Consensus or herding? (A: probably herding.) https://t.co/NbJX64TzUA pic.twitter.com/NJsmKmxOSs— Nate Silver (@NateSilver538) November 7, 2016
and when HRC wins by 4-5%, will it still be herding? https://t.co/Jh8fEoYKwD— Steve Saideman (@smsaideman) November 7, 2016
Clinton's swing state lead is tight. https://t.co/0fy6yWLITV pic.twitter.com/XlVyyKHrwY— FiveThirtyEight (@FiveThirtyEight) November 7, 2016
More clickbait. How about this: "Trump is behind in several states when he needs all of the swing states to go his way" https://t.co/qMk8bPHwo5— Steve Saideman (@smsaideman) November 7, 2016
Maybe I should be glad that he is pushing back against complacency, but I am just annoyed that most of his tweets have been aimed at getting attention with panic, rather than being clear that Trump needs to run the table of the swing states to win. And that is very unlikely to happen.
People are annoyed that Silver seems biased against HRC. That is not my complaint--that his predictions are more conservative, having been burned by Trump's win in the primaries. I am simply frustrated with how the stuff is being communicated--that the chances of Trump winning is like the Cubs coming back. No, not at all. The two events may have the same % probability assigned but the processes are so very different that the analogy is less than helpful. Winning three games in a row is hard, where each one can be decided by a bad pitch or fielding mistake. What is the equivalent of that in this race where the candidates have been mostly apart by 5% with noise causing that number to go up or down a bit? Being uncertain because one's model builds uncertainty into things is different than the actual history of baseball where teams have come back down three games to one. I have not been able to articulate an alternative metaphor than the lotteries and sports and Russian roulette that people use to discuss uncertainty and probability in this campaign. All I know is that the current set sucks.
And the responsibility for communicating this stuff is on Silver and those of his ilk. I wish they would get it right. But that might be my confirmation bias speaking. Mine ... or Silver's.
2 comments:
Lee Drutman suggested that in future campaigns, FiveThirtyEight should have an extra option on their forecast that would calculate the final result using the 2008/2012 model. I understand why Silver's doing this, but he's fighting shadows to the detriment of his reputation for accuracy. (When the Democrats have to actually send out emails to hyperventilating supporters not to update the page every 10 minutes, you're doing something uniquely wrong.) From this point onward, any US polling firm that doesn't even try to properly sample bilingual Hispanics, Asians/"Other" demographics and white women has to be nuked from orbit.
And yes, the aggregation world is basically in shreds over the difference between a 3% and 5%+ HRC victory. We're all morons.
IMO, Sean Trende and RealClearPolitics deserve the most blame for peddling 'missing white voters', false realignment theories and flooding their average with partisan polls.
Post a Comment