Why the polls are BS

I have seen some "OMG the sky is falling" talk here lately, O the media is spinning this against Kerry, no bounce, no bounce bull. Lets together take a moment to examine the polls and  think about this, shall we?

First, Mr John Zogby.  His poll stating that there was no bounce from the convention cited that from July 27-29 there was no real change. Kerry went from 2 up to 5 up, but his number didnt rise, Bush's Fell.  48-46 --> 48-43

http://www.zogby.com/news/ReadNews.dbm?ID=848

well, here's a thought! from July 27-29, what hadn't happened? Oh Yeah, KERRY'S SPEECH!  the fact the Bush fell when all that can be shown from this poll is the Clinton speech for 3 days and the Edwards from 1 day means that there was any movement is a great fucking sign.

Mr. Zogby then goes on to give us more information, even which he says means that the horserace numbers are suspect.  He shows Bush getting creamed with women, singles, hispanics, young voters,was tied with men and southerners and had a 25 pt lead with people who didnt vote last time.  If you look at this page, its almost tempting to call it for Kerry. (Bush had 19% hispanic support to Kerry's 69!)

http://www.zogby.com/news/ReadNews.dbm?ID=849

Now, the folks at Newsweek.  they give us a ginormous 2 day sample poll, one day of which, like Senor Zogby, was before the Kerry speech.  I dont know about you, but post convention bounces should probably be polled "post" convention, no? A further look at the internals here is also cherries jubilee! The post convention sample was Kerry 50% Bush 40%, only factoring the pre speech segment (which even these people said means that the poll is shit) led to the 7 point lead (49/42) Plus there are a number of down screen sub polls that shout, "WTF are you talking about?! No bounce my shiny metal butt!!!"

http://www.msnbc.msn.com/id/5568072/site/newsweek/

I dont have enough info on the CNN Gallup poll, but that there was huge discrepencies between Registered and Likely voters is telling, as is the fact that the poll hasnt been released yet and people are taking it for gospel.  

Now, see, the bounce is there, and unless you think that the spin will be enough to remove it before Mr Mcswerveandcrash and Dick "fuck yourself" Cheney get their Neo-fest, go get a Mt dew if you're under 21, or a PBR if you're over it, and RELAX!

(Insert soothing Ocean noises here)

•  Jar for Tipping

  and Tipping in a Jar

  by Matt in Wisconi on Sun Aug 01, 2004 at 07:17:17 PM PDT

•  The way

  The way they determine likely voters is simple, I'm afraid.

  They simply ask, "are you likely to vote this year?"

  by beholderseye on Sun Aug 01, 2004 at 07:27:09 PM PDT

  •  No, more complicated with Gallup

    Some folks do use the method that you noted, but not all. Some of the more "complex" likely voter models are based on several screening questions involving the respondent's past voting history. Gallup also applies certain weights and makes assumptions about turnout in producing its LV model. Highly, highly suspect. Not to mention that Gallup uses a small sample for LV's--763 in this case, so the margin of error (+-4) is always higher than for the RV number (+-3). The gallup LV model has always--in every single poll--magically lumped numbers onto Bush's total. In my opinion--and Ruy Tuxeira's--it is largely to be ignored.

    Kerry leads 50-47 amongst registered voters. Now, that also doesn't show a bounce, but I'll take that number over the LV one. It has Kerry at the magical 50. And in the internals, Kerry has made major gains on leadership qualities. That being said, I think it's a bad sample, and I think USA Today knows it--that's why they've extended the poll for another two days.

    -Newsweek showed a very solid bounce, considering that only half the poll was taken after Kerry's speech. And the second night of polling, the bounce was much larger.

    -Rasmussen has now also shown a bounce. Kerry is at 49 today--that is only the second time in the entire poll that a candidate has been at 49 (the other was Kerry, shortly after the Edwards pick)--and with leaners, Kerry is at 51-46, tying the highest number recorded by this always tight poll.

    -Zogby was also showing the makings of a bounce before Kerry's speech.

    I highly suspect that CNN/Gallup is an outlier. Kerry may not show large gains in every poll, especially the ones in which he was already ahead.  But I think it's pretty evident that the convention helped him.

    by mattb25 on Sun Aug 01, 2004 at 07:40:31 PM PDT

    •  Dont get me wrong

      I think that the Gallup poll was an outlier too, but I am suspicious that the bounce was not as much as we probably all wanted it to be. Dont take my negativity for pessimisn, it leaves much room to increase those numbers... but the REASON gallup extended its poll wasnt to increase the sample size, but instead, to run individual polls in battleground states if you read the article carefully.

      Regardless, those battleground state polls will be the ones that matter. Maybe bush gained extra constituances in Texas, but that really wont mean crap when it comes to the electoral college.

      The real poll ALL of us should watch, and mark my words.. isnt newsweek, Rasmussen, or Gallup, but instead, the Zogby poll that will come out August the second, tommorow, that will show us the state-by-state breakdown.

      Those will be the important numbers.

      by beholderseye on Sun Aug 01, 2004 at 07:45:18 PM PDT

      •  Oh, I know that's there stated reason

        though I suspect it's also a decent way of covering their ass if the rest of the polls show something different.

        My guess is we'll see about a 5 point bounce in most polls, and that would be very solid for a challenger who was already head in an election in which most folks have made up their mind.

        Gallup may not have even made a sampling error. But when you try and make all those likely voter assumptions about a sample of 763 people, funny things can happen--and they always seem to do so in these gallup polls. Their reputation has really taken a hit--they've been pretty off the mark in the most recent presidential races.

        by mattb25 on Sun Aug 01, 2004 at 07:48:49 PM PDT

    •  CNN

      What gets me is that CNN reports the Bush win poll as gospel without explaining the whole truth about the numbers.  Of course, this is CNN I'm talking about...  I think they'll dismiss any other poll as not true, since their numbers are "right".

      by mlk on Sun Aug 01, 2004 at 07:53:32 PM PDT

      •  CNN is a disgrace

        But I was surprised that they don't have this poll plastered all over the website. I looked to me like even Blitzer didn't really believe the numbers.

        by mattb25 on Sun Aug 01, 2004 at 07:55:16 PM PDT

    •  Polls are BS when pols don't like the numbers

      But the Gallup (actually USAToday) result is interesting. [CNN, Gallup and USAT all publish results from the same Gallup sample, but all three can publish divergent results based on divergent screening and weighting models.]

      The 50-47 RV, 47-50 LV result implies their LV model had pro-Bush respondents 13% more likely to vote than pro-Kerry respondents. That strikes me as rather high, especially considering Kerry's likely advantage among seniors. Kerry's sample will be heavy with minorities and first-time voters ... and maybe that's where the tsunami is lurking.

      That's 13% if the percentages were exactly 50.00-47.00 etc. Published results -- rounded to the nearest percent -- could be hiding a 50.5-46.5 ratio, or a 49.5-47.5 ratio, so the Bush/Kerry LV turnout factor could be anywhere from 8% to 18%.

      by RonK Seattle on Sun Aug 01, 2004 at 07:55:49 PM PDT

•  New poll out

  Bush 50
  Kerry 47
  I hope it's garbage.
  http://www.pollingreport.com/wh04gen.htm
  

  by jfern on Sun Aug 01, 2004 at 07:31:42 PM PDT

  •  That isn't listed on the site

    Is this the NBC/WSJ poll? We knew that would be tilted towards the GOP.

    by JamesB3 on Sun Aug 01, 2004 at 07:38:20 PM PDT

    •  It's Gallup

      by paperbag on Sun Aug 01, 2004 at 07:46:43 PM PDT

    •  Nope

      That's the CNN poll we've already covered ad nauseum.

      Look at things this way: this is a very tight race.  Regardless of a convention bounce or no convention bounce, this thing is going to come down to turnout.

      by Jonathan on Sun Aug 01, 2004 at 07:49:40 PM PDT

      •  We need some kind of bounce

        because it looks awful for Democrats every time CNN and the other media whores keep saying there was nothing. They are saying that Democrats are losers.

        After the Thug hatefest, when the media will go crazy praising them to the skies 24/7, they will make sure he gets a huge bounce. They will say this was Bush's "turning point" and the "moment he became the frontrunner", and all the usual garbage. That's why we need some sliver of a positive vibe, because the conventional wisdom is becoming that this is all a big disaster for Democrats. That they gave their best, and the public didn't want it.

        by JamesB3 on Sun Aug 01, 2004 at 07:58:06 PM PDT

        •  The histrionic terror alerts

          which are all over AOL and all the major broadcast media, and being pushed as THE terror alerts for the ages, bigger than anything else, are also having a big effect for Bush.

          by JamesB3 on Sun Aug 01, 2004 at 08:00:20 PM PDT

          •  They are?

            James, do you have any polling data to support this?

            If not, then don't assume the worst!

            I repeat, this thing is going to come down to turnout.  If Gore (a lackluster candidate who ran a terrible campaign) could beat Bush in 2000, despite all the media advantages Bush had then, Kerry can win this thing.  And don't tell me about the advantages of incumbancy: incumbancy comes with a record, and Bush's sucks.

            Let's fight this thing, no matter what the polls say, like we're going into the election dead even!

            by Jonathan on Sun Aug 01, 2004 at 08:05:12 PM PDT

            •  9/11 and terror

              are still the key factors for Bush. That's why the GOP is playing them for all they are worth, and getting the results they want.

              by JamesB3 on Sun Aug 01, 2004 at 10:20:17 PM PDT

              •  I don't see the results

                Look, Bush has been losing ground in every poll on the topics of terrorism and who would make a better commander-in-chief.  How is playing up terrorism going to help him under those conditions?

                My call is this:  we're reaching a point where further fear-mongering regarding terrorism is going to inspire a desire for a change in leadership.

                I gotta tell you, I have enough trouble not getting depressed with the whipsaw horserace.  I don't know how you stay sane being this pessimistic, James ;^)

                by Jonathan on Sun Aug 01, 2004 at 10:30:48 PM PDT

  •  New poll out

    Kerry 51%
    Bush 45%

    I hope it true.

    Post-Convention Polls Mixed on Kerry Bounce

    by assyrian64 on Sun Aug 01, 2004 at 07:44:34 PM PDT

    •  Newsweek

      by paperbag on Sun Aug 01, 2004 at 07:45:51 PM PDT

      •  These

        These are not new polls. These have been polls that have been out for quite a while.

        by beholderseye on Sun Aug 01, 2004 at 07:48:06 PM PDT

        •  Well...

          ...they are new polls, but they've been discussed previously.

          Everyone should relax, IMHO.  We won't have a full picture until week's end how the convention impacted things.

          by Jonathan on Sun Aug 01, 2004 at 07:51:16 PM PDT

•  I don't think CNN has too much faith in the poll

  Look on their politics splash page.  There is no link to it.  It seems to me that if they thought they were really on to a ticket that is worse that McGovern-Eagleton, it would be the lead story:

  http://www.cnn.com/ALLPOLITICS/

  by DCDemocrat on Sun Aug 01, 2004 at 08:11:05 PM PDT

•  watch what happens on election day

  I predict high turnout, something the polls are for the most part letting fall through the cracks.

  I predict that this year will be one of those years that turns decades of reliable polling methods and data on their collective head.

  by Sauceman on Sun Aug 01, 2004 at 09:17:46 PM PDT