Nine hundred and thirty-one days after the end of the collection and calculation of data for the 2022 demographic census, the IBGE (Brazilian Institute of Geography and Statistics) has not yet published the microdata of the research, which allows more detailed analyzes for academics and public decision-makers. The institute’s rationale is a “high risk” of privacy violations among respondents, driven by advances in artificial intelligence.
A resolution of the IBGE board of directors, from November, consulted by Leaf, indicates a vulnerability if microdata is disclosed according to current disclosure protocols. The document also highlights the need for a new privacy policy.
In an interview with LeafMarcio Pochmann, president of the IBGE, said that the microdata is ready. He reiterated that the postponement stems from the analysis of an internal technical committee, which assessed the risk of re-identification of people using AI tools.
“The law that the IBGE must respect is that of statistical secrecy. Several institutions around the world are reviewing (processes), because advances in artificial intelligence make it possible to work with anonymous data and identify them,” he said, adding that information can no longer be disclosed as in the past. “We are evaluating internally what we can disclose, if we can disclose everything, otherwise,” he told reporters at the institute’s conference held in Salvador.
Microdata represent the lowest level of disaggregation of the survey and come from the more in-depth questionnaires applied by the IBGE. Although the files do not contain the names of the people interviewed, it is possible to identify a person by combining different data. This may involve possible discrimination, invasion of privacy or targeted attacks.
Research institutes have increased their alert after the general data protection law of 2018. During the previous census, in 2010, this did not exist. The standard determines “the use of reasonable technical means available at the time of processing” to guarantee the anonymity of the data, so that the data loses the possibility of being associated, directly or indirectly, with a person.
The release of census microdata was scheduled for December 4. These are the microdata of the sample, obtained by means of a large questionnaire applied to part of the population, and the so-called weighting zones, smaller territorial units used in the sample calculations. The IBGE, however, postponed the publication until 2026, but without defining a new date.
Although motivated by the risk of breach of confidentiality, the delay reflects a series of crises that the IBGE has faced, which erupted during the Covid pandemic and which meant that the 2022 census was already born in unfavorable conditions, with a series of changes in the calendar.
The census, the largest statistical survey in the country, was only carried out because in April 2021 the Maranhão government went to the Federal Supreme Court against the Union and the IBGE to challenge the decision of then-President Jair Bolsonaro (PL) not to carry it out that year. Despite pandemic restrictions, there were guarantees that this would be possible.
Considered a public foundation with a strong technical reputation, the IBGE began to experience various setbacks – funding cuts, suspension of research and the discrediting of methodology, such as that which calculates unemployment, as a farce. During the discussion of that year’s budget, the census resources, which were BRL 2 billion, decreased to BRL 53 million.
At the STF, the Government of Maranhão affirmed that the lack of current data on the population of Maranhão could generate a loss of tax revenue and make the creation of public policies impossible, a reality in other states and municipalities.
The Supreme Court accepted the request and ordered that the research be carried out the following year, in 2022. The Union did not release the necessary budget and the research was carried out with the support of parliamentary amendments.
Faced with the hostility of a part of the population who threatened or refused to answer, the census takers began the field interviews in August and finished them 10 months later. The questionnaire has also been edited. One of the elements removed was international emigration, which would measure the departure of people from Brazil to other countries.
Already under the Lula (PT) government, the Ministry of Planning made a contribution of 380 million BRL in the form of an extraordinary credit so that the institute could complete the investigation.
Since 2023, the IBGE has published aggregated data on various topics, such as population, households, race, religion and mobility. THE Leaf Institute members said this was the strategy they found to be able to publish the data in the face of adversity.
The delay in disclosure, particularly of pending microdata, has slowed the publication of scientific articles. At the Center for Metropolis Studies, one of the national references in the analysis of political and socio-economic transformations of cities, linked to USP (University of São Paulo), some research is pending, as well as the interactive systems used with IBGE data made available to the public for consultation.
“Microdata is irreplaceable. We have not been able to study to what extent metropolises have transformed in terms of social structure over the last decade, which has been a very negative social dynamic with a very long economic crisis, a pandemic, the Bolsonaro government”, explains Eduardo Marques, political scientist and director of the CEM.
In the area of demography at Unicamp (State University of Campinas), students are encouraged to use other sources of information while microdata are not published, but they suffer losses. “A student who is interested in studying fertility today uses data from the Ministry of Health. With the census, we can break down information on fertility by religion, for example, which is unachievable on the basis of the ministry,” explains Joice Melo Vieira, professor in the institution’s demography department.
Some articles, because they only contain data up to 2010, are no longer of interest to scientific journals. “We are waiting for the 2022 census to be released so we can update it and then submit it.”
In June, the IBGE revealed evangelical growth in the country, but information on the progress of each denomination is not yet available. For Lívia Reis, a researcher at the Institute of Religious Studies, only microdata explains which churches have added more believers and which have lost popularity.
“These figures are causing a sensation in the religious field. But the IBGE is taking essential precautions, because it understands that the data is not only important for researchers, but that it can be used as an instrument of internal conflict within the religious field.”
While dealing with outstanding questions for 2022, the institute’s technicians are preparing for the 2030 census, which will pose new collection challenges.
Along the way, there are two more major censuses: the agricultural census, which should take place every five years, but the last edition dates from 2017; and research on the homeless population, the first without household collection dynamics. The forecasts are for 2027 and 2028 respectively.
About The Author
