Opinion – Helio Beltrão: Chamber errs in wanting to criminalize electoral research

The Chamber of Deputies last week approved the urgency of the bill (“PL”) that will punish polling institutes if they point out voting intentions that differ from the results of the polls. The final text has not yet been defined, but the initial idea is to impose fines and imprisonment (from four to ten years) for statisticians and administrators of institutes (and contractors) who publish research, in the 15 days prior to the election, that deviate from the ballot boxes beyond the stated margin of error.

It is astonishing that the institutes have been consistently wrong in the last three electoral cycles: in the first round of the 2018 presidential election, in the 2020 municipal elections and in this first round. Unusually, the error invariably occurred for the same side, overestimating the vote on the left and underestimating the vote on the right. The revolt of the disadvantaged leaders and the distrust regarding the honesty of the institutes, therefore, is justified. However, criminalizing the institutes’ mistakes will not solve the problem. Research in Brazil is already heavily regulated by the TSE.

First, it is necessary to clarify that polls do not predict how voters will vote at the polls. Predictions and guesses are carried out by speculators: political analysts, ordinary citizens or even those who put money in online bookmakers. Honest polls are a “picture” — blurred and imperfect — of “voting intentions” in a given collection period.

Therefore, there is a fundamental epistemological problem in punishing the photographer of a portrait on today’s date if another photo, two weeks later, turns out differently. In the final stretch of any election, new facts occur at breakneck speed; everything can change. The PL assumes that the institutes have a crystal ball. Its most harmful effect, however, is less transparency for the voter. After all, institutes can choose the safe path of not publishing, avoiding arrests and fines, and selling only tracking reserved for parties.

Additionally, the sources of errors committed by the institutes are varied. There are numerous value judgments involved. For example, institutes must decide on what proportions of income and education ranges to adopt in their sample. Some choose to follow the proportions of the most recent IBGE survey, while others calibrate the proportions considering, for example, that the survey is outdated or that part of the citizens lies when declaring their income to the government.

Finally, there is also the problem of shamed voting, possible flaws in questions, people who choose not to respond to the survey, and the interviewer’s gender bias, which can affect the answer. It is evident that there are fraudulent surveys aimed at helping a particular candidate, whose responsible should be punished by the legislation in force. But the factors listed above are the result of legitimate mistakes.

There is nothing more important to a research institute than its credibility, its greatest asset. Those who deliver unreliable results will tend to lose ground with clients, who are not just media companies, but political parties, banks and class associations.

The solution involves more competition between institutes, less bureaucracy and independent assessments based on first-rate statistical and methodological analysis, such as those by the Americans AAPOR and FiveThirtyEight.

But Brazil is still addicted to state interventions.