The polling company Median Strategies admitted that its election surveys were fabricated as part of a ‘short-term social experiment’. The company said it wanted to see how polling claims could enter coverage and spread without independent verification.

One false survey showed Los Angeles Mayor Karen Bass leading challenger Nithya Raman by 12 points before their runoff. Median Strategies claimed it had surveyed 560 Los Angeles voters between July 30th and August 5th and reported a 4.1% margin of error, giving the numbers the appearance of a professional poll.

The firm also published a Wisconsin survey showing democratic socialist candidate Francesca Hong leading David Crowley by 23 points. Hong went on to lose the Democratic primary to Crowley by less than one point, exposing how far the invented numbers were from the actual result.

After reporters pressed the company for details, Median Strategies acknowledged that the surveys were not genuine. It withdrew its releases, stopped publishing polls, and warned journalists, campaigns, and voters not to cite any of its previous findings as real data.

Campaigns use surveys to decide where to spend money, which messages to emphasize, and what type of staff a race deserves. Donors and advocacy groups may also redirect funding toward a candidate who appears to be surging or abandon one who looks certain to lose. Under this pretense, polling, even fake polling, can have a heavy influence on campaigns and election results.

A fabricated lead creates momentum and attracts endorsements, while also discouraging supporters of the trailing candidate from donating, volunteering, and ultimately voting for a particular candidate because the contest appears settled. In a primary decided by less than one point, as Wisconsin’s was, even micro shifts in behavior can be significant.

Median Strategies made its polls appear legitimate by publishing a specific sample size and margin of error on a professional-looking website. Such a presentation was enough for the numbers to spread before anyone confirmed who conducted the surveys or whether the respondents and data even existed. As primaries are wrapping up and November midterms are around the corner, campaigns will now have to scrutinize unfamiliar polling firms before repeating their findings.

Treat any poll from an unfamiliar organization as unverified until its methodology can be examined. A release should identify who paid for the survey, when interviews occurred, how participants were contacted, and whether the sample reflects the electorate. A margin of error proves nothing because fabricated data can include technical details.

Before accepting a dramatic result, compare it with established pollsters' averages. One survey showing a candidate far ahead while every other poll shows a close contest deserves scrutiny. Search for the firm’s previous work, named leadership, and downloadable questions or cross-tabs. If none exists, the numbers should not guide your expectations.

In the digital age, misinformation spreads quickly, especially with the rise of AI. Verifying information before repeating or reposting it can help prevent fabricated polls like this, as well as other misinformation, from influencing what voters think and, ultimately, election outcomes to a degree.