The arguments of the pre-campaign of (-RJ) to request the suspension of the TSE (Superior Electoral Court) are considered fragile by two experts consulted by the Sheet.
Although they see no signs of manipulation or induction of the main results, they make some technical reservations regarding the survey.
The survey shows a drop of six percentage points for Flávio in voting intentions in a second round scenario against the president (PT). With this, the PT member would beat the son of () by a score of 48.9% to 41.8%.
According to Flávio’s pre-campaign, “the research reveals a serious manipulative precedent and failed to observe the neutrality expected in electoral surveys intended for dissemination.”
She argues that the survey not only measured voters’ opinion, but presented “stimuli capable of influencing the respondent’s perception before questions about image, rejection and electoral viability”.
A Sheet interviewed Antonio Lavareda, political scientist and honorary president of Abrapel (Brazilian Association of Electoral Researchers), and Raphael Nishimura, statistician at the University of Michigan, about the PL’s questions.
The only aspect on which experts agree with the party refers to the fact that questions about rejection and image perception are arranged after a block of questions about Banco Master and .
Nishimura also takes issue with the format of the online questionnaire on a single scrollable page, with the ability for a respondent to change their answers after reading a question presented later.
The Atlas institute defends the criteria adopted, including the order of the questions in the questionnaire.
Voters who collaborated in the online survey were asked 48 questions (voting intention is among the first). 5,032 voters were interviewed, from the 13th, when Flávio’s dialogues with Vorcaro were held, to the 18th. Flávio’s rejection measured by the institute in April was 49.8% and now it was 52%. The margin of error is 1 percentage point.
For Lavareda, the question about rejection could not be contaminated by the insertion of “any other question that could lead some respondents in another direction”. “That’s a mistake,” he says.
Lavareda adds, however, that he does not see technical consistency in the PL’s other arguments. For him, even if one or another question could have been better formulated, there are no serious problems with them.
Regarding the Atlas/Bloomberg survey questions on voting intentions, Lavareda considers there to be no impact, as they were some of the first in the survey.
Although this question does not appear in the PL representation, Nishimura criticizes the fact that the questions are on a scrollable page, with “potential for the respondent to change their answers”.
But, for him, although the wording of some questions could be improved, in the points raised by the PL, there are no signs of infringement of technical rules.
Both Lavareda and Nishimura see no problem, for example, in comparative questions between Lula and Flávio.
In a note to Sheetthe Atlas institute stated that it is “common for topics of high public salience to be inserted before political assessment blocks to capture the effect they have.”
“Asking about prior knowledge of a notorious case before measuring an image is not equivalent to induction, as long as there is no presentation of persuasive or argumentative content – which there really wasn’t in the research.”
Furthermore, the institute states that it carries out frequent tests “to ensure the integrity of the sample”. He adds that including multiple questions on the same screen is “a very common practice to reduce abandonment, reduce interview time, and improve the respondent’s experience.”
“We see that an eventual return and change of previous responses is a rare behavior, and it is a controlled phenomenon”, he says.
The only question, among the 48 in the questionnaire, asked on a separate page (from which it was no longer possible to change previous answers) was the one that generated the most repercussion on social media. In it, the interviewees were presented with the audio sent by Flávio to Vorcaro about the financing of the film “Dark Horse”.
In the piece sent to the TSE, Flávio’s campaign does not oppose, in this case, the order of the question, but contests that the audio and its transcription were not attached at the time the institute registered the research with the Electoral Court. Flávio’s campaign argues that there is “no chain of custody and no demonstration of authenticity of the audio broadcast”.
HAS Sheet Atlas states that, in the TSE system in which the questionnaire is registered, there is no way to attach audio or video files and that the audio was public knowledge and notorious and was presented in full. “The published results also prove this”, says the note.
After the audio was broadcast by Intercept, Flávio did not question its veracity and .
