It’s been clear for a while that some more-uncouth-than-usual ratfucking supported lefty candidates like Francesca Hong and Abdul El-Sayed during primary season. I wrote about this back in August, when we knew about campaign-funded polls being treated like normal ones and rumor-mongering from lefty pundits. 𝕏Many 𝕏other 𝕏people were on this beat — knowing that we were being manipulated was kind of like knowing that Liberace was gay, in that you did not need to be a Poirot-level sleuth to figure it out.
Now, after the polling and analysis site * Fifty Plus One *started looking into the legitimacy of a group called Wedgewood Polls, Wedgewood admitted that they’re as fake as a candidate’s accent when courting the “urban” vote. But Wedgewood claims to be
Do you buy that? If you do: Please Venmo me all your money, because I need it to buy…let’s say medicine for my sick…veterans. This is actually the * second time* that a fake polling firm has tried the “social experiment” excuse: A group called Median Strategies, which produced fake polls favorable to LA mayoral candidate Karen Bass, tried the same move two months ago. So, we’re either in a golden age of Andy Kaufman-esque social experiments or a golden age of online bullshit!
What was this, actually? Some overzealous lefties? A Republican ratfuck? An attempt to manipulate betting markets? Maybe a better question is: Is it possible anymore to disambiguate those three? Republicans have no stronger ally than online lefty freaks, and once you’re committing fraud, you might as well get on Kalshi and try to score big. The biggest divide in American politics is no longer liberal/conservative; it’s normal person/nihilist Batman villain, and whoever did this is clearly on the Batman villain side of the divide.
Wedgewood clearly favored whichever candidate in a race was furthest left. As I write this, Wedgewood’s Twitter account is still live, and it’s full of breathless announcements touting favorable polls for left-leaning candidates. Here are some of their most notorious entries:
The El-Sayed/Stevens/McMorrow poll appears to have had a real impact on that race: 𝕏McMorrow claims that the poll led to donations drying up and increased calls for her to drop out. It might be true that Republicans who wanted to pick their opponent (El-Sayed) also wanted to pick their opponent’s opponent, and therefore tried to make McMorrow look unviable.
Nate Silver — whose polling model incorporated Wedgewood polls, and who is 𝕏writing an article about the fallout — found that Wedgewood polls “tended to be ~1 point more favorable for Ds than the consensus”. I did my own as-good-as-you’d-expect-for-some-fucking-comedian analysis and found a +1.8 Democratic tilt (the abundant shortcomings in my methodology are described in this footnote1). But Nate and I are looking mostly — and maybe entirely, because I don’t know which polls are in Nate’s analysis — at *general election *polls. The dirtiest of Wedgewood’s work appeared to be in the primary — they had Crowley losing by 29 points! Wedgewood may have been mostly interested in the primary and kept the ruse going through the general election because disappearing in a puff of smoke *during election season *would give away the ~~scam~~ experiment.
It is common for a poll to have a partisan lean. It is not common for a fake poll to have a partisan lean — if you’re making up numbers, you should make up believable ones! The fact that Wedgewood’s numbers weren’t just *fake *but odd — so odd, in fact, that people eventually sniffed out the ruse — suggests that their motivation was to affect the race. Wedgewood says that they “never intended to sway an election or permanently mislead the public” — I say “bullshit”. And I wonder what a jury in a criminal trial for fraud might say.
I found 18 Wedgewood polls compiled by the site politicalbetting.com. These were all *general election *polls — I couldn’t find any compilation that included primary season. I wanted to compare Wedgewood polls to other polls around the same time, but that created two problems:
Sometimes, there just wasn’t enough credible polling for a comparison. So, I threw out Wedgewood polls for six races: MS senate, KS senate, KS governor, NE senate, NE governor, and NV governor. That left 12 Wedgewood polls that could be compared to other polling data.
There’s ambiguity in the phrase “around the same time”…how far back and forward should I go? And how do I correct for changes in the average caused by the fake Wedgewood polls? This part was as much art as science, but I did my best.
In the 12 polls I could analyze (MI senate, MI governor, GA senate, GA governor, ME senate [two polls], IA senate, IA governor, OH senate, OH governor, TX senate, TX governor), Wedgewood polls had a D-leaning bias that averaged out to 1.8. In only one race — TX governor — was the Wedgewood poll more favorable to the Republican (by about a point) than similar polls.
