[LS]

You apparently don't know how this works. Pollster finds out what their PARTY affiliation was for 3 previous presidential elections, not "who they voted for." They don't even ASK "who did you vote for" in the last election.

According to Ras, if the person was a registered R for three consecutive presidential elections, he's a reliable R for polling purposes.

All I can tell you is that the Rove model that we followed in 04 was dead on. I found ONE registered Republican in all my door to door activity who wasn't going to vote for Bush.

On the other hand, if AS MOST POLLSTERS DO, you say, "did you vote Democrat or Republican in the last election?" or "What is your political affiliation, Democrat or Republican?" you will get much higher Dem % this time around---meaningless.

[muawiyah]

Having worked with America's largest statistical sampling systems for several decades, and also having dealt with marketing pollsters, and having studied the field, all I can say is this about that:

There are only so many ways to stuff a porkchop, but there are a million ways to stuff a cat.

The polling firms constantly change their methodologies because, as it turns out, none of them are perfect, and even those that are reliable a few times will suddenly fail to be even reasonably close.

So, what is the point you wish to make ~ that Rasmussen did so and so, and somebody else did something else, and then...............

There's no perfect approach to polling public opinion, and there's no one way to ask the same question. I can ask you how you describe your preference (a common way), or ask you who you voted for BY NAME, BY PARTY, BY SEX, BY RACE ~ etc.

There's a lot of variation in the methods.

I think I told you the other day I'm thinking of moving into data mining to see if we can track public opinion that way. The secret may be in the trendlines, and if we can catch trends early we can all get rich!