- We will not know which polling firm had the best turnout model until after the election.
- Based on previous history, having a good turnout model in previous election cycles does not necessarily mean that you will have a good turnout model in this one.
- A poll is not accurate simply because it agrees with you; in fact, a poll is often not accurate even when it agrees with the end result.
- Every criticism about the assumptions made in a particular turnout model will itself be based on assumptions.
And everybody in the business will agree to the above, with the private caveat Except for my favorite polling firm, of course.
PS: What? No, there’s been no new polling, particularly. But this has to be brought up sometime.