Rasmussen Reports Kerry 48% Bush 48%

Rasmussen Reports ^

[renotse]

Kerry 48% Bush 48%

[Paradox]

Thanks! I am glad that this is more than just wishful thinking.

[MNJohnnie]

Saturday raw data
Bush 43.7
Kerry 48.1

Sun
Bush 48.3
Kerry 47.8

Mond
Bush 51.4
Kerry 47.5



Where do you get this? I have been giving myself an ulcer until I realized that the data had to flawed. Thanks a LOT for the confirmation!

GO W!

[rocklobster11]

Wherever you are getting your numbers from for individual days is incorrect. Rasmussen explicitly stated yesterrday that each of the three days of polling had Kerry in the lead. Thus Sunday was not 48.3-47.8 in favor of Bush. I'm betting it was very close to a tie with a very slight Kerry lead.

Someone must be calculating these daily numbers, because I see the same numbers repeated over and over, but they are incorrect.

[MNJohnnie]

but SurveyUSA has been the best state pollster along with Mason-dixon. Of course SurveyUSA could be doing something wrong this election.



Source for this? All I can find is that Survey USA does polls for hire. Who is paying for the Survey USA polls? I cannot find any 3rd party data that backs up this claim for Survey USA

[True_wesT]

Just crapping out the weekend sample.

[rocklobster11]

If you look at the chart you posted, it is obvious that the "weekend bump" for Kerry is a myth. The chart doesn't show the high or low points coming at regular intervals. Rather, it shows the polling randomly bouncing within the margin of error with Bush most likely at 48% and Kerry most likely at 46%. The numbers from Saturday were obviously a bad sample, which makes sense given a poll with a 95% confidence level.

[rocklobster11]

This is raw data (after he weights everything) that I've been tracking since January.

You sure this is actual raw data from Rasmussen, or is it your estimate of the raw data.

[neutrality]

I don't have any related articles bookmarked, but reading their election track record scorecard should at least give them some credibility:

http://www.surveyusa.com/electiontrackrecord.html

http://www.surveyusa.com/SurveyUSA2002ScorecardSenate.PDF

[AmishDude]

Weekend falling away...

Zzzzzzz...

Call me on Thursday...

[AmishDude]

It wouldn't be a push poll if they asked the draft question last. At least in terms of the poll results.

[LS]

Well, it's absolutely clear that this was the "October Surprise," and for whatever reason the NY Slimes jumped the gun and gave us time to detect the lie before it did damage. If I'm Kerry, I'm very worried: absolutely NOTHING he has done can get him the lead two days in a row with ANY pollster at ANY time.

[neutrality]

It's a myth that just won't die. I challenge anyone to find a pollster supporting it.

[LS]

His Florida analysis was VERY encouraging to me. Basically, the pros are going into those counties and they will NOT underperform in 2004!

[Petronski]

I don't think it's a myth at all. I see a pretty consistent pattern of the thick red line peaking whenever a vertical dotted line passes by. No one said it showed perfect rhythm, but there is a periodicity generally consistent with a 7-day cycle.

[MNJohnnie]

I don't have any related articles bookmarked, but reading their election track record scorecard should at least give them some credibility:
http://www.surveyusa.com/electiontrackrecord.html

http://www.surveyusa.com/SurveyUSA2002ScorecardSenate.PDF


I'm sorry but I cannot except their word that they are great pollsters. That is just marketing. Survey USA doesn't show up on any of the Harris Historical studes of Pollster's accuracy. Did they have a different name or have they always been Survey USA?

[neutrality]

I'm not sure how you can dismiss their track record so quickly, but here's what Polipundit says about SurveyUSA:

http://polipundit.com/wp-comments-popup.php?p=1963&c=1

[Eva]

No, you are right, we don't know the order of questions, but what would be the purpose of the question being asked at all. What % of the people would you expect to answer affirmatively? There would be no other reason to ask this question than to influence voters.

[hchutch]

The GOP has a ground game this year, unlike 2000.

If Bush is ahead in Gallup, and has a job approval of 50+%, then he is in good shape, I think.

[MNJohnnie]

Now there’s a University of Oregon study that compared the accuracy of Zogby and SurveyUSA polls and came to these conclusions:


One important, although not surprising, finding is that published partisan polls reflect a partisan bias. The other fairly strong finding here is that despite the criticism to which they were subjected, Zogby International and SurveyUSA performed at roughly the same level as other nonpartisan polling organizations in 2002. By most measures, Zogby did just slightly more poorly than the norm, while SurveyUSA did somewhat better.


Doesn't seem much of an endorsement for Survey USA. "Zogby did just slightly poorer"

[rocklobster11]

I don't think it's a myth at all. I see a pretty consistent pattern of the thick red line peaking whenever a vertical dotted line passes by. No one said it showed perfect rhythm, but there is a periodicity generally consistent with a 7-day cycle.

Okay, I put together a little graphic to overlay each weeks samples for 7 weeks:

The vertical lines (9/3 and 9/10) were Fridays, so you could argue that Bush has his worst day on Wednesday (which includes Sun,Mon,Tue samples) and his best day on Sunday (which includes Thu, Fri, Sat samples), which are conversely Kerry's worst days.

You could also argue that it is all just statistical noise, and 7 weeks isn't a good enough period to analyze, especially considering the Debates which occurred throughout the sampling period.