Skip to main content

View Diary: Senate Snapshot, October 8th: Ugh (247 comments)

Comment Preferences

  •  Um, I can't imagine (11+ / 0-)

    where you are getting your numbers on the Patty Murray vs. Dino Rossi race. Just because a Rassmussen telephone poll that switched to "likely" voters recently showed Rossi with a 3 point lead? It's so effing dumb, it's beyond belief.  Rassmussen has a TERRIBLE record with its "likely" voter model in Washington. They are always wrong.

    Show me on the doll where Rahm touched you.

    by taylormattd on Fri Oct 08, 2010 at 06:56:07 PM PDT

    •  I thought from (0+ / 0-)

      here, but unless I'm reading that wrong, it has Murray up .9%.

      "Gussie, a glutton for punishment, stared at himself in the mirror."

      by GussieFN on Fri Oct 08, 2010 at 07:50:29 PM PDT

      [ Parent ]

    •  You don't have to imagine (0+ / 0-)

      In the diary, there's a link to the methodology.

      And if you click that link, there's another link to the source data.

      •  Oops (0+ / 0-)

        Never mind...that appears to be the data supporting the model, and not the actual poll data.

      •  Polling data (as stated in the diary) (0+ / 0-)

        is from Pollster.com, over the past 25 days.

        So there are 6 polls for WA-SEN (Rossi/Murray):

        Rasmussen (10/6) 49/46
        Rasmussen (9/28) 48/47
        Fabrizio, McLaughlin (R) (9/26-27) 48/42
        FOX (9/25) 47/48
        SurveyUSA (9/19-21) 48/50
        Rasmussen (9/14) 46/51

        Means are Rossi 47.6, Murray 47.3

        •  But if you go to (0+ / 0-)

          Rasmussen, you find that they don't release any information about the likely voter model that they are using for their polls of that race. We can assume that they are tilting it heavily in favor of Republicans for two reasons, 1) Rasmussen has always had better numbers for Republicans, and 2) Almost all pollsters are assuming higher Republican turnout in this election and their likely voter models reflect that assumption.

          For all we know they could be assuming that Republican turnout will be up 25% from 2008 and that Democratic turnout will be down 25% from 2008. Obviously if their top line reflects such an assumption, it is going to severely depress the Democratic candidate's numbers in the top line of that poll. You have to regard the top line with a lot of skepticism because we don't know what kind of likely voter model they are using.

          •  Agreed (0+ / 0-)
            I wasn't commenting on the validity of the data, just trying to answer the "where you are getting your numbers" question.

            That said, the explanation in the methodology link indicates that in past years, the inclusion of Rasmussen and other such polls yielded greater overall accuracy than other models out there.

            Whether that's true this year (or in a particular race prediction made up entirely of Rasmussen/SUSA polls, as with the current WA-SEN) is open to debate.  I happen to agree that this particular prediction is off, and that Murray will win.

      •  When you average corrupted data (1+ / 0-)
        Recommended by:
        wdrath

        you get corrupted averages.

        Agricultural hemp is "Jobs, Jobs, Jobs."

        by ben masel on Sat Oct 09, 2010 at 12:48:59 AM PDT

        [ Parent ]

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site