3 ms·

Nate Silver has done some good work on what exactly went wrong, and it was a combination of factors: - (some) polls were systematically off, mostly because the

Nate Silver has done some good work on what exactly went wrong, and it was a combination of factors: - (some) polls were systematically off, mostly because they underestimated white non-college turnout and overestimated black turnout. Note that poll errors were actually lower than for any other election–only this time they actually got the winner wrong. - Most of the poll aggregators operated under the assumption of statistical independence of state-level results. This was a serious mistake, and both Huffpost as well as The Upshot (to a lesser extend) should be somewhat ashamed. 538, by the way, did this right and had Trump at 3X% on election day. - There actually was some significant change in voters' intent in the last days, possibly fuelled by all the news stories the FBI directors was generating. It's highly likely that an election two weeks earlier or two weeks later would have ended differently. It's also important to note that newspapers aren't really in the business of predicting the future. If they "relied too much on social media" for any facts they reported, I'd love to see some examples.

humanrebar 9y ago
Beyond making bad predictions and tonally describing the election poorly (it didn't seem to be coming down to the wire based on the coverage), the NYT did an atrociously poor job of explaining the mind of a Trump voter to its readership. To this day, there is a lot of reliance on imagery of be-mulleted bogeyman that is both facile and harmful if America wants to live in harmony with itself again some day. That is, we're not just talking about getting the numbers wrong. The "narrative" itself is way out of whack as well.
kobeya 9y ago
2X% actually (I remember watching it in real time), but the point stands as it was way higher than anyone else thought.
- scott_s 9y ago
  And Nate Silver got a lot of flak for that in the week before the election.
  - zaphod12 9y ago
    He did indeed! He published an article in the last week that listed 4 outcomes for the election, each with ~25% confidence. 3 Had a Clinton victory and 1 did not. I recall him also making some pretty emphatic condemnations of Sam Wang's Princeton Election Consortium, arguing that the numbers absolutely did not support PEC's >95% confidence in a Clinton win. Wang, for his part was so confident that he offered to eat a bug if Clinton lost. Wang's statistical method turned out to be a disaster, but he did follow through.
    - kobeya 9y ago
      Having actually eaten a bug before (yay Boy Scout survival badge), I'm going to remember that one. The follow-through will impress people but it's not actually dangerous or even that unhygenic (as long as you eat the right bug).
lettergram 9y ago
The system I've been working on actually predicted a Trump win: https://projectpiglet.com/ https://projectpiglet.com/ The problem was so many people just hated Clinton, and at the time, Trump had more people positive about him.
- matt4077 9y ago
  Then you should be very sceptical of your system. "History proves me right" works for Fidel Castro, but not for lottery winners.
  - lettergram 9y ago
    I am fairly skeptical, but given it's been pretty consistently accurate, I'm becoming more confident. I look at net Promotor Score on any type of entity. It's predicted the rise of Bicoin, and various other currencies, but it's not always perfect. For example, it didn't predict a brexit. It could just be I don't have enough data. It's a work in progress.
1wheel 9y ago
Silver's analysis probably overstates the impact of the Comey letter: > it’s now clear that Mrs. Clinton was weaker heading into Oct. 28 than was understood at the time. Several other polls were conducted over the same period that showed Mr. Trump gaining quickly on Mrs. Clinton in the days ahead of the Comey letter. And the timing of these polls — particularly the gap between when they were taken and when they were released — has probably helped to exaggerate the effect of Mr. Comey’s letter on the presidential race https://www.nytimes.com/2017/05/08/upshot/a-2016-review-theres-reason-to-be-skeptical-of-a-comey-effect.html https://www.nytimes.com/2017/05/08/upshot/a-2016-review-ther...

humanrebar 9y ago

Beyond making bad predictions and tonally describing the election poorly (it didn't seem to be coming down to the wire based on the coverage), the NYT did an atrociously poor job of explaining the mind of a Trump voter to its readership. To this day, there is a lot of reliance on imagery of be-mulleted bogeyman that is both facile and harmful if America wants to live in harmony with itself again some day. That is, we're not just talking about getting the numbers wrong. The "narrative" itself is way out of whack as well.

kobeya 9y ago

2X% actually (I remember watching it in real time), but the point stands as it was way higher than anyone else thought.

scott_s 9y ago
And Nate Silver got a lot of flak for that in the week before the election.
- zaphod12 9y ago
  He did indeed! He published an article in the last week that listed 4 outcomes for the election, each with ~25% confidence. 3 Had a Clinton victory and 1 did not. I recall him also making some pretty emphatic condemnations of Sam Wang's Princeton Election Consortium, arguing that the numbers absolutely did not support PEC's >95% confidence in a Clinton win. Wang, for his part was so confident that he offered to eat a bug if Clinton lost. Wang's statistical method turned out to be a disaster, but he did follow through.
  - kobeya 9y ago
    Having actually eaten a bug before (yay Boy Scout survival badge), I'm going to remember that one. The follow-through will impress people but it's not actually dangerous or even that unhygenic (as long as you eat the right bug).

lettergram 9y ago

The system I've been working on actually predicted a Trump win: https://projectpiglet.com/ https://projectpiglet.com/ The problem was so many people just hated Clinton, and at the time, Trump had more people positive about him.

matt4077 9y ago
Then you should be very sceptical of your system. "History proves me right" works for Fidel Castro, but not for lottery winners.
- lettergram 9y ago
  I am fairly skeptical, but given it's been pretty consistently accurate, I'm becoming more confident. I look at net Promotor Score on any type of entity. It's predicted the rise of Bicoin, and various other currencies, but it's not always perfect. For example, it didn't predict a brexit. It could just be I don't have enough data. It's a work in progress.

1wheel 9y ago

Silver's analysis probably overstates the impact of the Comey letter: > it’s now clear that Mrs. Clinton was weaker heading into Oct. 28 than was understood at the time. Several other polls were conducted over the same period that showed Mr. Trump gaining quickly on Mrs. Clinton in the days ahead of the Comey letter. And the timing of these polls — particularly the gap between when they were taken and when they were released — has probably helped to exaggerate the effect of Mr. Comey’s letter on the presidential race https://www.nytimes.com/2017/05/08/upshot/a-2016-review-theres-reason-to-be-skeptical-of-a-comey-effect.html https://www.nytimes.com/2017/05/08/upshot/a-2016-review-ther...