Reply #79: Well, the reason it is

[Febble]

a "difficult concept" is this:

If the pollsters state, in advance, that they are going to project the counted vote, and that the method they will use will involve incorporating vote-returns into the projections, and then that is exactly what they do...

why would anyone think that what they actually intended to do was to project the result from the polls alone, but that for some reason, instead, they did what they stated they were going to do, and and use the vote-returns in addition?

But I do appreciate the tone of your post.

I know that it is hard to believe that the exit polls were not designed as a check on the election process, and I rather agree that, American being something of a banana republic, perhaps international observers from the UN should do some independent monitoring, possibly using exit polls...but it is hard to see why anyone would think that the reason the 2004 exit polls were conducted was not for the reasons stated before they were conducted.

Except that I admit that probably not that many people read the E-M FAQ before the election. But I did.

It didn't, nonetheless, stop me considering, very seriously, that the reason the around-close-of-poll exit results were discrepant from the final results might have been because the count was wrong. Or from putting a vast amount of effort into trying to figure out whether it might be possible to disambiguate the two possibilities (error in the poll; error in the count); nor does it stop me remaining outraged by the broken state of your democracy. But it did, I admit, stop me considering that the projections were intended to be independent of the count, because it was stated in black-and-white that they wouldn't be, ergo, that their designed purpose was to predict the counted result, not act as a check on it.

But it doesn't mean that the pollsters were not interested in how anyone voted. They were (or their clients were) very interested. And while crosstabs performed on the National Exit Poll give you slightly different answers depending on whether you use weighted or unweighted data, the percentage point differences don't radically affect the information.

What the weights radically affect is whether what was a narrow margin between the two candidates put Bush in the White House or Kerry. Not why, to a reasonable approximation, particular groups voted for either.

Because even if you are right, and the poll was nearer to the true vote of the American public than the count, the fact is, tens of millions of people voted for Bush. The polls remain informative about who those people might have been, whether or not you are convinced that the total numbers are correct.

And I would add that I still believe that various forms of voter and vote suppression probably deprived Kerry of many thousands of votes, including votes in Ohio, which would have won him the election. But mostly those wouldn't show up in the exit polls.

In other words, while I don't think that the exit poll evidence provides good evidence that the election was stolen, I see plenty of evidence, as Peace Patriot outlines upthread, that the election was unjust, and that the injustice was disproportionately visited on Kerry's voters.