Skip to main content

View Diary: More accurate than Nate Silver (308 comments)

Comment Preferences

  •  Aggregation vs. polling (2+ / 0-)
    Recommended by:
    Angie in WA State, Onomastic

    I heard someone from Gallup interviewed on NPR a few days ago, asking about the success of Nate Silver, over Gallup's projections. He made a good point, which I suspect you (Markos) have already thought about, but might not be apparent to others who haven't made statistics their life's passion.

    Nate's and your methods are largely by sampling (in a statistical sense) the reported results of various for-profit polls. Those polls, individually, were each less accurate than a statistically massaged composite, as pointed out above. That's the power of statistics (and a well-designed Monte Carlo simulation, especially.)

    Those polls are very expensive to conduct. Four years from now, the demand for these expensive polls will be driven by the media's recollection of Nate Silver's success. (Very few, other than the present readers, know of Markos' success.) I hope we don't see the demise of non-partisan polling; there will be fewer polls for Nate and Markos to sample!  And the quality of remaining polls will be suspect, since they will be necessarily driven by political parties with a very specific agenda.

    I think we have to be pretty careful about comparing these statistical analyses to polling results. Each has an important role.

    Anyway, just my 2 ȼ.

    -5.38, -2.97
    Romney went to France instead of serving in our military, got rich chop-shopping US businesses and eliminating US jobs, off-shored his money in the Cayman Islands, and now tells us to "Believe in America."

    by ChuckInReno on Wed Nov 14, 2012 at 10:18:12 AM PST

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site