Thursday, November 3, 2016

Nevada: A Crazy Polls vs Reality Case Study

Here we go again. Polls are all over the place with less than a week to go and people are freaking out. However, my recommendation remains completely unchanged: Keep Calm and Hillary On!

As a case study about why you shouldn't let every poll result cause a stress attack, let's look at Nevada. On Wednesday, CNN/ORC (a seemingly reputable source) produced an outlier poll that showed Donald Trump winning there by six points...when not even Republican-leaning pollsters Remington and Gravis had him up that high.

The ridicule was swift: Now let's see what Nate Silver's esteemed FiveThirtyEight currently thinks of this poll:

It gets weighted higher than any other poll and barely gets "adjusted". Meanwhile, a poll from (A-rated) Marist taken shortly before has their tie transformed into a +5 Trump lead thanks to 538's "adjusted leader" voodoo. I'm sure they have their reasons for this unskewing, but it doesn't make much sense to me. Whatever the case, this has helped to transform Nevada into a total toss-up or even GOP-leaning in the space of a couple of days. It even led 538's Harry Enten to post this terrifying scenario:
Meanwhile, Jon Ralston, the undisputed political guru of Nevada is nearly calling it already for Clinton based on early vote numbers:
On the presidential (and maybe applies to the U.S. Senate, too), some math still holds:

----Both candidates get 90 percent of base and split indies: Clinton by 4, 29,000 votes

----Both candidates get 90 percent of base, Trump wins indies by 10: Clinton by 2, 17,000 votes

----Both candidates get 90 percent of base, Trump wins indies by 20: Clinton by 3,000 votes

Note: Trump is not going to get 90 percent of the GOP base, and with all of those votes banked even before the Comey letter, it's almost impossible for him to win indies by 20. (Romney won indies by 7.) You see his challenge.

If Clinton holds her base here (data I have seen shows she is, and minority turnout is going up) and turnout patterns don't dramatically shift in the last two days of early voting, she can't lose Nevada. Solid lean Clinton right now.
So what's more likely: These early vote numbers translate to a solid Nevada win that resembles Obama's 7-point 2012 margin...or this one widely-ridiculed poll happens to be right?

I'm strongly betting on the former.


1 comment:

  1. Read also this article about case study and how to handle it with relative ease.