Monday, October 04, 2004

Amen! Well, sort of...

At first, I agreed with the headline... that changed in about 2 seconds.
From the NOW Action update email: Tell Time/CNN and USA Today to Reject Biased Polling Data!

Action Needed:

Recent polls of likely voters conducted by The Gallup Organization have substantially over-sampled Republicans and under-sampled Democrats and women -- overestimating the support for Bush.

Time/CNN and USA Today -- two of the country’s most prominent news organizations – routinely commission and report (and thus legitimize) Gallup’s poll results.
Because we don't agree with them, we are going to imply that Gallup is a corrupt evil organization that forces women to stay home and be barefoot and pregnant when they could put their kids into day care and work like a drone!

Please send messages to Time/CNN and USA Today to protest their use of The Gallup Organization’s election campaign polling data until Gallup can prove it has eliminated such biases from its data sampling.

Background:

If you are having a difficult time believing some recent polls that show George W. Bush leading John Kerry by a double-digit margin, especially since other polls indicate the race is a dead heat -- trust your instincts! A number of reports indicate that recent Gallup polls commissioned and reported by major news organizations such as Time/CNN and USA Today have substantially over-sampled Republicans and under-sampled Democrats and women, erroneously creating a double-digit lead for Bush.

For example, a Gallup poll conducted Sept. 13-15 reported that Bush leads among likely voters by 13 points -- 55% to Kerry’s 42%. Despite the fact that Democrats voted at 39% in the last two presidential elections, Gallup assumed that only 33% of those turning out to vote will be Democrats.

The result? Democrats were under-sampled by six points.

Gallup also assumed that 40% of November voters will be Republicans -- yet Republicans only voted at 35% in 2000 and 34% in 1996, the last two elections.
I think it is a logical assumption since it is their party that is in power though not many Republicans are insanely supportative of GWB the way many Democrats just cannot get over Clinton. Though according to the journal American Demographics, the numbers might be even further in the Democrats favor (though it still doesn't add up to 100% or even 77% that say they are registered to vote so, I must be missing something in how they intrepreted the data.)

Also, how people are registered to vote and who they vote for are two completely seperate things. For example, I am un-affilated with any major party in Tennessee. I would, in this survey probably be listed as an independent. They ask, "who do you plan to vote for in the next gubenatorial election?" I may say "Phil Bredesen" - who is a Democrat. They may ask "Who do you plan to vote for in the next Senate race?" I may say "Marsha Blackburn" - who is a Republican. They won't ask why because that cannot be qualified as a percentage.. but my flip-flopping will wreak havoc on their pretty little theory! Mmwwahhhahahahaha!

More than anything - it sucks being a statistic.
The result? Republicans were over-sampled by five to six points.

Furthermore, Gallup under-sampled women, who are more likely to support the Democratic candidate. Cumulatively, these errors created a double-digit lead for George W. Bush out of thin air.

George Gallup, Jr., is an evangelical Christian who said in a speech earlier this year: "The most profound purpose of polls is to see how people are responding to God" and "When I ask a question on these subjects, what I'm always trying to find out is: 'Are we doing the will of God?'
"And this has what to do with his polling abilities? Non sequiter anyone? This paragraph is COMPLETELY out of place in this email... but I wouldn't expect NOW to use logic.

It is the media’s responsibility to provide accurate data to the public, and to reveal any biases in the data collection or sampling. By reporting biased polling data, Time/CNN and USA Today have failed in this obligation and need to hear from people who care about accurate reporting and polling.

Again, please send messages to Time/CNN and USA Today to protest their use of The Gallup Organization’s election campaign polling data until Gallup can prove it has eliminated such biases from its data sampling.
Yes - completely un-biased questions like "If all the money were going to support education, would you vote for a income tax?" or "If it were down to John Kerry, Hitler, and your mother-in-law, who would you vote for in the next Presidential election?"