The organization behind those numbers has now admitted that most of its election surveys were fabricated.
Wedgewood Polls acknowledged Wednesday that it had manufactured polling results as part of what it described as an experiment designed to expose serious weaknesses in America’s political polling industry.
The admission raises troubling questions about how easily questionable election data can enter the national conversation, influence campaign narratives, and potentially shape political spending.
And in Mississippi, where Democrats are hoping to pull off an upset, the timing could hardly have been more significant.
A Suspicious Poll Suddenly Made Mississippi Look Competitive
On October 14, Wedgewood published numbers suggesting Hyde-Smith held a narrow 52% to 48% advantage over Colom.
That four-point difference painted a dramatically different picture of a state that has consistently favored Republicans in statewide federal elections.
For Democrats hoping to expand the Senate battlefield, such numbers offered an encouraging narrative.
For Republicans, they suggested a race that might require more attention and resources than previously anticipated.
But the apparent breakthrough came from an organization that has now acknowledged manufacturing much of its supposed research.
Wedgewood attempted to explain its actions in a public statement.
“Wedgewood Polling, an experimental project to demonstrate the vulnerabilities, lack of oversight, and overall low barrier to entry in the modern political polling industry, makes this public statement. Having previously worked in the polling and research sector, we watched as the public’s trust in polling eroded alongside a rise in unverified, low-quality methodology,” the organization said.
The group insisted that its objective was to demonstrate how easily questionable information could gain credibility in political circles.
“We wanted to test a simple hypothesis: How easy is it for an entirely fabricated polling firm to gain traction, be taken seriously, and expose how little scrutinization exists in media and public commentary?”
That explanation does little to resolve the larger issue.
A polling company was able to distribute invented election numbers while presenting them as legitimate research.
And the public had little reason to know the difference.
Independent Researchers Uncovered the Fabrication
The scheme began unraveling when researchers at FiftyPlusOne examined the underlying data behind Wedgewood’s Mississippi survey.
Rather than simply accepting the published results, investigators scrutinized the information supposedly collected from actual voters.
Wedgewood had identified Pollfish as the platform used to gather responses.
But when researchers compared the supplied records against authentic Pollfish information, serious discrepancies emerged.
Respondent identification numbers and transaction timestamps displayed irregularities that did not match genuine survey activity.
Those findings raised fundamental questions about whether the reported interviews had ever occurred.
FiftyPlusOne confronted Wedgewood on Tuesday.
By Wednesday, the organization had acknowledged that most of its polling data had been generated rather than collected from voters.
Wedgewood maintained that one survey involving Nebraska’s Senate contest had been conducted legitimately.
The rest of its operation, however, demonstrated how easily an organization could create the appearance of credible political research without actually performing the work.
The revelation also exposes an uncomfortable reality for media organizations and political commentators.
A professional-looking polling release can quickly become a headline, even when the underlying methodology has not been independently verified.
Democrats Were Preparing to Spend Big in Mississippi
The fabricated survey emerged as Democrats were already taking a closer look at Mississippi.
Senate Majority PAC, a major outside spending organization supporting Democratic Senate candidates, reportedly planned to invest $1.2 million in efforts benefiting Colom.
That decision came amid polling suggesting Hyde-Smith’s advantage might be smaller than expected.
Other surveys conducted by Data for Progress and Impact Research also showed the Republican incumbent leading by single-digit margins.
Those separate findings mean Wedgewood’s false numbers cannot automatically be blamed for Democrats’ spending plans.
Nevertheless, the episode illustrates why polling accuracy matters enormously during competitive election seasons.
Campaigns use surveys to determine where to invest advertising dollars, deploy staff, and concentrate fundraising efforts.
Political donors rely on polling to decide which candidates have realistic opportunities to win.
News organizations use the same information to determine which contests deserve national attention.
When fabricated numbers enter that system, the consequences can extend well beyond a misleading headline.
And while Wedgewood described its activities as an experiment, the organization has not established that its actions were harmless.
Another Polling Firm Admitted Making Up Election Numbers
Wedgewood was not the only organization facing questions about fabricated surveys.
Median Strategies also acknowledged producing false polling data.
One of its surveys suggested Wisconsin Democratic gubernatorial candidate Francesca Hong enjoyed a commanding lead of more than 20 percentage points over primary opponent David Crowley.
The actual August 11 primary produced a dramatically different outcome.
Hong lost by less than one percentage point.
That discrepancy highlights the danger of treating individual surveys as definitive evidence of electoral strength.
An invented poll can create the impression that a candidate is cruising toward victory when the actual electorate is almost evenly divided.
Such distortions can affect fundraising, media coverage, and public perceptions of campaign momentum.
They also risk undermining confidence in legitimate polling organizations that follow established research standards.
The admissions by Wedgewood and Median Strategies raise an obvious question: How many other questionable surveys have escaped meaningful scrutiny?
There is no evidence in the information disclosed that the two organizations coordinated their activities or that Democratic campaigns knowingly relied on fabricated data.
But their admissions make the need for independent verification difficult to ignore.
Who Holds Political Pollsters Accountable?
The most disturbing aspect of the Wedgewood controversy may be how little apparently stood between invented survey results and public acceptance.
The organization did not need to demonstrate that hundreds of voters had actually participated before distributing its findings.
Its numbers could be discussed alongside surveys produced by established research firms.
Only after outside investigators examined the raw records did the deception come to light.
Whether Wedgewood’s conduct violated any particular law would depend on facts not yet established, including how its surveys were represented, distributed, and used.
Fabricating a political poll is not automatically the same as committing a prosecutable election offense.
But the absence of an obvious criminal violation does not eliminate serious ethical concerns.
Americans deserve to know whether polling numbers presented as factual research are supported by actual interviews and verifiable methodology.
Republicans and Democrats alike have reason to demand that standard.
For Hyde-Smith and Colom, the Mississippi Senate contest will ultimately be decided by voters, not polling companies.
Yet the controversy has exposed a vulnerability in the political information system at precisely the moment when Americans are relying on election coverage to understand the races ahead.
Wedgewood set out to prove that fabricated polling could be taken seriously.
Its own admission suggests the experiment succeeded.
The unanswered question is why independent researchers had to uncover the deception before the public learned that much of the polling was never real in the first place.
