Imprimir Republish


Study estimates how many years scientific collaborations last in Brazil

UFABC survey analyzed articles written as coauthorships in 81 fields

Felipe Mayerle

Scientific collaborations between Brazilian medical researchers can last around five years—that is roughly how long scientists in the field tend to work together, coauthoring academic papers. In nuclear and biomedical engineering, the average length of coauthoring relationships is eight years. In history, languages, literature, philosophy, and the arts, it is much shorter, at around two years—in these fields, there is less of a tradition of cooperative work and most papers are authored by a few people at most, and often one person alone.

This unprecedented analysis of the longevity of scientific partnerships is part of a survey on the academic output of Brazilian PhD holders in 81 fields of knowledge, based on data from the Lattes résumé platform run by the Brazilian National Council for Scientific and Technological Development (CNPq). “Collaborations that result in coauthored articles rely on subjective variables, such as compatibility and trust, and on the publication culture of each discipline. We wanted to determine the probability of researchers grouping together and establishing this type of relationship,” explains Jesús Pascual Mena-Chalco of the Center for Mathematics, Computing, and Cognition at the Federal University of ABC (UFABC).

The study was the result of the ongoing doctoral research of UFABC computer scientist Andréia Gusmão, supervised by Mena-Chalco, and was published in the scientific journal Em Questão in 2022. Articles, conference presentations, books, and book chapters by more than 200,000 PhD holders who updated their profiles on the Lattes platform between 2019 and 2020 were compiled. The researchers behind the study extracted three indicators from this huge volume of data. One was the longevity of academic coauthorship: the time, in years, that collaborations between groups of researchers continue to publish articles.

The second indicator was the scale of academic coauthorship: the number of people who participate concurrently on the same paper. Articles with single authors—which accounted for more than half of the papers in areas more resistant to scientific partnerships, such as education, law, and languages and literature—were excluded from the calculation. To prevent distortions by outliers, studies with more than 40 coauthors were also not included. Physics, a field in which there are many international research consortiums with thousands of members, was the discipline with the most omitted papers.

The third indicator, dubbed the “coauthor journey,” was established by the group itself and is determined by the relationship between the longevity and scale of the coauthorship. The smaller the group of researchers involved, the longer the journey tends to be—after all, it is more difficult to keep a team of four or more coauthors together over time than a group of two or three. But the pattern varies greatly between disciplines. “In mathematics, groups are smaller, made up mostly of advisors and students or alumni, but they generally stay together for several years. In areas of health, such as medicine, pharmacology, nursing, and dentistry, the groups are often much bigger, but they tend to work together for a shorter period of time,” the paper states. There are always points outside the curve, however, and the longest journey identified by the survey was in medicine: two researchers who wrote papers together for 46 years, in what was classified as a lifelong partnership. The study observed similar types of collaboration in disciplines that had a common origin, such as physics and astronomy, biology and medicine, and computing and electrical engineering. “They are sister fields, which have been separated but continue to share publishing practices,” says Mena-Chalco.

There has long been a consensus that collaborative research is essential to the progress of science. Many fields of research have become so complex and specialized that they require cooperative and interdisciplinary efforts to achieve advances in knowledge. In some areas, international consortiums have been formed, leading to a phenomenon not addressed by the UFABC study known as hyperauthorship, when articles list more than a thousand authors, each responsible for a minuscule contribution—in data collection or analysis, for example (see Pesquisa FAPESP issue no. 289).

Alexandre Affonso / Revista Pesquisa FAPESP

“Public policies have been implemented worldwide to encourage and expand collaboration. Identifying the characteristics and efficiency of these partnerships has therefore become important to evaluating the results of a researcher’s work,” says information scientist Solange Santos, production and publication coordinator of the SciELO Brasil journal collection and a coauthor of the article by the UFABC researchers. “Recently, even in fields that are more averse to collaborations, there has been a progressive increase in the number of coauthored papers.”

Previous studies by Mena-Chalco and colleagues have addressed other characteristics of collaborations in Brazil, making use of new technologies to extract and analyze massive volumes of information from the Lattes platform. One aspect they identified was the increasing frequency of collaborations: just over one million collaborations were recorded between 2008 and 2010, in contrast to just 63,944 recorded between 1990 and 1992 (see Pesquisa FAPESP issue no. 218). The influence of geographical proximity on the formation of cooperative networks was also studied (see Pesquisa FAPESP issue no. 241). In the latest study, the team looked for clues about how research groups stay together over time and what factors might influence their longevity.

The researchers are not only looking at data from the Lattes platform. Last year, at the XXII National Information Science Research Meeting (ENANCIB) in Porto Alegre, Mena-Chalco and Gusmão presented the results of a study in which they surveyed the types of collaborations adopted in information science. They analyzed articles from 1968 and 2021 indexed in the Web of Science database, which does not cover as much literature as the résumés on Lattes and focuses on traditional and well-established international journals. As expected, the results showed that smaller groups publish together for longer and that institutional collaborations, characterized by groups of researchers from the same institution, are the longest lasting, with coauthor pairs lasting for an average of 10 years. In international partnerships, the longest period of collaboration between two researchers was seven years.

According to the UFABC computer scientist, the indicators are objective but do not measure collaborations in all their complexity. “Not every collaboration results in the publication of an article, just as there are papers with a large number of coauthors between whom the relationship was not necessarily that of a collaboration,” he says. “We analyzed the social interactions established when researchers share article authorship, but there are many more people involved in the process, such as advisors, article reviewers, and journal editors. Behind each article is the cooperative work of a much larger number of people.”

One aspect addressed in the Em Questão article were the interactions between advisors and students, which are maintained for varying lengths of time depending on the field. “In economics, for example, former students commonly stop collaborating with their advisors, unlike in many disciplines,” says Mena-Chalco.

In another article published in Em Questão in 2021, a group of researchers from the School of Arts, Sciences, and Humanities at the University of São Paulo (EACH-USP) showed that papers coauthored by students and their advisors have grown in volume in all fields since the 1980s, especially in the biological and agricultural sciences, and more recently in disciplines such as languages, literature, and linguistics. “Even in the humanities, where coauthorships were rare because theses are seen as belonging to the student and not their advisor, there has been a notable growth in articles authored by professors and graduate students in recent years, even though they are less common than in other fields,” says computer scientist Luciano Digiampietri, a professor of information systems at EACH-USP and one of the authors of the article. “Generally speaking, the continual increase in Brazilian scientific output over the last 40 years has resulted in an increase in collaborative studies.”

The UFABC group is now looking to evaluate coauthorship journeys in greater depth. “We are going to analyze the issue of gender to find out if these journeys are different in more homogeneous or more diverse groups,” says Mena-Chalco. Another focus is the longevity of collaborations seen in broader categories, such as institutions or fields with which the coauthors are affiliated. “It would be interesting to know the relationship between the longevity of collaborations and the international presence of Brazilian researchers in large fields—in high-energy physics, for example, where coauthorship networks are huge—and to generate data that will help funding agencies decide where to invest.”