The Roar of the Crowd

03/23/2015 - 15h36

Folha's headline on Monday (March 16): "'Get Out, Dilma' (referring to President Dilma Rousseff) Brings Together 210,000 in São Paulo and Multitudes in the Country."

This headline was wrong, and not exactly (or only) for the reason that provoked almost hundred of protests to the ombudsman. The scolding by readers was focused on the calculation of the multitudes, considered ridiculous given the million people reported by the police, but the entire story is unfortunate.

The first one was to crowd into a single slogan the diversity of reasons they were in the streets. There is no doubt that the protests were against the Workers Party (PT) administration, but throwing the whole world into the same bag with "Get Out, Dilma" is an improper inference, as shown in the statements by demonstrators interviewed.

The second problem was to put in the number (which always is an estimate and not an exact count) instead of the scope: it was the biggest political act in the city since the "Direct Elections Now" protests.

To cap things, the newspaper was mistaken in reporting that the biggest political act had been the demonstrations for direct elections in the Sé Plaza (in downtown São Paulo) in January 1984. That was in the Anhangabaú Valley in April of the same year, which was reported at the time to have drawn 1 million people.

In 1994, 10 years after the "Direct Elections Now" protests, Datafolha measured the area and calculated that there were no more than 400,000. In 2011, since the invention of georeferencing software, a new measure adjusted the number to 379,400 people.

The same revision set the numbers on Paulista Avenue at 949,000; on Consolação Avenue at 564,000 and the Bagatelle Plaza (across from Mars Field Airport) at 1.15 million - all with seven people per square meter, the standard crowd on the subway at peak hours.

Since then, Datafolha has used the method to measure the Gay Parade, the Evangelical March, and the Pope's appearance at Copacabana Beach. In all these cases, the discrepancy between estimates provoked a torrent of complaints, mainly from organizers accustomed to reporting grandiose numbers.

The trouble is the result of schizophrenia practiced by the newspaper itself, which, most of the time, publishes estimates by police and only in some events does it invest in its own measurements. The results are a crazy mixture, which helps to carry things to the point of lack of confidence.

Ironically, the same newspaper which invested time and money in seeking the closest parameters to reality continues to publish news such as that on Jan. 2: "New Year's in São Paulo With 2 Million is Peaceful."

The headline does not say so, but the calculation was made by police. Does it make sense to report the number uncritically, without the caveat, according to Datafolha, that so many people don't fit in that space? (On the police side, 2 million people at a party who don't occupy the whole street and 1 million at a protest?)

Could there be a mistake in the calculations? Certainly, by both sides. Every methodology has limitations which affect the result, according to assertions by two experts contacted by this column, Moacyr Duarte, senior researcher at Coppe, institute of research engineering at Rio de Janeiro Federal University, and Marcelo Zuffo, professor at the Polytechnic School at São Paulo University. To correct it, they use tools which try to compensate the distortions.

On Friday afternoon (March 20), the Public Safety Department agreed to send the ombudsman some photos taken from an Eagle helicopter, which had been denied to the newspaper.

There are few and concentrated at the most crowded points, locations where police should be stationed because they are the most subject to problems. The images and the explanation of the two methodologies were sent to the professors for their evaluations.

Zuffo and Duarte made great observations, but they are too long for this space and there was not enough time to contact Datafolha and the police afterward. This will be done next week, and the result will be posted on the ombudsman's page online. Maybe the discussion will help sharpen the tools. This brutal discrepancy makes estimates useless and only feeds disinformation.

Translated by JOHN WRIGHT

Read the article in the original language