Were the numbers gathered in and around Washington, DC? If so, does that potentially reflect the nation? That would be my first quesiton of this poll. But it is good to see that SigEps are working on shedding light on these numbers. I just hope that there was a minimal amount sampling error and sampling bias injected into this survey.
Will Hull - NV Alpha '03 | Homepage | 10.29.08 - 7:39 am | #

Good questions. Any poll is nothing more that a snapshot. Often the questions, sample technique, and all the other factors can allow a clever pollster to "photoshop" the results.

I have been following the daily tracking polls at gallup.com and a site for the true number geeks called fivethirtyeight.com which aggregates as many polls as they can find for each state to try to get a red on the electoral vote count (538 total electoral votes, get it?). The numbers from all over do seem to be trending in one direction lately, though. (Sorry about that, Will. )
Bryce Giesler | Homepage | 10.29.08 - 8:28 am | #

I wasn't second guessing the direction that the polls are going, but it is the representativeness that I was after. I am pleased, as a matter of fact, that they are going in that direction, but it is just to be fair and representative that I ask such questions.

It probably also helps that I am currently taking a graduate level statistical methods course at GMU on my way to my MPA, as it is enlightening me to never take a poll at face value. Who knows, we could see the Bradley effect and what secretly racist/sexist people say and what they do in the polling booth could be two entirely different things altogether blowing all polling data out the window on Election Day.
Will Hull - NV Alpha '03 | Homepage | 10.29.08 - 9:28 am | #

Will, not to go too number-nerdy on you but if you have touched on meta-analysis in your studies the fivethirtyeight.com site will interest you.

They essentially take all the polls that have been published, weight them by age and quality (sample size, technique, etc.), and then run 10K iterations on a state by state basis to determine probabilities of any given outcome. Fascinating, in a geeky sort of way.

Oh, and I stand corrected on your preference. I got you confused with another regular commenter who works for the RNC.
Bryce Giesler | Homepage | 10.29.08 - 10:10 am | #

Yeah, I know that guy. It's not his fault, he's that way, but I still must respect him for it. Really good guy. We all know who we're talking about. For those who don't, probably don't know him.

Thanks for the 538 reference.
Will Hull - NV Alpha '03 | Homepage | 11.03.08 - 11:41 am | #