Skip to main content

View Diary: Gravis Marketing Exposed As A Fraud (via Democratic Underground) (108 comments)

Comment Preferences

  •  It doesn't take much to make robocalls. (6+ / 0-)

    Tools used? A computer, a cell phone, and a web hosted company to put out the calls (www.voicebroadcasting.org is one).

    In other words, it doesn't take much to make automated calls.

    Or do polling.

    Sorry, I do not see a smoking gun.

    BTW here's a link to some images from their Twitter account: https://twitter.com/...

    •  Robo call mechanics.... (13+ / 0-)

      is one thing, but what takes a bit more skill is the crafting of poll questions.  Good pollsters know from experience that the way you word questions and the order you ask them can have unforseen impacts on results if you don't know what you are doing.

      Kos always makes a big deal out of putting all the data out there for others to see in trying to analyze which pollsters are legitimate and which are cooking the books.

      Has anyone seen detailed data from Gravis with questions and cross tabs, etc.?

      Free markets would be a great idea, if markets were actually free.

      by dweb8231 on Sun Oct 07, 2012 at 01:25:15 PM PDT

      [ Parent ]

      •  Gravis goes overboard with crosstabs (9+ / 0-)

        Some of their reports run over a hundred pages, with endless pages of chi square tables for every possible variation (seriously, I found one for how Latina Jewish women were voting in the Nelson/Mack Senate contest in one of their Florida reports a few weeks back). They present their numbers to the hundredth of a percent ... a ludicrous (and meaningless) degree of precision.

        I sometimes whine about pollsters who don't release enough details and cross-tabs, but the first time I clicked on a Gravis report it struck me as ridiculous beyond belief ... like someone trying to snow folks by presenting tons and tons of meaningless data in order to make it look like they are serious.

        I don't know that this DU article proves anything, but any reasonable gut check on Gravis and Doug makes it clear they reek with suspicion.

        My friends, love is better than anger. Hope is better than fear. Optimism is better than despair. So let us be loving, hopeful and optimistic. And we’ll change the world - Jack Layton

        by terjeanderson on Sun Oct 07, 2012 at 03:00:02 PM PDT

        [ Parent ]

        •  more likely they actually did do the poll (4+ / 0-)

          and didn't know or care how to refine the tabs.

        •  Are you serious? Do you have a link? (12+ / 0-)

          No reputable pollster or other researcher would report results to hundredths of a percent unless the MOE is under 1 percent, because to do so implies a level of precision which simply isn't present.  And they simply CAN'T have polled enough Latina Jewish women to have produced a number on which the MOE isn't so high that whatever number they come up with is anything other than completely meaningless.

          To me, what you just pointed out is, if anything, even more damning than what is in the diary.

          Bin Laden is dead. GM and Chrysler are alive.

          by leevank on Sun Oct 07, 2012 at 03:17:34 PM PDT

          [ Parent ]

          •  Here's a recent one (5+ / 0-)

            I don't think this is the one with Latina Jews... that one had a report over 100 pages, this one is just 44. The ridiculous cross-tabs start on page 9 and run until the end.

            Gravis Florida 9-29It is a PDF.

            I don't find this is necessarily proof of anything (except stupidity) ... I suspect that Doug simply ran the computer analysis and doesn't know enough about polling to understand what is and isn't significant in the numbers that spew out. He then throws them all up in a report thinking they give him credibility.

            My friends, love is better than anger. Hope is better than fear. Optimism is better than despair. So let us be loving, hopeful and optimistic. And we’ll change the world - Jack Layton

            by terjeanderson on Sun Oct 07, 2012 at 03:26:16 PM PDT

            [ Parent ]

            •  It's worse than that (10+ / 0-)

              They say the poll had a sample size of 914 respondents.  One person would therefore be 0.109% of the sample.  Rounded to the nearest hundredth of a percent, that would be 0.11%.  They give several categories as 0.10%, which is simply impossible, unless they rounded down.  But OK, maybe they rounded down, although why they'd do that I have no idea.

              But it gets worse.  They give some percentages that are 0.05% or even lower.  For example, on the tablulation of likelihood to vote by Zimmerman guilty, they report 0.07% as being somewhat likely to vote and "don't know" on Zimmerman's guilt, plus another 0.07% as being somewhat likely to vote and "unsure" on the Zimmerman guilt question.  Either there was nobody who was in either of those categories, or there was one person, and the percentage should have been 0.00% or 0.11% (or possibly 0.10%).

              Similarly, they report 0.05% of respondents as being Jewish and of "Other/Unsure" race.  They report 0.06% as believing America is on the right track and of "Other/Unsure" race, and 0.06% of respondents being Muslim and that the U.S. is headed in the right direction.  They report 0.07% of respondents as being Muslims who believe America is headed in the right direction.

              Bin Laden is dead. GM and Chrysler are alive.

              by leevank on Sun Oct 07, 2012 at 04:08:22 PM PDT

              [ Parent ]

          •  sounds to me that their spreadsheets... (1+ / 0-)
            Recommended by:
            terjeanderson

            ... were hastily done with some arbitrary default value for the number of digits after the decimal place.  

            "Minus two votes for the Republican" equals "plus one vote for the Democrat." Arithmetic doesn't care about their feelings either!

            by G2geek on Sun Oct 07, 2012 at 11:09:45 PM PDT

            [ Parent ]

    •  If i ever get calls from robots with... (0+ / 0-)

      .... cellphone audio, it's 100% certain that I will hang up on them without wasting even a single word.  

      "Minus two votes for the Republican" equals "plus one vote for the Democrat." Arithmetic doesn't care about their feelings either!

      by G2geek on Sun Oct 07, 2012 at 11:06:54 PM PDT

      [ Parent ]

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site