Overview
- TPSI publicly acknowledged late Thursday that analysts identified a problem in a text-message survey panel and moved some respondents who had picked Jay Collins toward James Fishback, and it failed to disclose that Fishback’s campaign had paid for some polling.
- FiftyPlusOne said it will permanently remove TPSI’s polls from its analytics after the company admitted it altered vote-choice data, and other aggregators and analysts have already excluded TPSI from averages.
- TPSI denied it fabricated raw responses and said it adjusted its final forecast using an experimental method called the “Coalition Expectation Overlay,” while also admitting it omitted its funding disclosure in some releases.
- The altered polls overstates Fishback’s support but did not change the result: Fishback finished third in the Florida Republican primary behind Jay Collins and winner Byron Donalds.
- Experts say the episode, following a separate fake-poll 'social experiment' by Median Strategies, highlights specific risks in modern polling—especially reliance on text-message panels and undisclosed sponsorships—and could push stricter vetting by monitors and reduced public trust in polls.