The polling firm The Public Sentiment Institute (TPSI) admitted to altering voter data in a survey for the Florida Republican gubernatorial primary to show higher support for a specific candidate. The admission followed a report by the polling monitor FiftyPlusOne, which stated on Thursday that TPSI had failed to disclose that the candidate, James Fishback, had commissioned the poll.
The incident occurred during the 2026 Florida primary cycle, where Rep. Byron Donalds (R-FL) ultimately won the Republican nomination for governor. Fishback finished third in that contest, trailing both Donalds and Jay Collins. FiftyPlusOne, an organization that aggregates and analyzes polling data, announced it would no longer include TPSI surveys in its future analytics due to the violation of industry standards.
In a post-election summary, TPSI acknowledged that it had identified respondents who originally selected Jay Collins but whose profiles suggested they might belong to a different voting group. The firm then reassigned those responses to James Fishback. TPSI stated this was done because Collins' support level in their raw data did not align with their internal read of the race, leading them to believe there was a data-quality issue with their text-message panel.
The scale of the issue extends beyond Florida, as this is the second instance of fabricated polling data reported in the current cycle. In a separate case, a group called Median Strategies admitted that its surveys in the Los Angeles mayoral race and the Wisconsin gubernatorial primary were part of a "social experiment" using fake data. These polls reached high-level campaigns, including that of Los Angeles Mayor Karen Bass, and affected perceptions of candidates like Francesca Hong in Wisconsin. The cumulative effect of these incidents, combined with significant errors in the 2024 presidential election and the 2026 Michigan Democratic Senate primary, may lead to increased skepticism among the public regarding the reliability of political forecasts.
For individual citizens, these developments mean that the information they see in news reports or on social media regarding election leads may not be based on actual voter sentiment. The knock-on effect includes polling aggregators like FiftyPlusOne tightening their verification rules, which could lead to fewer polls being considered legitimate by mainstream analysts. As of August 20, TPSI has been removed from major aggregation platforms, and Median Strategies has shut down. Future primary and general election cycles will likely face heightened scrutiny from independent monitors to prevent undisclosed campaign-backed or fabricated data from entering the political information ecosystem.