Some Things to Watch for in the Contest for the House

[Time for change]

As we all know, pollsters are predicting numerous losses of Democratic House (and Senate) seats. Nate Silver devised a model based on a composite of all available (and what he considers credible) polling and other data to predict the likelihood of each House and Senate seat going one way or the other. While Democrats currently hold a lead of 77 House seats, he is predicting that by the end of today’s election the Republicans will have a lead of 29 – a net loss of 53 Democratic seats and a net gain of 53 Republican seats.

Nevertheless, polling is based on numerous assumptions, some which may or may not be accurate. Therefore, despite the gloomy forecast, there is a possibility that the polls have been biased in favor of Republicans. Silver gives the Democrats a 16% chance of hanging on to a House majority (In the Senate he is predicting a loss of approximately 7 Democratic seats, which would leave the Democrats with a 52-48 majority – and a 7% chance of a Republican takeover of the Senate, assuming that Lieberman does not cross over).


Seats to watch in the House

There are 22 House seats which are rated between a 20% to 40% chance of going Democratic. Of these, 21 are currently held by Democrats. Silver’s model predicts that the Republicans will win approximately 16 of these 22 seats. If they win close to that many seats, or more, that would be a very good indication that the statistical model is reasonably valid, and would be a good sign that the Democrats would have almost no chance of hanging on to the house this year. On the other hand, if the Democrats win approximately half of those seats, that would be a very good sign that the model was biased in favor of Republicans, and that therefore the Democrats would have approximately an even chance of hanging on to the House. If they win substantially more than half of those seats their chances of hanging on to the House should be very good. Those 22 seats are:

20% - 40% likelihood of Democratic win – closing times at 7:00 or 8:00 p.m. Eastern Daylight Savings Time
AL-2
FL-22
FL-25
IN-9
IL-17
OH-16
MS-4
NJ-3
PA-7
PA-8
PA-11
TN-4

Later closing times
AZ-5
CA-11
CO-3
MI-17
NM-2
NY-19
NY-23
OR-5
SD-1
WI-7

Then there are 15 seats for which Democrats are given between a 40% and 60% chance of winning. All of these are currently Democratic seats. Democrats need to win 10 or 11 of these seats to have approximately an even chance of hanging on to the House:

40% to 60% likelihood of Democratic win
CA-20
GA-2
KY-6
IL-14
MA-10
MO-4
NH-2
NY-24
NC-2
NC-7
NC-8
OH-6
TX-23
VA-9
WV-1


Factors associated with likelihood of Democratic wins in the House

Almost all of the competitive House seats this year are currently held by Democrats. Of 83 reasonably competitive seats (with a probability of Democratic victory between 10% and 90%), 78 are currently held by Democrats and only 5 are held by Republicans. In addition, there are 191 seats that are currently rated as 90% or greater likelihood of Republican victory (including 124 with a 99.9% or greater likelihood), which includes 19 current Democratic and 172 current Republican seats). At the other end of the spectrum, 161 seats are given a 90% or better chance of going Democratic, and only 2 of those are currently held by Republicans. The average Democrat is given a 76% chance of hanging on to his/her seat, while the average Republican is given a 98% chance of hanging on.

It would be nice to understand the factors associated with the likelihood of House seats going one way or the other. There are two factors that stand out. One is the Cook Partisan Voting Index (PVI), which is a measure of how the Congressional Districts voted in the last two presidential elections. The PVI alone explains 64% of this year’s House predictions. Another important factor is which Party currently holds the seat, which alone explains 70% of this year’s House predictions. When the two are added together they explain 80% of the House predictions (It is less than the total of the two because the two factors are highly correlated with each other). That means that all other factors combined, including the unique characters of all the candidates, explain only 20% of the predictions independent of the PVI and which Party currently holds the seat.

The importance of the PVI is seen in the fact that the good majority of Democratic Congresspersons who are predicted to lose their House seats this year come from districts with Republican PVIs. Of 66 Democratic Congresspersons considered more likely than not to lose their House seat this year, only 15 are from districts with Democratic PVIs, while 49 are from districts with Republican PVIs.

I also evaluated votes on Obama’s TARP bill, and on his health care bill. Those factors added nothing towards explaining the current polling data. House members from conservative districts were much more likely to vote against the health care reform bill, and they are also much more likely to lose their House seats – mostly because of the conservatism of their district rather than because of a particular vote.


Voter suppression

One thing you never see mentioned in election polling articles is voter suppression – even though it is a major factor in many national elections in the United States these days, and probably accounted for George W. Bush’s election “victories” in both 2000 and 2004.

Information on what kinds of voters actually turn out to vote comes from exit polling. So when pollsters make assumptions about registered Democrats being less likely to vote than Republicans, they base those assumptions largely on exit poll data. But with intensive voter suppression activities, many voters who come out to vote, or who attempt to vote, don’t actually vote and aren’t included in the exit polls either.

All types of dirty tricks are used. Massive organized efforts are made to “challenge” voters in Democratic polling places, thereby slowing down the voting process, causing long voting lines, and causing many voters to leave, either because of time constraints or because of the intimidation itself. In many poor neighborhoods, flyers are passed out with incorrect information on voting times and places, or with threats of imprisonment in order to scare voters away from the polls. This year “Tea Party Patriots” have put out an ad that puts a $500 bounty on “fraudulent voters”. And they also use fraudulent computer programs to make voters ineligible to vote as they did to tens of thousands of Florida voters in 2000 and hundreds of thousands of Ohio voters in 2004.

The vast majority of voter suppression is conducted by Republicans, targeting Democratic voters. And they’re at it again this year, with Tea Party zealots playing a major role:

Numerous reports have documented how state GOP chapters, local Tea Party groups and organizations like Americans for Prosperity are mobilizing across the country – holding training sessions and posting instructional videos on their websites about how to challenge suspicious voters. But the right's concern about widespread voter fraud has virtually no basis in empirical reality; a 2007 study by the Brennan Center for Justice found that "an American is more likely to get struck by lightning than impersonate another voter at the polls."

Thus it is that massive organized efforts by thugs to repress voting end up getting interpreted by pollsters as lack of interest in voting by poor and minority Democratic voters.