Link? Although I guess we'll know in a few hours anyway.
In any event, my present understanding is that different polling firms are using different turnout models that make different assumptions. This part of the reason poll averaging is a thing to begin with and part of the reason that different polling firms have persistent house effects that can be measured and corrected for. In any event, if it is the case that polling firms use the last general election turn-out model every new election then we should have enough election data going back to 1968 to understand how that effects projections and correct for it using regression analysis.
TL;DR: I think this is grasping at straws.




Reply With Quote