Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
Stats United States Politics

Nate Silver's Numbers Indicate Probable Obama Win, World Agrees 881

An anonymous reader writes "The state-by-state election outcome probabilities today on Nate Silver's 538 imply a 97.7% probability for Obama to win 270 or more electoral college votes this coming Tuesday. A site that allows anyone but U.S. citizens vote seems to indicate that the rest of the world hopes these numbers are accurate. "
This discussion has been archived. No new comments can be posted.

Nate Silver's Numbers Indicate Probable Obama Win, World Agrees

Comments Filter:
  • by paiute ( 550198 ) on Monday November 05, 2012 @10:46AM (#41881005)
    Now does anyone have data on whether the forecasting of a win discourages the supporters or opponents of the projected winner from actually voting?
    • by Sloppy ( 14984 ) on Monday November 05, 2012 @11:34AM (#41881841) Homepage Journal

      Now does anyone have data on whether the forecasting of a win discourages the supporters or opponents of the projected winner from actually voting?

      Well.. tons of anecdotal evidence (whatever that's worth). Lots of people say they won't vote Libertarian or Green because it would be "throwing their vote away." They say it's throwing-away because polls always indicate the person they'd like to vote for, is very likely to lose (so they vote for someone else who has higher polling numbers, instead).

      Apparently the thinking is like this: if you vote for someone who lost, then your vote "doesn't count." From that I conclude that since all the losers' votes votes didn't count, the winner is always unanimously elected. You can't get a stronger mandate than that, so it's our way of telling the winner that 100% of America agrees with them on 100% of issues.

      For reasons I don't understand, after the election, though, over half the people say they don't agree with whoever won. It's very strange, almost as though they don't really believe that losing is the same as not counting. Go figure.

  • uhh (Score:5, Interesting)

    by nomadic ( 141991 ) <nomadicworld.gmail@com> on Monday November 05, 2012 @10:46AM (#41881009) Homepage
    The link to FiveThirtyEight says Silver predicts an 86.3% chance of an Obama victory. The "97%" link is to an anonymous python script and output at a different site. Could we get some context here?
    • Re:uhh (Score:4, Informative)

      by 140Mandak262Jamuna ( 970587 ) on Monday November 05, 2012 @11:08AM (#41881393) Journal
      Princeton Election consortium is projecting 99.9% win for obama. Dr Sam Wang is a very well respected professor of statistics. His methods are public. Votamatic has been projecting Obama win for a long time. Obama was leading in Nate's estimate for a long time. The high water mark for Romney was about 40% chance immediately after the first debate.
    • Re:uhh (Score:5, Informative)

      by Anonymous Coward on Monday November 05, 2012 @11:55AM (#41882259)

      Nate's aggregate numbers allow for the possibility of systemic bias in the national and state by state polls, based on their historical distribution around the real election results.

    • Re:uhh (Score:5, Funny)

      by Remus Shepherd ( 32833 ) <remus@panix.com> on Monday November 05, 2012 @12:06PM (#41882493) Homepage

      I think the original poster is saying that an 86.3% chance will come true 97% of the time.

      Between him, you, and me, one of us doesn't understand statistics.

  • by Guano_Jim ( 157555 ) on Monday November 05, 2012 @10:47AM (#41881021)

    The combination of the headline and TFS might be construed as "Nate Silver says that Obama's got a 97.7% chance of winning the election," which isn't quite true.

    I think it's more accurate to say that Nate Silver predicts an 86.3% chance of Obama winning 270 electoral votes.

  • by AdamHaun ( 43173 ) on Monday November 05, 2012 @10:48AM (#41881043) Journal

    Nate Silver puts the odds at around 87%, largely based on the chance of there being a systematic offset in the polling data. Still looks pretty good for Obama, though. Fingers crossed.

  • by eldavojohn ( 898314 ) * <eldavojohn@gma[ ]com ['il.' in gap]> on Monday November 05, 2012 @10:49AM (#41881053) Journal

    A site that allows anyone but U.S. citizens vote seems to indicate that the rest of the world hopes these numbers are accurate.

    Okay so you're talking about roughly six and a half billion people. As of the writing of this post, votevotevote.net's page says:

    1050 VOTES have been received

    Furthermore can someone point me to, say, a Chinese version of votevotevote.net's page? I mean, surely you'd want to represent the largest population of the world or are you simply relying on the rest of the world to speak English? And you're going to then utilize that as evidence that the rest of the world hopes that Obama wins? Surely this site isn't even worth mentioning in a news context.

    • by Anonymous Coward on Monday November 05, 2012 @11:04AM (#41881315)

      Population size has nothing to do with the sample size required for statistical significance. Effect size does, and with the votes currently standing at about 85% in favour of Obama, 1000 or so is more than enough, by a very long way. There are going to be strong biases in the sample, but that's a problem of sampling method, not sample size.

      But then, most Slashdotters know bugger all about statistics and will mod you up anyway. Complaining about sample sizes is pretty much a guarantee of +5.

    • by Simmeh ( 1320813 ) on Monday November 05, 2012 @12:19PM (#41882739)

      And you're going to then utilize that as evidence that the rest of the world hopes that Obama wins? Surely this site isn't even worth mentioning in a news context.

      Please take it as given that the rest of the world is crying out for Obama to win. Both your parties are crazy, but the Democrats are somewhat closer to being actual human beings.

  • by TwobyTwo ( 588727 ) on Monday November 05, 2012 @10:50AM (#41881085)

    According to today's actual posting from Nate Silver, the same data leads him to conclude an 86.3% chance of an Obama win in the electoral college. Still high, but your "Nate Silver's Numbers Project..." headline is true if parsed carefully, but very misleading. If you want to say "I conclude from Nate Silver's numbers...", well fine.

    Silver's Nov. 4 post is at: http://fivethirtyeight.blogs.nytimes.com/2012/11/05/nov-4-did-hurricane-sandy-blow-romney-off-course/ [nytimes.com] (paywall :-( )

  • As a Canadian (Score:4, Insightful)

    by bravecanadian ( 638315 ) on Monday November 05, 2012 @10:53AM (#41881147)

    I can only hope that Nate is correct.

    I'm not saying that Obama has done a great job. I think he squandered a lot of the extreme level of public goodwill that he had coming into office. However, he did drag the US kicking and screaming into the first world by passing healthcare reform (even if it didn't end up single payer thanks to campaign contributions to even his own party) and he did manage to blunt some of the economic disaster he was left with.. on the other hand he definitely should have thrown a ton of the wall streeters into jail instead of inviting them right back into the White House, but that is how the game is played now. It is the golden rule. Those with the gold make the rules.

    Romney on the other hand is hard to pin down. He has taken every stance available on every issue. For the slashdot crowd, the fact that his numbers just don't add up should be a big red flag as well. In his desperation to get elected he just tells whatever crowd he is in front of exactly what they want to hear.. facts be damned.

    The polarization, name calling, and divisiveness in politics is at an obscene level in the USA right now and unfortunately Canada isn't far behind. Truth seems to have gone right out the window.

    The spending on elections is disgusting.

    I'm so glad it will be over either way..

    • Re:As a Canadian (Score:5, Informative)

      by sydbarrett74 ( 74307 ) <sydbarrett74@gma ... minus herbivore> on Monday November 05, 2012 @11:16AM (#41881501)

      Romney on the other hand is hard to pin down. He has taken every stance available on every issue.

      I'm not crazy about Romney, but Paul Ryan ('Eddie Munster') positively scares the fuck out of me. He really is an odious, red-eyed demon. If Romney wins and decides to delegate a lot of (albeit constitutional) power to Ryan, we could be in for a world of suck.

      • Re:As a Canadian (Score:5, Informative)

        by bravecanadian ( 638315 ) on Monday November 05, 2012 @11:32AM (#41881811)

        Romney on the other hand is hard to pin down. He has taken every stance available on every issue.

        I'm not crazy about Romney, but Paul Ryan ('Eddie Munster') positively scares the fuck out of me. He really is an odious, red-eyed demon. If Romney wins and decides to delegate a lot of (albeit constitutional) power to Ryan, we could be in for a world of suck.

        Agreed. Anyone who thinks that highly of Ayn Rand and Atlas Shrugged needs their head examined.

        Even one of the disciples, Alan Greenspan, finally admitted that so much of what he believed for so long about rational self interest didn't pan out in reality.

    • Re:As a Canadian (Score:5, Interesting)

      by jjohnson ( 62583 ) on Monday November 05, 2012 @11:29AM (#41881749) Homepage

      As Canadians, we tend of overestimate the power of the presidency because we equate it to being the prime minister, and having a friendly congress to having a majority in Parliament. It's similar, but crucially it's without any party discipline. Members of your own party are likely to agree broadly with you in general, but there's no guarantee they'll vote with you, and they can actually be terrible burdens. A big part of what happened to Obama was that Blue Dog Dems, realizing they were swing votes, could command a high price for their support (e.g., Bart Stupak, a centrist Dem from Michigan, trying to add pro-life clauses to the Health Care Reform bill). Had the Dems in Congress shown any kind of unity, they could have steamrolled the Republicans. But while the Republicans have shown more discipline, it's still herding cats.

      A Canadian PM with a majority can pass pretty much any legislation that he wants that doesn't cause a PR uproar that threatens re-election. Simple as that. No American president has ever had that much power domestically.

  • He's probably right. (Score:5, Interesting)

    by yog ( 19073 ) * on Monday November 05, 2012 @10:54AM (#41881165) Homepage Journal

    But winning the battle won't win the war. Mr. Obama will be weakened by the divisive campaign; the electorate is bitterly split, and he will find Congress harder to work with. The members of Congress will be acutely aware that 48 or 49% of the popular vote went to his opponent (and he may even lose the popular vote). They will be less willing to go out on a limb to support his policies unless they are from strongly pro-Obama districts, and the average district will be closer to a 49-51 split.

    This year's elections reflect a very divided country that is uncertain how to proceed. As the wars wind down, the economy will be the foremost topic on most people's minds, and Mr. Obama has only a minority of the people's support on economic issues. Probably, we will have four years of deadlock and uncertainty followed by the 2016 presidential elections which will either vindicate Mr. Obama's big government approach, or relegate him to the history books.

    Just my humble opinions :)

    I hope that everyone votes tomorrow, regardless of your choice. The best possible outcome is that everyone votes; that way, the elections more fully reflect the will of the people, so that we can put this nastiness behind us, let bygones be bygones, and move on. Democracy -- gotta love it! The worst possible form of government, except for all the other forms of government (Winston Churchill).

    • by vlm ( 69642 ) on Monday November 05, 2012 @11:13AM (#41881459)

      This year's elections reflect a very divided country that is uncertain how to proceed.

      Sure its not the opposite? A nearly 50/50 means both sides played well and almost perfectly figured the exact lies to tell to get 51% of the electoral vote.

      Using made up numbers, its like the stereotypical Right Left value to get 50% of the vote is exactly 7.1, so the right guy played at 7.095 and the left guy played at 7.105 and its all down to who looks nicer on the camera or whatever. I guess its extremely clear the winner will only be 0.05 off the R/L scale compared to what the electorate wants. I claim this is today's situation.

      No idea what to do is when its a blowout. Lets say one side has a professional politician and the other side has the village idiot and the same 7.100 R/L gameplay position will win half the votes. So the pro aims at a 7.105 score and the moron goes for 3.995. Well its probably going to be a blow out win for the pro, but all we know "for certain" about what the population wants, is the aggregate will of the electorate is somewhere between 4.000 and 7.100, who really know for sure. I claim this is like Reagan's second term.

      Nearly 50/50 means both sides know what lies the population likes to hear. Its only a blowout when one side has absolutely no idea what lies the population likes to hear.

      There is also a temporal anomaly in that at least on the R side you need to go hard core neo to get the nomination and then intense middle ground to have a hope of election... so which lies were true? Is he a complete loon or merely slightly right of center? In other words the challenger is always less predictable than the incumbent because of the nomination process.

  • by mrquagmire ( 2326560 ) on Monday November 05, 2012 @10:56AM (#41881221)
    Does this model account for any direct or indirect vote tampering?
  • by Thud457 ( 234763 ) on Monday November 05, 2012 @11:03AM (#41881305) Homepage Journal
    This shit is ridiculous. In this age of modern speed-of-light communications, it takes TWO YEARS to choose a president. In George Washington's time, it took less time to visit every town on horseback and make his case.


    Oh, and thanks to Florida's genius governor, Rick Scott, cutting early voting short, you've got a good chance of another six weeks of court battles an recounts.
  • by rgbatduke ( 1231380 ) <rgb&phy,duke,edu> on Monday November 05, 2012 @01:02PM (#41883573) Homepage
    I've been tracking http://electionanalytics.cs.illinois.edu/ [illinois.edu] for some time now -- a University managed site that uses Bayesian analysis of all of the available polls to come up with their estimates. I tend to think that Universities are less likely to biased in their poll meta-analysis in the first place, and of course I'm a big fan of Bayesian analysis for large multifactorial problems with many levels of conditional and marginal probability. Anyway, this site is a lot less ambiguous -- it currently calls it for Obama at the level of 99.6% probability, with an expected electoral vote count of almost 303 for Obama (where "EEV" isn't an integer valued function but rather reflects the expected mean outcome of many "elections" held assuming that there is some sort of unbiased iid process underlying the poll noise). Romney's numbers are dropping, fairly rapidly, over the last week -- it looks like Obama is very likely to be 99.9% likely to win going into the actual election on Tuesday.

    You won't see this on the major news stations, of course, as they long ago gave up any pretence at objectivity in the election, and besides, if the election were statistically certain going into Election Day (as this one appears to be) it might actually influence the outcome, just as the stations that persist in claiming that Romney HAS a chance, or HAS "momentum" (whatever the hell that means outside of the context of physics in an environment where his polling numbers are almost without exception going down) are trying to influence the election, just as the stations that make the opposite claim for Obama are trying to influence the elections.

    The other nice thing about the election analytics site is that it also predicts the cumulative outcome of the Senate and House races. The Senate race currently suggests that Obama will win by enough to have some coattails to catch and help out in close races there, although the House races appear to be a lock at this point. We'll see if any house races end up as surprises -- possible if Obama beats 300 EV by a substantial margin, possible if the fact that the election is "over" in many states and districts causes the Republicans to just stay home and not bother to vote "only" for a congressional candidate where a lot of democrats show up to ride the Obama wave home.

    rgb

"Pok pok pok, P'kok!" -- Superchicken

Working...