You lose sixteen points, and whaddya get?
A fat load of crap from the Nashville Tennessean
, that's what. The headline shouldn't be about how the differential in Tennessee is Bush by 16... it should be why there hasn't been a consistent polling result in the whole few months running up to this election.
Bush may be ahead here, but I'm certain that the lead is single-digit.
Here's the methodology:
The Tennessean and Chattanooga Times Free Press poll was conducted by the national polling firm of Mason-Dixon Polling & Research Inc. of Washington, D.C.
It included a sample of 625 registered voters who said they are ''likely'' to vote in the Nov. 2 general election. Interviews were conducted by telephone over four days from Sept. 11-14.
The sample included people whose phone numbers were randomly selected and then weighted by exchanges to ensure statewide coverage and by county to reflect voter turnout in prior elections. (emphasis mine)
The poll includes a margin for error of plus or minus 4 percentage points. That means there is a 95% probability the results of the poll are within 4 percentage points higher or lower than the result if the entire population was sampled. For example, if a candidate scores 15%, there is a 19 out of 20 chance the actual result will fall between 11% and 19%.
Here's why their conclusions are wrong, wrong, wrong:
Tennessee's population of 5.8 million is not evenly distributed throughout the state. This poll has a built-in bias that will tend to skew down the potential Democratic turnout. The population is heavily urbanized, and the turnout in 3 counties -- Davidson (Nashville), Shelby (Memphis) and Knox (Knoxville) could easily determine the outcome of the election. Hell, the total margin of victory for Bush in 2000 was roughly 80,000 votes (a 6 point spread, for what it's worth) -- so one game at Neyland Stadium at 80% capacity could have swung the election.
Now, based on 625 people which were pulled from a geographically weighted sample of the state's 95 counties, they're trying to say that Bush would win by over twice that?
Even given Bush's dismal record? Even despite the fact that every other poll for the past couple of months has shown a statistical dead heat? Even given the phenomenal increase in Democratic activity, in voter registration, and in a visible public sentiment that is a lot more anti-Republican than it has been in at least four years? Even given that Tennessee's state house and senate are majority Democrat? And that the governor is a Democrat? And that the two largest metros in the state are heavily Democrat?
Sorry, Mason-Dixon. That dog don't hunt. This election may be determined by "geos," but not the way you're reporting them. Rural turnout may skew Republican by percentages, but you're taking from a smaller pool numbers-wise. If you win a county of 100,000 by a spread of 60-40, you still get beaten (and outweighed) by a county of 1,000,000 which splits closer to even. Hell, you gotta have five or six counties with a much bigger spread in order to beat back the huge metros.
And if Shelby County breaks for George W. Bush, I'll put up an elephant in my cubicle at work.