Main Menu

Predicting the election

Off topic.

There was a neat piece on the PBS NewsHour last evening on trying to figure out who will be elected President in November. They talked about three approaches: polls, models and markets.

Watch Political Polls, Professors Predict the Presidential Race on PBS. See more from PBS NewsHour.

I don’t think any of those can take into account things like Romney’s “47%” remark, or some surprise world event. Also they do not take into account campaign “get out the vote” efforts on election day.

I have my own forecasting method that literally takes everything into account. I asked, “Will Barack Obama be re-elected?”

image

While I got the photo above from the Internet, this is the answer I received in the single trial I made with my real Magic 8-ball™.

35 Responses to Predicting the election

  1. avatar
    richCares October 13, 2012 at 1:49 pm #

    Two highly reliable and accurate predicton sites have been and are
    1. Intrade
    http://www.intrade.com/v4/home/
    .
    2. 538
    http://fivethirtyeight.blogs.nytimes.com/
    .
    both predict Obama wins, plus every electoral vote prediction favors Obama.
    ex: http://www.electoral-vote.com

  2. avatar
    Dr. Conspiracy October 13, 2012 at 1:59 pm #

    The PBS program talks about Intrade and another market site in Iowa.

    richCares:
    Two highly reliable and accurate prediction sites have been and are
    1. Intrade
    http://www.intrade.com/v4/home/
    .
    2. 538
    http://fivethirtyeight.blogs.nytimes.com/
    .
    both predict Obama wins, plus every electoral vote prediction favors Obama.
    ex:http://www.electoral-vote.com

  3. avatar
    foreigner October 13, 2012 at 2:31 pm #

    also betfair,Ladbrokes, … (other bookmakers)
    ideosphere.com
    there are also pages that list all the actual prices at the major bookmakers on one page

    interesting are the intrade-prices during a debate

  4. avatar
    richCares October 13, 2012 at 2:42 pm #

    Interesting, In Oct of 2008 Nate Silver correctly predicted every state but one (Indiana). More than likely 201 will follow the same pattern.
    http://fivethirtyeight.blogs.nytimes.com/

  5. avatar
    Scientist October 13, 2012 at 4:42 pm #

    richCares: Interesting, In Oct of 2008 Nate Silver correctly predicted every state but one (Indiana). More than likely 201 will follow the same pattern.
    http://fivethirtyeight.blogs.nytimes.com/

    The one caveat that I would point out is that those were predictions made a couple of days before the election. Predictions that Nate or anyone else makes today are obviously less accurate. The same is true of pollsters-they tend to get judged on the last poll before the election, not those 2 or 3 months out.

    I think Obama will win, but I am less sure than I was a week ago. One thing to keep in mind is that the Obama folks are thought to have a superb ground operation in the swing states. If so, that will likely win the day. Who votes generally determines close elections, which this one is.

  6. avatar
    Dave October 13, 2012 at 6:21 pm #

    A couple quick points:
    1. People often say that Silver correctly predicted most states in the last election — but to say that is to misunderstand the value of his work. Silver’s whole schtick is making empirical models of probabilities, which is very much not the same thing as predictions.
    2. There are good theoretical reasons to suppose prediction markets do a good job of predicting, but the theory depends on there being high volume, which no actual prediction market has acheived. A way of think about this is to ask yourself, how much money would it cost to make the market say what you want. To do this to, for example,. the price of IBM would be exhorbitant; but to do it to the Obama share on intrade would be a pittance on the scale of presidential campaign funding.

  7. avatar
    misha marinsky October 13, 2012 at 7:20 pm #

    I would like to recommend the Iowa Electronic Markets:

    http://iemweb.biz.uiowa.edu/quotes/Pres12_Quotes.html

    This is an aggregator:

    http://www.predictwise.com/politics/2012presidentindividual

    Intrade has an accuracy of ~90%. They were correct about Obama in 2008. The only times there were wrong recently was Sharron Angle, Tancredo and the Supreme’s health care decision.

    PredictWise has Obama at 61.4 %; Mittens at 38.1 %.

    My feeling is that these markets are more accurate than a poll.

  8. avatar
    Dr. Conspiracy October 13, 2012 at 9:54 pm #

    In the case of the health care decision, I think they were pooling ignorance.

    misha marinsky: Intrade has an accuracy of ~90%. They were correct about Obama in 2008. The only times there were wrong recently was Sharron Angle, Tancredo and the Supreme’s health care decision.

  9. avatar
    Dr. Conspiracy October 13, 2012 at 9:58 pm #

    The thing that I have to keep in mind is that pollsters do not predict anything; they just estimate what the current state of the electorate is. The models are predictive. My 8-ball is also predictive, and so far it has never failed to correctly state the outcome in any electoral contest.

    I wonder if the psychic octopus can help, or this one, or maybe this one.

    Scientist: The same is true of pollsters-they tend to get judged on the last poll before the election, not those 2 or 3 months out.

  10. avatar
    donna October 13, 2012 at 10:30 pm #

    doc:

    have you looked at allan lichtman’s “13 keys”/

    he’s correctly predicted every president dating back to reagan

    ‘Never Wrong’ Pundit Sticks By Obama in 2012

    http://www.usnews.com/news/blogs/washington-whispers/2012/10/03/never-wrong-pundit-sticks-by-obama-in-2012

  11. avatar
    Dave October 13, 2012 at 11:24 pm #

    Lichtman’s method is kind of dubious. When you construct a model by tweaking it to fit past data, what you wind up with does not necessarily have any predictive power. Nate Silver has a critique of LIchtman’s model here, where he raises several other objections.

  12. avatar
    Wile October 13, 2012 at 11:46 pm #

    misha marinsky:

    Intrade has an accuracy of ~90%. They were correct about Obama in 2008. The only times there were wrong recently was Sharron Angle, Tancredo and the Supreme’s health care decision.

    Isn’t Intrade eventually “correct” about all of their listings?

    http://www.intrade.com/v4/markets/contract/?contractId=745353

  13. avatar
    foreigner October 14, 2012 at 12:55 am #

    I made the experience with the nominee markets that intrade predictions were
    better than 538, intrade contracts reacted earlier to news and trends

  14. avatar
    richCares October 14, 2012 at 2:42 pm #

    I forgot the 7-11 poll, they sell coffee in blue cups and in red cups. On the past 3 elections these sales predicted it correctly. Currently blue cups are at 62% red at 38. it’s over!

  15. avatar
    misha marinsky October 14, 2012 at 3:11 pm #

    richCares: I forgot the 7-11 poll, they sell coffee in blue cups and in red cups. On the past 3 elections these sales predicted it correctly. Currently blue cups are at 62% red at 38.

    http://www.inquisitr.com/359972/7-11-cup-poll-pegs-obama-as-likely-winner/

    October 11, 2012 – …the 7-11 coffee cup poll has correctly predicted the winner of the last three elections.

    And who does 7-11 peg as the current leader via its unofficial coffee cup polling? Barack Obama, with a 60 percent lead over Mitt Romney’s 40 percent, even in some GOP strongholds like South Carolina, Texas and Utah.

  16. avatar
    Keith October 14, 2012 at 5:17 pm #

    misha marinsky: And who does 7-11 peg as the current leader via its unofficial coffee cup polling? Barack Obama, with a 60 percent lead over Mitt Romney’s 40 percent, even in some GOP strongholds like South Carolina, Texas and Utah.

    The problem with that poll is that it might just be that Obama voters can stomach bad coffee better than RMoney voters.

  17. avatar
    James M October 15, 2012 at 1:39 am #

    richCares:
    Two highly reliable and accurate predicton sites have been and are
    1. Intrade
    http://www.intrade.com/v4/home/
    .
    2. 538
    http://fivethirtyeight.blogs.nytimes.com/
    .
    both predict Obama wins, plus every electoral vote prediction favors Obama.
    ex:http://www.electoral-vote.com

    I have lost count of the number of sincere offers I have made to Republicans who express total certainty in Mr. Romney’s election, for them to purchase shares on InTrade where I will match them in shares for Obama.

    I have received a handful of responses, but all rejections of the proposition.

  18. avatar
    James M October 15, 2012 at 1:43 am #

    Keith: The problem with that poll is that it might just be that Obama voters can stomach bad coffee better than RMoney voters.

    The 7-11 coffee poll’s record is three for three.

    On the other hand, Mormons do not drink coffee.

  19. avatar
    misha marinsky October 15, 2012 at 2:00 am #

    James M: On the other hand, Mormons do not drink coffee.

    Yeah, they go in for hard drugs instead.
    http://abcnews.go.com/Health/MindMoodNews/story?id=4403731&page=1#.UHulBW_7L4Q

    When Express Scripts issued its first national survey of prescription drug use in 2002, it sparked a heated debate across Utah about what, if any role, the church played in the state’s high dependence on antidepressants such as Prozac and Zoloft.

    “Three years into my marriage my husband was drinking, using drugs and stepping out on me…”

  20. avatar
    misha marinsky October 15, 2012 at 2:04 am #

    James M: On the other hand, Mormons do not drink coffee.

    Google “mormon drug utah.”

  21. avatar
    gatsby October 15, 2012 at 2:46 am #

    James M: On the other hand, Mormons do not drink coffee.

    But, as we know from Mitt Romney himself, they do eat coffee ice cream.

  22. avatar
    foreigner October 15, 2012 at 6:59 am #

    Germans do it by octopus
    http://en.wikipedia.org/wiki/Paul_the_Octopus

  23. avatar
    misha marinsky October 15, 2012 at 9:19 am #

    foreigner: Germans do it by octopus

    I’d like to be under the sea
    In an octopus’ garden in the shade
    He’d let us in, knows where we’ve been
    In his octopus’ garden in the shade

    http://www.lyrics007.com/The%20Beatles%20Lyrics/Octopus's%20Garden%20Lyrics.html

  24. avatar
    LW October 15, 2012 at 7:44 pm #

    Whenever I visit the main page here now, this video starts automagically. Is anyone else having the same problem?

    My provisional theory is that this is due to some bad Flashblock/Adblock craziness, where the video isn’t blocked but the control widget is. I think I’ve seen something like this before.

  25. avatar
    foreigner October 16, 2012 at 9:12 am #

    why do they let the whole nation vote ? Just a representative sample would
    be enough. The polls are doing it anyway. And there could be elections
    every month or such. Update the constitution – the founding fathers didn’t
    know about computers,internet,TV,

  26. avatar
    JPotter October 16, 2012 at 9:49 am #

    foreigner: why do they let the …TV,

    That’s about 4 facepalms capped by a, “well, duh.”, around a decades-late call for an e-revival of the town meeting form of gov’t. Let’s here it for e-Democracy. Until that glorious day, we’ll always have Call to Power II

  27. avatar
    foreigner October 16, 2012 at 11:30 am #

    I don’t understand.
    But I realized it’s not so easy, we must ensure that selected voters
    are not bribed – should still be possible

  28. avatar
    JPotter October 16, 2012 at 2:01 pm #

    What you’re describing is … a republic. We already have “selected voters.” At several levels, town, county, state, federal. In the case of a Presidential election, we have the Electoral College. The people don’t vote directly for the President. Although, thanks to various laws binding electors in most states, the people are closer than ever to doing so.

    Before calling for a revision of the Constitution (such ‘revisions’ would be called Amendments), please master the its current form.

    We are moving in the exact opposite direction that you are recommending.

    But hey, let us know when you find your philosopher kings, Plato.

  29. avatar
    Paper October 16, 2012 at 6:25 pm #

    Cheese Louise, why not just go back to the time when the state legislatures selected senators and the electoral college. *^#%!!!

    foreigner:
    I don’t understand.
    But I realized it’s not so easy, we must ensure that selected voters
    are not bribed – should still be possible

  30. avatar
    foreigner October 17, 2012 at 2:47 am #

    the system is outdated. We have telefone and internet now and the population is >300M.
    Voting is unreasonable, the cost for traveling to the ballot is bigger than the benefit
    from the desired candidate winning multiplied by the chance that your vote is decisive.
    (negative expectation value)
    In other areas of decision making we already do it differently.
    The election campaigns are directed to the majority of uninformed voters
    and looks strange already with the straw polls, Perry gaffe, Cain skandal,
    Trump birther issue and such things. That’s not what policy is about.

  31. avatar
    James M October 17, 2012 at 3:17 am #

    foreigner:
    Voting is unreasonable, the cost for traveling to the ballot is bigger than the benefit
    from the desired candidate winning multiplied by the chance that your vote is decisive.
    (negative expectation value)
    In other areas of decision making we already do it differently.
    The election campaigns are directed to the majority of uninformed voters
    and looks strange already with the straw polls, Perry gaffe, Cain skandal,
    Trump birther issue and such things. That’s not what policy is about.

    There is nothing in the Constitution that says the legislature of a state cannot literally flip a coin or throw darts at a board, or use any other discernment method that they “may direct.”

    Good luck with your whole redesign!

  32. avatar
    Paper October 17, 2012 at 8:46 am #

    See Oregon.

    http://en.wikipedia.org/wiki/Vote-by-mail_in_Oregon

  33. avatar
    JPotter October 17, 2012 at 9:30 am #

    foreigner: don’t understand.

    I can see that. THe scope of that lack is becoming more apparent. Please, keep posting.

    foreigner: Voting is unreasonable,

    Now that is hysterical! Have you voted in this country? It is indeed a terrible burden. On the appointed day, I make a right turn on the way home instead of a left and go about 100 yards out of my way. There’s never a line. It takes only as long as I wish to take in reading the ballot. (The quadrennial election is the Big One, makes for a long ballot).

    Not everyone has it so easy. But over the past 200 years, it has been getting easier, and restriction on franchise has been much reduced, rather than increased. You’re arguing against history.

    Notice how much commentary there is on the civics of … whatever country you’re claiming to be from goes on here?

    You are coming up with some awesome voter suppression slogans tho!

  34. avatar
    foreigner October 17, 2012 at 9:46 am #

    JPotter,
    I assume “terrible burdon” was meant ironic.
    from your desciption I estimate your voting cost at ~$1,
    the estimated benefit from your candidate winning at ~$5K
    and the probability of your vote being decisive at ~5e-7
    for an expectation value of -$0.99.75

  35. avatar
    JPotter October 17, 2012 at 10:38 am #

    I see the concept of collective action escapes you.

    And your estimates are based on …. nothing. Have you been talking to Monckton?