Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

Poll geek humor

September 4th, 2008, 9:08pm by Sam Wang

Many of you won’t get this joke by Nate Silver over at FiveThirtyEight…

In the live thread leading up to McCain’s speech, Silver writes:

7:36 CDT: [Nate] Gallup has the score Giants 20, Redskins 6. Rasmussen has them ahead 14-7. But Zogby Interactive has it Redskins 11, Giants 5.

I found this to be absolutely hilarious.

I think I spend too much time looking at polls.

(By the way, the score at the time was actually Giants 16, Redskins 7, third quarter.)

Tags: 2008 Election

6 Comments so far ↓

  • Independent

    Most of us diletantes share your sense of the surreal when it comes to polling organizations. When and how do you think they can be made to shape up?

  • Blaise Pascal

    What’s the reported margin of error on those scores?

  • Sam Wang

    When their customers (media organizations) decide they are better off with more accurate information. However, with small error bars and contradictory polls it is possible to squeeze more stories out of the data. So my guess is never.

    In the meantime, keep reading sites that combine polls…

  • Michael

    This (funny) post brings to question something that I have always wanted to ask both you and Nate. Can you say something about the economics of polls? How exactly does Gallup or Zogby make money off a poll? (I have been guessing it’s in selling more detailed information to campaigns. Is that right?) Do you have any sense of how much money is at stake? Do different polls make money in different ways?

    If this is outside your field — or too much to take on here — just ignore.

  • Mark Powell

    Reminds me of a modeling joke:

    2 + 2 = 5 for very large values of 2 and very small values of 5.

  • Sam Wang

    The question of polling economics is a fascinating one, but unfortunately I don’t know much. I will look into this. For now…

    Polling organizations can make money by making their results available early to their media organization customers, including background details. Long-term partnerships are obviously a source of stable income.

    It’s a great question about whether partisan campaigns are a major source of income to polling outfits. They only directly sponsor a small fraction of polls. In general, polling doesn’t seem to be all that expensive. The liberal site DailyKos has commissioned several polls, which I recall cost only a few thousand dollars to do.

    Exit polls are their own animal – media organizations pool resources to commission those. I think what they are buying is the ability to call races with large (>20%) margins early in the evening. They’re generally not used for really close states because they are not precise enough.

    I’ll definitely read more about this. Thank you for the question.

Leave a Comment