right or wrong some polls do accept it and factor it in apparently.
Agreed they factor it in, but as I said, you can't just accept the answer.
I would ask who you voted for last time. If you DID vote last time, that would make you a tad more likely to vote this time.
So it would be quite a bit more complex than just asking if they intend to vote. Heck, registering to vote would seem an intent to vote, but it is proven from the voting record that not all who register to vote, do vote.
The point being they have to ask some questions and do some deduction to figure it out.
The next question is what happens when they poll 645 people and 600 of them say they're for Romney? By luck of the draw, they polled 600 republicans and 45 democrats. The electorate will not look like that in November. So they have to weight the results so it's more like 323 republicans and 322 democrats, assuming a near 50-50 split among those who go out to vote. How do they come up with 50-50, or 40-40-20 (20 independent), or something else?
Going back a step, they also figure out 90% of democrats are voting for Obama and 96% are voting for Romney. Apply that to the 323 & 322 numbers and you come up with their likely voter result (Romney 49, Obama 48 or whatever).
So how do they figure out 90% of democrats are voting for Obama?
(This isn't a partisan exercise, but one of understanding what the polls are telling us and what the complaints are)