Reply #22: Stop shouting and listen, [View All]

If you are talking about the exit polls, it is simply because the precincts polled in each state tended to go for Kerry more than in the precincts which were not polled.

I do wish you would stop shouting, but as you never do stop shouting, I will shout too:

ALL VOTERS WERE NOT CHOSEN RANDOMLY. PRECINCTS WERE CHOSEN FIRST. PRECINCTS WERE ALSO NOT CHOSEN RANDOMLY. THIS IS NORMAL PREDICTIVE POLLING PROCEDURE.

EVEN IF PRECINCTS WERE CHOSEN RANDOMLY, PRECINCTS DIFFER SO MUCH FROM EACH OTHER, THAT THE POPULATION MAY NOT BE REPRESENTATIVE OF THE TOTAL POPULATION

(All this easily discovered from the Mitofsky web page)

Mitofsky says that the precincts were chosen according to a predetermined formula, and predictions could be made after considering what was known about those precincts. This formula was likely to have been the same for each state, and therefore most states would show the same apparent bias in the poll data.

Basically, this is a double layer of sampling. They sample the precincts, and then they sample the voters in those precincts. The latter procedure is random, the former is not. If you want to do mathematical calculations, you MUST.. I shall repeat MUST first calculate the probability of the precincts being representative of the population as a whole. Due to the structure of precincts, with many small ones, and a few large ones, it would be very difficult to get a group that would be 100% representative, and the likelihood is that urban precincts will predominate. Mitofsky would know this perfectly well, and would be able to deal with it mathematically when making predictions.

However YOU DO NOT KNOW HOW THE PRECINCTS WERE CHOSEN, and therefore your calculations are worthless. If you happen to know even which precincts were chosen, perhaps you could let us in on the secret, as there are thousands of anxious statisticians all ready to begin. (actually, I don't think that would be too difficult to find out, but what we really need to know is what Mitofsky knows about those precincts, and why they were chosen)

If you still think voters were chosen randomly, have a look at the number of people polled and just ask yourself how they would possibly do it in a state like California. It would have meant sending somebody out to every precinct and canvassing every two hundred voters or so (probably even fewer than that: I don't have the figures with me). Some precincts are so small and remote that you would have to hire someone just to canvas a couple of voters all day. Obviously far too expensive.

The sad thing is that the final figures may well indicate fraud. However, nonsensical and incomplete calculations only serve to distract us from the importance of the real thing.

I very much doubt if you have read this far, but if you have, Happy New Year. Perhaps your New Year Resolution might be to stop shouting. Thank you.