I promised warned you there’d be more. And where we last left off, I talked about how elite pundits were motivated to call this election wrong before it even began.
How could a Daily Kos nobody such as myself produce a prediction 3 weeks out far more accurate than those paid millions to do the same?
For the record, here was my prediction:
I sprained my arm patting myself on the back.
This isn’t to brag, but to show how dangerously wrong elites with access to insiders and news media apparatus were about this election, and how easy it was to call this election right based on the data.
Garbage In…
Let’s start with the obvious. Republican pollsters flooded the zone with garbage polls.
What the above table tells you is that non-partisan pollsters were calling this election precisely; media elites were motivated to ignore that. And when it comes to partisan pollsters, some were more egregious than others.
Trafalgar has had an A- rating from Nate Silver. We’ll get to Nate in a minute. From the recent Slate article highlighting every single miss from Trafalgar this cycle:
A 2020 New York Times article about the company—published before Biden’s victory, even—noted that “Trafalgar does not disclose its methods, and is considered far too shadowy by other pollsters to be taken seriously.” But it does release crosstabs, i.e., breakdowns of polling responses by subgroup, which allow observers to see under the hood at least a little bit to find out how it gets its overall numbers.
But Trafalgar wasn’t the only pollster playing games this cycle. Data for Progress, an ostensibly left-wing polling operation, was notorious for also predicting a more rightward outcome of this election than the nonpartisan pollsters predicted. Data for Progress’s head lost $50,000 betting on races based on his own polling.
Data for Progress has had a B rating from Nate Silver. And don’t forget patriot polling, which turned out to be two high school preppies just making stuff up.
Was the high school kids’ work included in Nate Silver’s and the RCP averages? You better believe it. Also, here are the kids in case you think I’m making this up.
Patriot polling
Why? Why produce bad polling? Shouldn’t a pollster be motivated to be as accurate as possible? From the Slate article again:
One can only speculate as to where Trafalgar found a set of Gen Z voters—89 of them, according to the other data it provided—that broke 2-to-1 for the MAGA candidate in a swing state like Georgia. But there are rewards available in the attention economy for doing so, which provides free advertising for Trafalgar’s paid, private work.
Trafalgar understands our attention economy. That’s all. Trafalgar founder Robert Cahaly knows that. And he made his money.
...and Garbage Out
And that brings us to our buddy Nate Silver, who decided a policy of Laissez-faire was the correct approach to obvious bad faith by partisan pollsters, thus rendering 538 the Facebook of polling aggregators. And no, it’s not like this issue wasn’t brought up to Nate Silver, who made the stunningly ignorant argument that because Democrats weren’t spamming the polling aggregates in response, their private polls were giving them bad news. This argument was even repeated (but not necessarily endorsed) here on the front page back on November 2.
First, to Nate Silver and the partisans making that argument:
Second, the issue was never about campaign internal polls, which Nate Silver knows are not included in the 538 average. Campaigns have all sorts of strategic reasons for not releasing their internals (most importantly, it reveals where the releasing campaign believes the race is). The issue always was partisan, third party polls, often with either no published crosstabs and methodology, or wildly implausible crosstabs, spamming the polling aggregates. This was a deliberate bad faith effort that was statistically apparent. Look what it did to Nate’s own Senate forecast!
I circled the time period when the spam polls really show up. Look what it does to Nate Silver’s forecast. See how the 538 forecast goes from right to wrong in the course of two months.
Nate Silver’s final forecast was astonishingly 51R/49D with Democrats having only a 41% chance of winning 50 seats. They may win 51. Nate Silver, ironically, would have been far more accurate if partisan polls were excluded from the average. Nate Silver’s defense is that he has no idea whether the partisan or nonpartisan polls would be more accurate. This is stunningly obtuse.
And RCP? Nobody should take this bunch of out of touch partisans seriously ever again.
