NV-TargetPoint Consulting/Washington Free Beacon: Sanders & Clinton tied at 45%

[Figueira]

Wow. Is it fair to say that the race is a toss-up if Sanders wins this?

I'd call it Lean Clinton if Sanders wins Nevada. It depends what the demographic breakdown looks like though.

[Figueira]

538 forecasts are pointless with one poll. But at the very it least it seems like they threw out the Clinton skew in their "polls-plus" forecast after they saw how little impact Clinton's endorsements had.

I never understood that anyway. If Clinton's enorsements had a huge effect, wouldn't that be reflected in the polls?

[Figueira]

538 forecasts are pointless with one poll. But at the very it least it seems like they threw out the Clinton skew in their "polls-plus" forecast after they saw how little impact Clinton's endorsements had.

I never understood that anyway. If Clinton's enorsements had a huge effect, wouldn't that be reflected in the polls?

No.  The idea is that, when they went back to look at what was predictive of election day results in presidential primaries past, yes, polling in the state was of course the most predictive thing.  But "party support" (as measured in this case by endorsements, since that's a quantifiable thing) also had some predictive power, in the sense that if two candidates polled the same a month before election day, the one with more endorsements was more likely to see his/her polling numbers hold up.  As you get closer to election day though, the endorsements become less predictive, while the polling becomes more predictive.

Ah, I see now.