Fake Polls Flooded Media – Right Before Midterms

Two weeks before the midterms, a polling firm admitted most of its election numbers were made up, and major outlets had already shared them.

Story Highlights

Wedgewood said the “vast majority” of its polls had no real surveys behind them.
Researchers uncovered mismatched data markers and confronted Wedgewood before the admission.
Major aggregators and outlets included Wedgewood’s figures, then removed or corrected them.
The case shows how easy it is to inject fake data into polling databases.

Public Admission Of Fabricated Polling

On October 20, Wedgewood said most of its released polls were not based on real surveys, calling the project a test of polling safeguards. The group claimed one Nebraska Senate poll was genuine but did not detail the rest. PBS reported the “vast majority” of data lacked survey support, and that description guided later corrections and removals by media and aggregators. The admission came late in the campaign season, when polling often shapes coverage and expectations for close races.

Wedgewood framed the effort as an “experimental project” that showed the low barrier to entry in modern polling. Axios reported that statement and described how the firm’s numbers entered well known datasets despite little public history. That mix of a confident message and thin documentation let the operation look credible for a time. The case highlights a basic weakness: systems that rely on self-reported methods without outside verification can be gamed by bad actors.

How Investigators Detected Problems

FiftyPlusOne researchers Mary Radcliffe, G. Elliott Morris, and Cooper Burton reviewed respondent-level files for a Wedgewood Mississippi Senate poll. They found mismatched electronic markers and patterns that did not align with normal survey records. They then checked those details with the vendor named in the files and brought the findings to Wedgewood. After that confrontation, Wedgewood publicly acknowledged the fabrication on October 20 in a statement shared with reporters and readers.

This sequence matters because it shows the admission followed concrete data checks, not rumors. The researchers documented where the files failed standard tests and where identifiers did not line up with platform records. Their post laid out each step in plain terms for readers. That approach gave editors and aggregators a clear reason to remove the data. It also gave professional pollsters a clean example to use in training screens for future submissions.

Media Use And Rapid Corrections

PBS reported that both The New York Times and FiftyPlusOne had included Wedgewood in polling databases, then removed it after the admission. Other outlets had cited Wedgewood figures, which led to corrections and retractions. Decision Desk HQ removed 17 Wedgewood polls from its averages and lists. Newsweek took down three articles and corrected others that had relied on the numbers, according to follow-up reports on the fallout.

Wedgewood Polls Revealed To Be Fabrication In Terrible Drop For Steyer pic.twitter.com/SPQLKvKklg

— The Pigeon Press (@PigeonThePress) October 21, 2026

These steps trimmed the visible footprint, but removal cannot erase what readers already saw. Aggregated polling can influence story choice, race ratings, and donor mood. Axios noted that the scandal shows how easily a new brand can appear, publish slick data, and get picked up by trusted platforms. That dynamic feeds wider anger about elites and institutions on both left and right. People see systems that move fast for headlines but slow for checks.

Why This Resonates Across Politics

Americans of many views already think powerful insiders bend the rules. A case where fake numbers flowed through respected channels confirms that fear. The episode does not prove that most polling is fake, and nothing here shows a clear partisan tilt driving the fraud. It does show that vetting can miss a polished fake, and that editors can repeat bad data under deadline pressure. PBS and Axios traced this chain clearly and tied it to recent similar cases.

What Comes Next For Polling Gatekeepers

Editors and aggregators now face a choice. They can add slower, outside checks on new firms, or keep relying on self-reported methods and quick questionnaires. Stronger intake rules could include verified invoices from fieldwork vendors, time-stamped raw files, and named staff with reachable references. FiftyPlusOne’s step-by-step exposure offers a simple screen: ask for respondent-level data and look for digital seams before posting top-line numbers.

Caveats On Scope And Impact

Wedgewood claimed one Nebraska Senate poll was real, but a full poll-by-poll audit is not yet public. No analysis has shown direct voter behavior changes from these releases. That said, corrections landed fast, and major platforms pruned the data. The clearer lesson is structural. One fake operation exploited light screening and got reach. Restoring trust will require proof of method, not just polished websites and confident press notes.