With votes still being counted, things are not looking particularly good for President Trump’s re-election chances. Some states stopped counting ballots altogether last night for inexplicable reasons only to have the resumption this morning bring about massive, sometimes improbable dumps of Biden votes. In one instance in Wisconsin, it appears Biden won 93% of a 150,000 vote dump when Hillary Clinton only won the county in question by 60% in 2016. Of course, all of this is so ambiguous that there’s little that can be done. Twitter will fight about it, and Republicans will fume, but the results aren’t likely to change. A similar story is playing out in Michigan, where Trump’s large lead has turned into a deficit because of a vote dump showing 100% of votes went to Biden.
Welcome to the People's Democratic Republic of Michiganzuela where, mysteriously, every single mail-in vote of the 100,000+ that was counted all went to Joe Biden! pic.twitter.com/cJbuD8mUN2
— Brandon Morse (@TheBrandonMorse) November 4, 2020
Regardless, the fact that we don’t know the official winner of the Presidential election doesn’t mean we can’t make some definitive conclusions about what has transpired. For example, we now know in unequivocal fashion that the polls were absolute garbage again this cycle.
Just how bad were they? Take a look at these misses on the polling averages.
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
These aren’t just misses. These are systematic, widespread issues with polling that can’t simply be explained by last minute voter surges or honest mistakes. In five of those states, Trump won well outside the margin of error compared to what the polling was saying. Note that the much vaunted 538 averages, which Nate Silver curates based on his great intellectual heft, are even worse than the straight averages from RCP. We wouldn’t want that dirty Trafalgar pollster included in the data, right?
At some point, 538 should probably explain how exactly they get their “ratings” for pollsters.
NYT/Siena, rated an A+ pollster by 538, overestimated Biden's support by:
6+ in FL
4+ in NC
Probably 6ish in MI when the count is in
10 ish in WI
10+ in IA
9+ in OH
— Tom Bevan (@TomBevanRCP) November 4, 2020
Wisconsin may have been the biggest miss of all though.
49.6-49% currently in a state that “quality pollsters” NYT Siena had Biden +11, WaPo Biden +17, and Reuters +10. The closest? At Biden +1 was… Trafalgar. https://t.co/m0ENdNNAlR
— Marcus (@MarcusHUSA) November 4, 2020
Ohio and Florida weren’t much better.
Again, for the record. The final (mainstream) Ohio polls:
Quinnipiac: Biden +4
Emerson: Biden +1
Quinnipiac: Biden +5
Trump, with 94% reporting, winning by +8
— Kimberley Strassel (@KimStrassel) November 4, 2020
Final @trafalgar_group poll in FL had Trump ahead by 2. The only poll to give Trump this much.
Trump is now ahead by 2.4% in FL with +90% votes in.
6 for Biden was Emerson’s final
4 Biden Reuters
3 Biden Siena
When will @Trafalgar get their due?
— Yossi Gestetner (@YossiGestetner) November 4, 2020
Of course, there’s always one poll that’s so out there that it’s hard to believe it’s not a parody. Quinnipiac comes close, but this has to take the cake.
It's early, but it's worth noting that Univision Network teamed up with Joe Biden's Latino pollster to claim Trump was *LOSING* Cubans by 15.
He's winning Cubans by ~40.
Univision's poll was off by 55 pts.https://t.co/fN2aRcvmMY
— Giancarlo Sopo (@GiancarloSopo) November 4, 2020
That Univision poll missed the mark by a whopping, probably unheard of 55 points in regards to Trump’s share of the Cuban vote. Whoever conducted that poll should never be allowed in the industry again. Actually, they should probably retire from all lines of work and go live in the woods somewhere just to be safe.
Other notable misses outside of the presidential election include Lindsey Graham and Susan Collins coming out victorious. Graham was supposedly tied with his challenger. In the end, it wasn’t close. Tom Tillis also won as an underdog in the polling. John James is over-performing in Michigan as well, holding a slight lead even though that race is likely to flip at the last second if the past is any indication.
In short, the polling this cycle was actually markedly worse than 2016, and I’m not sure I thought that was possible. This is an industry that needs to be burned to the ground, not given yet another chance to set narratives and influence elections. Of course, the pollsters aren’t the only problem. The polling aggregators like 538 are perhaps even worse given the supposed authority they place behind their data, but that’ll be for my next post.
(Please follow me on Twitter! @bonchieredstate)