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

Home stretch activism: two Senate races

October 27th, 2012, 11:15pm by Sam Wang

With nine days left to the national election, many races have fallen into place. Where should activists put their effort and money?

As longtime readers know, I believe that a practical application of poll aggregation is to determine where one should apply one’s efforts: races that are on a knife’s edge (20-80% win probability). My attitude was shaped by the extremely close Presidential race of 2004, in which George W. Bush and John Kerry traded the lead several times during the campaign. That year, several Presidential swing states were essential.

In 2012, I direct your focus downticket, to Senate and House races. The question is this: Given the efforts that are already under way, what will have the largest marginal effect on the 2013 political scene? For example, in the House of Representatives I advocate get-out-the-vote activity in knife-edge districts (“House: prediction update and GOTV advice,” Oct. 24).

Tonight I highlight three areas to donate. A final blast may make all the difference – to either side. For Democrats, the ActBlue site at left. For Republicans, Crossroads GPS.

1. North Dakota (Heitkamp v. Berg). North Dakota is strongly Republican, but also has an individualist style of politics somewhat separate from the national scene. Its economy is booming, and there has been an influx of new residents. Heidi Heitkamp (D) and Rick Berg (R) have been neck and neck all season. Close observers think that Heitkamp has run a very strong campaign. October polls ( show a tie (median margin +0.0 +/-  3.0%, n=5).

Heitkamp has been outspent by Berg, who now has $1.2 million cash on hand compared with Heitkamp’s $0.6 million. Conservative groups American Crossroads and Crossroads GPS have paid for ads against Heitkamp. This is an area where home-stretch donations can make a difference.

2. Montana (Tester v. Rehberg). The dirt farmer Senator, Jon Tester (D), has been battling a strong challenger, six-term Congressman Rep. Denny Rehberg (R). In their last debate, the Affordable Care Act (Obamacare) was a main point of contention for both candidates.  Rehberg, a climate change denialist, is a member of the House Science Committee. October polls ( show Tester +1.3 +/- 0.7% (n=3).

3. House races. As I’ve written, Republicans are highly likely to lose seats this election, endangering their majority. Conditions are fluid, and national organizations like the DCCC and NRCC are deploying resources where they will make a difference. The appearance of massive PAC spending highlights the need to support party-based efforts.


Finally, a note on the races I have not mentioned. My examination of other close Senate races suggests that in a number of them, one candidate has taken a clear (if sometimes small) lead. For this reason I am de-emphasizing them. I will give an overview of Senate conditions in the coming days, an update of my most recent assessment.

Tags: 2012 Election · House · Senate

78 Comments so far ↓

  • Brian

    Totally off-topic, but I would really, deeply appreciate an informed opinion on this article:

    It’s deeply disturbing, if true, but I’m not in a position to evaluate the claims being made.

    • Sam Wang

      I think it is crazy. Matt McIrvin and Froggy have addressed it in a satisfactory fashion. I might have more to say, but that’s my basic reaction.

    • Erin O'Daly

      Thank you very much for saying that, Dr. Wang. I would love to see a more in-depth analysis of exactly why it is crazy, though. With electronic voting machines that leave no paper trail, I worry that these records may be manipulated without a trace. Some are concerned this already happened in 2004 in Ohio and in the 2012 primaries. I would very much like to believe that rigging elections is beyond the pale for either side, and a thorough debunking of the linked article would be welcome.

    • Olav Grinde

      Although it seems possible there may be the occasional vote flipping, it seems to me that vote rigging on this scale is exceedingly unlikely. I would be astonished if it were true.

      A key question that I have: What would be the telltale signs of vote flipping? My question was much better formulate by Futurist in his comment yesterday:

    • Amitabh Lath

      Looking at it from the point of view of someone who might want to tamper with voting machines can be instructive. How would you do it and not get caught? How many people involved? Would all of them stay quiet?

      The code that goes into these machines is pretty heavily scrutinized. Something obvious would be caught right away (and any subroutine or function or method you are thinking of right now is obvious).

      Anything that turns a D to an R at some random fractional rate would have to know how to tell a D from an R (not as simple as it sounds). Also, the extra CPU cycles it takes to make this decision would be a red flag.

      There are other checks. I do not know exactly, but something like a random pattern test where you send in a string of D’s, R’s, I’s etc. and check that it comes out exactly. Somewhat like the chkdsk utility. There are probably other, more intensive tests. Computer scientists can look at core dumps of the machine state and read it like Sam could an fMRI of your cerebral cortex.

      Finally, there are the post-voting statistical tests.
      These can be quite powerful, your tampering would have to be small enough not to trigger these tests. You wouldn’t know a-priori how small.

      The punishment for being associated with vote tampering on this scale would be immense. I do not see cautious rich people like the Romneys having the guts.

  • HWAV

    I have demographic polling data from North Dakota Senate. There are only really 3 polls (of 14) that released crosstabs, but I entered gender and party affiliation when available.

    There isn’t enough data for the sub-demographics so some of the graphs look pretty screwy, but you can mess around with the technique to get decent results.

    There is one poll from Essman/Research that seems to be an outlier that sort of skews the local regression technique.

    I’m not up to date on MT-Sen, yet.

    • HWAV

      I’d like to add…

      I think Indiana Senate would be the “other” close race, but I think that changed after Mourdock’s comments about rape during the final debate.

      Akin is/was absolutely getting destroyed among females (down about 10% among females).

      There really isn’t enough polling from Indiana Senate to draw any conclusions yet, but if Akin’s slide is any indication, I think Donnelly gets the victory.

    • 538 Refugee

      I’m not sure what it is about male Republican senate candidates that seems to make them so preoccupied with pregnancy by rape. I’m not sure I really want to know what is going through their minds but find myself hoping more of them speak out. ;)

      On a more serious note I’m sure these ‘gentlemen’ will be glad to help pay the quarter of a million dollars it will take to raise “God’s will” and put her/him through college.

  • Logan

    Keep an eye on the race in Nebraska. Two recent polls show Bob Kerrey within reach. No one really thought that would be the case. But perhaps at the end, he’s closing well. I’d keep an eye on it.

  • Brian

    Thank you, Sam. For what it’s worth, I did check Snopes, et al, before bringing it here.

  • Show Your Work

    Thanks so much Sam for all of your continued work and insightful commentary.

    I hope this storm proves but a glancing blow and that you can safely enjoy watching your predictions be validated!

  • Arsalan

    You should create a graph showing how steeply the hits that this site gets fall after the election is over.

    • Ken Dogson

      I had a student (HS) once submit “The Growth of a Mountain Over Time” as an example of how to make a line graph.

      Very funny stuff.

  • Lupin

    I was wondering how you reconciled you analysis with the recent number-crunching by Red State’s expert:

    I am not a mathematician; yet both approaches seem convincing, yet reach diametrically opposed conclusions.

    • KariQ

      I’m not a mathematician either, but I can see a few problems with article after a brief review, but the most crucial one is that he is relying on only two pollsters to prove his claim that Republicans will turnout in huge numbers relative to Democrats this year: Rasmussen and Gallup. Not coincidentally, this means he has to completely ignore or discount every other pollster that has reviewed voter’s partisan identification, all of which find a significant Democratic advantage:

      Of course, he’s correct that if it turns out that Rasmussen and Gallup – and only Rasmussen and Gallup – have it right, then Obama’s going to lose. But how likely is it that two pollsters with rather poor recent track records have it right this time, while the pollsters that have done better in 2008 and 2010 are suddenly completely wrong?

    • Ed

      I just scanned the link so I can’t post an in-depth reply. However, much of his argument seems to be based on his conclusion that Romney will win Ohio. But Obama doesn’t need Ohio to win re-election. He has many paths to victory. Winning Ohio is the easiest path but not the only one.

    • 538 Refugee

      Lupin, this is an easy one. is cherry picking results they WANT to be true and ignoring the rest. Keep in mind this is obviously a site that is there to support a point of view. I don’t know if their claim that Gallup and Rassmusen are the two largest polling firms is correct or not but that is what they are using to validate this approach. We see repeats of the underlining assumptions that independents will break mostly for Romney, Republicans will turn out in numbers (strange since he admits he was always lukewarm about Rominey) and Democrats will stay home and a good percentage of those that do vote will vote Republican. All these subjects have been beat up here before.

  • Mike B.

    Mike above, mentioned a potential flaw in this model. And it is not Dr. Wang or the math.

    It is pollster. The model relies on them accurately reporting polls. They have already missed the past few days Grove polls in NC and WI. They have included Grove polls before.

    They have now failed to report a NV poll that is O+4:

    The poll is legit.
    Nate Silver has given it a bunch of weight. I have emailed pollster in the most polite terms about this problem and of course have gotten no repsonse.

  • John

    Dear Sam,

    Slightly off topic as well. I was wondering what you made of the Rand panel-both generally as a polling methodology and more specifically as regards the considerable gain that Obama has made (from plus 2 to nearly plus 7) following on from the 3rd debate. Is this an outlier comparable to the Gallup findings or is it measuring some movement the other polls have missed. Thanks again for all your fascinating observations-for this Democrat abroad, it has had a much needed calming effect.

    • IBP

      There are a few forecasting sites that show O getting around 51.5% of popular vote right now. Which is what around what Rand is showing right now. Coincidence?

      Most accurate forcast market since 2000.

    • E L

      @john: I’ve been following the comments on PEC for several months now on the Rand poll. The consensus here seems to be that Rand’s methodology is new and diverges so much from the traditional polling methods that we’ll just have to wait till after the election to see if Rand’s method has any validity. To sum up my impression of opinion here: interesting but untested.

  • Alfred G. Cuzán

    I observe that the Median EV Estimator appears to have settled at between 1 and 2 percent. Far from “returning to the mean” as was expected, then, the current plateau is about where Obama’s previous low was before the conventions. Comments?

    • 538 Refugee

      The Republicans and the independents that lean to the right liked what they saw in the shiny new Mitt in the first debate and the enthusiasm gap closed a bit. There was a segment always looking for a reason/excuse to vote against Obama and they finally got ‘cover’ for their decision. The doubt about the economy and the obvious short term voter memory we saw in 2010 are still factors.

    • 538 Refugee

      Would pollsters switching from registered to likely voters as we get closer to the election affect the meta-margin since this typically seems to favor Republicans?

    • xian

      I observe that every time someone makes a curve-fitting assessment about the direction or pattern of the meta-margin (including our esteemed proprietor), the MM al ost immediately belies that claim.

      In this case, the number is now over 2, so not between 1 and 2, as you just said. Now that I’ve said that, by the same token, I have jinxed it.

    • orchidmantis

      It was striking at 538 how much the current graphs of EV and percentage chance of win look like June and July. Despite the millions spent and all the actual events, we’re where we were before the conventions.

  • Mike B.

    Could someone do the math for me in layman’s terms and explain why WI is O+2?

    I would be most grateful.
    I seem to be missing something from the data on pollster.

    Froggy, you have been a big help.

    • ChrisD

      Ths WI glitch is discussed in the prior thread, as well as one in CO. Let’s see what the noon update brings.

  • 538 Refugee

    With 27 electoral votes, Ohio is considered the most crucial of battleground states.

    Anyone want to take a stab at how this effects the electoral math? ;)

    • Olav Grinde

      Well, look at that poll in context. Here are the other recent polls:

      In short, Fox News is continuing its tradition of cherry picking. Mr Murdoch’s “fair and balanced” news media has long been striving to dictate a pro-Republican narrative in this election, and chanting the “Romney is ahead and has momentum” mantra — even when that clearly goes against the facts.

      I would be willing to offer Fox News the following bet:

      — If Romney wins, then I donate the sales price of my house to charity.

      — If Obama wins, then Fox News is sold and they close shop (permanently!) — with the sales price donated to foreclosed homeowners.

    • JamesInCA

      Unless it is followed by several similar results, no effect at all.

    • BlairMac

      Well we all new Fox News isn’t about fact checking, verifying or validation. This idiot doesn’t even know that Ohio has 18 electoral votes. Propagandist don’t care about getting thing rights. They just have a message that they want to get out before they steal the vote. Its tied in Ohio… Romney squeezed by Obama at the end.

    • BlairMac


    • 538 Refugee

      BlairMac. It is only 18 votes if Obama wins the state but 27 if Romeny wins. That is why I wondered how it effects the current electoral math models.

    • A New Jersey Farmer

      Any article that uses surge and momentum is suspect.

  • Rey Howard

    Sam, slightly off-topic but nonetheless important: Often, I post a link to one of your posts as part of a comment to a friend’s oh-my-God-Romney’s-gonna-win freak-outs. In such situations, one doesn’t have the option of choosing which thumbnail will appear; Facebook selects one automatically from among all those on the page. And the one it selects, every single time, is… Crossroads GPS! Might there be some setting on your site that’s causing this, which can be tweaked to point to a more worthy image?

  • bb

    I love the simplicity of the meta-margin to present what must be some very complicated math. Very cool.

    I’m looking around this morning and I can’t understand the reason there was a jump from 1.64 to over 2. Did I miss a big state poll?

  • ChrisD

    WashPost has a new VA Senate poll today: Kaine 51, Allen 44. Reverse coattails!

    • DrOrbit

      I am not so sure about that. IF you look at the 2008 race map of Va. Obama won by a lot of the PV but look at the county map. There was a lot of red. Yet Mark Warner ran for Senate there and the map for him was a sea of blue. Va is notorious for splitting the ticket. However, I do think Obama will win Va, barely.

  • DrOrbit

    I still think O is going to win Va by around 1%. Most of these models give it to R, but admit that is barely. The latest PPP, Gravis and WaPo polls show him ahead by 4 or a little more. Any input?

    • Justin S

      No one has been factoring Virgil Goode in. Surely he cuts into Romneys numbers particularly in his formner district. In a tight rac he very well could flip the state

  • DrOrbit

    @Justin, I concur. They also do not count Johnson either.

  • Vicki Vance

    Thank you for this site, Sam – it has kept me from going crazy since the first debate. I also took your advice in two ways. Yesterday I went to NC to volunteer with the Obama campaign to GOTV (not much I can do in SC), and this morning I donated as you suggested using your actblue link. The Obama operation in NC is impressive and I felt like I had made a real contribution. Altogether, I really feel much much better and I have you and your site to thank for it.

  • BillSct

    So how does Ohio have 18 EV for Obama but 27 for Romney?

    • DrOrbit

      I think it is that new Romneymath they have been discussing. Maybe they are claiming if he wins OH, then he wins CO too?

    • 538 Refugee

      I only posted the Fox News link because I knew people would get a chuckle out of it. It was a pretty silly error to make so we are just having some fun with it.

  • JackieFla.

    What is the thinking, by Prof. Wang and/or others, about the possible impact of Hurricane Sandy on the election — given its potential for, say, curbing GOTV efforts by the Obama campaign as well as causing problems for polling places and poll workers, and generally depressing turnout?

    • DrOrbit

      Va is now allowing early voting due to the Storm. Fortunately, it passed through a week before the election. Paper ballots and generators are set up in Va. Now as for Pa, they have no early voting. And the brunt of the storm will turn on them. So hopefully, they can get things cleaned up and ready for Election day.

  • AllInForObama

    Hi, Dr. Wang. The site is great and I appreciate the way that you take the time to respond to questions that visitors pose to you. I have this question. So far today, the site has indicated that the prospects for President Obama’s re-election have improved as the day has worn on. I have some trouble understanding that, given that there was a credible (e.g., not Rasmussen or Gravis Marketing) poll that showed Ohio tied and another credible poll that showed Obama up by only three points in Minnesota (!). The poll yesterday that showed Obama up by four in Virginia was certainly welcome news but I figure that you had already priced that into your projections by this morning. Is there something going on that I don’t know about? As far as I can see, there doesn’t appear to have been much good news today in terms of state-level polling (not to mention the “Des Moines Register” endorsement debacle). Thank you very much and keep up the outstanding work!

    • 538 Refugee

      The model is set up so it doesn’t panic/obsess about individual polls like we do.

    • Dean

      If this is any consolation to you, the Detroit Free Press just endorsed President Obama, as did the New York Times.

      The Ohio News poll that was released this morning showing a tie is an old poll (10/18-10/23 polling dates). I wonder why it was released five days later. Newer polls show the president leading in Ohio.

      I understand that feeling, like we’re going to see that one poll that confirms our deepest fears, or like we will see that one poll that gives us confidence for victory.

    • Dean

      The president got two very nice endorsements today, from the Detroit Free Press and from the New York Times.

      The Ohio News Poll that was released early this morning showing a tie is five days old. Newer polls show Obama leading in Ohio. Why did the pollster wait five days to release the results?

      As far as legislative races, one race in my state is interesting. I just learned that a Super Pac ended its funding for Rep. Joe Walsh. Tammy Duckworth is leading him by 10 in a recent poll, and that’s good news.

  • DrOrbit

    The Des Moines Register is rather meaningless. The economy in Iowa is better compared to most of the nation. They have made huge gains in Wind power manufacturing and ethanol, things Romney said he was against. Also, go to the DMR site. They appear to be getting ripped up by their local readers for the endorsement. The state polls were not too bad today. Minnesota is not going to go Romney. And I question the OH poll that has it tied when over a dozen other polls have Obama leading 2 or more.

    • AllInForObama

      No, Mitt Romney is not going to win Minnesota but an Obama lead of three points there is not a good sign of how things are trending in the Midwest generally. (Although the President retains a one-point lead in the IBD/TIPP national poll, down from two points yesterday, the poll showed his eight-point lead (and, the day before that, his nine-point lead) in the Midwest has completely evaporated.) As for the “Des Moines Register,” naturally they’re going to hear disproportionately from irate Obama supporters (such as me) . It’s hard to imagine a ton of messages flowing into their offices saying, “I applaud you, ‘Des Moines Register,’ for endorsing Mitt Romney.” People are moved to write to a newspaper when they’re angry, not when they’re satisfied. As for the Ohio poll, I think it’s more worrisome than you suggest because, as Nate Silver says today, it’s a highly reputable poll and it previously showed Obama with a five-point lead in that state. Now, if the President really is up by four points in Virginia, Ohio becomes significantly less meaningful because there are several ways that Obama could win the election without it. Let’s hope that he really is winning there.

  • wufwugy

    Wow RAND has Barry winning 53% of males. No way that’s a true reflection, but I will reiterate that I think this cycle is when we see the polls significantly underestimate the mobile vote

  • Sam Wang

    I really can’t get you guys to focus on downticket…too bad.

    • ChrisD

      Since many of us are obsessive poll-watchers, how about creating and putting a Senate Power of Your Vote table in the right-hand sidebar with links to’s state polls?

    • bsk

      Give it a few more days of MM > 2% and I’m sure we’ll all calm down :)

    • Froggy

      I guarantee that if you had something akin to MM and EV estimator for Senate races, updated as frequently, people here would pay more attention to it.

      The presidential race is really important this year, in that whatever party wins the White House gets credit for the inevitable economic recovery, and thus a big leg up on winning in 2016. Plus there’s a good chance of a couple of appointments to the Supreme Court (more if you factor in the probable 2016 win), which is very important.

      The presidential race, besides offering more frequent polls as data points, also offers more complicated strategy in terms of combinations of states, and possible strategic moves and countermoves. This is seductive to the human mind, I think. (You’re the expert on this.) If we got the North Dakota Senate race in terms of polls of individual counties, we all might be a lot more intrigued to try and figure out what is happening and what is going to happen.

    • DrOrbit

      I am watching the down ticket. I am seeing that the latest poll in Va has Kaine up by around +7. Good news. But Va is notorious for voting split ticket in Presidential elections. 2008 is case in point. Obama carried the state, but Warner actually received almost 500,000 votes more than Obama did.

  • Amitabh Lath

    Okay, okay, downticket discussion. Bill Foster (IL 11) is trending up. Cook and Sabato have it leaning D. Before Sam highlighted this race, Foster’s opponent was favored. Good going Sam.

    More physicists in congress please!

    • Sam Wang

      There’s an internal poll from the Foster campaign showing him four points up on Biggert. I feel good about this one!

    • Amitabh Lath

      Unfortunately, it looks like redistricting is going to make NJ03 a tough one. Cherry Hill got moved out of NJ03.

    • Show Your Work

      I would like to second the motion for more physicists (and any other hard science) in congress. As their fields are all likely to be deemed “lies from the pit of hell”, we could use more reasoned voices to drown out the crazy ones.

    • orchidmantis

      It’s interesting how heavily MA Senate is coming down to control of the Senate as a reason to vote for Warren. (Admittedly it’s mine. But I was surprised to find much of the state in here with me.)

      In an unexpected twist, a close House race in MA6 is undercutting the “MA delegation all Dem” argument, since we might flip a House seat to counter the Senate.

  • ChrisD has 7 Senate polls in the last week for OH. Their median is Brown +5, and they include two Rasmussen polls (B+4, B+5). The Cincinnati Inquirer endorsed both Brown and Romney yesterday. As in VA, these two Senate races might afford a nudge to Obama at the top of the ticket.

  • A New Jersey Farmer

    If you really want to go downticket, consider giving to Marie Corfield who is running in a special NJ Assembly election. She’s a great friend to education and a solid Progressive.

  • KJL

    Being from WI, I’m greatly interested in the Thompson-Baldwin Senate race. Two recent occurrences, last week: debate, and controversial over-the-top ad from Thompson citing Baldwin’s nay vote on a 2006 9/11 commemorative bill as a “slap in the face” to veterans.

    The ad campaigns have been “overwhelmingly negative” ( and the Crossroads GPS ads are hallmarks of character assassination in my estimation. This race will be a real test of whether one side can win by completely going all out on the attack against the other.

Leave a Comment