Wednesday, January 20, 2016

Nate Silver's shifting line

Big Data is getting egg all over its face this election cycle. Agnostic has already eloquently articulated as much, so let's go ahead and pile on.

Just two months ago, Nate Silver, head proprietor of 538, wrote:
For my money, that adds up to Trump’s chances being higher than 0 but (considerably) less than 20 percent.
Contender for 2016's most 
punchable shitlib face?
To his credit, he does allow--in cowardly CYA weatherman fashion--that Trump's chances are greater than 0%!

Silver's now on to explain how Trump can't win the general election because he's unpopular with voters. If he had any shame, he'd concede that his November assurances that Trump was a long shot at best for the GOP nod is now very much the contrarian position not only based on polling data--which he discounted in favor of prediction markets--but also based on... prediction markets.

The odds as of January 20th:

Typical Trump rally
Yes, betting against ¡Jabe! offers a potential 20% annual return (if purchased now and cashed out in mid-July when the Republican national convention occurs), though the market takes 5% of a participant's withdrawals.

Bizarrely, Silver asserts that while Trump's poll numbers are high, electoral participation by his supporters will not be correspondingly impressive.

Googling images for "[candidate] rally crowd" makes it abundantly clear that Trump's rally crowd advantage dwarfs even his lead in the polls, and he's holding several of these rallies per week, sometimes even per day. The presumption that rally attendance inversely correlates with electoral participation is a curious one to say the least. "We'll drive an hour and then wait in line outside in the bitter cold to see, from the nosebleed section, a tiny figure on the dais that the jumbotron reveals is in fact Trump, but we're not swinging by the polling location at the church four blocks down the road after work to vote for him in our state's primary!"

Unsurprisingly, Silver has nothing to say about social media activity either, another front on which Trump's dominance is uncontested.

Typical Rubio rally
As pleasantly surprising as it is to be able to crush the DJIA by betting against ¡Jabe! or against Chris Christie (who was at 7% when I bought), Marco Rubio's purportedly strong showing is the one I find most puzzling. He's been relentlessly promoted by the establishment, and it's safe to presume that the residual in those 5:1 odds comes from the assumption that said establishment will do something untoward to thwart the will of the Republican electorate and destroy the spirit of the nominating process by sneaking him in by hook or crook (or brokered convention) if Trump is chosen by actual primary voters, but Rubio's polling numbers have been--and continue to be--terrible. The most recent Reuters/Ipsos five daily national tracking poll among likely Republican primary voters:

He does even worse among all self-identified Republicans, garnering just 8.1% and placing fifth (behind ¡Jabe!).

While we're doing our best to wreck desperately hoped for self-fulfilling prophecies media narratives, let's tackle the one about Trump not being popular among young people.

Republican partisan affiliation is positively correlated with age, and consequently Republicans aren't popular among young people generally, but Trump comfortably beats the rest of the bunch. Again, the most recent five day national tracking poll results for those aged 18-34 among Republicans (all Republicans, as sample sizes for likely GOP voters in the age range are too small to be reliable):

Rubio, the market's putative runner-up, gets handily beaten by "wouldn't vote". Ouch.


Thursday said...

Silver is right. Trump's chances of being the nominee are still low. The establishment, and voters, who gave us McCain and Romney are going to rally around someone, even if it is Cruz.

Audacious Epigone said...

There is a long paper trail and it's going to reveal a lot of people to be off one way or the other. While I'd like to think that matters, pundits rarely suffer for any incorrect predictions they make. To the contrary, they often make a killing getting everything spectacularly wrong.

chris said...

28 y.o Iranian immigrant pushes 20 y.o German girl to her death in front of train.

Ukrainians fleeing war in Ukraine denied asylum in Germany. Just not brown enough.

13 year old German girl gang raped for 30 hours by "foreign nationals"

Anonymous said...

Silver has offered the best mea Culpa one could hope for from someone who shits rainbows. Next, of course, will be a data-authority plea of why it's shameful to vote for Trump. SJWS and all...