September 7, 2018

Election polling experiment

From the New York Times

For the first time, we’ll publish our poll results and display them in real time, from start to finish, respondent by respondent. No media organization has ever tried something like this, and we hope to set a new standard of transparency. You’ll see the poll results at the same time we do. You’ll see our exact assumptions about who will turn out, where we’re calling and whether someone is picking up. You’ll see what the results might have been had we made different choices.

avatar

Thomas Lumley (@tslumley) is Professor of Biostatistics at the University of Auckland. His research interests include semiparametric models, survey sampling, statistical computing, foundations of statistics, and whatever methodological problems his medical collaborators come up with. He also blogs at Biased and Inefficient See all posts by Thomas Lumley »

Comments

  • avatar
    Megan Pledger

    I can see weird things happening. Say they call between 6-8 pm local time so it starts off with the Dems ahead as New York dominates, then the Repubs get ahead as the South and Mid-West start to dominate and then Dems come back again as California and the West Coast starts getting called.

    I don’t know that this will be helpful.

    6 years ago

    • avatar
      Steve Curtis

      Its ‘poll results’ not election night results.
      “It starts tonight, when we’ll publish the first New York Times Upshot/Siena College polls of the most competitive battlegrounds in the fight for Congress”

      6 years ago

      • avatar
        Steve Curtis

        The numbers were in for their first ‘transparent poll’- a single congressional district in Orange County Los Angeles. The numbers that stunned me were these:
        ‘We made 34,782 calls, and 501 people spoke to us.’ A contact rate of 1.4%
        https://www.nytimes.com/interactive/2018/upshot/elections-poll-ca48-1.html
        They may rule people who werent registered to vote, but still last election 306,000 people voted.

        6 years ago

        • avatar
          Thomas Lumley

          Yes, one of the outcomes of this project will be more people knowing what the response rate to opinion polls is like.

          It’s impressive, given these numbers, that the big health surveys have response rates near 50% — they do *lots* of callbacks to get that.

          6 years ago

      • avatar
        Megan Pledger

        But people generally ring when most people are home – 6 pm – 9 pm. Or else you get a sample dominated by the unemployed and the retired.

        6 years ago

  • avatar
    Ben Brooks

    FiveThirtyEight covered this in their latest podcast. It’s fair to say their feelings were mixed:

    https://fivethirtyeight.com/features/politics-podcast-inside-the-trump-administration/

    6 years ago