You're not making any sense, or I'm not understanding you. You're right you need a correct sampling rate, but BEFORE the election. How do you do this? By examining voter rolls, and coming up with an approximate value.
What you're saying is that "as more and more demographic information was gleaned from actual votes, they could more accurately weight their exit poll sampling"!!!!! What sense does it make for a poller to determine a sample while they are taking their poll? It's absolutely useless. You determine your sampling rates beforehand with the best information you have (you will have a margin of error as you mentioned).
So you're basically admitting that CNN changed their sampling rates while polling was still going on.
Example of what you're saying:
1) hmm, we estimate that there will be 50/50 democrats and conservatives voting tonight, this will be our sampling rate
2) oh look, so far we have polled 10 people and 7 of them are for bush and conservative, and 3 of them are for kerry and democrats. we should change our sampling rate to 70/30 conservative.
oh wow, this gives us an estimated exit poll of 70%bush/30%kerry. we're right on! how did we do that?
It seems fishy to me that the two states with computerized balloting and no paper trail, had Kerry winning in the exit polls, but the outcome was decidedly different. In fact these two states had the highest discrepancy in exit poll vs. final poll numbers.
Slashdot Mirror
You're not making any sense, or I'm not understanding you. You're right you need a correct sampling rate, but BEFORE the election. How do you do this? By examining voter rolls, and coming up with an approximate value. What you're saying is that "as more and more demographic information was gleaned from actual votes, they could more accurately weight their exit poll sampling"!!!!! What sense does it make for a poller to determine a sample while they are taking their poll? It's absolutely useless. You determine your sampling rates beforehand with the best information you have (you will have a margin of error as you mentioned). So you're basically admitting that CNN changed their sampling rates while polling was still going on. Example of what you're saying: 1) hmm, we estimate that there will be 50/50 democrats and conservatives voting tonight, this will be our sampling rate 2) oh look, so far we have polled 10 people and 7 of them are for bush and conservative, and 3 of them are for kerry and democrats. we should change our sampling rate to 70/30 conservative. oh wow, this gives us an estimated exit poll of 70%bush/30%kerry. we're right on! how did we do that?
It seems fishy to me that the two states with computerized balloting and no paper trail, had Kerry winning in the exit polls, but the outcome was decidedly different. In fact these two states had the highest discrepancy in exit poll vs. final poll numbers.
LOL. you're right, we would all like to be cunning linguists.