Newsweek Poll Horribly Weighted

There have been three major studies on party identification within the past year, one released by Harris in late February, one conducted by Pew in June and July, and the other conducted by the National Annenberg Election Survey during the first three weeks of July. Combined, the three surveys interviewed nearly 30,000 people, with Pew coming in at 19,118, Harris weighing in at over 6,000 and NAES at 3,715 registered voters out of 4,275 interviews. All three surveys had very low margins for error and weighted their results to reflect national demographics. Their results were as follows:
	DNC  RNC  IND  NA
Harris	33   28   24   15
Pew	33   29   --   38
NAES	34   32   24   10
The reliability and sheer size of these three polls make one thing pretty clear. Right now, Party ID in this country, when leaners are not pushed, is 33% Democrat, 29% Republican, and 38% independent / not aligned. Considering this, much of the bounce from the Newsweek poll is generated by poorly weighted Party ID numbers.
	   Party ID  Bush Kerry    Und/Oth
Republican   38%       94    4	   2
Democrat     31%       14   82	   4
Independent  31%       45   40	  15
Total	    100%      52   41	   7
However, if the sample was properly weighted, here are what the results would look like:
	   Party ID  Bush  Kerry   Und/Oth
Republican   29%      94      4	   2
Democrat     33%      14     82	   4
Independent  38%      45     40	  15
Total	    100%     49.0   43.4     7.6
The number one mistake a pollster can make is to conduct a survey from an unrepresentative sample population. That is exactly what Newsweek has done. Of course Bush is going to lead in a poll with a sample that is disproportionately made up of Republicans. In reality, in the Newsweek poll Kerry is only down by 5.6 in a three way trial heat, and thus is probably only down by 4 or 5 in a two way trial heat.

Five points is a lot less than eleven, especially when one considers that this is probably Bush's peak (almost every candidate reaches their peak immediately after their convention). Kerry is clearly in the game. Spread the word--this poll stinks.

Tags: General 2008 (all tags)

Comments

19 Comments

Re: Unfortunate
ABC did exactly the same thing after the DNC convention. They released a poll with 39-29 Dem to GOP ID, which is bogus. Not surprisingly, that poll showed a large lead for Kerry. In my more cynical moments I wonder if some firms do this intentionally just so that get more press by producing wild results.
by Chris Bowers 2004-09-04 12:03PM | 0 recs
Re: What about the Time poll?
Newsweek isn't wrong when their results are properly weighted. It is somewhat better for Bush than the others, but it was taken only on Thuirsday and Friday, unlike ARG or Zogby. They also pushed leaners harder than those other two. Supposedly Rasmussen himself has claimed that the same bogus weighting and pushing took place in the Time poll as took place here. I'll be on the lookout for internals to see if this is true.
by Chris Bowers 2004-09-04 12:07PM | 0 recs
Re: Pew Poll
Thank you very much. Very useful.
by Chris Bowers 2004-09-04 03:33PM | 0 recs
Re: This is a golden opportunity for Kerry
If Kerry's "great closer" rep is to believed, we may be in for a fight.
by Geotpf 2004-09-04 03:49PM | 0 recs
Re: Polls
Zayin, still think Nader is going to drop out?

(He's an astrologer-he was posting over at the Rep-leaning Election Projection site before the guy running that site shut down the comments.)

by Geotpf 2004-09-04 03:52PM | 0 recs
There's a 100% accurate way to find out party ID
Data on which party somebody is registered to is public information in most, if not all, states.  Just look up the actual numbers-you don't have to poll anybody.
by Geotpf 2004-09-04 03:54PM | 0 recs
Not in Missouri
In Missouri voters do not register by party. There is no party ID on the voter file.
by Michael Bersin 2004-09-05 01:53AM | 0 recs
Re: Party Registration
And that is also not an accurate indication of real party ID. For instance, I am a registered Republican. Why? Because I live in Wyoming where you can register at the polls and, also, switch your registration. Because we are such a Repub-dominated state, many of the most interesting and significant races are in the Republican party. Last month, I switched over to vote against Barbara Cubin -- I'll take every opportunity they give me. They claimed not to have enough forms to allow me to switch back after I voted, but I suspect it was retaliation for my "Somewhere in TX a village is missing its idiot" t-shirt. So I will switch again but, in Wyoming, I know you could find a lot of people at different times who are registered in a party that they feel very little affinity for.
by kainah 2004-09-06 04:06PM | 0 recs
Re: the polls
Don't expect Rasmussen to grow at all. Tomorrow the Wednesday sample that was so good for Bush is being removed. The Bush lead should drop to around 3 points.
by Chris Bowers 2004-09-04 03:59PM | 0 recs
Re: Polls
zayin you claim to be an astrologer who declared that it was in the stars that kerry would win. so tell me what happened ? doesn`t look very plausable now.i guess astrology is just bunk. a curious thing though a website www.starlightnews.com also called for kerry to win.hey who know`s miracles can happen.
by EXBUSHVOTER 2004-09-04 04:52PM | 0 recs
polls
you are missing the point. it doesn`t matter if the polls are correct  or not it`s the perception that bush has an 11 point  lead that counts.
by EXBUSHVOTER 2004-09-05 05:01AM | 0 recs
WOW !!
 
 Posted on Sun, Sep. 05, 2004

Graham book: Inquiry into 9/11, Saudi ties blocked

By FRANK DAVIES

fdavies@herald.com

WASHINGTON - Two of the Sept. 11, 2001, hijackers had a support network in the United States that included agents of the Saudi government, and the Bush administration and FBI blocked a congressional investigation into that relationship, Sen. Bob Graham wrote in a book to be released Tuesday.

The discovery of the financial backing of the two hijackers ''would draw a direct line between the terrorists and the government of Saudi Arabia, and trigger an attempted coverup by the Bush administration,'' the Florida Democrat wrote.

And in Graham's book, Intelligence Matters, obtained by The Herald Saturday, he makes clear that some details of that financial support from Saudi Arabia were in the 27 pages of the congressional inquiry's final report that were blocked from release by the administration, despite the pleas of leaders of both parties on the House and Senate intelligence committees.

Graham also revealed that Gen. Tommy Franks told him on Feb. 19, 2002, just four months after the invasion of Afghanistan, that many important resources -- including the Predator drone aircraft crucial to the search for Osama bin Laden and al Qaeda leaders -- were being shifted to prepare for a war against Iraq.

Graham recalled this conversation at MacDill Air Force Base in Tampa with Franks, then head of Central Command, who was ``looking troubled'':

``Senator, we are not engaged in a war in Afghanistan.''

''Excuse me?'' I asked.

''Military and intelligence personnel are being redeployed to prepare for an action in Iraq,'' he continued.

Graham concluded: 'Gen. Franks' mission -- which, as a good soldier, he was loyally carrying out -- was being downgraded from a war to a manhunt.''

Graham, who was chairman of the Senate Intelligence Committee from June 2001 through the buildup to the Iraq war, voted against the war resolution in October 2002 because he saw Iraq as a diversion that would hinder the fight against al Qaeda terrorism.

He oversaw the Sept. 11 investigation on Capitol Hill with Rep. Porter Goss, nominated last month to be the next CIA director. According to Graham, the FBI and the White House blocked efforts to investigate the extent of official Saudi connections to two hijackers.

Graham wrote that the staff of the congressional inquiry concluded that two Saudis in the San Diego area, Omar al-Bayoumi and Osama Bassan, who gave significant financial support to two hijackers, were working for the Saudi government.

Al-Bayoumi received a monthly allowance from a contractor for Saudi Civil Aviation that jumped from $465 to $3,700 in March 2000, after he helped Nawaf al-Hazmi and Khalid al-Mihdhdar -- two of the Sept. 11 hijackers -- find apartments and make contacts in San Diego, just before they began pilot training.

When the staff tried to conduct interviews in that investigation, and with an FBI informant, Abdussattar Shaikh, who also helped the eventual hijackers, they were blocked by the FBI and the administration, Graham wrote.

The administration and CIA also insisted that the details about the Saudi support network that benefited two hijackers be left out of the final congressional report, Graham complained.

Bush had concluded that ''a nation-state that had aided the terrorists should not be held publicly to account,'' Graham wrote. ``It was as if the president's loyalty lay more with Saudi Arabia than with America's safety.''

Saudi officials have vociferously denied any ties to the hijackers or al Qaeda plots to attack the United States.

Graham ran unsuccessfully for the Democratic presidential nomination and then decided not to seek reelection to the Senate this year. He has said he hopes his book will illuminate FBI and CIA failures in the war on terrorism and he also offers recommendations on ways to reform the intelligence community.

On Iraq, Graham said the administration and CIA consistently overplayed its estimates of Saddam Hussein's threat in its public statements and declassified reports, while its secret reports contained warnings that the intelligence on weapons of mass destruction was not conclusive.

In October 2002, Tenet told Graham that ''there were 550 sites where weapons of mass destruction were either produced or stored'' in Iraq.

''It was, in short, a vivid and terrifying case for war. The problem was it did not accurately represent the classified estimate we had received just days earlier,'' Graham wrote. ``It was two different messages, directed at two different audiences. I was outraged.''

In his book, Graham is especially critical of the FBI for its inability to track al Qaeda operatives in the United States and blasts the CIA for ``politicizing intelligence.''

He reserves his harshest criticism for Bush.

Graham found the president had ''an unforgivable level of intellectual -- and even common sense -- indifference'' toward analyzing the comparative threats posed by Iraq and al Qaeda and other terrorist groups.

When the weapons were not found, one year after the invasion of Iraq, Bush attended a black-tie dinner in Washington, Graham recalled. Bush gave a humorous speech with slides, showing him looking under White House furniture and joking, ``Nope, no WMDs there.''

Graham wrote: ``It was one of the most offensive things I have witnessed. Having recently attended the funeral of an American soldier killed in Iraq, who left behind a young wife and two preschool-age children, I found nothing funny about a deceitful justification for war.''

by Orsurgeon 2004-09-05 10:32AM | 0 recs
Plus Hurricane Frances
I don't think that many residents of Florida's East Coast were answering phone polls on 9/2-3, and I have to believe that they skew Democratic.
by EvanstonDem 2004-09-05 11:51AM | 0 recs
Re: Newsweek Poll
But non-whites do vote at considerably lower rates than whites. They may still be undersampled in this survey, but not hugely.
by EvanstonDem 2004-09-05 11:53AM | 0 recs
Re: Warning: Dissenting Viewpoint
Good point.

Kerry needs to convince America that it is safe to vote Bush out.

The debates are key.

by Geotpf 2004-09-05 05:33PM | 0 recs
Re: Newsweek poll: 42%
Does anyone know what percentage of US households actually fit that description? Offhand, I'd say ~12%
by Jerome Armstrong 2004-09-05 07:13PM | 0 recs
Re: SOME ACTUAL POLL DATA: Military Demographics
Good job on catching the bias in the Newsweek poll.

FYI:

The military represents 1.3% of U.S. households and had 2.2% of U.S. babies in 2002  according to the Military Demographics web site.

http://www.militarycity.com/advertising/MTMilitaryDemo.pdf

by Hanuman 2004-09-05 11:44PM | 0 recs
Newsweek poll: 42% of the households were military
Good insight.

According to Military Demographics, The military represents 1.3% of U.S. households in 2002 (not the 42% Newsweek stacked the deck with in their biased poll).

http://www.mydd.com/story/2004/9/4/154842/1919#68

by Hanuman 2004-09-05 11:52PM | 0 recs
Re: fix in on Newsweek poll
I absolutely believe the fix was in for this poll, especially after watching Gallup-CNN-USAToday perform precisely the same function for the bushies after the Dem convention. Their poll showing Kerry having lost ground to Bush came out at almost the same exact time in the cycle as this Newsweek poll. The script was written long ago on coverage of the conventions.

We have to begin focusing on the fact that the same is almost certainly true for the debates. How do we begin to counter that spin before it takes off?

What keeps me somewhat on an even keel is remembering how badly the press treated Gore -- and he still won!

by kainah 2004-09-06 03:51PM | 0 recs

Diaries

Advertise Blogads


----------- myDD - skin -----------