re: #153 JamesWI
Yeah, that’s because the weight gets reduced the older the poll is. But the recent Gravis ones are generally fully-weighted.
But since I don’t know enough about how he’s doing the “house effect” stuff, I guess I don’t know if he’s taking those polls as seriously as the full weight would suggest.
Like I said earlier his model automatically assigns new pollsters a slightly better than average rating. I’m not sure but I don’t think he can assign a house effect until a pollster has been through at least one election.