Skip to main content

View Diary: ABC/WaPo: O 49/ R 46 LV (410 comments)

Comment Preferences

  •  Pew polls stink... (6+ / 0-)

    I don't want to see another one of their shitty polls ever again if I can help it.  The whole reason Mitt had any bounce was from their crappy outlier last week that set the media narrative.


    by LordMike on Sun Oct 14, 2012 at 10:00:45 PM PDT

    [ Parent ]

    •  Pew is actually usually reliable (8+ / 0-)

      That's part of the reason why the poll caused so much chaos. If it was Rasmussen nobody would've blinked an eye. I think Pew just caught a huge enthusiasm gap and had an outlier on top of that enthusiasm gap.

      •  They're usually reliable until they're not... (7+ / 0-)

        They gave Bush a 15 point lead after his convention, too... a crock of shit.  Zogby was the best pollster out there until he screwed us all believing in a Kerry win, and now he has to do email polls for Newsmax 'cos he sucks so badly.  Pew's credibility is shot now, too, IMO.


        by LordMike on Sun Oct 14, 2012 at 10:05:50 PM PDT

        [ Parent ]

      •  They're usually just very susceptible to swings... (1+ / 0-)
        Recommended by:

        I'm not sure why, exactly.. they have a very good reputation but they have polls just showing massive swings in both directions. According to RCP they showed Obama up +15 in late October 2008! Their final poll was near perfect at Obama +6, but along the way they showed huge numbers both ways. Did anyone really believe the Obama +8 number they were showing before the Romney +4?

      •  Pew was the worst outlier in Obama's (1+ / 0-)
        Recommended by:

        direction 3 weeks prior to their latest poll.   Their Obama +8 poll was the most lopsided one in Obama's direction, and right after that shocker came out (and was widely regarded as an outlier) the whole unskewed laugher started on the right.   I guess Pew is an equal opportunity offender.  Not a single other polling firm showed as large an Obama lead pre-debate, and not a single other polling firm showed as large a Romney lead post-debate (among known pollsters.)    I also remember Pew had some head scratchers in 2008 as well.  

    •  Please stop (9+ / 0-)

      Pew is one of the best in the business.  You can't reject legitimate pollsters just because you don't like the result.  That's a very wingnut thing to do.

      •  True, but they are very sensitive to (3+ / 0-)
        Recommended by:
        brn2bwild, DefiantOne, Supavash

        temporary emotional blips....

      •  Do you believe their result last week was accurate (1+ / 0-)
        Recommended by:

        Did any other poll corroborate it?  No.  It was a crap poll, period.  Yes, even the best firms have outliers, but this one was pretty extreme.  

        As for being the best in the business.   Zogby once was the best in the business, too.  Now he polls for soap companies.  Past performance does not guarantee future results.


        by LordMike on Sun Oct 14, 2012 at 10:17:36 PM PDT

        [ Parent ]

      •  Even the best pollsters ave outliers (4+ / 0-)

        from time to time. If a pollster never has any outliers they would be suspect IMO. I'm hoping that R +4 was Pew's outlier this season and their nest poll will show an Obama lead.

        •  uggh, my spelling is terrible. (1+ / 0-)
          Recommended by:
        •  It's hilarious. (0+ / 0-)

          Pew posts a national poll in line with Daily Kos's own poll, but its an oulier. WaPo/ABC posts a poll different from every other national poll, but that's spot on legit!


          I think I might go over to the right wing sites posting about this poll and compare them to some exact same comments over here about the Pew Poll.

          You can't pick pollsters you like when they show you stuff you want to hear. You gotta pick reputable pollsters and monitor trends.

          •  oh really?? (0+ / 0-)

            Just how is this ABC poll different then what we are seeing so far this week? The trackers like IBD,Reuter's and Rand are showing the same thing as the ABC poll. You also have to figure in the pretty good state polls from last mid-week period on.

            •  No they aren't. (0+ / 0-)

              Tracking polls and field polls are not the same thing. Not only is the order and number of questions different, but weighting and sample sizes are done differently.

              I examine the Rand poll closely on a daily basis and it never gave a Romney lead but definitely noted a significant gain in support. I'm neutral on the Reuters poll and pay no attention whatsoever to IBD.

          •  You misunderstood me. (0+ / 0-)

            I'm not crapping on Pew or their poll. I was hoping its an outlier. An outlier does not meanits not legit btw. And I dot think any other national poll has shown Romney +4.

      •  Sorry, but going from Obama +8 to Romney +4 (3+ / 0-)

        in the span of 3 weeks is insane.  A 12% swing?  Come on.  No other firm had anything close to that.  Trackers showed slight movement, even Ras.  Other polls were fairly close to their previous findings (showing at most 3, 4% movement towards Romney, often less. )  But 12% is unrealistic.  It makes BOTH the Obama and Romney polls strong outliers.   How anyone can claim that "Pew is one of the best in the business" when you are looking at an obviously ridiculous spread here, is beyond me.   Polling firms sometimes have the magic sauce, sometimes they don't.   In this cycle Pew has looked terrible.  It happens.  

        •  The technical term is "wacky" data. (0+ / 0-)

          If you do a poll at the wrong time, the data can be thrown way off kilter by a bump in either direction.

          •  Both of their polls have been outliers (2+ / 0-)
            Recommended by:
            DefiantOne, GoogleBonhoeffer

            in either direction.   When that happens you simply don't have it.  The race has been fairly close, bouncing from +2 to -2 in all the trackers - Rasmussen, Reuters/Ipso, IBD/TIPP, Rand, Gallup.   So, a +8% Obama poll is just as crappy as a +4% Romney poll, and Pew managed to hit both of those outlier results in the span of less than 3 weeks.   Prior to that?  The Pew poll before that from mid-July showed Obama with a 10-pt. lead, another gross outlier, the same week Democracy Corps showed a 4-pt. Obama lead, Gallup showed a 2-pt. Obama lead, and Rasmussen showed a 3-pt. Romney lead.  

        •  95% confidence interval (1+ / 0-)
          Recommended by:

          1 in 20 polls will fall outside that MoE.

          An MoE of 4.0 would mean any tie race would, 95% of the time, show up anywhere from R+4 to O+4.

          One in twenty would find a result > +4 for either candidate in that scenario.

          Maybe they've just been really unlucky? Maybe they take so many snapshots that they have a lot more data in the 5%? Maybe I'm too tired to come up with reasons?

          •  Maybe they are not as good as a lot of (1+ / 0-)
            Recommended by:

            people think they are.   All 3 polls this cycle have been complete crap, made no sense.  O + 10 in July, O + 8 in September, R + 4 last week.   In 2008 their polling was very crappy as well, with outlandish results, O +15, O+12.

            We give Rasmussen crap for having bad polls all season long just to adjust them a week before election day to claim they were amongst the closest, if Pew has a good poll right before election day, should that excuse threir completely terrible polling throughout the cycle?    Personally, I don't think so.  And, I say that even if they show a "great" poll with Obama suddenly surging to a 10 point lead tomorrow.  

    •  When Obama was leading by 8 (6+ / 0-)

      I thought the numbers were a little bit too high, even though I liked them.

      Then suddenly it's Romney + 4

      A 12 point bounce? Really? Not credible at all.

    •  Pew polls take a long time to answer... (0+ / 0-)

      ...which is why they tend to swing wildly in their results. If you're demoralized you don't want to finish the poll. If you're excited, you tell everybody about it. That's why Pew swings so much - self selection bias. It's a measure of excitement, for sure, but the beta is too high.

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site