And yet Cohn's complete "analysis" of the LA Times poll is that its skewed because it weights 2012 voters and according to him 2012 Romney voters are more likely to report they voted for Obama four years later. No data to back that supposition up, and no checking in with the USC research team for validation either.
He admits its a different methodology, but it must be flawed because he doesn't understand it.
When a poll is an outlier, it ought to be questioned. This LA Times poll has had Trump leading for six months.
That's way outside the polling norm.