Fake Polls Fooled Big Media!

Businessperson holding a tablet displaying the word VOTE — Photo: Michal Chmurski / Shutterstock

A polling firm openly confessed it faked “polls,” and major outlets still ran the numbers anyway.

Story Snapshot

Median Strategies admitted its posted “polls” were fabricated and withdrew them.
Fake toplines copied real poll features like sample size and margin of error.
One bogus claim had Francesca Hong up 20+ points; she later lost by under one point.
Wedgewood Polls also said most of its releases had no real data and still got aggregated.

What actually happened, not the spin

Median Strategies said its published results were fabricated and told readers not to treat them as real polls. The admission came with a shutdown and a claim that this was a “short-term social experiment” about how unverified polling spreads. The firm’s Wisconsin and Los Angeles releases looked like standard polls. They listed field dates, a sample of 560, and a 4.1-point margin of error, but the data was not genuine. That is not a gray area; it is a direct confession.

The fake numbers had real reach before the correction. The New York Times reported that one release said Francesca Hong led her primary by over 20 points. She lost by less than one point, which shows how far off the narrative was from reality. That should bother every voter who relies on polls to plan donations, turnout, or attention. Fabricated data can warp the scoreboard, and the scoreboard changes behavior.

How a “poll-shaped object” fooled the system

Median’s posts wore all the right clothes. They used the language of polling and the visual cues people trust. That made them easy to share and hard to challenge on a quick read. Wedgewood Polls went further. It said most of its releases had no history, no credentials, and no real data. Still, it claimed its work landed in big-name aggregators and got treated as legitimate. Columbia University’s statistics blog described how Wedgewood sent purported raw data to an aggregator on October 19, 2026.

These cases prove two narrow truths. First, at least two outfits faked polls. Second, aggregators and media can amplify fakes before vetting them. They do not prove the whole industry is rotten. They do show a gatekeeping problem that rewards speed over proof. The American Association for Public Opinion Research has long called for basic disclosure: who fielded, how sampled, what questions asked. If that had been required and enforced, the fakes would have struggled to pass first contact.

What motives explain this, and which ones do not

Both groups framed their work as a “social experiment.” Median Strategies denied making or trying to make money on betting markets tied to their posts. The statements still leave open questions most readers would ask: who paid, who built the sites, and why hide identities if the goal was education? Until investigators answer those questions, the most solid facts remain the admissions and withdrawals, not the motives behind them.

From a conservative common-sense view, intent matters less than impact. If you publish numbers you know are not real, and they shape public talk or campaign choices, you crossed a line. Calling it a lesson about media literacy does not undo the damage. It also does not excuse aggregators or reporters who ran with attractive numbers without the basic checks that any newsroom should demand.

Fixing the scoreboard before the next game

Voters need three practical changes. First, treat any poll without full methodology as a rumor, not a number. If there is no questionnaire, sponsor, sample source, weighting notes, and contact method, it is not ready for prime time. Second, push aggregators to post audit trails. If they ingest, they should log when, why, and how they verified. Third, demand raw data or respondent-level summaries from new or unknown firms, even if delayed and anonymized.

Media outlets need a two-switch safety system. One, a preflight check that blocks publication of toplines without required disclosures. Two, a red-box correction protocol that flags removals as loudly as the original headlines. Aggregators should rate pollsters on transparency and reproducibility, not just volume. A small number of fakes can still bend averages if they slip in near an election window.

The bottom line for readers who do not have time for games

These episodes show that fake polls can look real, spread fast, and take too long to die. They do not prove that all polling is fake, and saying so would be lazy. They do prove that trust must be earned, not assumed. If you would not buy a used car without a title, do not buy a poll without its paperwork. Until the industry locks the doors, keep your hand on your wallet and your eye on the methods page.