Politics and Public Opinion, Elections, Pethokoukis

A sign Nate Silver’s FiveThirtyEight blog may have skewed way off course

103012538Intrade

There’s been a lot of controversy about the FiveThirtyEight election forecasting blog run by Nate Silver at the The New York Times. According to Silver’s analytics, day by day President Obama has been incrementally closing the deal over the past two weeks. (See above chart). That, despite national polls showing Romney with a lead and Team Obama scrambling to shore up its Midwest firewall. Yet prediction market Intrade is showing no such thing. While it does give Obama the edge, the final verdict still clearly remains in doubt.

In addition, the economic numbers continue to point to a narrow Romney win. Models have their role, but it always best to do a reality check. What are the fact on the ground? Sometimes you shouldn’t listen to the GPS. The Machine doesn’t always know best.

30 thoughts on “A sign Nate Silver’s FiveThirtyEight blog may have skewed way off course

  1. Real Clear Politics averages all polling data and predicts;
    Obama 290, Romney 248
    http://www.realclearpolitics.com/epolls/2012/president/2012_elections_electoral_college_map_no_toss_ups.html

    The Princeton Election Consortium predicts
    Obama 303, Romney 235
    http://election.princeton.edu/2012/10/29/nerds-under-attack/#more-8151

    Votamatic predicts
    Obama 332, Romney 206
    http://votamatic.org/

    Ezra Klein studies six electoral models, mostly academic and compares them to Nate Silver 538 blog, four predict Obama victory, one predicts Romney win.
    http://www.washingtonpost.com/blogs/ezra-klein/wp/2012/08/31/forecasting-the-election-most-models-say-obama-will-win-but-not-all/

    A University of Colorado prediction by two professors predict a Romeny win, 320 to Obama 218.
    http://www.theblaze.com/stories/electoral-college-model-predicts-romney-will-win-big-in-2012-and-its-been-right-since-1980/

    And if you look at the current polling data for the states, there is nothing to indicate a Romney win is likely.

    Romney’s path to an electoral victory is much more difficult than Obama’s, check out;

    http://www.buzzfeed.com/zekejmiller/how-each-presidential-candidate-can-win-in-two-ch

  2. CloseThis Deal…

    Gallup now says Obama has plunged 22-points versus where he was in the early-2008 vote count.

    That is a collapse of monumental magnitude.

    The NY Times is no more reliable than Stalin’s Pravda.

    • After his 22-point plunge, Gallup says Romney is now 7 points ahead…52-45 nationally in the early-vote count. He’s ahead in Ohio too. And the Dems say the early vote is where they always clean-up. Looks like the only thing the dems will have to clean-up are their boxer shorts and bloomers.

      NPR now has Romney up 48-47–that’s NPR, Big Bird’s house.

      When does Obama give his concession speech??? Landslide !!!

  3. Electoral-vote.com
    Obama 280, Romney 235
    http://electoral-vote.com/

    Polyvotes predicts popular vote totals,
    Obama 50.92, Romney 49.08
    http://pollyvote.forecastingprinciples.com/

    Electionprojection predicts;
    Obama 290, Romney 248
    http://www.electionprojection.com/blog/tags/2012_presidential_election.php

    Electoral Map forecast;
    Obama 281, Romney 257
    http://electoralmap.net/2012/intrade.php

    Pollicymic
    Obama 293, Romney 245
    http://www.policymic.com/articles/12379/election-prediction-10-reasons-obama-will-win-in-a-landslide

    University of Illinois
    Obama 291,4, Romney 246.6
    http://electionanalytics.cs.illinois.edu/election12/index.html

    Article, “5 reasons the Colorado University prediction is wrong” – it predicts Romney win
    http://www.nerdwallet.com/blog/markets/2012/cu-election-model-wrong/

    • Nate Silver aggregates and unskews polls. That’s it. What “content” do you want? Silver has a little black box unskewer that you can agree with or not. Just like the polls that he’s putting into his black box have already been unskewed by the pollsters themselves (controlling, e.g., for sex, age, race, region, etc.). I love to see the left rail against people who unskew polls using their own parameters, but protect Nate Silver at all costs.

  4. The RED LINE looks more like a rising wish line than it does a trend line which is declining…the unambiguous trend is declining in the opposite direction of the rising wish.

    But then again….isn’t that what Hope and Change was all about in the first place? Next Tuesday, the architect gets his payoff.

  5. Rasmussen now shows Romney winning the electoral college at 271. Nate Silver’s 538 basically missed the story of the election by relying on polls with overly optimistic turnout models biased toward Obama and adding in a feature that – up to the last minute – weighted in the performance of the economy.

  6. Silver must have been in charge of the dems early-exit polling in 2004 that had the media conclude and then announce to America that Kerry beat Bush.

    Pollsters using 2008 turnout rates, and thereby oversampling dems by 7-8 points, is producing the very same phony picture.

  7. Have you ever wondered what all of this talk about polls being “skewed?” Here’s an explanation with examples:
    “Typical Examples of 2012 Presidential Poll Skews”

    http://gulfcoastcommentary.blogspot.com/2012/11/examples-of-typical-2012-presidential.html

    Romney is likely ahead by a comfortable margin. People will call it a “late surge” or something like that, but in reality it’s been clear all along if you analyze the data.

    Ok, thanks.

  8. You guys are too funny. Gallop predicted the election wrong 4 years ago and now you’re using them as your “expert” source?????

    …..Drinking the Fox News coolaid. LOL!

    BTW, who was the best predictor in 2004? You guessed it…..Nate Silver.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>