LInks for the Week of November 6, 2016

On the Excellences of Nate Silver and http://fivethirtyeight.com

Has Trump Already Lost Nevada FiveThirtyEight

This is the kind of story that I remember Nate Silver and his http://fivethirtyeight.com doing right over and over again in 2012 and 2008. Take the model--a very useful poll-aggregating model, even with what I regard as its unhelpful Bayesian frame--and use it as a springboard for a really smart, really high quality moderately-deep dive into some aspect of the situation that can be illuminated by the data.

It is my sense that this election cycle http://fivethirtyeight.com has to some degree lost its way. That is, I do not remember reading nearly as many pieces along these lines as I seem to remember reading in the past:

  • Is this a failure on my part--have I not been paying attention to what they have been writing because I am submerged in a tide of drivel?
  • Is this a failure on the website's part--that it doesn't guide me to what I, at least, regard as the high value parts either because they are publishing too much stuff without sufficient organization or because they are missing crosslinks?
  • Is this a failure on the market's part--that what gets eyeballs that can be sold to advertisers is not of great interest to me?

I do not know...

Harry Enten: [Has Trump Already Lost Nevada? | FiveThirtyEight][]: "In Nevada... polls suggest a tight race.... [But] Democrats, thanks in large part to high turnout on the last day of early voting... have built a large lead...

...What exactly does Clinton’s early-vote advantage in Nevada tell us? And if Clinton does win the state, how does it affect her chances in the Electoral College overall?... Of everyone who has voted early so far in the Silver State, there are 6 points more registered Democrats than Republicans. That’s huge because most voters in Nevada vote early... 70 percent.... Still... the early vote numbers don’t guarantee the polls are wrong.... But let’s say the polls are wrong in Nevada, and, as Nevada political reporter Jon Ralston put it, Trump “needs a miracle in Vegas on Election Day — and a Buffalo Bills Super Bowl championship is more likely” to win the state. How much of a difference would that make to Clinton’s chances of winning the White House?

Well, it depends on whether the polls are underestimating Clinton only in Nevada or whether there’s a more systemic problem. Let’s say the error in Nevada was more specific to Nevada.... If we give Clinton Nevada, her chances of winning the Electoral College increase from 65.5 percent to 68.1 percent....

Even if Nevada isn’t a sign that the polls are underestimating her everywhere, a Clinton win in the Silver State does keep an important path open for her.... Without Florida or North Carolina... [and] New Hampshire... Clinton falls back below 270.... Clinton can, though, substitute in Nevada and its six electoral votes for New Hampshire....

A polling miss in Nevada is far more significant if it means the polls are off in other states in the same direction. In 2012, the polls underestimated Obama in Nevada and all over the map. That led to an easy Obama victory, despite pre-election polls pointing to a close race. In the FiveThirtyEight polls-only forecast, which assumes that if one state’s polls miss then a polling error in that direction is more likely to occur in other states too, Clinton wins 88 percent of the time with a victory in Nevada. That’s obviously a big difference....

For now, though, we don’t really know what the early vote in Nevada portends for Clinton nationally...

Comments