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

Likely governing scenarios for 2013

August 29th, 2012, 5:00pm by Sam Wang

Analysis of Presidential and Senate races gives a current long-term outlook for divided government at 34% probability, 1 in 3. Including the House would make the likelihood even higher. A House prediction will come tomorrow.

First, here is a histogram of Senate outcomes given current polls and a shift of up to +/-3% (uniform distribution) between now and November.

Senate control 65-35 D-R

From this histogram, Democrats will retain Senate control with 65% probability, an uptick from my previous estimate of 52%, mainly because of Rep. Akin (R-MO)’s comments on “legitimate rape.” The probability estimate is very rough. Anything between 20% and 80% should be considered a knife edge.

What are the likeliest outcomes? Cases in which President Obama is re-elected are easier to estimate because his current November win probability as 88%. This covers most possible futures, and control of the Senate will be largely a statistically independent event from an Obama re-election.

This leads to the following scenarios, in descending order of likelihood.

Scenario 1: President Obama and a Democratic Senate: 57%. Under current rules requiring 60 votes for many actions, the Senate would be nearly paralyzed. However, Majority Leader Harry Reid has said that reforming the filibuster is a priority for him in the new Congress.

Scenario 2: President Obama and a Republican Senate: 31%. Reid said he’d also be in favor of filibuster reform in this case too. He’d have little say in the matter, though, and no good argument for opposing whatever the new majority wants. Although Obama has recently said that Republicans would be more likely to play ball once he’s re-elected, I am having trouble imagining that. I am having an easier time imagining continued polarization. Could be nasty.

The remaining options involve Romney winning the election. This would require a swing of at least 2.0% from current conditions. It would almost certainly be a close election, despite a silly model from a remote ivory tower.

Scenario 3: President Romney and a Republican Senate: 9%. In this case, Majority Leader McConnell would be under heavy pressure to do away with the filibuster. We would see lots of action, that’s for sure: repeal of Obamacare, continued tax cuts across the board.

Scenario 4: President Romney and a Democratic Senate: 3%. For similar reasons as Scenario 3, I think interest in reforming the filibuster would be greatly reduced.

Scenarios 2 and 4 make a total probability of 34%, or 1 in 3.

And…if Republicans retain the House, that is yet another possible path to divided government.

It appears near-certain that the Republicans will lose House seats. But how many? That is a murky question. Current signs indicate that one party has a slight edge in gaining/retaining control. But which one? I’ll post that tomorrow.

Tags: 2012 Election · House · President · Senate

13 Comments so far ↓

  • Tapen Sinha

    Does your scenario one include 50-50 plus VP tie-breker?

    Once elected, it is hard to knock people off the Senate. It is amazing what happened in 2010.

    I cannot see how the Republicans could lose majority in the House – Akin notwithstanding.


  • Matt McIrvin

    I am sort of amazed that the probability of Democrats retaining the Senate could be that high.

  • Alan Cobo-Lewis

    If you assume that presidential election and senate control are independent events, then P(Dem Senate | Obama wins) should equal P(Dem Senate | Romney wins). But you’re proposing P(Dem Senate | Obama wins) = 57/88 = 0.65 but P(Dem Senate | Romney wins) = 3/12 = 0.24. That’s not consistent with statistical independence.

    Is it possible that the estimated probabilities on Scenario 3 & 4 should be flipped? That would make the probability of divided pres/sen about 40%.

    Alternatively, have you assumed statistical non-independence (that polls might move uniformly toward Dems, or toward Republicans, up and down the ballot between now and election day)?

    • Sam Wang

      Yes, I did the latter. Opinion will move by an unknown amount in the coming ~70 days. If Romney wins that gives us a prior that things moved 2 points or more in his direction.

      My general approach is to use the Presidential race as an anchoring prior, on the grounds that it is based on more data and is validated from 2004 and 2008. When later I calculate combined P(Senate and House | Romney) scenarios, I will do the same. Obama/Senate/House does not require this if P(Obama) ~ 0.9 persists.

      Further comment welcome.

  • Sam Champion


    Please explain your comment that Dems would control House and Senate if election were held today.

  • Olav Grinde

    @Sam: “If the election were today, Democrats would control both houses of Congress.”

    Dr Wang, is that conjecture based on your meta-analysis of palls in all the House and Senate races?

    In the Senate, does this mean you expect Heitkamp, Tester, Baldwin, Berkley and/or Donnelly to win? What about Warren, who is trailing in Massachusetts? Or Mark Clayton in Tennessee, who although ostensibly a Democrat is something of a rogue candidate?

    If my count is correct, there are 102 House seats being contested in this election, and the Democrats have to win 48 more seats than the Republicans to regain control. That’s a score of 75–27. Isn’t that exceedingly optimistic? Do current polls support such optimism?

    • Sam Wang

      My calculation is based on August (some July) polls using median-based statistics. The methods are transparent. If you find them optimistic, this says your outlook does not reflect the available data.

      Baldwin and Clayton are not competitive at this time. See a past post for Senate probabilities. I do not have that automated.

      Large swings in the House are not surprising. The 2012 election was a 15-point swing in the national vote. It was a swing back from a comparable Democratic majority. Swings back and forth are also not surprising.

  • Olav Grinde

    Thank you! I very much look forward to seeing your Meta-Analysis of the Senate and House elections. I understand this is coming soon.

    • Sam Wang

      I am probably not going to post details each time. If you wish to inspect the Senate data, I treat as decided (for now) 47 D/I, 44 R. Races that are analyzed in detail (i.e. medians and probabilities calculated) are CT, IN, MA, MO, MT, ND, VA, and WI. MO and WI are not competitive at present, but one never knows.

      Current Senate margins, obtained from RealClearPolitics, are as follows. A positive margin indicates a Democratic lead. No pollster is used more than once. It’s in MATLAB but I imagine you can decipher it. More recent polls are listed first.

      margins_ct=[-3 4 -3 8]; %
      margins_in=[-2 0]; %
      margins_ma=[-5 2]; %
      margins_nd=[6 -9]; % includes DFM 7/24-26
      margins_va=[0 0 2]; %
      margins_wi=[-6 -9 -5 -11]; %
      margins_mt=[-4 5 -3]; %
      margins_mo=[9 10 -1]; %

      The House margin can be seen by eye in the “House outlook” post when high-noise or high-bias polls are excluded. The interface allows you to play with this. Or you can calculate the median, a more tedious procedure that yields the same result. Most of what I have to say about House control is in that post. The House is a case where I think a race-by-race analysis to fill in missing data might be of some use. However, I do not know which hobbyists will be doing that this year. If they are reading this, I welcome their response.

  • Frank

    Sam, what do you think the odds are of a 50-50 Senate?

  • Karen Andres

    I feel that you are the third person in the bed*, I am checking your site so often.

    *4th if you count the dog

Leave a Comment