Why a fake poll got into mainstream poll trackers, how it was discovered and the common thread behind both fake poll scandals this year.
On Tuesday, a group calling itself Wedgwood Polls said in a since-deleted social media post that it published fabricated polls in midterm election races, claiming that it was an “experiment” to test whether media outlets and aggregators would report on the data. That announcement came after Wedgewood was asked to comment on the findings of an investigation into its work by FiftyPlusOne, a website founded by former staff of the data-journalism site FiveThirtyEight.
It feels like déjà vu from just two months ago, when a group calling itself Median Strategies admitted to publishing fabricated results, also calling it an experiment testing the media’s gullibility. But while most media and poll trackers shunned Median Strategies, more than a dozen polls published by Wedgewood were listed by poll aggregators that vet the legitimacy of polls, including the New York Times and FiftyPlusOne.
Why some aggregators listed Wedgewood’s polls
In short, it’s because of a lie.
The Times published an editor’s note Wednesday saying Wedgewood “answered a series of questions regarding the identities of the pollsters; the methods they were using to draw their sample, contact respondents and weight their polls; and the source of their funding.”
FiftyPlusOne described reaching out to Wedgewood to ask who they were and their background in polling.
FiftyPlusOne article by Mary Radcliffe, G. Elliott Morris and Cooper Burton
“We were told the organization was a bipartisan team, including Republican Jack Sims and Democrat Andrew Prater, and were provided information about their funding sources and methodology. Sims had previously worked for other pollsters that met our standards, which we independently confirmed, and the methodology statement on their website appeared to be industry standard.”
RealClearPolitics said in an email Wednesday to The Washington Post that its poll-tracking pages did not include Wedgewood polls. RCP’s director of polling aggregation on why it avoided the group:
Kavon Nikrad, RealClearPolitics director of polling aggregation
“They were a new pollster with no track record to evaluate, had a ‘placeholder-style’ Google Sites website, had no staff directory on that website, and no social media history to look back on to establish that they were an actual, legitimate polling firm. Meeting these criteria does not mean automatic inclusion in RCP’s Avgs, but they are a start.”
How they were caught
The staff at FiftyPlusOne was suspicious about a Mississippi poll Wedgewood released showing Republican Sen. Cindy Hyde-Smith leading by only four points just hours after a prominent election handicapper changed its rating of the race from “Safe Republican” to “Likely Republican.”
FiftyPlusOne requested the raw polling data, which Wedgewood provided, and noticed the data was “surprisingly clean” for a survey using an opt-in online sample, closely mirroring the population by race and other groups despite Wedgewood saying they didn’t use quotas.
Suspicious about the sample, FiftyPlusOne paid $900 to commission a poll of Mississippians using the same questionnaire and online sample provider, Pollfish, to check if they got similar results. The raw sample was far less representative than the Wedgewood data, with Black and White respondents each making up 48 percent of the sample, compared with 67 percent of Wedgewood’s respondents who were White.
But the poll trackers found the smoking gun in two forgettable columns in the data called “respondent id” and “trans id,” with the latter tied to the order of completion. The newly commissioned poll showed respondent IDs ranging from nine to 10 digits, some including dashes.
Wedgewood’s data only showed respondent IDs with nine digits, none with dashes. The FiftyPlusOne survey found the “trans id” correlated strongly with the time stamp for when a respondent completed the poll — as they expected — but in Wedgewood’s file the trans id had no connection to the time of completion.
Bottom line: The fact that metadata in the poll FiftyPlusOne purchased from Pollfish differed systematically from the Wedgewood data, supposedly also from Pollfish, suggested the latter may be fabricated.
FiftyPlusOne asked Wedgewood about these discrepancies on Oct. 19, and one day later, the group fessed up on X.
What can poll trackers do differently?
Wedgewood ran a much more elaborate scheme than Median Strategies, which fessed up when a Los Angeles Times reporter asked to speak with their pollster. How can poll aggregators weed out groups that provide reasonable answers to questions about methodology and even provide raw datasets?
In their article, FiftyPlusOne said that AI-powered tools “have made it easier for those willing to act in bad faith to produce potentially bogus documentation of results that match the quality of those produced by legitimate research firms.” As a result, they intend to screen pollsters more vigorously, “including meeting them in real life or via a video call, and potentially reviewing firms’ microdata — to hopefully avoid problems like this in the future.”
The New York Times cited its editor’s note, but did not directly respond to The Post’s question about what changes it might make to poll-vetting procedures.
Anonymity is a common thread in 2026 polling fraud
Wedgewood made elaborate efforts to persuade the New York Times and FiftyPlusOne they were legit, providing answers to detailed methodology questions about how their surveys were conducted and claiming to work with specific professional pollsters. But the humans behind Wedgewood — and Median Strategies before them — have faced no public accountability for publishing fraudulent data.
Last week’s Pollwatcher raised concern about Wedgewood and other pollsters that don’t name staff on their websites, which together account for 14 percent of polls tracked by the New York Times that measure national support for Congress or support for Senate in competitive states.
There’s one positive update on this front: One firm that didn’t list staff on its site last week, Quantus Insights, now features names and photos of five staff members. Co-founder Jason B. Corley emailed The Post, identifying himself and noting that his name appears at the end of the firm’s polling slide decks and his social media accounts identify him as the firm’s lead pollster.
That leaves just one entity producing election polls in competitive races that does not name pollsters on their websites: SoCal Strategies. Wedgewood’s representatives claimed to previously work at SoCal Strategies, according to Zachary Donnini from the poll aggregator VoteHub. SoCal did not reply to messages sent Wednesday to an email address on its website or direct messages to its X account. On Thursday morning, their website, Substack and X account disappeared.
Worth your attention
Democrats lose faith in mail voting — While the Supreme Court blocked a U.S. Postal Service plan to screen ballots mailed to voters this year, the news may have sapped enthusiasm for mail voting. A Pew Research Center poll finds a 21-point drop in Democratic confidence that absentee or mail ballots will be counted as voters intend, from 85 percent in 2024 to 64 percent this year. Republicans remain far less confident in mail voting at 42 percent, while about 8 in 10 voters in both parties are confident that in-person ballots will be counted as voters intended.
Arab Americans shift left — Arab American registered voters favor Democrats by 58 percent to 34 percent over Republicans in support for Congress, according to a Washington Post-Schar School poll conducted in August and September. The poll finds that 41 percent of Arab American voters report that they voted for Kamala Harris in 2024, while 39 percent backed Donald Trump, in line with previous estimates.
Americans spooked by the midterms — “Scared” and “apprehensive” are the most common words Americans used to describe how they feel about November’s midterm elections, named by 29 percent of likely voters in a New York Times-Siena College national poll released Wednesday. Democrats and Republicans were about as likely to say they felt scared; Democrats were more likely to feel happy or excited, while Republicans were more apt to feel angry or disappointed.
Want this in your inbox?
Sign up for The Pollwatcher newsletter here.
What do you want to know?
Send your questions, comments and ideas to [email protected].
The post The Pollwatcher: Another fraudulent pollster is out of business, and a third group just disappeared appeared first on Washington Post.
