Wedgewood Polls admitted most of its surveys were fabricated after major aggregators listed them. Here is what happened and how to read the next poll.
Ask almost anyone and they will say they do not trust the polls. Then a number lands in their feed, and the same people share it and argue over it, often without opening the link to see who ran the survey. Donors and campaigns have done the same with real money this year. That habit is the weakness Wedgewood Polls says it set out to test. On Oct. 20, with Election Day a week away, the firm admitted that most of its releases were fabricated.
FiftyPlusOne, a polling aggregator, had asked the firm about its data. Researchers who reviewed raw data from a Mississippi survey found irregular respondent IDs and timestamp patterns that did not match genuine records from Pollfish, the survey platform Wedgewood said it used. Wedgewood answered with a public statement calling the project an “experimental project to demonstrate the vulnerabilities, lack of oversight” in polling. It said it wanted to test how easily a fabricated firm could gain credibility. Its answer was “Far too easy.”
Wedgewood also said the project “was never intended to sway an election or permanently mislead the public.” Its motives remain a subject of speculation, and an email to the address listed on its X account bounced back.
Who Used the Numbers
The New York Times removed 15 Wedgewood polls from its tracking, and Decision Desk HQ removed 17. Decision Desk HQ said its Senate and House seat totals and control probabilities did not change. VoteHub and Silver Bulletin also dropped the polls. POLITICO said its own morning polling newsletter mentioned Wedgewood twice.
FiftyPlusOne, which had listed Wedgewood, said it asked who was behind the firm, was given two names and confirmed that one had prior polling experience. It said it had no reason at the time to doubt the firm. FiftyPlusOne also said Wedgewood had collected donor money for crowdfunded polls. Wedgewood said one crowdfunded Nebraska survey was genuine. That claim has not been verified.
What Fake Polls May Have Cost
A Wedgewood poll of Michigan’s Democratic Senate primary, taken June 25 to 28, put Abdul El-Sayed ahead of Haley Stevens, 46% to 39%, with Mallory McMorrow at 9% among 353 likely Democratic primary voters.
McMorrow suspended her campaign on July 5. She has said the Wedgewood poll “dried up” donations, left volunteers “devastated” and led to numerous calls to exit the race. “It killed us,” she wrote on X. Jennifer Holdsworth, a Democratic operative who ran a super PAC supporting McMorrow, said donors used the poll to pull back funding while the group was being outspent by $80 million. Coverage of her exit noted that she had lagged in recent polling and had not built a strong voter base, so the poll’s role cannot be measured. El-Sayed’s campaign cited the Wedgewood polls during the primary.
In Mississippi, Wedgewood posted a poll showing the Senate race within reach, shortly before Democrats put money into the state. Democratic candidate Scott Colom sent the poll to his press list as proof of “massive momentum.” Election watchers argued that outside groups were using such data to decide where to spend. A spokesperson for Senate Majority PAC, the main Democratic Senate super PAC, answered on X: “Absolutely not. We don’t make spending or strategic decisions based on external polling.”
Wedgewood’s numbers reached Texas as well. A July survey showed Democrat James Talarico leading Republican Ken Paxton among Hispanic voters, 64% to 30%. The polls in WECU’s Texas Senate report come from established pollsters, and none of them is among the firms named here.
A Mayor’s Poll and a 21-Year-Old
In August, a site called Median Strategies posted a poll claiming Los Angeles Mayor Karen Bass led challenger Nithya Raman by nearly 12 points ahead of the Nov. 3 runoff. It claimed to have surveyed 560 voters from July 30 to Aug. 5. Bass’s campaign posted the result on X and later deleted the post. A campaign spokesperson said leadership judged the poll legitimate because “the poll was reported on by multiple news outlets.”
After questions from the Los Angeles Times, the firm said on Aug. 17 that the poll was not real. It called the effort a “short-term social experiment” about how unverified polling spreads. Median had also posted fake polls in Wisconsin and Nevada. The Wisconsin poll showed Francesca Hong ahead by more than 20 points in the Democratic primary for governor. Hong lost. Several credible polling organizations did not aggregate Median’s numbers because they could not verify its methods.
The Guardian identified the man behind the site as Rahil Prakash, a 21-year-old recent college graduate, using domain registration records. “I wanted to see if fake polls could really penetrate the ecosystem that easily,” Prakash told the paper. “And as it turned out, it could.” He said he built the site with artificial intelligence, acted alone and apologized to the campaigns that were affected. He said he was inspired by a polling miss in Michigan’s Senate primary.
The American Association for Public Opinion Research called the fabrication “a serious breach of ethics.”
The American Association of Political Consultants said Median is not one of its members and that publishing fabricated survey data as independent research is a deception aimed at voters, the press and campaigns. “Professional polling is accountable polling,” said AAPC President Kyle Roberts. “Researchers who stand behind their numbers put their names on them, disclose how the work was done, and accept that their record will be judged against results.”
A Third Firm Goes Dark
SoCal Strategies, which published polls on competitive races, went offline on Oct. 22 after FiftyPlusOne raised questions. Its website, Substack and X account disappeared. FiftyPlusOne said it had received credible claims of wrongdoing from multiple sources but has not proven fabrication. It removed two SoCal polls whose authenticity was disputed and will stop collecting the firm’s surveys.
FiftyPlusOne described SoCal as an associated pollster of Wedgewood and said it could not verify the firm’s legal information or the people involved. Wilson Rochford, SoCal’s lead pollster, denied data problems in a call with FiftyPlusOne, according to the report. FiftyPlusOne also heard allegations that he was not involved at all.
How to Read the Next Poll
• Find out who is behind it. A real firm names its people and explains how it reached voters.
• Look for the basics. The political consultants’ association says a public poll should disclose who conducted it, who paid for it, how respondents were reached, when it was in the field, the sample size and how the data were weighted.
• Be wary of one surprising number. Decision Desk HQ said removing 17 Wedgewood polls did not change its seat totals or control probabilities, which shows why an average of many polls is a better guide than any single survey.
• Ask questions. A pollster that will not answer is a warning sign. Mark Baldassare of the Public Policy Institute of California put the advice in three words: “Call the pollster.”
The Bottom Line
Wedgewood said aggregators failed to vet it. FiftyPlusOne calls that false and says it asked who was behind the firm and saw no reason to doubt it. Either way, the cases point to the same gap. A poll that looks professional can reach campaigns, donors and headlines before anyone sees the raw data.
Until verification catches up, treat a single poll as a clue and an average of many as the better guide.
Sources:
• POLITICO, “How a fake pollster may have moved some 2026 elections” (Oct. 22, 2026)
• FiftyPlusOne, “FiftyPlusOne just caught a pollster” (Oct. 21, 2026)
• FiftyPlusOne, “More pollster funny business uncovered” on SoCal Strategies (Oct. 22, 2026)
• Lynnwood Times, “FiftyPlusOne exposes fabricated poll data, Wedgwood says it was an ‘experimental project’” (Oct. 21, 2026)
• KMPH, “Fake polling firm tricked media, claims fabricated surveys were for ‘social experiment’” (Oct. 2026)
• Decision Desk HQ, Wedgewood Polls Michigan poll page (June 29, 2026)
• Bridge Michigan, “Mallory McMorrow ends Senate campaign” (July 5, 2026)
• ABC7 Los Angeles, “Widely shared poll results for LA mayoral race were fake, ‘polling firm’ says” (Aug. 18, 2026)
• The Guardian, “Website behind widely shared fake election polls run by 21-year-old” (Aug. 20, 2026)
• CalMatters, coverage of the Median Strategies fake poll (Aug. 2026)
• American Association of Political Consultants, “Fabricated Polling Undermines Voters, Press and Campaigns” (Aug. 19, 2026)
• WLOS, “Polling firm disappears amid media questioning surveys, employees” (Oct. 22, 2026)
• WECU News, “Could Texas Elect a Democratic Senator for the First Time Since 1988?” (Oct. 26, 2026)
Have a correction or tip? See our
corrections policy or contact the newsroom.
