Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

The risk to the GOP’s majority in the House

October 8th, 2013, 7:03am by Sam Wang

In the current shutdown, John Boehner might be acting out of fear of losing his position. If he doesn’t appease the hardliners who are willing to take the government and economy over the brink for their goals, he could be ejected from the Speaker’s seat. However, is there some chance that he’ll lose his position another way – by Republicans losing the House majority in January 2015?

Usually, the President’s party loses House seats at the midterm election. The normal expectation would be for Republicans to gain seats in 2014. But exceptions can happen, for instance in 1998, when Democrats picked up five seats. That came after a rough few years during which Speaker Newt Gingrich led the way to the last substantial shutdown.

On the other side of the coin, House Republicans enjoy an exceptional advantage in the form of gerrymandered districts. In the 2012 elections, Democrats won the national popular vote by 1.5%, but they needed a 7.3% margin to take control. So broadly speaking, opinion would have to swing by about 6% or more for control of the House to become competitive.

Where is public sentiment today? Before the shutdown, three generic Congressional preference polls taken Sept. 23-29 (Quinnipiac, Rasmussen, PPP) show an average of Democrats +6.0+/-1.5%. That’s a swing of less than 5%…before the shutdown.

A provocative set of district-level polls was conducted for MoveOn by PPP. These are partisan organizations, but I note that of major pollsters, PPP had the best accuracy in 2012. Also, even the worst house biases are no more than three percent. So keep that in mind.

PPP surveyed 24 Congressional districts currently held by Republicans. They asked voters to choose between their current representative and a generic Democrat (data as PDF). Here are the margins they got, plotted against last November’s election result:

The swing was toward Democrats for 23 races (below the red diagonal) and toward the Republican for 1 race (above the diagonal). The key piece of information is the gray zone. If more than half the points are in that gray zone, then that predicts a swing of >6% and a Democratic takeover. Currently, 17 out of 24 points are in the gray zone.

Individually, the district-by-district swing is quite variable, +4% to -23% (where + indicates a swing toward Republicans). But the average is clear, -10.9+/-1.5% (mean+/-SEM). That predicts a national popular-vote margin of D+12.0%.

Since the election is over a year away, it is hard to predict how this will translate to future seat gain/loss. If the election were held today, Democrats would pick up around 30 seats, giving them control of the chamber. I do not expect this to happen. Many things will happen in the coming 12 months, and the current crisis might be a distant memory. But at this point I do expect Democrats to pick up seats next year, an exception to the midterm rule.

Note that in these calculations I did not even include the worst of the news for Republicans. In a followup series of questions, PPP then told respondents that their representative voted for the shutdown. At that point, the average swing moved a further 3.1% toward Democrats, and 22 out of 24 points were in the gray zone. That would be more like a 50-seat gain for Democrats – equivalent to a wave election. An analyst would have to be crazy to predict that! However, it seems like mandatory information for a Democratic campaign strategist – or any Republican incumbent who won by less than 20 points in 2012.

Tags: 2014 Election · House

55 Comments so far ↓

  • adrianannette

    My new Poblano! Thank you- this is exactly what I have been looking for.

  • E L

    “In the current shutdown, John Boehner might be acting out of fear of losing his position.” I somewhat disagree. First, who would become Speaker if Boehner goes? I don’t see any candidate. Of course, Ted Cruz might try.

    Second, I think Boehner really fears cracking the House Republicans and, then, the national Republican party into two permanent warring factions and then into two permanent electoral parties running against each other in most red states. Such a rupture would condemn the national Republican party to dissolve two fractured national parties, neither one of which is strong enough to win most election.

    What’s happening to McConnell in his primary is a forewarning to Boehner of where the Republican party could be headed. Boehner doesn’t want to be known as the Speaker who destroyed the Republican party forever by pushing its destruction in a House vote.

    • Sam Wang

      Well, that is all true too. About every 50 years or so, one party or the other shifts in character – or disappears entirely. You have described the larger dynamic quite well.

    • me


      As for Boehner, at some point he’s going to have to decide whether party or country comes first.

      However, the GOP adopted the right wing agenda with or without Boehner. He’s obviously powerless.

      We’ll know the fracture happened when more than a trickle of GOP reps begin to convert to Independents or to [aghast] the Democratic Party.

    • Olav Grinde

      “…who would become Speaker if Boehner goes? I don’t see any candidate. Of course, Ted Cruz might try.”

      Ted Cruz is a Senator. It’s highly unlikely he could become Speaker of the House. ;)

    • Claire

      Ted Cruz can’t be Speaker of the House – he’s a senator.

    • Sam Wang

      That is not true. The Constitution is silent on the issue of whether the Speaker must be a member of the House.

    • E L

      I brought up Ted Cruz as Speaker because he’s been the leader of the 30 member Tea Party caucus in the House (Yes, the House.). He has had frequent meeting with their caucus to formulate strategy. Part of Boehner’s nightmare is this unprecedented interference in House affairs by two Senators, Cruz and Lee. Normally, McConnell, the minority leader, would restrain them, but he’s helpless with a Tea Party opponent in the KY senate primary who has embraced Cruz and Lee.

  • MAT


    Are there structural differences in the turnout for off year elections that would change this analysis, being as you are using 2012 (a Presidential year) as a baseline? In other words, is a likely voter in 2014 different than a likely voter in 2012?

    • Sam Wang

      MAT – Yes, midterm voter intensity is harder to predict. For instance, pre-election polls in 2010 underestimated the size of the Republican wave by, I forget, something like 10-20 seats.

  • mediaglyphic

    thanks for this Prof Wang. Sabato and company published this a while back

    There were 25 seats that the President did well in but still went Red for congress. I saw someone else make the point that Dems are not getting good candidates. I know that OFA has lots of money but apparantley the DNC is bleeding. I think we need execution to take advantage of this

  • Patrick

    I’ve wondered for a long time about turnout by district. I’d expect that blowout districts, those with very lopsided votes, to have low turnout. As far as I can tell that is true. But there also seems to be some sign that R districts get lower turnout than D districts.
    This could just be that the smaller voting sample in low turnout districts reflects the district as well as in high turnout districts. But it could also be that D voters in R districts (and R voters in D districts) are discouraged voters. They are voters who don’t believe they can make a difference.
    So some data oriented questions, for data I can’t find on the web:
    If you plot the fraction of eligible/registered voters who voted (turnout) by vote share, is this symmetric around the 50% mark?
    If not, where are the districts that are shifted to low turnout that also have close races (races where vote share to the parties is close to 50%)?
    Are the low turnout districts regionally clustered?
    Do they have expensive media markets?
    Are the low turnout districts getting worse in their turnout? Ie do voter discouragement efforts work? It would probably be better to ask that at the Count rather than district level since counties are stable and districts have shuffled.

    My big question is: Can focused effort on voter engagement in selected low turnout districts make a difference?
    Republicans seem to believe that this is true. If they thought their districts really were secure, why expend so much effort on voter discouragement?

    • Sam Wang

      In an off-year, typical turnout patterns favor Republicans. It will be interesting to see which way things go next year. At this point Democrats seem likely to be fired up. But so is the Republican base.

      Certainly a major action item for both sides is to focus on districts whose races are likely to be within 10-15 points. The question is figuring out which races those are.

    • greensleeves


      “The question is figuring out which races those are.”

      Ask Emily’s List. They seem to know & are already working on developing candidates for them.

    • MAT

      If you want to use NC as a test case, there is a good amount of data on the State Board of Elections website on turnout by party affiliation. Here is a link to turnout per Congressional District:

  • Dave Kliman

    I fear that new voter suppression laws and the consolidation of the GOP citizens united Financial advantage will set the bar even higher for the Democrats. The GOP is pulling out all the stops in the name of power.

    • wjca

      The thing about Citizens United is, it gives more influence to big businesses. And the big businesses are looking at the current behavior of the House GOP and saying “Wait a minute! This isn’t what we were signing up for!”

      They were, of course, but like anyone who fails to read the small print, they probably didn’t realize what they were getting. And now that they know, their influence may no longer be running towards the Republicans; certainly not as strongly.

  • Olav Grinde

    Dr Wang, I have been eagerly awaiting your take on this! Just posted to Politico, with a link to your article today.

  • Amitabh Lath

    The consensus of the polls does seem to indicate a drift towards blue, but with large uncertainties. Would it be possible to Monte Carlo this? Assign a mean and sigma to each district, and run 10k psuedoelections?

    • Sam Wang

      I don’t think a Monte Carlo simulation is needed. We can formulate it like the Presidential meta-analysis from last year, as follows.

      The calculation I have given produces a mean and SEM which gives us a sharp view of the current national popular sentiment, D+12.0%. The pre-shutdown number was D+5.0%. That gives two points that can help us guess the distribution of popular-vote margin on Election Day 2014.

      Let’s then say the Election Day 2014 popular vote drifts from today’s conditions: mu=12%, and use sigma=5% since it’s such an uncertain situation. Using either a t-distribution (3 d.f.) to allow long-tailed outcomes, the probability of a Democratic takeover is 79%. If we assume mu=9%, which corresponds to assuming that things are extreme today and will subside, then we get a probability of 62%.

      That is totally counter to historical midterm patterns, so consider this as provisional for now.

  • William

    Hey — why isn’t the NYT getting you as a replacement for Nate Silver? Because I would refresh every hour or so if they did!

  • Another Holocene Human

    Sam, with all due respect, I don’t believe the Republicans are fired up.

    They were manic before November 2012. After election day, they were angry. Since the immigration reform battle, they’ve been silent. The moderate Republicans are afraid for the future of their party, while the more radical ones are quiet, or focusing on positive stuff in their lives instead of politics.

    They’ve been burned. (Whether this lasts is an open question.)

  • Bruce Thompson

    A couple of questions:
    1. In the calculation of the total number of Democrats vs. Republican votes in a Congressional election, what is done with districts where one or the other party does not field a candidate? In such cases, the total for that party would be at or near zero. Is some estimate made of the likely vote if there were a candidate?
    2. Is there some estimate of what per cent of Tea Party members get social security checks? Not raising the debt limit would seem to directly strike at this part of the Tea Party if these checks are delayed–and much worse if the Republican proposal to prioritize bondholders were adopted.

  • Avattoir

    Olav Grinde –

    1. The Constitution requires that the House “shall chuse [sic] their Speaker”.

    2. By custom, the Speaker has always been a member of the House.

    3. The Constitution does NOT REQUIRE the House “chuse”, or choose (or chews) their Speaker from the House, Congress, elected officials, U.S. citizens, humanity, or this planet.

    4. Of those categories, Senator Cruz qualifies under at least b), c) and f).

    5. When the Democratic party ceded control of the South to regional white demagogues, its reward was overwhelming, seemingly perpetual control of Congress.

    6. Under those conditions, it made sense for Republicans to strive to concentrate power in the executive, and by that also in the courts.

    7. In such circumstances, an ambitious young white Republican male might more reasonably harbor dreams of gaining the presidency, or failing that, a seat in court.

    8. When the Democratic party ceased accommodating southern white supremacists in favor of pursuing a broader coalition with an eye to the future demographics of the electorate, it made sense, at least for a time, for the Republican party to respond with its Southern strategy.

    9. Going with its Southern strategy inevitably reduces to what now empowers the Republican party in Washington, D.C.:
    a) big money to buy control of states,
    b) gerrymandering,
    c) vote suppression,
    d) propaganda,
    e) obstruction and
    f) brinkmanship.

    1o. Under these conditions, an ambitious young white Republican male might far more reasonably harbor dreams of achieving the Speakership of the House, or failing that, as a lobbyist or political newsertainer.

    • John Parenteau

      So, yer sayin’ they may “Chuse Cruz?” Sorry, I couldn’t help myself.

    • Olav Grinde

      Avattoir, thank you for your correction and clarifications! I was not aware of this Constitutional fine point.

  • Keith

    Hello Prof. Wang do you have any sense what the intensity of polling will be for the midterms?

  • rd

    Not sure if this switching between swings and margins is just confusing me, or confusing Sam.

    “House Republicans enjoy an exceptional advantage in the form of gerrymandered districts. In the 2012 elections, Democrats won the national popular vote by 1.5%, but they needed a 7.3% margin to take control. So broadly speaking, opinion would have to swing by about 6% or more for control of the House to become competitive”

    1.7% 2012 Dem margin means 50.85 to 49.15
    7.3% Dem margin to win means 53.65 to 46.35. That means that the Dems would need a swing of an extra 2.8% – that is, a 2.8% swing changes the margin by 5.6%. A swing like that sounds less dramatic.

    • Olav Grinde

      Sam, yes, I was wondering about this too.
      Could you please clarify?

    • Sam Wang

      “Swing” is defined as a change in the margin. The 2012 vote, which was D+1.3%, would have to change, uniformly across all districts, to D+7.3% to make the House competitive. That involves 3.0% of Republican-voting people changing their minds. It could also involve changes in Democratic or Republican turnout, in which case a change in 1% of voters leads to a 1% swing.

  • Tlaloc

    Prof. Wang,
    Is there any kind of good rule-of-thumb for comparing questions about generic politicians to performance of actual politicians? For instance do 95% of actual pols perform within, say, 4% of the generic, or something of that nature? Obviously the charisma and discipline of a politician can play a big role in how they perform (Akin, Murdock being the two most recent examples).

    • Olav Grinde

      Seems to me that this is going to be incredibly skewed by gerrymandering, which aims to amply demographic variations far and beyond the mean.

    • Sam Wang

      That is a very good point. Good catch, Olav.

      Of the 24 districts surveyed by MoveOn, 11 8 are from partisan-gerrymander states (FL/MI/OH/PA/VA, swing toward Democrats, 13.9+/-8.9% 14.6+/-9.6% SD) and 13 15 are not (swing toward Democrats of 8.5+/-5.2% 8.3+/-4.9%). Basically, swings are magnified in the gerrymandered districts, which is interesting, perhaps not unexpected.

      Assuming the latter districts reflect national sentiment, the actual preference right now is D+9.8% D+9.6%, not D+12.0%. That might make the gain in Democratic seats smaller, but it’s still over the magic D+7.3% threshold.

      However, there is a second problem. The seats-votes curve is also not smooth because of gerrymandering, which adds a further layer of complication. In fact, there’s a visible hump in the curve which I never got around to publishing. Clearly this calls for a more sophisticated analysis. I’m on it!

  • Amitabh Lath

    Sam, thanks for the quick calculation. A sigma of 5% is conservative enough, but I don’t think we can apply mean shift of D+12% uniformly (can we?)

    Your plot above shows the mean drifting by 5% to 20% for different districts.

    • Sam Wang

      That is true, there has to be a way of dealing with that. My inclination would be to figure out its approximate distribution. Its kurtosis is 1.96, so it’s actually narrower than a Gaussian, whose kurtosis is 3. Then fold that into the model.

      If we assume Gaussian then it all adds in quadrature, which is nice and easy. I am lazy that way, I want to do the whole model with a pencil, paper, and a MATLAB command line.

    • Amitabh Lath

      Olav points out gerrymandering, which does shift the means of some districts beyond 5sigma. No amount of mean shift is going to help in those districts.

      The polls sample D and R in a certain ratio. Historically, the turnout in midterms is skewed >>R. And among R’s the GOP’s tactics are actually approved of.

      Also, your post above is now up on Andrew Sullivan’s Daily Dish blog.

    • Sam Wang

      No, it’s the other way around. The swing is *larger* in gerrymandered districts. Perhaps contrary to the intuition you are expressing, I think it is the “packed” (i.e. Democrat-heavy) districts that are impervious to swings.

      The extra districts that Republicans have squeezed out are all close. Based on the MoveOn data, they have larger swings. Perhaps they have barely enough Republicans to get over the top, and more independents. Frankly, I have to build my own intuitions here.

  • John Parenteau

    My crystal ball says: 1. Voters’ memory is short, this effect I would expect to fade before the next election, and once it does we are unlikely to see a repeat close to the 2014 election. But 2. If the Tea Party presses the issue and a repeat does occur, it seems to me the Republicans very much resemble the 1850’s Whig Party which was torn apart by Radicals. (The Tea Party members are, after all Radicals, not Conservatives.) The Republicans may well split.

  • Kalee

    It’s hard to say this far in advance, but the time the next election rolls around things could be completely different again.

  • Ken

    One way or the other I cannot see Boehner being Speaker of the House next term.
    Either the D’s sweep the election and Nancy is back or the R’s hang on and the tea party elects one of its own to take the gavel.
    A campaign against Boehner based upon his likely resignation and need for special election after he is relegated to the back benches ( ala Newt) coupled with other campaign issues may actually create a vulnerability for him in his district.
    The vulnerability may come from the right ( most likely) or the left.
    If his business constituency reaches a disgust level with his failure to prevent economic choas by allowing shut down and possible default he has no back up plan.
    The ideologues see him as the sort of “professional” D.C. pol they blame for everything and the left won’t bail him out like they did Murkowski in Alaska.
    Election season for him won’t be any fun at all.

  • mediaglyphic

    Dr. Wang do we not also have to deal with the 25 or so seats that the President won and Dems lost in congress. There seem to be other variables at play (not sure if its candidate quality or some other impact). I haven’t seen a list of districts that Romney won and the Dems won in congress. I wonder what that number is (is it symmetrical?)

    • Sam Wang

      Agreed. It is important to examine the composition of individual districts. If you look at the MoveOn data, you may notice that the largest swings are from partisan-gerrymandered districts. There is more to say on that. I don’t know the answer to your question but I bet people like Dave Wasserman do.

  • Mark Jamison

    Professor, have you factored in California’s new electoral system of jungle primaries? Unlike gerrymandered districts like mine here in NC it’s hard to primary someone from the extreme in the California system.

  • Amitabh Lath

    “The swing is *larger* in gerrymandered districts. ”

    Wow. Yes, this is completely counter-intuitive for me. I have to change my mental image of these highly R districts.

    Certainly the MoveOn data shows that even districts like NV03 and OH06 disapprove of the shutdown (more than they disapprove of govt default, apparently).

    If it was just due to small sample size then you would expect a scatter, and they all seem to be leaning +D. So it is either real, or some systematic effect in the poll.

    Could this be due to overweighting D? If this is real, what is (are) the mechanism(s)?

    • CMeier

      > “The swing is *larger* in gerrymandered districts. ”

      > Wow. Yes, this is completely counter-intuitive for me. I have to change my mental image of these highly R districts.

      This seems perfectly intuitive to me. Gerrymandering involves stuffing a few districts with large numbers of your opponents while creating many districts with a bare majority of your own kind. Those districts are really only safe in the average election. Only you need to do to lose big, however, is to upset a smallish number of voters in these safe districts who would usually vote for you. In a district that is heavily packed with your opponents, there aren’t many voters to swing, so you will see a lower percentage swinging A “safe” district, however, with a bare majority of your voters will have a larger group of voters who could potentially swing to the other side.

    • Amitabh Lath

      Cmeir, you are correct, and Sam has just added a post to this effect. I was assuming that Republicans would have to change their minds for these districts to be vulnerable, and that was not likely. However, the gerrymandered districts depend on Independents, who are prone to be angry about the current unpleasantness.

    • mediaglyphic

      we need to create a finite element model with each voter having three or four dimensions. Wouldn’t even need a quantum computer to model it!!

  • Dr Mark

    First, the GOP legislatures got greedy and created too many “close” districts. Moreover, as we move farther through the decade, migration and young reaching voting age (others dying) always dilutes the original advantages of gerrymandering (especially the waves of Hispanic migrants). Finally, everyone is wrong about this being way too long before the 2014 elections to affect outcomes — it’s just long enough to find and fund Democratic challengers!

  • Roger Paul

    I will remember the shutdown, I blame the Republicans, and next election I will be doing everything I can to get those jerks out of office. It seems like they have decided to repeal Obama care now instead of waiting till next election and letting Americans decided for themselves.

Leave a Comment