NV Emerson: UTDH +7

[DataGuy]

Hate to rock the boat, folks, but what if this whole "Unbeatable Titan Dean Heller" thing actually ends up being true? I'm not making an affirmative prediction that he will win, but now it's pretty obvious this is not in the bag for anyone, Jacky included.

Yes, I know, Nevada polls always underestimate Democrats and all that, but on average it's only been by 2 or 3 points, and in the 2016 Senate race Democratic overperformance was only +0.6. So at the very least I'd say we have a tied race here.

And say what you will about Emerson, but they are a B+ rated pollster and actually have a D+0.1 mean-reverted bias. I agree that we should be skeptical of their polls when they have something like 25% undecided, but in this poll the undecideds stand at a fairly normal 11%.

[DataGuy]

Hate to rock the boat, folks, but what if this whole "Unbeatable Titan Dean Heller" thing actually ends up being true? I'm not making an affirmative prediction that he will win, but now it's pretty obvious this is not in the bag for anyone, Jacky included.

Yes, I know, Nevada polls always underestimate Democrats and all that, but on average it's only been by 2 or 3 points, and in the 2016 Senate race Democratic overperformance was only +0.6. So at the very least I'd say we have a tied race here.

And say what you will about Emerson, but they are a B+ rated pollster and actually have a D+0.1 mean-reverted bias. I agree that we should be skeptical of their polls when they have something like 25% undecided, but in this poll the undecideds stand at a fairly normal 11%.

538 increased their rating because they added online surveys to supplement for their lack of calling cell phones. Ive said multiple times how I think that was wrong on 538s part.

Regardless, 538 doesn't have a bad track record predicting Nevada. Despite Trump slightly leading the polls there in 2016, they favored Clinton to win the state and got pretty close to the final margin. They were also correct on the Senate race, even as they missed some other races badly. So when they have Heller as a 60% favorite, as they do now, that should probably mean something.

[DataGuy]

I think others have pointed out the problems with this "poll" quite well. Instead of my usual shtick about how a polling error here, which has happened before, is more likely than NV not only being immune to the blue wave, but hosting a giant red wave, let me talk about something else...

This absurd electoral truism of "Nevada bucks the trend." It really ONLY applies to the Senate races, and it's a prime example of correlation not equaling causation. In 2010, Harry Reid's opponent was Sharron Angle, who was a complete lunatic, and Republicans should have known that someone with ties to their state as strong as Reid wasn't going to go down easily. Had he faced a stronger opponent, he might well have lost. In 2012, Dean Heller's opponent was scandal-plagued, and basically written off after polls showed her consistently down by about 5%. He still just barely eked out a win against her with just 46% of the vote. Had he faced a slightly better opponent, he almost definitely would have lost. Then there's 2016, where Nevada really did not buck the trend, unless your criteria for "bucking the trend" is based only on who wins. In a slightly Republican-leaning year, a Democratic-leaning state narrowly went Democratic. It still swung Republican, just not by enough for it to flip in a year that was only somewhat Republican-leaning.

And why on earth would Nevada buck the trend this year? Heller's not a popular incumbent like Baker, Scott, or Hogan, Nevada's not a Republican stronghold like North Dakota, and while Rosen might basically be a generic Democrat, it's not like there's anything so offensively off-putting about her as to turn off a large segment of the population.

If 2018 is a good year for Republicans, sure Heller could win. But there's really no reason to believe Heller will survive a blue wave, unless you believe the polls. And if you do, I'll direct you to IceSpear's posts or my signature.

I understand why people are skeptical about NV polls. And the fundamentals do favor Rosen. Based on my calculation of the fundmentals, Rosen starts with an advantage of about 4.5. But whatever you think of their reliability in predicting the specific margin, it is very clear the polls have shifted recently from a small/moderate Rosen lead to a small/moderate Heller lead. Even assuming that Heller's being overestimated by the average 2-3 points, that makes the race a tie. I would rank this a pure tossup right now.

[DataGuy]

I thought it might be interesting to see how some of these pollsters being discussed now performed in 2016. Emerson's last poll of NV showed both Clinton and Cortez-Masto leading by 1, and they both won by 2.4. That's not a huge error. In fact, it's perfectly normal if not a little better than average.

And in the presidential race, the most accurate NV poll was, wait for it ... Gravis. Yes, Gravis. Their last poll showed Clinton up two, and she won by two. Their Senate poll actually overestimated Cortez-Masto by 4 points or so. This year, they have most recently shown Rosen +2.

I know it's common practice to look at crosstabs to discredit polls that don't "fit," but the crosstabs always have large margins of error anyway and are not supposed to be viewed as very predictive. As Nate Cohn discussed recently in a Twitter thread, some of us might know too much for our own good. https://mobile.twitter.com/Nate_Cohn/status/1050579929203642368