Republicans had their South Carolina Primary on January 19th. In the evening of Jan.19, CNN published an exit poll around 8 p.m., but "adjusted" that same exit poll a few hours later as more and more results were coming in.
With a supposed change of only seven(!) more respondents (from 1648 to 1655) the adjusted exit poll showed amongst others Mike Huckabee rising 3% in the female vote, Rudy Giuliani's womens percentage dropping in half (from 4% to 2%), Ron Paul shrinking in the male vote from 8% to 5% and collapsing in the youth vote from 13% to 7%.
These polling numbers cannot be scientifically correct. It seems that they have been adjusted to match the actual results better.
Here the first CNN exit poll of the SC Primary (R):
http://good-times.webshots.com/...
or
http://image63.webshots.com/...
(screenshot made on Jan.20, 02:15 GMT+1, i.e. Jan.19, 08:15 pm EST)
And now the second, adjusted, CNN exit poll of the SC Primary (R):
http://good-times.webshots.com/...
or
http://image62.webshots.com/...
(screenshot made on Jan.20, 12:36 GMT+1, i.e. Jan.20, 06:36 pm EST)
Source: http://edition.cnn.com/...
(of course only the adjusted exit poll is still available online)
Some questions are arising:
Why did CNN change its exit polls?
What does CNN fear if the exit polls do not match the results as they were supposed to do?
Will CNN adjust its exit polls tonight again?
P.S.:
As I don't suppose to make it to the recommended list, I hope those of you concerned with voting security are having a closer look on the various exit polls, as there were some major discrepancies in 2004.