As far as my comment about usless info, let me clarify: the information itself is not useless, if it were accurate. I certainly think age groups, race and gender are important to track - if tracked factually. Giving or taking 4 points on a poll is just not accurate enough - until I see an exit poll tracking at least 10% of the people who vote at every single polling location, these numbers will continue to be nothing more than fluff used by the media to show how much they "know".
No worries, guys, we're not debating... just trying to spread democratic love by learning the facts!
Snick, these exit poll companies (specifically the ones I mentioned above) take painstaking care to ensure that their error of margine is extremely low. These polls are all based on statistics (I know,yuck) that calculate the 'statistical power' needed to achieve numbers with a standard of error low enough to be pretty darn accurate.
I am not sure how you arrived at a sample size of 10% of the Wisconsin population to quell your fears of an inaccurate representation of the demographic data? I agree that, the bigger the sample size, the closer you will be to the true number of black voters for Kerry or Bush... but in the real world that just isn't possible.
In fact, almost all of the facts that we hear are based on a small sample size (yet large enough to yield significant results when compared to control groups). All scientific studies are conducted on a small sample size used to represent the total population.
Every drug we take, whether it be birth control, aspirin, cough suppressant, penicillin, etc. was only tested on an incredibly small percentage of the total human population (especially given how many people take these kinds of popular drugs). I know I'm belaboring the point, but just imagine if a drug company had to test out one of it's drugs on 10% of the human population before it was made accesable to the public. The world population is roughly 6,397,510,180 (wow)... which means a company like Abbott or Baxter would have to test the effects of their drug on 639,751,018 people... yikes.
I think the problem is not the small sample sizes of these exit polls (because like I said, they end up being pretty darn accurate in the end) that make them problematic... it's that the numbers are reported very early
... before most of the population has cast their vote... which can screw with the statistics. People do not take these numbers with a grain of salt and are too anxious to wait the 24-48 hours until all votes are in and all intended "pollees" have been polled.
Exit polls are actually good things, as long as we understand how they work and how we can use them to get the most accurate information out of them.
Again, no debating or fighting... and I hope I didn't send anybody under the bed huddling in the fetal position when I started talking about statistics...
And believe me, I definitely agree with you that if we could poll every single voter that would be the best, most accurate representation of the voters.
Damn that enormous, ever increasing population...