Unlikely Voter

Poll Analysis and Election Projection

Updating my Senate projection

It’s now been nearly two months since I made a projection of the US Senate elections. Since then we’ve had a number of primaries, television ads, polls, and other developments, so let’s see where we stand now.


Senate right now: 59 D-41 R.

Seats up right now: 19 D-18 R.

Seats therefore not up: 40 D-23 R.

Minimum result needed for majority: -9 for Democrats, +10 for Republicans.


I’m going to work off the polls when they’re available, and make guesses otherwise. That’s usually easy because the seats that aren’t polled are typically safe. Changes from last time will be in boldface.


AL: Richard Shelby: Primaries done. No polling I know of. 99.9% R victory.

AK: Lisa Murkowski: Primaries in progress. Only polling I can find is of a hypothetical Murkowski v Sarah Palin primary. 99.9% R victory.

AZ: John McCain: Primaries in progress. McCain up 15-20 over Glassman. 99% R victory

AR: Blanche Lincoln: Primaries done. John Boozman up 20+ in polling. 99% R victory.

CA: Barbara Boxer: Primaries done. Polling mixed. 45% R victory (from 30).

CO: Michael Bennet: Primaries done. Polling mixed. 55% R victory (from 67).

CT: Chris Dodd (retiring): Primaries done. McMahon making steady gains on Blumenthal. 15% R victory (from 1).

DE: Joe Biden (resigned): Primaries in progress. Mike Castle’s lead down to the teens. 95% R victory (from 99).

FL: Mel Martinez (resigned): Primaries in progress. Polling mixed. 55% R victory (from 30).

GA: Johnny Isakson: Primaries done. Isakson lead in the 20s. 99.9% R victory.

HI: Daniel Inouye: Primaries in progress. Inouye nearing 70 in poll. 0.1% R victory.

ID: Mike Crapo: Primaries done. Crapo lead hovering around 40. 99.9% R victory.

IL: Barack Obama (resigned): Primaries done. Illinois hates both candidates, but polling close. 60% R victory.

IN: Evan Bayh (retiring): Primaries done. Dan Coats lead growing. 99% R victory (from 95).

IA: Chuck Grassley: Primaries done. Grassley rebounds from one bad poll. 99% R victory (from 95).

KS: Sam Brownback (retiring): Primaries done. Republican Jerry Moran leading all polls. 99.9% R victory.

KY: Jim Bunning (retiring). Primaries done. Randal Paul ahead in most polls. 75% R victory (from 70).

LA: David Vitter: Primaries done. Vitter well ahead in most polls, but down to 9% lead in one poll. 90% R victory.

MD: Barbara Mikulski: Primaries in progress. Mikulski near 60 in polling. 0.1% R victory.

MO: Kit Bond (retiring): Primaries done. Roy Blunt maintaining modest lead. 75% R victory (from 70).

NV: Harry Reid: Primaries done. Reid gains slim lead over Angle. 45% R victory (from 70).

NC: Richard Burr: Primaries are done. Polling still volatile, but Burr leads the whole way. 70% R victory.

ND: Byron Dorgan (retiring): Primaries are done. John Hoeven leads by so much he breaks the graphics in my analysis tool. 99.9% R victory.

NH: Judd Gregg (retiring): Primaries in progress. Kelly Ayotte still has led the whole way, but opponent Bill Binnie has surged. Hodes has never led. 80% R victory (from 90).

NY: Chuck Schumer: Primaries in progress. Is this race even contested? A hypothetical poll against Larry Kudlow has Schumer up 45. 0.1% R victory.

NY: Kirsten Gillibrand: Primaries in progress. Gillibrand leads every matchup. 1% R victory.

OH: George Voinovich (retiring): Primaries done. Portman takes lead over Fisher. 70% R victory (from 45).

OK: Tom Coburn: Primaries done. Coburn lead approaching 40. 99.9% R victory.

OR: Ron Wyden: Primaries done. Wyden comfortably ahead. 10% R victory.

PA: Arlen Specter (primaried): Primaries done (especially for Specter). Polls close and volatile with Pat Toomey taking a small edge. 65% R victory.

SC: Jim DeMint: Primaries done. No polling I can find. 99.9% R victory.

SD: John Thune: Primaries done. Unopposed. 100% R victory.

UT: Bob Bennett (primaried): Primaries done. Lee up 30. 99.9% R victory.

VT: Patrick Leahy: Primaries in progress. Leahy up 35. 0.1% R victory.

WA: Patty Murray: Top two primary in progress. Murray keeping small lead on Rossi. 40% R victory.

WV: Robert Byrd (died): Special election in progress. Manchin leads by 16. 1% R victory (from 0)

WI: Russ Feingold: Primaries in progress. Polling mixed. 45% R victory (from 40).

The simulation

100,000 trials using the probabilities above.


A 6 (from 5) seat Republican gain is most likely according to these estimates, with 5 and 7 virtually equally likely after that. Those three outcomes account for 59% (58,661/100,000) of the curve. 2% (1,992/100,000) of the trials gave Republicans a Senate majority. 0.03% (25/100,000) gave Democrats a gain.

So we have a slight gain from last time for Republicans, with 6 seats overtaking 5 as the favorite outcome. The biggest drop for Republicans was in Nevada, but that was more than offset by Florida, Ohio, and other seats. According to the polls, though, a majority for Republicans still seems most unlikely.


5 Responses to “Updating my Senate projection”

  1. Yes, Schumer has a (probable) GOP opponent. I think he’s an ex-CIA or ex-intelligence guy or something. He doesn’t have much of a chance, but at least he’s in it.

  2. Does your simulation of the 6 seat gain being most likely outcome assume that the variables are independent? Is there any reason to assume that?

  3. That certainly makes sense if the only bias in the polls comes from independent sampling discrepancies.

    If you are assuming that then of course your simulation will give you approximately the right probabilities.
    There is however a fairly easy way of doing it in closed form (if I’m safe neglecting the 99% percent ones.) Namely, you expand a product of a lot of terms like (.55 Boxer + .45 Fiorini). If you have say 12 competitive races, then that’s only 4096 outcomes and you can easily write code to calculate the probability of each one and sum up outcomes that lead say to +6 Republican.

    However, I think this has limited value. The real uncertainty is how the races will develop over the next 11 weeks. Since we can’t predict the future, we could view the shifts in each race as being random variables. But they’re far from independent.

    • Well duh the race will change.

      That doesn’t diminish the value of looking at where we are RIGHT NOW, though.

      It’s impossible to predict where we will be if we don’t know where we are.

Write a Comment

Comments are closed.