A little-known US polling firm whose numbers were cited by national news outlets and loaded into major poll aggregators admitted on October 20 that it had fabricated most of its surveys, two weeks before the November 3 midterm elections.
Wedgewood Polls cast the hoax as an "experiment" meant to expose how easily a fake pollster could win credibility. But FiftyPlusOne, the election data site that had been investigating the firm, says Wedgewood went public only after it was asked to answer for anomalies in its data, hours before an exposé was due to run.
Wedgewood Says Faking Credibility Was Far Too Easy
In a statement posted on X, Wedgewood said its team had worked in polling and research and had watched public trust in surveys erode as low-quality, unverified polls multiplied. It set out to test whether a firm that did not really exist could be taken seriously. The answer, it said, was "far too easy."
"Despite having no history, formal credentials, or real data behind the vast majority of our releases, Wedgewood Polls was added to major polling aggregators, including The New York Times, FiftyPlusOne and VoteHub, and repeatedly treated as a legitimate data source," the firm said.
Wedgewood said the project was designed "to test whether media outlets, aggregators, and commentators would report on data without verifying the source, methodology, or underlying firm legitimacy." It said only one release, a crowdfunded Nebraska poll, was conducted with a real methodology and was used as a benchmark against its invented "synthetic" data. The firm insisted it had no intention of swaying any election.
Parts of that account check out. Straight Arrow News (SAN) reported in August that Wedgewood's polls appeared on The New York Times polling page, though the newspaper flagged the firm as one that did not meet its standards for select pollsters. Wedgewood also claims its numbers were cited on air and online by MS NOW, formerly MSNBC. That claim has not been independently confirmed, and neither MS NOW nor The New York Times had publicly responded at the time of writing.
A Mississippi Upset Poll Drew National Headlines
The release that ultimately unraveled the scheme was a Mississippi Senate poll Wedgewood said it fielded from October 12 to 14. It claimed to have surveyed 511 likely voters and showed Republican Senator Cindy Hyde-Smith leading Democratic challenger Scott Colom by just 52% to 48%, within a 4.3-point margin of error.
A competitive race in Mississippi, one of the most reliably Republican states in the country, was instant news. Newsweek ran a story on October 15 framing the numbers as a sign Democrats could pull off a historic Senate upset in the deep-red state. Political news site TPM cited the same poll that day as one of the few independent surveys not linked to Colom's campaign, and Colom's team put out a press release declaring the race tied.
FiftyPlusOne noted that the poll landed only hours after Sabato's Crystal Ball, a widely followed election forecast, shifted the race from "Safe Republican" to "Likely Republican." Mississippi outlet Magnolia Tribune also pointed out that a Wedgewood poll of Louisiana's Republican Senate primary in June missed the final result by nearly 13 points. According to conservative site Frontpage Mag, Wedgewood had also published a Kansas Senate poll showing Democrat Adam Hamilton ahead of Republican Senator Roger Marshall by 2 points.
FiftyPlusOne Rebuilt The Survey And Found The Fakes
Acting on a tip, FiftyPlusOne asked Wedgewood for the raw data behind its Mississippi poll. It then spent about $900 running the same questionnaire with 600 respondents on Pollfish, the online survey platform Wedgewood said it used, and compared the two datasets. The site published several red flags:
- An implausible sample: Wedgewood's unweighted sample was 67% white and 31% Black, with an unusually low design effect of 1.18, even though the firm said it did not use quotas. FiftyPlusOne's own sample came back 48% white and 48% Black, with a design effect of 1.66.
- Respondent IDs that do not occur naturally: In genuine Pollfish data, roughly half of respondent IDs contain a hyphen and only about 40% are exactly nine digits long. None of Wedgewood's IDs had a hyphen, and every one was nine digits. FiftyPlusOne put the odds of that happening by chance at close to zero.
- Timestamps that do not line up: In real data, the platform's transaction numbers track survey completion times almost perfectly, with correlations between 0.94 and 1. Wedgewood's data showed a correlation of negative 0.03. The transaction numbers suggested the responses were collected around July 2026, not in October.
- Signs of AI generation: The fonts and writing style of Wedgewood's reports and crosstabs bore hallmarks of AI tools, though FiftyPlusOne cautioned that legitimate pollsters use AI too and that this alone proves nothing.
FiftyPlusOne said it asked Wedgewood to respond on October 20 and received only an abusive message, followed shortly by the firm's public confession on X. Cooper Burton, a data scientist at FiftyPlusOne, told Raw Story the site had planned to publish its findings the next morning and moved them up after Wedgewood preempted the story.
Michigan Democrat Says Fake Polls Sank Her Campaign
The fallout reached well beyond Mississippi. During Michigan's Democratic Senate primary, Wedgewood released a poll on June 24 showing Abdul El-Sayed, a former Wayne County health director, at 46%, US Representative Haley Stevens at 39%, and State Senator Mallory McMorrow far behind at 9%. The same poll suggested McMorrow was the only Democrat who would lose to Republican nominee Mike Rogers, a former congressman. El-Sayed went on to win the nomination, and McMorrow left the race on July 5.
Writing on X on October 20, McMorrow said the "experiment" had real consequences for her campaign. "Donations dried up. Volunteers were devastated and pleading for guidance on what to do," she wrote. "Reporters hounded me with calls to drop out. Sure, a point was proven. Polling is broken. But tens of thousands of people poured their hearts, time, energy, hard-earned money into something they believed in," she added, saying the effort "was treated like a game."
McMorrow said a separate primary poll from another firm, which she described as an open online survey link that people could take multiple times, had already shown her far behind, and that her campaign had publicly called it out. When Wedgewood's numbers followed, she wrote, "It was a one-two punch. Then all of the sudden it was a pattern, and people believed it had to be true. It killed us." She stressed that she still backs El-Sayed against Rogers, while urging the media to "take a hard look at what's legitimate and what's not."
Nate Silver Strips Wedgewood From His Midterm Forecast
Wedgewood's polls had quietly fed into some of the most closely watched midterm forecasts. Election analyst Nate Silver, founder of FiveThirtyEight, said on October 21 that he had removed the firm's surveys from his Silver Bulletin models and banned it permanently. He said the fake polls had leaned roughly 1 point more favorable to Democrats than the broader consensus.
The largest shift came in Michigan, where El-Sayed's projected lead over Rogers narrowed from 3.6 points to 3 points. Silver noted that the change also reflected a newly added Marist poll showing El-Sayed ahead by 2 points. Democrats' share of his generic congressional ballot average slipped from 51.7% to 51.5%, while Republicans moved from 42% to 41.9%.
"Overall, our forecast is little changed, though that's partly because some strong generic ballot polls entered this AM for Democrats," Silver wrote on X. He also defended the aggregators that had included the firm: "Previously, Wedgewood responded to inquiries from multiple polling aggregators and answered questions about their identities, methods, and funding, which is why their polls were subsequently aggregated."
According to Newsweek, between 15 and 17 Wedgewood polls had been cited across various sites before they were pulled. Newsweek said it has corrected and retracted stories that relied on the firm's data.
Red Flags Surfaced Months Before The Confession
Questions about Wedgewood were raised as early as August. SAN reported that the firm's website was built on Google Sites and listed no staff names, phone number, or address, and that its only social media presence was an X account opened in June. Wedgewood described itself as a two-person team, one progressive and one conservative, but refused to identify its members or appear on camera, citing past doxxing of one partner. Its web domain was registered on June 26, a day after it released its Louisiana poll, and the firm acknowledged at the time that it had not yet formally incorporated.
FiftyPlusOne rejected Wedgewood's claim that aggregators had done no vetting. The site said the firm identified its founders as Jack Sims, a Republican, and Andrew Prater, a Democrat, and that it independently confirmed Sims had worked at an established polling firm. Wedgewood's published methodology also appeared to meet industry standards. FiftyPlusOne added that Wedgewood had solicited donations from the public for "crowdfunded" polls.
Aggregators Tighten Vetting After Another Polling Hoax
FiftyPlusOne has deleted every Wedgewood poll from its database and averages and will no longer include data from any firm founded or run by Sims or Prater. Going forward, it plans to require video or in-person meetings with new pollsters and may ask for respondent-level raw data before adding them.
The case is not the first of its kind this year. According to Raw Story and SAN, a firm called Median Strategies released a Los Angeles mayoral poll in August and, when challenged, told the Los Angeles Times it had been a "short-term social experiment," an explanation strikingly similar to Wedgewood's. Earlier this year, FiftyPlusOne also dropped a small Florida pollster after finding it had altered some respondents' answers. Cheap, fast online survey panels have made it possible for almost anyone to publish a poll, and the hoaxes suggest that the gatekeepers have struggled to keep pace.
Why The Hoax Resonates In Taiwan
The scandal lands as Taiwan heads into its own vote. Ahead of the 2026 Taiwan local elections on November 28, parties, campaigns, and media outlets are releasing a steady stream of surveys, from party-commissioned telephone polls to AI simulations of virtual voters. In August, President Lai Ching-te alleged that fake polls and Chinese election interference were circulating in the Kaohsiung mayoral race, a claim critics disputed.
Taiwan has already prosecuted at least one fake poll case with a cross-strait dimension. In August 2024, the Taichung District Court sentenced the head of an online media outlet to eight months in prison over a fabricated poll on that year's presidential election, which prosecutors said was commissioned with funding and instructions from Chinese officials. Taiwanese law also bars the publication of election polls in the final 10 days before voting. The Wedgewood case shows that verifying who stands behind a poll, and whether its raw data is real, matters as much as the numbers themselves.
Social Media Links & Editor Tag - Penny