Bush Bounce

[CBL4]

The Bush convention "bounce" was apparently due more to a polling sample error than reality. It was widely reported to be 10% but it was probably closer to 2%.

According to the Economist, the bounce was 10% by Newsweek, 4 by Time, 2 by Gallup and 2 by the Economist. However "Newsweek's poll ... used an odd sample—38% Republicans, 31% each for Democrats and independents, when current party registration has Democrats with 33% and Republicans with 29%." In other words, the most reported bounce was simply to do a polling anomally and the real bounce was probably 2 or 3%. This is still bigger than Kerry's negative bounce but small by historical standard. In all the polls, the bounce among likely voters was even smaller.

Here is a link to the story in the Sept 9th, Economist but you probably need to be a subscriber to see it.
http://www.economist.com/World/na/displayStory.cfm?story_id=3177113

The moral, never trust one or two polls.

CBL

 

[Charlie Monoxide]

I'd like to see a poll where the question is "Do you lie to pollsters?"

What would be a true response?

Charlie (answer yes or no, have you stopped beating your wife?) Monoxide

 

[Cleon]

All of this is why I pay attention to polls about as much as I pay attention to my horoscope. They seem to have about the same bearing on reality.

 

[corplinx]

I posted on this very topic in a thread called "double digit dookie".

 

[CBL4]

I posted on this very topic in a thread called "double digit dookie".

You made a brilliant observation. Seriously, I was not reading the election thread when you posted it.

CBL

 

[Skeptic]

The moral, never trust one or two polls.

Indeed so. But why trust more than one or two polls? Why should there be collective wisdom in individual ignorance?

The problem is, they used a different poll to prove the preious poll is wrong. Want to bet that by next week someone will find out that that the no-bounce poll was itself distorted, and the "truth" is that there was a 20% Bush "bounce"? And the week after that, THAT poll will be attacked as biased, and... you get the picture.

 

[rikzilla]

Still bouncing per these polls!

 

[Matabiri]

I like this chart:

(But it's only about Bush)

 

[aerocontrols]

The Bush convention "bounce" was apparently due more to a polling sample error than reality. It was widely reported to be 10% but it was probably closer to 2%.

According to the Economist, the bounce was 10% by Newsweek, 4 by Time, 2 by Gallup and 2 by the Economist. However "Newsweek's poll ... used an odd sample—38% Republicans, 31% each for Democrats and independents, when current party registration has Democrats with 33% and Republicans with 29%." In other words, the most reported bounce was simply to do a polling anomally and the real bounce was probably 2 or 3%. This is still bigger than Kerry's negative bounce but small by historical standard. In all the polls, the bounce among likely voters was even smaller.

I think it's important that what is really going on here is that polls differ in their methodologies, and it's not clear who is correct and who is incorrect. People switch loyalties, and such switches would show up in polls before they show up in official party registration counts.

If one wants to argue that the the pollsters who believe that the 33/29 weightings are correct and who weight their polls accordingly are getting better results will need to explain things like this

In the last few weeks, Kerry campaign officials have been nervously eyeing polls that show an erosion of the senator's support among women, one of the Democratic Party's most reliable constituencies. In a New York Times/CBS News poll conducted last week, women who are registered to vote were more likely to say they would vote for Mr. Bush than for Mr. Kerry, with 48 percent favoring Mr. Bush and 43 percent favoring Mr. Kerry.

In 2000, 54 percent of women voted for Al Gore, the Democratic nominee, while 43 percent voted for Mr. Bush.

Those women who voted for Gore in 2000 and who plan on voting for Bush in 2004 - any chance they're telling pollsters that they're Republicans now? If they (and others) are doing so, and the pollsters are then 'weighting' their polls back to the predetermined party registration weighting, then it's those pollsters who are making the error.

There is also this poll that's showing switchovers to Bush by African Americans since 2000. I suppose the line "surveys that get results I don't agree with are crap" can be used here.

MattJ