TSE postpones decision on censoring poll showing Flávio Bolsonaro's decline

Brazil's Superior Electoral Court (TSE) has delayed its ruling on whether to censor a controversial opinion poll. The poll, conducted by Atlas/Bloomberg, reportedly showed a decline in support for Senator Flávio Bolsonaro, prompting his team to request its suspension.

The case reached the TSE after Justice Kassio Nunes Marques initially voted to confirm the censorship. However, Minister Estela Aranha subsequently requested additional time for a thorough review of the matter, leading to the postponement of the decision. The poll in question was released on May 19 and surveyed 5,032 individuals using the Atlas RDR digital recruitment method between May 13 and May 18.

Senator Bolsonaro's legal team argued that the poll's methodology, including the "use of associations" between Flávio Bolsonaro and Daniel Vorcaro, the owner of Banco Master, unfairly influenced respondents. They claimed that the "disposition of questions and themes" within the survey "contaminate and induce the answers of the interviewees."

AtlasIntel, the firm behind the poll, defended its scientific rigor. The company stated that the vote intention data was collected without reproducing audio during the questionnaire's application. According to AtlasIntel, the material was presented to users in a later stage, preventing them from returning to questions or altering recorded answers. One of the lawyers representing Bolsonaro's team is Maria Claudia Bucchianeri, a former TSE minister.

Court documents reveal that the poll included a video with audio of a conversation between Flávio Bolsonaro and Daniel Vorcaro, presented as the final item after 48 questions. Interviewees were asked to evaluate the content by swiping left for a negative assessment and right for a positive one. Justice Marques noted that these circumstances lend plausibility to the argument that the poll might have "exceeded the limits of regular statistical measurement" by potentially using "inductive stimuli capable of contaminating subsequent responses."

“

possible utilização de estímulos indutivos aptos a contaminar as respostas subsequentes relativas à imagem, rejeição e intenção de voto, reforçando a plausibilidade jurídica da tese de que a pesquisa possa ter extrapolado os limites da regular aferição estatística