Bogus Numbers Warped Races – Industry Rattled

Voters standing in line at a polling place entrance — Photo: Andrey_Popov / Shutterstock

Two polling outfits admitted they published fake “polls” that fooled media and aggregators, exposing a deep verification failure.

Story Snapshot

Median Strategies confessed its “polls” were fabricated and withdrew them as non-genuine data.
Wedgewood Polls said most of its releases had no real data and still reached major aggregators.
A bogus Los Angeles poll showed field dates, sample size, and a margin of error, mimicking real surveys.
Fake numbers shaped race stories, including a Wisconsin primary that flipped on Election Day.

Admissions of Fabrication by Median Strategies

Median Strategies posted a public statement in August 2026 saying its releases were fabricated and should not be treated as real polling. The firm withdrew its posts and framed the operation as a short-term “social experiment.” The New York Times reported the admission and noted the removal of those materials from the firm’s website. These facts confirm the core point: numbers that looked like professional polling were not genuine, even as they circulated widely first.

Associated reporting described how one phony release used all the trappings of standard survey work. A Los Angeles result cited a sample of 560 voters, a field window of July 30 to August 5, and a 4.1-point margin of error. Those details gave it a veneer of authority, despite later confirmation that the numbers were bogus. That shows why many readers, reporters, and even campaigns might have taken the release at face value before the retraction.

Wedgewood Polls and the Aggregator Problem

A second group, Wedgewood Polls, said most of its releases lacked real data or formal credentials. It claimed the project was meant to test how weak oversight lets fake polls pass as real. Reporting says Wedgewood’s outputs were added to major polling aggregators and treated as legitimate before deeper checks occurred. This admission broadens the concern beyond one firm and highlights a process failure in media and data pipelines.

Columbia University’s Statistical Modeling blog described how Wedgewood shared allegedly fraudulent raw data with an aggregator on October 19, 2026. That account underscores how fake or synthetic numbers can enter trusted databases with little friction, then ripple through coverage and campaign chatter. While the record does not quantify voter impact, the pathway for false data to spread is now documented in multiple named reports.

Real-World Effects on Race Narratives

The fake outputs did not just sit online. The New York Times reported Median Strategies had a fabricated claim showing Francesca Hong leading Wisconsin’s Democratic gubernatorial primary by more than twenty points. She later lost by less than a point. That swing shows how bogus numbers can warp expectations, drive storylines, and leave voters confused when reality collides with hype. This is exactly how trust erodes—people see headlines that later crumble.

The episode also reached campaigns. Reports say a Los Angeles result tied to Median Strategies drew amplification before being deleted. When public figures and outlets share unvetted “polls,” it lends them weight they do not deserve. That is why verification before publication matters. Once false numbers hit social feeds and aggregators, corrections rarely travel as far or as fast as the original claim.

What This Means for Readers and the Industry

These cases prove some polls were faked; they do not prove the entire industry is fake. The available reports focus on two entities—Median Strategies and Wedgewood Polls. Broader claims would require evidence about other firms, methods, and audits that we do not yet have. Still, the confirmed pathway for fabricated data into respected outlets and averages is a serious warning for voters, reporters, and campaigns alike.

Common-sense steps can help. Demand basic facts with every poll: sponsor, full questionnaire, sampling source, respondent counts, weighting rules, and contact method. Look for independent replication or at least consistent results across multiple credible firms. Treat single, flashy outliers with caution, especially if they shape a favored narrative. Good polling exists. But as these admissions show, gatekeepers must prove it with transparency before they ask the public to believe it.