Why The Polls Stunk Up The Place
The Stream - Wednesday November 9, 2016
by William M Briggs
Boy, were the pros’ polls and predictions lousy. And not just lousy, malodorous. And worse than malodorous, embarrassing. Let’s see.
Swing and a Miss, and a Miss, and …
The LA Times on the 6th put out a “final” election map giving Hillary 352 electoral votes. Ouch.
Famed pollster Frank Luntz as late as 6:43 PM EST yesterday tweeted, “In case I wasn’t clear enough from my previous tweets: Hillary Clinton will be the next President of the United States. #ElectionNight” Pretty clear, Frank, pretty clear.
Some of the state polls make for high comedy. Poll averages had Hillary up by 2 in Pennsylvania, and she lost by 1, a significant swing. Hillary was supposed to win by 7 in Wisconsin. She lost by 3, an enormous turnaround.
A pollster with even more elite cred, Nate Silver, has been unequivocal all the way. Hillary would win, he said, with varying, precise numerical probabilities, always large. The most he would concede was that his last models showed “more uncertainty,” but that “Clinton will probably win anyway.”
The totals are still not final as of this writing, but it looks like Trump might win the popular vote, too. Yet most polls (on Monday) had Hillary up by at least 2 points, with some as many as 6. Only the IBD and LA Times tracking polls had Trump besting Hillary. You wonder if at the LA Times the tracking poll people were on speaking terms with electoral-college-prediction people.
The New York Times started the day yesterday giving Hillary more than an 80% chance of winning the election.
Some of the state polls make for high comedy. Poll averages had Hillary up by 2 in Pennsylvania, and she lost by 1, a significant swing. Hillary was supposed to win by 7 in Wisconsin. She lost by 3, an enormous turnaround.
We could go on, but there’s no reason. The performance of our elite media was a bust. Why?
Nothing to See Here
Part of the reason is good, old-fashioned confirmation bias; and part is worship at what I call the Cult of Measurement. Last, but most important, is a phrase you’ll be hearing a lot these next few days: “Nobody saw this coming.”
Not so. Some people predicted Trump would win way back on New Year’s Day. The signs were there if you knew what to look for. Problem is, the pros saw the signs but dismissed them as uninteresting.
The mainstream press didn’t want to believe there was a “shy” Trump effect. Yet the evidence for hidden support was there and in plain view.
What happened with Brexit wouldn’t apply here, they insisted. The “migrant crisis,” especially events broadcast out of Sweden and Germany, the car bombings and mass rapes, and where large areas of cities are now so barbarous the police don’t dare enter, taught no lasting lessons. Open borders -- Hillary’s self-professed “dream” -- would be welcomed by all but a handful of “racists.” Oh, and deplorables.
Killing sprees in France, Belgium, and on American soil, too, were seen as important, yes, but they were dismissed as small matters that could be dealt with by, for instance, creating a no-fly zone in Syria. The punditocracy didn’t see that a good chunk of Americans had no desire for a war with Russia.
Yes, elites knew of some dissatisfaction with Obamacare, which scarcely any elite uses, but a few small tweaks would fix it.
The Project Veritas videos showing Democrats were paying people to create violence at Trump rallies, among other improprieties, were dismissed by some as “fakes” and “frauds.” The media thought the best way to handle Wikileaks’ steady stream of damning revelations was to ignore them as much as possible.
The attitude to money was startling. When the results were first coming in yesterday and, if you squinted, it looked good for Hillary, a snide professor at the University of Maryland tweeted out, “Huh. I guess uneducated white voters don’t participate in prediction markets.” No awareness on his part that working-class folk don’t have money to blow on such frivolities.
Elite pundits really do live in self-created opaque bubbles. They didn’t know how much their vision for the world was disliked. The Pauline Kael syndrome is real. They knew no one who could challenge their vision.
Seeing What You Want to See
Confirmation bias is when you only see evidence that supports your belief. Reread the tweet by the professor above, which he meant as an insult. But it was a true statement! The man should have recognized that large numbers of people were not being accounted for in prediction markets.
The mainstream press didn’t want to believe there was a “shy” Trump effect. Yet the evidence for hidden support was there and in plain view.
Technicalities led to other errors. Democrats were over-sampled and many who were supposed to be likely voters were missed because it wasn’t recognized how different this election was from others.
Everybody around elites -- namely other elites -- were saying Hillary was going to take it. It’s difficult to discount the opinion of all your colleagues. Wishcasting, a type of confirmation bias well known to meteorologists, also played a role. Wishcasting is when the evidence that supports a strong desire is given more weight than it deserves. Is there any doubt the whole of the elite media were desperate for Hillary to win?
Seeing Too Much
The last flaw is technical (read about it here), yet it’s easy enough to grasp in broad strokes.
Pollsters sometimes make two mistakes: they believe everything can be measured, and therefore can be mathematically modeled. This is the Cult of Measurement. The enormous work put into their “rigorous and scientific” models causes pollsters to fall in love with them. And they therefore don’t realize that their models, which by definition cannot account for what cannot be measured, miss important details.
This is why Nate Silver and Frank Lunz failed so spectacularly. They didn’t realize that this election was not like others. Since they couldn’t quantify and model the changes, like those identified above, they didn’t think the changes were important.
We’ll see in 2020 if they learned their lesson.
Read More Here
No comments:
Post a Comment