A major polling aggregation service announced Thursday that it is severing ties with a survey firm after uncovering evidence that the pollster allegedly manipulated data in its coverage of Florida's Republican gubernatorial primary.
FiftyPlusOne said it is “permanently banning” any future use of survey results from The Public Sentiment Institute (TPSI). The decision comes after an internal review found that TPSI failed to disclose that it was paid by the campaign of James Fishback, a candidate in that race, and reportedly altered vote-choice data for certain respondents.
“This fabrication appears to have boosted the poll numbers of the political candidate that paid TPSI for polling services,” FiftyPlusOne wrote in a statement on its website.
TPSI acknowledged the lack of transparency regarding its client but strongly denied any deliberate falsification. In a statement to The Hill, the firm said: “No respondent was removed, no response was altered, and no data was fabricated. When the raw data was reprocessed following the election, it reproduced the same measurements, and every candidate’s raw response share fell within the applicable margin of error of the certified results.”
However, in its own “accuracy review,” TPSI admitted that an “experimental layer” was applied to the survey after initial results “did not align with our read of the race.” The firm said it identified respondents who had selected Lt. Gov. Jay Collins but whose broader profile aligned with a competing coalition, and “leaned those respondents toward James Fishback.”
That adjustment moved Fishback into second place in the poll, behind Rep. Byron Donalds (R-Fla.), and dropped Collins to third. Donalds ultimately won the primary on Tuesday, defeating both Fishback and Collins.
TPSI said the experimental layer has been “discontinued and will not appear in any future TPSI release.” The firm added: “We accept responsibility for both the disclosure failure and the methodological error. We do not accept the characterization that either amounts to deliberate data falsification, and the underlying record does not support that conclusion.”
The controversy adds to a turbulent week for polling credibility, following a separate incident in which a firm called Median Strategies admitted to fabricating data in a Los Angeles mayoral race as part of a “short-term social experiment.” Austin Park, co-founder of Vote Hub, noted that his firm had already excluded TPSI from its averages for months. In a social media post, Park wrote: “The American Association for Public Opinion Research just released a statement on fabricated polling following the Median Strategies issue. Disappointing to see TPSI effectively fabricate their data as well.”
In addition to banning TPSI, FiftyPlusOne said it is temporarily suspending use of data from another firm, Patriot Polling, due to similar concerns about undisclosed Fishback campaign sponsorship. The organization stated: “We have discovered evidence that suggests both firms conducted surveys for political campaigns without disclosing their clients publicly.” All existing surveys from these pollsters will be labeled as partisan, as the aggregator says it cannot be confident they were conducted on a nonpartisan basis.
The Hill has reached out to Patriot Polling for comment. The developments come amid heightened scrutiny of polling practices, especially in high-stakes primaries like Florida's. For context, see Trump's reaction to Vindman's Florida Senate primary defeat and Nixon's Florida primary win energizes the left.
