More CBS/YouGov "likely" voter polls that are really registered voter polls
These recent YouGov state polls are demonstrating a consistent flaw, reporting a registered voter poll as a likely voter poll. Registered voter polls, according to Nate Silver, almost always favor Democrats. This probably represents the most likely explanation for the CBS pollster using this tactic.
According to the Ohio poll released today, 994 of the 1000 registered voters they surveyed were deemed likely to vote. Included in the pool of likely voters were registered voters who deemed themselves a "maybe" as to whether they will vote or not, and also those who said they will "probably" vote. These voters combined to represent seven percent of the likely voter pool. They were all deemed likely voters. Most pollsters throw those voters out of the likely voter pool.
Opening up the likely voter pool to include essentially every registered voter they talked to resulted in a +7 Democratic voter turnout assumption in a state that is even, with a one-point Republican party ID edge as recently as 2014. Not to mention, the intensity is on Trump's side, so nothing justifies assuming such a high Democratic turnout.
It was no better in the Florida poll, where they assumed only seven of the 1200 registered voters they interviewed would stay home, despite the fact that around 72 of those registered voters said they were a "maybe" or a "probably," as opposed to definitely. Again, this is not what most pollsters assume.
This poll assumed Democrats would have a +2 turnout advantage on election day, a much more reasonable assumption than Ohio. That assumption is in line with recent elections. The poll also had a much more reasonable outcome, but among registered voters, not likely voters.
Follow my new account on Twitter for daily updates on the election!
Clinton +0.20 Points in All Ref Battlegrounds
There are now four realisitic scenarios where Trump can win without Pennsylvania.