Poor PPP, too non-credible outliers in a row.
What is the other non-credible outlier they did?
The one with Hillary up by 26.
Yeah. These were in a row in the sense of being consecutive PPP polls of Pennsylvania; I suppose they released polls of elsewhere in between. Of course, the first of those was at the height of the Wright spamfluff, which partly explains it.
I'm not going to judge them until they offer their final poll. For Ohio, Texas, and Wisconsin they released a poll two days before the election and they pretty much nailed all three of those. Ah, but the question is: Is that because of genuine movement, or because their sampling model works only once everybody has made up their minds? Given that polls are always perceived as sort-of-predictions, and that as soon as you use sampling, likely voter scans etc, this perception isn't even
altogether unfair ... a good pollster (as opposed to decentish) should have semi-decent results much earlier than that.
Eh. Winning by just 5 points would be equivalent to losing by considerably more than 10 in Wisconsin - ie a good point for a decent exit from the race. Sort of doubt it, though. More likely about 10-12 points.