O listens to at least 2,000 people aged 16 or over to portray the voting intentions of the Brazilian electorate in each survey. The interviews are carried out in person, at points of flow on the streets, and follow statistical criteria so that the sample represents the entire population.
Created in 1983 as the research department of Grupo Folha, the institute adopted the name the following year, when it carried out its first electoral survey — about the indirect election that led Tancredo Neves to the Presidency.
Currently, Datafolha conducts interviews in at least 113 small, medium and large cities, in all regions of the country. There are around 160 researchers in the field to collect data. The institute does not carry out commissions for politicians or parties. The surveys are contracted by media outlets with a contractual obligation to publicize the results.
How interviewees are chosen
The selection is made in stages. First, the municipalities that will be part of the survey are drawn. Then, the neighborhoods and points where the interviews will be carried out. In the final phase, the institute uses proportional quotas for gender and age, based on data from and (Superior Electoral Court).
Other sociodemographic variables, such as education, income, occupation and religion, are monitored by the institute based on its research history. The sample must reflect the total Brazilian electorate.
With each new round, Datafolha draws up cities and points of approach. The objective is to prevent parties from operating in research locations and prevent the same residents from being interviewed more than once, which would bias the sample.
The interviewer cannot know in advance who will be approached, and no person can offer to answer the questionnaire.
Why on the street, and not by phone
Datafolha opted for a face-to-face approach at flow points. According to the institute, telephone surveys representative of the total electorate are unfeasible in Brazil because a significant portion of Brazilians do not have a landline at home.
“Despite practically everyone having a cell phone, people don’t answer calls and not everyone has the same availability to answer calls throughout the day”, says the institution’s general director, Luciana Chong.
The street approach also avoids the difficulty of accessing residences in condominiums, buildings and favelas.
Quality control
Datafolha checks a 20% sample of each researcher’s material.
“For this check, we use artificial intelligence, which compares the interviewees’ answers in audio with the answers written down by the researcher”, says Chong.
When a problem is identified, the quality team checks the total number of interviews carried out by the researcher and, if necessary, this material is redone. Furthermore, there is monitoring considering the city’s electoral history and the results obtained in the sample. All interviews are recorded.
What is the margin of error
Every sample survey has a difference — the tolerated difference between the measured value and the real value to be measured. National Datafolha surveys generally have a maximum margin of error of plus or minus two percentage points, with a confidence level of 95%.
This means, in a simplified way, that if 100 samples were taken simultaneously from the same population, 95 of them would have results within this range.
Technical draw and fluctuation
A technical tie occurs when the difference between two candidates is within the margins of error — that is, when the confidence intervals overlap. By the same principle, if a candidate’s voting intentions varied within a margin from one survey to another, it is said to have fluctuated, not to have grown or fallen.
Order of questions
The arrangement of questions in the questionnaire may influence the answers. Therefore, Datafolha does not ask questions that mention the names of candidates, parties or government evaluations before questions about voting intentions.
Registration and disclosure
Institutes must register their surveys up to five days before publication. The record informs who contracted the research, the amount paid, the methodology, the period of completion and the statisticians responsible. The TSE does not have prior access to the results.
Datafolha only records your searches before carrying them out. The institute opposes registration after collection, a practice that opens up space for the strategic use of surveys as a political marketing instrument.
Datafolha publishes the complete statistical bases and detailed segmentations on its website after .
