I was just over on TPM reading that Gallup's LV model assumed the electorate would look like 2010 based on race. In other words, 78% white. I understand that one way to build a LV model is to base it on past experience. Leaving aside the argument of using a non-Presidential year as the basis for a Presidential year, I have another question about LV models.
Aren't they at least partially based on questions the pollster reaches that help determine whether the person being polled is likely to vote? I am certain I have read that pollsters ask questions to determine voter intent and then classify as LV vs RV. Is this data not used to determine the LV turnout model? Did Gallup just ignore it?
I suspect this question has already been raised in many diaries given all the unskewing discussion here on DKos. I read all the time here and rarely write or comment but I got a really, really, really low User ID :-), so please forgive me and explain this to me again.
Thanks,
Steve