Aggregating and analyzing polling data has become a bit of a cult industry the last decade or so. It really took off in 2012, where Nate Silver rose to fame for predicting an election my dog could have guessed correctly. Regardless, sites like Real Clear Politics, which does straight averages of polls, and figures like Dave Wasserman have also become popular election cycle staples. Meanwhile, you have outfits like Cook Political Report which specialize in categorizing races as being leaning, strongly leaning, safe, etc. for one party.
What do all of the above have in common coming out of Election Day 2020? They were all proven be hilariously bad at their jobs.
Want to know how badly Silver’s 538 missed the mark? Here are a few examples.
FiveThirtyEight 20 hours ago: Biden wins Florida 69 in 100
NYT now: Trump > 95% chance of winning Florida pic.twitter.com/2o1nrO2ku9
— Peter J. Hasson (@peterjhasson) November 4, 2020
And now a look at how much Trump overperfomed the 538 averages in swing states:
A really rough night for polls and a disastrous night for certain pollsters such as Quinnipiac who showed Biden doing so much better.
— Josh Jordan (@NumbersMuncher) November 4, 2020
When it became clear that 538’s analysis was useless garbage, Silver got more combative throughout the evening. At first, he was insistent that Florida was an outlier and not proof of widespread systematic polling errors. As the night dragged on, it became obvious how wrong he was. Pollsters ended up off by 7-17 points in places like Wisconsin, Michigan, Ohio, and Iowa. The same people who were making fun of the Selzer poll earlier in the week (which ended up exactly correct in Iowa) were made fools of once again.
Cook Political Report was also laughably off on their House, Senate, and swing state ratings. They had a Democrat wipe out in the House coming. In the end, the GOP hugely over-performed their predictions.
Here’s the kind of arrogance we saw prior to the election from these people.
Would I rather trust those five polls? Or the 20+ October district-level polls I've seen, conducted by both Dem/GOP groups making major $$ decisions, showing Trump running 7-12 points behind his '16 margins in nearly all corners of PA?
Think I'll go with the latter.
— Dave Wasserman (@Redistrict) October 29, 2020
And here was Silver when the stuff really started to hit the fan.
You missed it huge again, and that's even if Biden sneaks this out. Your analysis is freaking useless. https://t.co/AYCjqoUBDZ
— Bonchie (@bonchieredstate) November 4, 2020
Trashing Trafalgar was perhaps the pollster bros’ biggest miss of the election. They insisted the methodology was flawed while giving A ratings to pollsters like Quinnipiac and NYT/Sienna despite their awful track records the last four years. In a word, the polling aggregate game was rigged and it blew up in their faces.
Eventually, it became clear they couldn’t just snark their way out of this, so Silver and Wasserman started reporting election returns and desperately gaming out how Biden could hang on, as if a close election win for the Democrat would somehow vindicate their previous, garbage analysis. It can’t and it won’t.
Anyone who believed the pollster bro narrative this year about double digit leads for Biden in states he ultimately lost should stop writing about politics. There's a point where you should realize: you're just not good at this.
— Ben Domenech (@bdomenech) November 4, 2020
The poll aggregate and analysis industry deserves to die a fiery death. It’s a useless barometer which takes trash data (i.e. polls) and tries to turn it into something authoritative. In the end, it’s a shell game. Any moron can assign a percentage chance to something and claim they were right all along no matter what the result is. Sites like 538 are absolutely pointless because the data they use is pointless. You don’t even have to assume bias on their part in this case. They simply don’t have the ability to do what they claim to be doing because accurate data simply doesn’t exist. Instead of admitting that, they continue to try to keep the grift going. These are people that should never work in the political field again.
Of course, we all know they will.
Today: "wow, the polls were pretty far off, maybe we should have taken them with a grain of salt."
Next Tuesday: "Here's Nate Entrails with a deep dive analysis of the latest tracking numbers from the NBS-Squippippinac Approve-O-Meter"
— David Burge (@iowahawkblog) November 4, 2020
(Please follow me on Twitter! @bonchieredstate)