A polling firm admitted it fabricated election data for two major U.S. races before shutting down its operations this week [1, 2].
The incident highlights the vulnerability of election cycles to misinformation and the ease with which fraudulent data can influence public perception of candidate viability.
Median Strategies released the fabricated results in early August [1, 2]. The fake data focused on the Democratic gubernatorial primary in Wisconsin and the mayoral race in Los Angeles, California [1, 2]. In the Wisconsin results, candidate Francesca Hong was shown leading by more than 20 points [2].
On Tuesday, the firm revealed the data was not based on actual surveys. A spokesperson for Median Strategies said the effort was a "short‑term social experiment" [1]. The firm said the goal was to test the spread of misinformation within the political ecosystem.
"We wanted to show how easily false results can be spread," the spokesperson said [2].
Following the admission, the company ceased all operations [3]. The firm did not provide details on who funded the experiment or how many people were surveyed to create the illusion of a legitimate poll. The fabrication targeted high-profile races, including that of Karen Bass in Los Angeles [1, 2].
The revelation comes as election officials and media outlets struggle to vet the surge of third-party data providers appearing during primary seasons. By creating a plausible but false lead for Hong, the firm demonstrated how a single data point can be amplified across social media and news aggregators before being verified.
“"short‑term social experiment"”
This event underscores a growing systemic risk where the appearance of professional polling can be weaponized to create artificial momentum for candidates. Because modern campaigns rely heavily on 'polling data' to secure donors and media attention, the ability of a defunct firm to manipulate these metrics suggests a need for stricter transparency and verification standards for polling methodology in the U.S.
