Datafolha followed Friday's national numbers with five state surveys, and the biggest electorate in the country goes the other way. In São Paulo, Senator Flávio Bolsonaro leads President Lula 47 to 42 in a second-round matchup, outside a margin of two points. In Minas Gerais the order reverses — Lula 46, Flávio 42 — but the margin there is three points, so Folha calls it a technical tie.
The other three are further apart. Rio de Janeiro has Flávio on 49 against Lula's 40. The Federal District has him on 51 to 39. Pernambuco runs the opposite way and by more than any of them: Lula 62, Flávio 29. All three carry a three-point margin.
For scale, Folha sets the 2022 results beside these. Lula lost São Paulo that year with 44.8 percent of valid votes to Jair Bolsonaro's 55.2, a gap of 2.7 million votes, and won Minas Gerais 50.9 to 49.1, a gap of 2.1 million. Those are shares of valid votes at an election; the numbers above are voting intentions with blanks and undecideds still in them, so the two do not sit on the same scale.
Poder360 published the Minas first-round detail: Lula 37, Flávio 31, Romeu Zema 10, then Renan Santos and Ronaldo Caiado on 3, Augusto Cury on 2, and four candidates at 1 or below; blank and spoiled ballots 7 percent, undecided 5. Adding Pablo Marçal to that scenario moves Lula to 36 and Zema to 8, with Marçal taking 4. The Minas survey cost 162,051 reais and was paid for by Folha and Globo, as the national one was.
Underneath the presidential numbers are the state races that supply each campaign its local platform. In São Paulo the governor, Tarcísio de Freitas, who is seeking re-election and backs Flávio Bolsonaro, leads the first round with 45 percent to Fernando Haddad's 27, and would win a runoff 54 to 35. Minas is more broken up: Senator Cleitinho Azevedo leads on 32, with Patrus Ananias and Alexandre Kalil — both former mayors of Belo Horizonte — tied on 12 each.
One methodological point is worth writing down, because it bears on how the whole release should be read. The five state surveys were not run over the same days. São Paulo was interviewed on the 18th and 19th, with 1,610 people. Minas Gerais, Rio and Pernambuco ran from the 18th to the 20th, with 1,204 each. The Federal District ran from the 18th to the 21st, with 910. So a week that produced one Datafolha headline after another produced it from field windows a day or two apart, in a week whose political weather changed daily.
Why it is on the board
Third on Folha's own most-read list when we collected.
This is the second part of a thread. The first, published here yesterday, was the national poll from the same institute in the same week — Lula ahead by six in the first round, and three published accounts that could not agree on when the interviewing stopped. Today's release names its own field windows, and there are three different ones inside a single set of surveys. What the national poll's window actually was is still not something we can settle from what has been published.
확인: Folha de S.Paulo · Poder360