Datafolha will release voting propensity analysis this Saturday

The Datafolha Institute will release this Saturday (24) a voting propensity model for the second round of the presidential election. The initiative aims to improve the x-ray of the electoral scenario, measuring not only the voter's declared preference, but their probability of going to the polls on Sunday.

Follow the results of the 2nd round on Folha in real time

The analytical model is built based on two fundamental variables: the proximity between the voter's home and their voting location, and the turnout history in the first round. The inclusion of these factors seeks to improve the reading of surveys, which capture the voter's intention, but not necessarily their ability or logistical willingness to vote.

"Considering that the dispute is very close, voter abstention will be a determining factor in the result. Therefore, Datafolha will present, in addition to the complete survey, the propensity to vote, taking into account factors that have a high correlation with the voter's commitment to going out to vote", says Luciana Chong, director of Datafolha.

According to the institute's most recent survey, released on Thursday (22), candidates Flávio Bolsonaro (PL) and Luiz Inácio Lula da Silva (PT) appear tied with 50% of valid votes each — the margin of error is two percentage points, plus or minus.

By crossing logistical convenience — such as ease or cost of travel — with recent voting habits, the methodology is able to adjust declared intentions and project a result closer to the reality at the polls.

The survey released on Thursday by Datafolha showed, for example, that high abstention and a voting location far from home tend to be more unfavorable to Lula.

The survey to be released this Saturday will be carried out on October 23 and 24, with 4,050 respondents and a margin of error of two percentage points. It will reflect voting intentions in the second round of the presidential election and for the government of the Federal District. It is registered with the Electoral Court under code BR-09540/2026.