What’s the point of having an ‘average’ of polls if most of them are atrocious?
They really need to toss out the polls that have no correlation to reality. Which basically just leaves Rasmussen.
No good purpose other than to dilute the influence of the ones that are most unreliable in his favor.
I prefer to stick with Ras. Meta-analysis and other data combination techniques have a place, but not when the methodology is misleading. And I do think it is misleading to poll people who dont vote or are unlikely to vote. Also, I wouldnt call it atrocious to use live interviewers. But I think Ras has a very good argument that you get a more honest answer with his automated polling when you are asking people about their opinion of the "first black president" (or any question where a person might be concerned about giving an answer that might be perceived as racist.)