October 24, 2012
Here are two maps for those trying to follow the U.S. presidential election.
Real Clear Politics calculates an average of recent polls for each state, on the basis of which they create the following map. They designate a state as "toss-up" if the candidate leads in the polls by 5% or less. For example, they're describing Michigan as a toss-up, even though President Obama leads in their poll average by 48.6% to 43.6%.
Alternatively, Intrade summarizes the current prices of their betting contracts on the outcome of each state, summarized in the map below. They call a race a "toss-up" if the implied probability of a victory is 55% or lower. According to the prices of their contracts, the race all comes down to Ohio. 270 electoral votes are needed to win the election. If you color Ohio blue in the map below, Obama gets to 277 electoral votes.
No Republican has ever won the U.S. presidency without carrying Ohio. The RCP poll average currently has Obama up 47.6 to 45.7 there. If you follow the link to the Intrade map, you'll see it updated every 60 seconds-- Ohio's been shifting back and forth between green and pale blue as I've been writing this post.
Posted by James Hamilton at October 24, 2012 11:20 AMdigg this | reddit
If Ohio is indeed the key to the election, there may be a slight arbitrage opportunity between the difference in price of Ohio and the price of Obama.
Posted by: Joseph at October 24, 2012 11:31 AM
Thanks James, that's a great post.
Posted by: Johannes at October 24, 2012 11:46 AM
I wonder if
a) there is a meaningful % of gamblers who vote with who they *want* to win, and if
b) gamblers tend to be, on average, republican or libertarian, and if
c) low volume contracts are meaningful in any case, (if I'm not mistaken the total volume for both Colorado contracts is less then 12,000 total (not today, total))
Posted by: idave at October 24, 2012 11:57 AM
As I noted in an earlier post, there is a small, but real, chance of a tie in the Electoral College, 269 to 269.
In that case, the new House elects the President, and the new Senate elects the VP, and there is some chance that we might have a Romney/Biden administration.
But the actual Electoral College vote in December would be very interesting. In the event of a tie, if one "Faithless Elector" (out of 538 electors) changes his or her vote, from the way their state voted, it would swing the election one way or the other.
Posted by: Jeffrey J. Brown at October 24, 2012 12:52 PM
Joseph, IN the last few days not much of one as the percentagages in Ohio and for Obama overall have been very close. The latter have tended to be a bit above those for OH as it remains possible for Obama to win even if OH goes to Romney, if he can get CO or VA and holds NH (as well as WI and IO). Nearly impossible for Romney to win without OH.
Posted by: Barkley Rosser at October 24, 2012 01:09 PM
The Intrade markets are incredibly thin and many people suspect that there may be deep-pocketed partisans trying to manipulate Intrade prices to create positive press coverage for Romney. You can see some discussion at:
Apparently, an investor overpaid just about $1k trying to boost Romney's odds artifically. Given the price of commercial advertising, that's not too expensive for the earned media coverage. It could also have instigated a herd of conservative traders to participate in the market.
Regardless, I'd take Intrade odds with a grain of salt at the moment. Investors don't like playing in a market where it's clear there's funny-business going on, polls are much harder to manipulate. Remember their odds on the ACA Supreme Court case?
Posted by: Tudor at October 24, 2012 01:13 PM
Follow the money. If enough people are speculating, then on average, their bets represent a synthesis of all publicly available information and probably some private info.
I honestly believe the polls are oversampling democrats by at least a point or 2 in most polls so that top electoral map is biased toward blue. For example, the Gravis Marketing poll used to get the Pennsylvania average has
Democrat 45% Republican 37%.
Posted by: tj at October 24, 2012 01:23 PM
I think its a mistake to characterize an intrade contract as an investment, or an intrade buyer as an investor.
Over sampling of Democrats is easily checkable. Look at the elections four years ago - state, local, all of them - and see which way the error ran. Same organizations have paid for, administrated, interpretted and publicized these polls every cycle. Clearly they have the ability to correct that error. If they chose not to, it would leave a trail, right?
Posted by: KevinM at October 24, 2012 07:36 PM
Rasmussen Reports has a good track record. See their Electoral map.
237 Obama, 235 Romney with 66 tossup.
Swing state tracking Swing State Tracking:
Romney 50%, Obama 46%
Michigan 48% to 48%
Nevada O 50%, R 48%
New Hampshire R 50% O 48%
Posted by: David L. Hagen at October 24, 2012 08:18 PM
OH has drifted back to a pretty steady light blue on intrade, and Obama is back to being at nearly 60%. This was a flukey day to look at it, one day after the weird Romney surge that was pretty clearly the work of one person spending only about $17,000 along with the frenzy over the anticipated Trump "bombshell" that proven to be a giant dud.
Posted by: Barkley Rosser at October 24, 2012 08:54 PM
Re: Electoral tie
Extremely doubtful. While in theory the math is there, in practice, the odds are very low. Further, the chance of one or more electors cross voting is fairly good - 26 states have no restrictions on how their electors should vote. One could easily see an elector from, say, Virginia (that doesn't have statute or pledge requirements), deciding to vote for the candidate that won the national popular vote.
Now, if it actually did go to the House. given the rule that each state delegation has 1 vote, Rommey would win easily; Republicans control 33 out of 50 delegations. However, imagine the level of pressure to vote for the candidate that won the popular vote; especially if it wasn't a thin margin.
As for the Senate. A Vice-President must be elected by a majority of the Senate. The constitution doesn't speak of what would happen if there was a tie. It's very unclear whether the sitting vice-president (Biden) would have tie breaking privilege as the vice-president does normally. It could get very messy.
Even if a three sigma event happened, I doubt that the electoral college would be abolished. The leverage it affords small states and various low population regions would stall any constitutional amendment abolishing. Yet, it's possible that the Maine/Nebraska solution might get approved: Those two states award 1 electoral vote for each congressional district won by a candidate, and the two extra (Each state gets electors that match the total number of Representatives and Senators), goes to the winner of the statewide vote.
That would really have an interesting effect on national elections, States such as California and Texas would find themselves "in-play", as both parties would be able to harvest a number of electoral votes from them.
I'll freely admit I'm biased, I can't think of any cogent argument in favor of the first-past-the-post, winner take all system that is our legacy from England.
Posted by: SecondLook at October 24, 2012 10:01 PM
Rasmussen Reports has a terrible track record. Check their results from 2008. They have a strong GOP bias.
There are several other election prediction markets besides Intrade.
Posted by: joe at October 25, 2012 01:00 AM
Re: Extremely doubtful regarding an Electoral College tie
I guess it depends on what probability you assign to one of the three following scenarios actually happening.
Hoping election doesn't end in Electoral College tie:
This year with the closeness of it all, Chuck Todd, host of MSNBC’s “The Daily Rundown” and chief White House correspondent for NBC, has put finger to iPad and come up with scenarios that could happen involving those 12 battleground states. It is assumed “red” and “blue” states vote as expected.
In one example, Obama gets Wisconsin and Virginia and Romney takes Florida, North Carolina, Ohio, Iowa and New Hampshire.
In another, Obama wins Wisconsin, Colorado and Virginia and Romney is successful in Nevada, Iowa, Ohio, North Carolina, New Hampshire and Florida.
The most likely of any, though, Todd’s mathematics show, is Obama gets New Hampshire, Wisconsin and Ohio with Romney winning North Carolina, Florida, Virginia, Iowa, Colorado and Nevada.
Posted by: Jeffrey J. Brown at October 25, 2012 06:42 AM
The only poll that matters is on November 6.
Check out the polls before 2010, then check out the results.
Posted by: Ricardo at October 25, 2012 06:51 AM
The point of a prediction market is that if someone does something like that it creates opportunity for traders to buy/sell at a favorable price until the price comes back down to equillibrium.
Posted by: Mike at October 25, 2012 09:16 AM
This is an intersting model based on economic data and not polling data.
The state-by-state economic data used in their model have been available since 1980. When these data were applied retroactively to each election year, the model correctly classifies all presidential election winners, including the two years when independent candidates ran strongly: 1980 and 1992. It also correctly estimates the outcome in 2000, when Al Gore won the popular vote but George W. Bush won the election through the Electoral College.
We continue to show that the economic conditions favor Romney even though many polls show the president in the lead,” Bickers said. “Other published models point to the same result, but they looked at the national popular vote, while we stress state-level economic data.”
Posted by: tj at October 25, 2012 09:51 AM
"The point of a prediction market is that if someone does something like that it creates opportunity for traders to buy/sell at a favorable price until the price comes back down to equillibrium."
I'm not arguing against prediction markets in general here. Although, I would say the effects of illiquidity and susceptibility to biases are broadly under-appreciated, particularly in dynamic information setting.
The InTrade market in particular these past few days has clearly been wandering in and out of equilibrium with a few agents taking advantage of the thin market to create news rather than reveal information.
As joe mentioned, there are several other prediction markets that have not suffered from this microstructure noise. Arbitrage opportunity? Maybe you could probably make a couple hundred bucks off of it but not a lot...remember the market moved almost 10% on less than a $20k bet.
Posted by: Tudor at October 25, 2012 10:02 AM
I'll go with Nate Silver's (the 538 blog guy) estimate of a tie, based on multiple simulations using best fit data: 0.5%
And, as I noted in my earlier post, even if that very slim probability occurs, the likelihood of one or more electors cross voting is very good. In the event of a tie, the candidate that gets the majority of popular vote will get selected by the college.
Why? As a matter of principle and conscience. Qualities that might be too rare in our political system, but not absent...
Posted by: SecondLook at October 25, 2012 12:37 PM
The Electoral College serves as a useful reminder that the Founding generation did indeed have feet of clay. They gave us slavery, states' rights, and the Electoral College. In fact, even after two bites at the apple they still couldn't get it right after the 12th Amendment. The Electoral College is an anachronism that only a neanderthal like George Will could love.
Posted by: 2slugbaits at October 25, 2012 07:11 PM
After I posted my note, I saw the 0.5% number on the 538 blog, which surprised me. I would have thought that Chuck Todd's Scenario #3 (Obama gets New Hampshire, Wisconsin and Ohio with Romney winning North Carolina, Florida, Virginia, Iowa, Colorado and Nevada), would have a higher probability than that.
In any case, what I find fascinating is the possibility that, if we have an Electoral College tie, after something like a billion dollars and countless man-hours have been expended, the election could swing on the vote of one of 538 electors, who might vote contrary to the way his or her state voted. And of course think of the Supreme Court ramifications (in regard to future nominees). One person could strongly influence the future makeup of the Supreme Court.
Given a tie, since a Romney election in the House seems very likely, so I would think that if an Elector votes contrary to the way his state voted, it would be from Romney to Obama (especially a secretly pro-choice, regarding abortion, Republican elector).
If an elector does change it from a 269-269 tie, it would seem to me that it would be circumventing the intent of the Constitution, i.e., it should go to the House & Senate, and I think that we may be underestimating the political response, if either Obama or Romney is elected via a "Faithless Elector." An "Illegitimate President" would probably be the kindest thing that about half the country would call the new president, especially if the losing candidate won the popular vote.
Posted by: Jeffrey J. Brown at October 26, 2012 05:47 AM
This Oct 1 - Oct 24 Gallup poll, comparing 2008 to 2012 voter demographics, is interesting. The 2012 demographics are all very close to the 2008 demographics except for party affiliation.
Likely voters who call themselves Republicans or lean Republican = 49. Democrat/Lean Democrat = 46. In 2008 it was 42 - 54 the other way. Democrats went from +12 to -4.
Posted by: tj at October 26, 2012 07:29 AM
Best predictors have been Nate Silver at 538 and the guy over at Electoral Vote.
Posted by: Gridlock at October 26, 2012 10:59 AM
People, you are assuming this will be a fair election.
Obama will win, period. He will manufacture about 500,000 bogus votes in key swing states, and Republicans are too stupid/spineless to investigate it.
Obama will win. He might have lost a fair election, but he will win this one.
Posted by: Darren at October 26, 2012 04:15 PM
To appreciate its numerous benefits, I recommend that you actually study the origin and intention of the Electoral College. William C. Kimberling, Deputy Director, FEC Office of Election Administration.
It was designed to protect minority rights together with proportional voting. It was designed to especially avoid foreign influence in electing the president, as was the requirement for "natural born citizen" of birth in the country of parents who were citizens.
Posted by: David L. Hagen at October 26, 2012 04:48 PM
Recommend you actually look at The Rasmussen Reports Track Record
Rasmussen’s final 2010 projections were published in the Wall Street Journal projecting that the Republicans would gain 55 or more seats in the House and end up with 48 or 49 Senate seats. The Republicans ended up with a gain of more than 60 House seats and 47 Senate seats.
2009— Rasmussen Reports was closest of all polling firms to the actual results in the New Jersey governor’s race.
2008-- Barack Obama won 53%-46%, and our final poll showed him winning 52% to 46%.
2004-- George W. Bush received 50.7% of the vote, while John Kerry earned 48.3%. Rasmussen Reports polling projected that Bush would win 50.2% to 48.5%. We were the only firm to project both candidates' totals within half a percentage point
Posted by: David L. Hagen at October 26, 2012 05:09 PM
In 2008, the Democrats had a 39-29 (D+10) advantage in hard party ID, and a 54-42 (D+12) advantage with leaners. In 2012 though, we’re in the post-TEA party era. Republicans now show a 36-35 (R+1) hard party ID advantage, and a 49-46 (R+3) lead with leaners. This gives us a range of party ID swings from 2008, from R+11 to R+15.
Posted by: David L. Hagen at October 26, 2012 05:21 PM
From my reading and understanding of what was the original intent when the Electoral College was created was that the electors would select a President based on merit, not on sentiment, or how on how popular a candidate might be in their respective home states.
Granted, it was an unusual bit of naivete for the members of the Constitutional Convention - and remarkable that the College was basically an unchallenged concept by them. However, that clearly was the "original intent" and therefore, any vote change would likely stand up in court.
Sadly, the Gallup organization reliability in polling isn't considered first class anymore. They are widely quoted and used much like in the manner that Xerox became the brand name for copiers, even past the time that competitors had come out with better machines.
If you want a better poll to support your argument (if not the best), try using Rasmusseun. Their comparative data on party affiliation shows currently: 36.8% Republican, 34.2% Democrat, 29% Other.
In 2008 at the same sample time (September), the numbers were 33.4%,3 39.0%,27.6%. By their numbers, the Democrats declined by 4.8% in affiliation, but the shift was split between the Republicans and Others.
By the way, in 2010, which saw major gains for the Republican in both Congress and in State races, the numbers were 33.1%, 34.6%, 32.3%. All of which suggests that it's the non-affiliated that really matter. As they always have. It isn't the engaged, committed, that decided our politics, it's the flexible, ambivalent, that do.
Posted by: SecondLook at October 26, 2012 06:55 PM
Romney is up in Rasmussen and Gallup and Pollster at Huffington post. Lets see what wanders can fake NFP produce on November 2nd.
Posted by: ivars at October 27, 2012 02:00 AM
It may be, as the Chinese saying goes, "An interesting election."
Of course, as Tom Brokaw said of the 2008 election, "The winner should have demanded an immediate recount."
Posted by: Jeffrey J. Brown at October 27, 2012 06:11 AM
...try using Rasmusseun. Their comparative data on party affiliation shows currently: 36.8% Republican, 34.2% Democrat, 29% Other.
Gallup has it 36R-35D so they are nearly identical to Rasumussen. You ignored the "lean" within Rasumussen's "other" category. That's the source of the largest change since 2008.
Posted by: tj at October 27, 2012 06:58 AM
The Electoral College is not the biggest distortion in American democracy. Rarely does the outcome differ from the popular vote.However, in the Senate, a citizen of Wyoming has 66 times the weight of a citizen of California. This is a gross distortion that affects the Senate every single day, not just a once-in-a-blue-moon presidential election.
Forty-one senators representing 10% of the population can overrule 59 senators representing 90% of the population. Even without the filibuster, 50 senators representing 16% of the population can thwart the other 84% of the population. That isn't democracy. That is just geographical happenstance. There is no justification for a rancher in Wyoming having 66 times the vote of the gardener in Los Angeles.
Posted by: Joseph at October 27, 2012 10:25 AM
Re Electoral College - is from the grand compromise balancing populus, states, minorities and continuation. The US is NOT a democracy. It is a democratic Republic.
SecondLook and TJ
On the change in voter ID and swing in the independent vote, Josh Jordan examines this further in: The Two Polls That Have Chicago Terrified
Taking the same scenario as Gallup and moving the independent results to match the current polling average changes Romney’s half-point victory into a 5.7-point victory. (As with Gallup, I’m assuming the Republican and Democrat voting margins stay the same as 2008.)
Posted by: David L. Hagen at October 27, 2012 07:09 PM
The "rotten borough" syndrome that exists in the Senate has long been noted by thoughtful observers. The extreme variation that you mentioned, Wyoming versus California, would probably have appalled the constitutional framers. the widest ratio at the time of ratification was about 13-1 (Virginia, including slaves, versus Delaware). Then, they didn't imagine the large numbers of relatively low population states being created out West; almost all of them for political advantage at the time.
The most equitable solution would be a partial reorganization of various states. Say, divide the four largest: California, Texas, New York, Florida, and combine the four of the smallest: Montana (44th in population) with Wyoming (50th), North and South Dakota (46th and 48th respectively), and Vermont with New Hampshire (49th and 42nd), and so on, that would create a fairer system without changing the fundamental nature of the Senate.
It will happen, perhaps, when we can genetically engineer pigs to fly...
Small footnote: Texas, by the terms of its being admitted to the Union, has the right divide itself into as many as 6 states, without needing consent of Congress. One would think that the enormous amount of clout they would gain would be appealing, but not the case...
Posted by: SecondLook at October 27, 2012 09:09 PM
Josh Jordan looks at trends in Ohio. Ohio is closer than you think
Posted by: David L. Hagen at October 29, 2012 04:43 AM
If Romney wins the states leaning his way (206) plus COL(9), FLA(29), VIR(13) IOWA (6) and Wisc(10) he has 273 electoral votes without Ohio.
All the major papers in Iowa just came out with their backing of Romney which should put him over the top there. First time the Des Moines register has not gone for the democrat in decades.
It's clear why attention is shifting to Wisconsin from Ohio. Now we know why Romney chose Ryan.
Rasmussen Reports issues our daily Swing State Tracking Poll to supplement, not replace, individual state polling. Romney leads by two in New Hampshire, Florida and Virginia. He’s up four in Colorado and six in North Carolina. Obama leads by two in Nevada, five in Pennsylvania and seven in Michigan. The candidates are tied in Iowa, Ohio and Wisconsin.
Posted by: tj at October 29, 2012 08:28 AM
Based on surveys from Oct. 1 through 24, [Gallup] finds 36 percent of likely voters call themselves Republicans, compared to 35 percent who are Democrats.
If those who are leaning toward a certain party are included, the GOP lead increases to 3 percent, standing at 49 to 46 percent.
But it was a much different story for Democrats in 2008, when they had 54 percent of people who identified themselves as Democrats or leaned Democratic.
Posted by: David L. Hagen at October 29, 2012 05:57 PM