to fear losing 25 seats and the House.
Types of analysis
Forecasting the overall seat change for the House of the Representatives can be done three ways: looking at national house polls, district-by-district, or a combination of both.
National polling gives a lot of bang for the buck and picks up early on trends, but it is inexact. District-by-district analysis is more labor-intensive, more difficult in years immediately following redistricting, and doesn't necessarily pick up on overarching trends quickly. But ultimately, it provides us with more information and is more accurate.
Looking at both gives us the best of both worlds. We can pick up on both national and local trends, and we are less likely to miss any important pieces of information.
National polling
Translating national polling to a seat estimate is rather easy. At the most basic level, you need to know the current number of seats the majority party holds and the expected national vote. Some models utilize variables such as presidential approval and popular vote form the previous election.
Simple models such as these capture a key element in House elections: the incumbent party will almost win more seats the generic ballot implies. Because of an inherent incumbent advantage, the party in charge will hold into the House in the case of a tie. In fact, most estimates from 2010 and 2012 indicate that the out-of-power party needs to win the national House vote by about 3 percentage points to win back the House.
The Democrats aren't even close to a 3-point lead yet. The simple Real Clear Politics average gives Republicans a 1.4 percentage point edge, HuffPollster puts it at 2.6 percentage points, and the TPM poll tracker sees the Republican lead at 3.1 percentage points. All those margins would indicate only small single-digit seat gains for the Democrats, or possibly even for the Republicans.
A potential problem with using these estimates, especially HuffPollster and TPM, is that they are heavily weighted by Rasmussen Reports. Rasmussen polls tend to be more Republican-friendly than the average pollster and are produced every week.
One way one potentially to avoid this issue is to create an average that takes into account these "House effects". It's also important to differentiate between registered voter and likely voter polls. With the notable exceptions of 2006 and especially 2008, the voting electorate was more Republican-oriented than the wider registered voter universe.
When we look solely at the likely voter polls since the beginning of February, and control for the influence of Rasmussen polls with a weighted average, the Republican lead is down to about 0.8 points. That's not very large, but that would still only net Democrats gains of somewhere between 5 and 10 seats.
When we utilize the same method with registered voters, we find Democrats up by about 0.8 points. That would point to a Democratic pick-up in the 10-15 seat range, which would be good – but not anywhere near the 25 needed to recapture control of the House.
My hunch is that the House effect-corrected, likely voter polls are closer to the truth at this point. It's important to remember that translating the national House vote to a seat estimate has a margin of error in the neighborhood of 15 to 20 seats. Therefore, even if the polls matched the national vote perfectly and assuming my preference for the likely voter polls, there is still a remote probability – between 1 in 10 to 1 in 25 – that the Democrats could take back the House.
If, of course, the Democrats gain momentum and the polling changes, then these estimates are obviously off.
District-by-district analysis
We have very little polling of the 435 individual districts. We also don't know how voters will react to their new "incumbent" congresspersons because of redistricting. Redistricting generally provided neither the Democrats nor the Republicans with an overall edge in the seat count, but it did shore up Republican incumbents.
One of the best analyses I saw of this phenomenon was produced on the right-leaning, but reliable Red Racing Horses, a conservative counterpart to Daily Kos Elections. The writer Left Coast Libertarian studied how the newly configured congressional seats voted in the 2008 election.
House votes in this nationalized political environment are highly correlated with past presidential results. There were very few Democratic incumbents who won seats in 2010 in districts won by John McCain in 2008. When you take a look at the chart below, you see the number of seats where McCain did better than he did nationally, R+, and number of seats where Obama did better than he did nationally, D+.
The number of highly Republican seats of at least +8R jumped by 17. The number of seats for the Democrats in the same category, at least +8D, has stayed steady. The number of seats that lean Democrat, +1 to +2D, has shrunk from 25 to 13 following redistricting. The number of seats that lean Republican, +1 to +2R, increased by 3.
The number of seats where McCain did better than he did nationally overall increased from 222 to 234. The number of seats where Obama did better than he did nationally decreased from 204 to 192. None of these numbers are very good for the Democrats and suggest that the national seat curve could potentially be underplaying the Republicans' hand.
Another way we can judge the House on the district level is by district-by-district expert analysis. These commentators – Charlie Cook, Stu Rothenberg and Larry Sabato – often have insider access to such information as internal polls and candidate quality. All of them are pointing towards minimal Democratic gains.
Cook currently has 185 safe Republican seats with an additional 24 seats likely to go the Republicans and 19 more leaning towards the Republicans. That is a total of 228 seats at least tending Republican. There are only 149 safe Democratic seats, 17 likely to be won by the Democrats, and 14 leaning towards the Democrats. Thus, only 180 seats in total are at least leaning towards the Democrats. And 22 seats are a toss-up.
Rothenberg gives a wonderfully concise forecast that a Democratic gain in the single digits is "most likely".
Sabato sees 185 safe Republican seats with an additional 25 likely going to the Republicans and 25 more leaning Republican. That is a total of 235 seats at least leaning towards the Republicans, at this point. There are only 149 safe Democratic seats currently, by his calculation, with an additional 19 likely to go to the Republicans and 19 more leaning that way. That is a total of only 187 seats at least leaning towards the Democrats right now, with13 seats a toss-up.
Finally, you might have heard of a "swing district poll" by Greenberg Quinlan Rosner (GQR), which has shown Republicans slipping. I'm not a huge fan of these swing district polls, because often, what is occurring nationally is also happening in the swing district aggregates. Individual districts can often vary with regard to the national tide, but swing districts as a group usually follow the trend. But even if they do, this poll isn't anywhere near as bad for the Republican party as the GQR memo suggests.
Yes, Republican incumbent margins are down about 5 points from 2010 in these swing districts, but they are still up 6 points overall. More importantly, that number has not changed by any statistical significance since September, which means that there is no momentum away from the Republicans.
Democrats would pick up about 10 seats if all the districts polled moved uniformly 5 points. Remember, though, that a number of these districts, such as Blake Farenthold's in Texas, have been dramatically altered by redistricting – and in this case, for instance, has become significantly more safely Republican.
It's also important to point out that Democratic incumbents were in far worse position at about this point in 2010. In a June 2010 GQR swing-district poll, Democratic incumbents were trailing their Republican opponents by 5 percentage points.
So, to make that point clear again, Republicans are currently winning by 6 points in their vulnerable districts, while Democrats were losing in theirs by 5 points at around this point in 2010.
All of these district indicators point to Republicans doing slightly worse in 2012 than they did in 2010, but not by much – and certainly nowhere near the Democratic election debacle.
Overall
My previous research had indicated that large changes in the House of Representatives are quite unusual when the White House and House of Representatives are controlled by different parties. As I wrote last October, "It would seem that in the past 60 years voters have been unwilling to reward or blame either party too greatly when faced with split government. They tend to like the status quo."
All the factors I look at seem to suggest that will hold. A modest Democratic gain of about 5-13 seats at this point seems the most likely scenario.