Two names emerge consistently across nearly every scenario tested
Com dezoito meses de antecedência, a disputa pelo Senado em São Paulo em 2026 já esboça seus primeiros contornos — não como certeza, mas como possibilidade. Uma pesquisa do Real Time Big Data, divulgada em 9 de março, testou oito candidatos em dez cenários distintos, revelando que Simone Tebet e Marina Silva carregam, por ora, a preferência mais consistente do eleitorado paulista. O levantamento não aponta um destino, mas ilumina um campo ainda em formação, onde a indecisão de uma parcela significativa dos eleitores lembra que toda corrida começa muito antes de seu desfecho.
- Tebet e Marina Silva aparecem como as candidatas mais estáveis da pesquisa, sustentando entre 15% e 20% de intenção de voto em quase todos os cenários testados.
- Fernando Haddad, quando incluído nas simulações, alcança 24% — o maior número registrado em qualquer cenário —, sinalizando que sua eventual entrada na corrida poderia redesenhar completamente o quadro.
- A volatilidade é real: entre 10% e 17% dos entrevistados declararam não saber em quem votar, e de 8% a 13% optaram por voto nulo ou em branco, indicando que o eleitorado ainda não se fixou.
- O campo de candidatos segue indefinido — a presença ou ausência de nomes como Haddad e Garcia altera sensivelmente os percentuais dos demais, tornando qualquer projeção prematura.
- A pesquisa, realizada com 2.000 entrevistados nos dias 6 e 7 de março, tem margem de erro de 2 pontos percentuais e está registrada sob o número SP-00705/2026.
O Real Time Big Data divulgou na segunda-feira, 9 de março, uma pesquisa que mapeia as intenções de voto para o Senado em São Paulo nas eleições de 2026. O levantamento testou oito candidatos em dez cenários diferentes — uma escolha metodológica que reflete a complexidade de uma disputa em que São Paulo elegerá dois senadores e o campo ainda não está definido.
Os nomes testados foram Simone Tebet (MDB), Guilherme Derrite (PP), Marina Silva (Rede), Ricardo Salles (Novo), Ricardo Mello Araújo (PL), Paulinho da Força (Solidariedade), Mário Frias (PL) e Rodrigo Garcia, sem partido. Em cada cenário, os eleitores escolhiam dois nomes, e os votos eram consolidados e normalizados a 100%.
Dois nomes se destacam com consistência: Tebet e Silva. No primeiro cenário, ambas marcaram 15% e 16%, respectivamente. À medida que o instituto testou combinações diferentes — acrescentando Garcia, Frias ou outros —, as duas mantiveram ou ampliaram sua vantagem. Com Garcia no campo, Tebet chegou a 20% e Silva a 17%.
O resultado mais expressivo surgiu quando Fernando Haddad (PT) entrou nas simulações. Em dois cenários, o petista liderou com 24% e 23%, os maiores percentuais registrados em toda a pesquisa. Silva permaneceu competitiva, entre 15% e 18%, mas a presença de Haddad compressa o espaço dos demais candidatos e sugere que sua eventual candidatura mudaria substancialmente a dinâmica da corrida.
O que os números também revelam é incerteza. A fatia de indecisos variou entre 10% e 17% conforme o cenário, e os votos nulos ou em branco oscilaram entre 8% e 13%. Com dezoito meses até a eleição e o campo ainda em aberto — Haddad não confirmou candidatura, Garcia não tem partido —, a pesquisa retrata menos uma corrida definida do que um terreno ainda em movimento.
The Real Time Big Data institute released polling on Monday, March 9th that maps out how São Paulo voters are thinking about the 2026 Senate race. The survey tested voter appetite for eight different candidates across ten distinct scenarios—a methodological choice that reflects the complexity of the contest, since São Paulo will elect two senators this cycle, and the field remains unsettled about who will actually run.
The candidates tested were Simone Tebet of the MDB, Guilherme Derrite from the PP, Marina Silva of the Rede, Ricardo Salles of the Novo, Ricardo Mello Araújo of the PL, Paulinho da Força from Solidariedade, Mário Frias also of the PL, and Rodrigo Garcia, who currently has no party affiliation. In each scenario, voters were asked to choose two names, and the institute consolidated those first and second votes, then normalized the results to 100 percent.
Two names emerge consistently across nearly every scenario tested: Tebet and Silva. In the first scenario, which included Tebet, Derrite, Silva, Salles, and Mello Araújo, Tebet pulled 16 percent and Silva 15 percent, with Derrite also at 15 percent. But as the institute tested different combinations—adding Garcia in one scenario, Frias in another, then Haddad in later matchups—Tebet and Silva held their ground or gained. When Garcia was added to the field, Tebet rose to 20 percent and Silva to 17 percent. When Frias was tested, Tebet stayed at 19 percent while Silva held at 18 percent, with Derrite also reaching 18 percent in that scenario.
The most striking results came when Fernando Haddad of the PT entered the hypothetical race. In the seventh scenario, which included Haddad, Silva, Garcia, Mello Araújo, and Paulinho da Força, Haddad led with 24 percent—the highest number any candidate achieved across all ten scenarios. Silva remained competitive at 18 percent. In the eighth scenario, Haddad again led at 23 percent, with Silva at 18 percent and Derrite at 17 percent. But when the ninth and tenth scenarios tested Haddad alongside a fuller field, his advantage narrowed. In the final scenario, which included Haddad, Derrite, Silva, Salles, Frias, and Paulinho da Força, Haddad led with 21 percent, Derrite was at 16 percent, and Silva dropped to 15 percent.
What stands out beneath these numbers is the volatility. Across all ten scenarios, the share of voters who said they didn't know or didn't answer ranged from 10 to 17 percent. Null and blank votes—those who either spoiled their ballot or abstained—ranged from 8 to 13 percent. This suggests the race is far from settled. With eighteen months until the 2026 election, a significant portion of the electorate remains genuinely undecided, and the field itself has not fully crystallized. Whether Haddad runs, whether Garcia stays in or drops out, whether other names emerge—all of these variables could reshape the landscape substantially.
The institute interviewed 2,000 people across São Paulo on March 6th and 7th. The margin of error is 2 percentage points in either direction, with a confidence level of 95 percent. The survey was commissioned by Real Time Big Data itself and is registered with the electoral authority under number SP-00705/2026. What the data reveals is a race in motion: two candidates with genuine traction, a field of others with real but secondary support, and a large pool of voters still making up their minds.
Citações Notáveis
The race is far from settled, with a significant portion of the electorate genuinely undecided and the field itself not fully crystallized— Analysis from polling data
A Conversa do Hearth Outra perspectiva sobre a história
Why does the institute test ten different scenarios instead of just asking voters who they'd vote for right now?
Because the field isn't fixed. Nobody knows yet who will actually run. Testing different combinations tells you which candidates have real appeal and which ones only look strong because they're running against a weak field.
So Tebet and Silva keep showing up at 15 to 20 percent no matter who else is in the race. What does that tell you?
It tells you they have a floor. They're not dependent on the field being thin. They have actual voter recognition and preference. That's rare at this stage.
But Haddad hits 24 percent in one scenario. Doesn't that mean he's actually the strongest?
It could. Or it could mean voters respond to a familiar name, a president's ally, someone with real institutional power. But notice what happens when you add more candidates—his number drops. He's not as durable across different field compositions.
The undecided voters are huge—up to 17 percent. Does that mean the race is unpredictable?
It means it's genuinely open. Eighteen months is a long time. Those voters will move. The question is whether they move toward the candidates already leading or toward someone new who hasn't entered the race yet.
What about the null and blank votes? Eight to thirteen percent is significant.
It's a signal of disengagement or protest. Some voters might not see anyone they want to vote for. That's a pool that could be mobilized by the right candidate or campaign, or it could stay home.