Lies, Damn Lies, and Statistics

Steve Soto has some interesting information regarding Gallup's polling data and methods--seems as if they've weighted Republican votes more significantly than hard data suggests. That probably why they keep showing a substantial Bush lead (8 points) even as other pollsters show that any sort of Bush bounce, post-convention, has evaporated as quickly as any hard questions about Iraqi WMD's.

Gallup thinks voters who identify themselves as Rethughlican will comprise almost forty percent of the electorate--which would be significantly higher than in 1996 or 2000.

Soto notes two things: first, the Gallup organization has been extremely gracious and open in providing data regarding their methods. Second, he cites Daily Kos to note that the CEO of Gallup is a large contributor to the GOP. Could this possibly influence their decision to give more weight to GOP voters than recent history would suggest?

Perception can be a significant factor in an election outcome. I'll let you decide for yourself if this story is true or not, but I was told a few years back by someone who claimed to be in the know on this: in the 1987 Louisiana Governor's election, Buddy Roemer's pollster took a single percentage point from five other candidate's totals, added them to his guy's numbers, and rode that to victory when Edwards conceded on primary night. Buddy showed "movement" when everyone else was stagnant.

So, I think Gallup's numbers should be taken with a dollup of sodium chloride--and, let's not forget, there might well be a number of NO voters on Bush also not reflected in the polling data--I don't know if anyone has factored in what I think could be a substantial number of people who don't pay attention to politics a whole lot, but who are sick and tired of the pResident running roughshod over basic American principles like the right to be left alone. We'll see, I guess.