Cycloptichorn wrote:Either the exit polls were massively tampered with, or the actual polls were, there's no two ways about it.
Or the exit polls were simply erroneous. Especially the early data. You will have noticed, if you have dug into the exit polls, that they were updated during the day (before the actual poll counts), and that the later batch of data often already showed clearly smaller margins for Kerry than the earlier batch. So the exit poll numbers were already correcting themselves to what the eventual result turned out to be as they got to be based on more of a complete sample. That seems to suggest to me that the early data were off-center and that the actual results are not all that unlikely, considering the exit polls were already lurching in their direction as the day progressed.
Furthermore, exit polls
can still easily be wrong, no reason why they couldnt. Its not like exit polls interview every single voter. They are based on the results of precincts that in the past provided a representative reflection of how a state as a whole voted. At least, thats how it works here. But of course, from election to election the make-up of the electorate and the balance between precincts can
change, and the "representative" precincts in result become less representative.
I.e., if some constituencies - say, the urban Republican constituency, or the rural conservative constituency, which will be more strongly represented in one kind of precincts than in another - turns out significantly more strongly than in the previous elections, then the "representative" sample of precincts the exit polls are based on doesn't work properly anymore.
That would also explain why "the exit poll results are not scattered about the mean", as the author you quote observes ominously.
And then there's the question of how many voters were willing to answer the exit pollsters questions and whether there was a skew involved there.
Quote:There is skew - but ONLY in states which the Republicans had previously stated to be target states in play. The skew is in the same direction every time; that is to say in favor of Bush.
How very conspirative - but there's just one hitch - it's not true.
The author shows that exit polls were considerably further off in 7 states than in 10 other states - suggesting, I guess, in his mindset, fraud in those 7 states. And those are "all target states in play"!, he notes - surely cant be a coincidence. But is that so?
The 7 states include North Carolina - in fact, that was the state where the "skew" was second-largest -- which hadnt been considered in play by either campaign for a long while already. The 10 states, on the other hand, include Colorado, Michigan, Iowa, New Mexico and Nevada - some of the most closely contested states, which all merited campaign stops till almost the very end. So? Another contention falls through.
Articles like this are invariably riddled with such blatant inconsistencies. I get around to signalling a handful and by then really no longer can muster up the faith to even think about whatever assertion comes next.
Basically, to quickly save a screenshot of preliminary raw exit poll data and uncritically assume that those must be right, even as one is questioning every detail about the actual voting results, evidences a highly selective application of critical thought - lets put it that way. I mean - to quote the page with exit poll data the author refers to - what exactly
is so difficult to understand about its warning that "it’s also important to consider that early polls are routinely unreliable"?
Cycloptichorn wrote:Seeing as the exit polls confirm the vast majority of the previous polling data up until the 2nd, it seems rather odd that EVERY poll would be that far off of the actual total...
Odd but hardly unusual at all. "Previous polling data up until the 2nd" in 2000, for 8 out of 11 pollsters, showed Bush winning by 2-9%; yet he lost the popular vote. In their final polls/projections, eight out of nine pollsters predicted Clinton would win the '96 elections by more than 8% (numbers ranging up to a 18-point victory) - yet 8% was what he won by. Every pollster in 1992 had Clinton winning by more than 6%, yet that was what he won by. Likewise, four out of five pollsters predicted Bush Sr. would win by 9-12% in 1988, but he won by only 7%.
Polling is an approximate art. It roughly indicates the range the result will be in. But the actual results will always throw in some surprises. And the polls don't count, in the end.
{One thank-you though: for pointing out that the currently published exit polls have been reweighed in order to fit the actual results. That makes perfect sense (considering exit polls are intended for analysing voter breakdowns with, not second-guessing the actual result) - and I had suspected as much - but I hadnt included that in my previous posts, when I was still blasting at people who said the exit polls were off so much by pointing them to how wholly correct the published results now are. Shame on me.}
