Human resources and scientific output in Brazilian science: Mapping astronomy, immunology and oceanography

Data on the Brazilian scientific community

These were collected from the Directory of Brazilian Research Groups, a Brazilian database designed and organized by CNPq in 1993 (CNPq, 2004). The agency has carried out six national censuses of Brazilian research groups and all this information has been incorporated in the database. Although researchers are not compelled to take part in the censuses, it is estimated that the database covers around 80‐90 per cent of the whole Brazilian scientific community. Thus, the database is one of the most important mechanisms in the country for:

• •

  planning national policies;

• •

  getting rapid and objective information; and

• •

  preserving an archive of science in Brazil.

Usually, to register a research group and participate in the census, the leader requests certification from the administrative research office of the institution where he or she is affiliated. That means that the researchers themselves cannot register their groups and so take part in the census. Not all such requests to register are granted. For those who receive certification, the leaders are allowed to register the group and return all the information requested, such as name, field, affiliation and details of the members and publications.

Although the censuses started in 1993, the method of collecting and the way data were compiled was changed after the second one in 1995. So, the time trend for the characteristics of the scientific community shown here include only the censuses of 1997, 2000 and 2002. The data from the 2004 census are not yet available.

Data source

The 20 journals with the largest number of articles (LNA) of the three fields were selected from the 2003 Journal Citation Report list available at the ISI web site. The idea to analyse the fields' scientific output from this set of journals is justified because Brazilian publications represent less than 2 per cent of the ISI database. This means that around two Brazilian articles would be found in a journal that publishes 100 articles per year and 20 in the journal that publishes 1,000 articles annually. Publication data from this set of journals (LNA) were then collected from the Web Of Science and gave a reasonable chance of finding some Brazilian articles in journals with the largest numbers of articles. The searches combined the filter address and filter source title. Each group of the 20 LNA selected in the three fields were, separately, combined with the address Brazil or Brasil. Publication data from these searches were saved and analysed with the help of Excel software. The resulting databases contain all information available at the Web Of Science such as full references, authors' names and address(es), accumulated citation numbers and type of publication.

Data on impact factor

Since the searches on the Web Of Science gave only the accumulated citations of the publications, data on citations had to be normalized to find a comparable impact factor. This was then measured by the ratio: number of citations to publications of year X (from year X up to October 2004), divided by the number of publications of year X and again divided by the number of years from publication to 2004. Thus, there were 26 Brazilian astronomy publications from year 1988 and they received a total of 510 citations up to October 2004, so the impact factor was calculated as 510/(26 × 17)=1.15. This method allows a comparison of citation data found for publications published in the five years studied.

The Brazilian scientific community – number of researchers, qualifications and sex

Details on the Brazilian scientific community can easily be found on the Directory of Brazilian Research Groups database. Table I shows the total numbers of researchers registered on the database in the three fields as well as in the whole of science for three censuses. The data indicate that the Brazilian scientific community is growing fast: from 1997 to 2002, the total number of Brazilian researchers increased from 33,675 to 56,891, or by 69 per cent. However, in immunology and oceanography, the increase was only about 40 per cent and in astronomy there was a reduction in the number of researchers. There are three possible explanations. Some astronomers may have declined to participate in the recent censuses; there are few new positions for astronomers at Brazilian research institutes and universities; there is a paucity of physics undergraduates and so of graduate students in astronomy and so of researchers. A detailed analysis would be needed to understand this tendency.

The scientific communities in the three fields show differences in the fraction of researchers with a PhD degree (Table I). Relative to the total number of researchers, astronomers seem to be the most qualified, followed by immunologists, while researchers in oceanography are only slightly more qualified than the average for Brazilian researchers. The high percentage of PhDs in astronomy may reflect the small size of the research community. In a country such as Brazil, where resources are scarce, fields with a small number of scientists tend to have the most qualified ones. Another factor may be the long tradition of astronomy in Brazil that has allowed it to gain international recognition and so require Brazilian astronomers to be more competitive. On the other hand, the recent consolidation of oceanography in the country, as well as its increasing interest for young Brazilians, are factors that may have contributed to the low number of PhD‐qualified oceanographers.

Concerning the sex of the researchers (Table II), the highest female/total ratio was found in immunology at about 58 per cent. Oceanography has an intermediate ratio at 41 per cent, while the lowest was found in astronomy at 22 per cent. The share of women in the three fields did not grow from 2000 to 2002. The different shares of women among the three fields may be related to the numbers of men and women enrolled in Brazilian universities, undergraduate and graduate courses. The high share of women in biological sciences as well as in social sciences and humanities compared with their low share in engineering and exact sciences has been frequently discussed worldwide (for example: Mcgregot and Harding, 1996; Lane, 2004; Tabak, 2002). In Brazil, as in most countries, including developed ones, women are still the minority in some undergraduate courses such as physics, mathematics and chemistry. The causes for such “exclusion” are complex and involve many factors, including cultural, social and economic ones.

The scientific publications registered in the Brazilian database

Most of the scientific articles published by developing countries, such as Brazil, are not read by the international research community and are therefore “invisible”, i.e. they are inaccessible and unknown outside the country (Gibbs, 1995). In Brazil, it is estimated that 80 per cent of the country's scientific literature is published only in domestic periodicals. In view of the limited distribution of Brazilian scientific periodicals, that most of them are in Portuguese and that the majority are not available electronically, most of their articles circulate almost exclusively within the main Brazilian university libraries.

On the database of the Directory of Brazilian Research Groups, researchers are encouraged to register details of all their publications, even these “invisible ones”. Thus, this database represents an important resource to map the visibility of the country's scientific output. The pattern of publications registered by Brazilian researchers (only by those who have a PhD degree) in the three fields is presented in Figures 1 and 2.

Among the three fields, immunology presents the largest number of publications registered in the database, with 6, 752 versus 1,962 in astronomy and 2,631 in oceanography. This is probably related to the number of researchers registered within each of the three fields (Table I). Although astronomy has the lowest number of publications, their distribution differs from that of the other two fields and of that found for all science in Brazil: almost 50 per cent of its publications are classified as “full articles”. On the other hand, immunology and oceanography seen to follow the pattern of the rest of science: “full articles” represent around 30 per cent of the total publications. Such difference may be a consequence of the qualifications of the researchers (Table I) or the size of the research groups. The number of people (mainly students) who are involved with the research may change the pattern of publications. A researcher that is, for example, advising only a single graduate student who sent an abstract (résumé in Figure 1) for a conference in a given year will populate the census with this information only. Another researcher, who is advising three or four graduate students, each of whom submitted an abstract, would account for more papers.

If the variable size of group may influence the pattern of publications, the same is not true when full articles are analysed according to the origin of the journals in which they were published (Figure 2). Among the information contained in the CNPq database, researchers are supposed to indicate whether the paper they are registering was published in an international journal or in a national journal. The first set of journals is defined as all journals that are not published or/and edited inside the country. Thus, it includes not only journals covered by ISI but by any other databases, such as MedLine, and some that are not covered by any database.

It is clearly the preference of astronomers and of immunologists to have their studies in international journals: 936 out of 959 (almost 98 per cent) and 1,594 out of 2,001 (almost 80 per cent), respectively. Conversely, the distribution of full articles in oceanography is similar to that found for all science in Brazil, around 50 per cent. The difference found here is probably due not only to internal practices and interests of the three fields but differences in their institutionalisation in the country. Clearly, research in astronomy, more than in the other two fields, is oriented toward international interests and peers. This explains the largest number of articles in international journals. In two previous studies (Figueira et al., 2003; Leta et al., 2005), it has been demonstrated that research articles (original papers) are the ones most frequently found among Brazilian articles published in international journals. But, reviews and case reports are the most frequent types of publications found in articles published in Brazilian national or domestic journals.

Brazilian scientific output registered in the ISI databases

A number of factors cause Brazilians not to publish in journals indexed in the SCI, such as the very low number of national journals covered; a supposed low quality of scientific production; language barriers; and varying editorial standards and submission practices among international journals. Despite all these difficulties, the number and the share of Brazilian publications in the ISI databases have increased remarkably over the last decades, thus increasing the visibility of its science. The increasing numbers of co‐authored publications as well as those of people trained for S&T are frequently indicated as the reasons for the growth of Brazilian publications in the ISI database.

In the present study, the original idea was to map the visibility of the three fields according to the type of journal: the 20 journals with the largest number of articles (LNA) and the 20 journals with the highest impact factor (HIF). But, as can be seen from Table III, there is a large overlap on the 20 titles of journals that were in both sets, LNA and HIF, in the three fields. Among the three fields, the lowest overlap of journals occurred in immunology: there are only eight journal titles common to the lists of LNA and HIF journals; the other 12 titles are in either the LNA journal list or the HIF journal list.

The time trend analysis indicates that Brazilian ISI publications (published in the 20 journals with the largest number of articles) in the three fields have been increasing, see Figure 3. Although the scientific community in astronomy is the smallest, and remains so (see Table I), publications from this field increased the most. Astronomy publications increased from 26 to 183 (×7.0) while in immunology and oceanography they increased from 20 to 136 (×6.8) and from 4 to 25 (×6.2), respectively.

The share of this set of Brazilian publications (Figure 3) of the ISI total publications, in 1988 and in 2003, is shown in Figure 4. Among the three fields, astronomy is the one with the largest relative share of publications. Brazil's share in this field, around 1.9 per cent, is higher than that for all Brazilian publications in the SCI, which, according to MCT (2004), was 1.55 per cent in 2002. On the other hand, the shares of Brazilian publications in immunology and oceanography are lower than that found for all science.

The data presented in Figures 3 and 4 may be related to specific characteristics of the fields. The superior performance found for astronomy may be a consequence of the fact that Brazilian astronomy publications are more oriented to international work which may be related to its tradition in the country and its qualified researchers. Moreover, the relatively small number of journals classified by ISI as being in this field (42) may also contribute to this difference. This would induce astronomers to concentrate their publications in just a few journals. For immunology, the large number of journals classified as relevant to this field (113) and the close relationship between immunology and other biomedical fields would mean that researchers in this area would make use of a larger range of journals. This would include journals not classified as immunology but in other biomedical sub‐fields.

Brazilian scientific output registered in the ISI database: journal distribution

The relationship between Brazilian publications found in the three fields and the total number of papers published in the journals where Brazilian publications were found is presented in Figure 5. This allows the hypothesis that Brazilian papers have a greater presence in journals with many publications (LNA) to be checked.

At first sight, this does not appear to be correct. However astronomy appears to be an exception. In this field, Astrophysical Journal and Astronomy & Astrophysics are the two journals with the largest number of articles (2,435 and 1,936, respectively – black squares) and, as can be observed in Figure 5, these are the journals where most Brazilian astronomers have published. This pattern does not apply to immunology and oceanography. In the first two journals in the immunology LNA ranking, Journal of Experimental Medicine (n=1,603) and Transplantation (n=1,154), Brazilians have published fewer than 20 articles in the five years studied (white triangles). In oceanography (white circles), Estuarine Coastal & Shelf Science and Limnology & Oceanography, the two journals with most articles, were not the ones where Brazilian oceanographers have published the most.

Table IV shows the five journals with the most Brazilian articles in the five years studied. Publications from astronomy tend to be concentrated in journals with the largest number of articles (numbers in parenthesis represent the ranked position of the journal). As for the total number of publications found in the five journals with the largest number of Brazilian articles, these were 70 per cent and 78 per cent of total Brazilian publications in astronomy and oceanography (378 out of 544; and 65 out of 83, respectively). In immunology, as publications are more spread, the 243 papers found in this set represent 57 per cent of its total.

Brazilian scientific output registered in the ISI database: the most productive institutions

It has been demonstrated that Brazilian science is concentrated in a few institutions, especially those from the public university system, which are supported by federal and state government funds. The University of São Paulo, USP, has been pointed out as the most productive institution overall (Leta and De Meis, 1996; MCT, 2004)

Table V presents the 10 Brazilian institutions that contribute most to publications in the three fields in the years studied. As can be noted, USP, the largest university of the state system, is the institution that has contributed the most to the Brazilian ISI publications in all three fields. One reason that pushes USP to the top of the leading Brazilian scientific institutions is the large number of graduate students and qualified scientists engaged in S&T activities developed in it. Another important reason is the continuous, stable and growing funds granted to USP's researchers by FAPESP, a state foundation responsible for funding S&T in the state of São Paulo (FAPESP, 2002)

Just as for all Brazilian SCI publications (Leta and De Meis, 1996), universities are also responsible for most of the publications in astronomy, immunology and oceanography. In astronomy, seven of the ten most productive institutions are from the public university system while in immunology and oceanography this ratio is six out of ten. In astronomy, two important research institutes linked to the Ministry of Science and Technology, the National Institute of Space Research (INPE) and the National Observatory (NO) are, after USP, the most productive institutions in the country. In immunology, the Oswaldo Cruz Foundation, in the city of Rio de Janeiro (FIOCRUZ – RJ) and in the city of Belo Horizonte (FIOCRUZ – MG), one of the oldest biomedical research institutes in the country and linked to the Ministry of Health, appears among the most productive institutions. In oceanography, unusually, an enterprise appears in the top ten institutions. PETROBRAS, the Brazilian Oil Company (see PETROBRAS, 2004), is one of the largest oil companies in the world, leading the sector in implementation of the most advanced deep‐water technology, for oil production. The company supports an important research centre “Centro de Pesquisas e Desenvolvimento Leopoldo Americo Miguez de Mello” or CENPES, which has revenues of approximately 1 per cent of the company's turnover.

Brazilian scientific output registered in the ISI database: citations and the most cited articles

One of the most frequently used pieces of information comprised in the ISI database is the number of citations an article receives each year. Such an indicator is often used to estimate the quality or the impact of the scientific output of a country or an institution or a field. Nevertheless, many critics have questioned this supposed relationship between citations and quality, as it is known that there are many reasons to cite an article. Despite the limitations of this indicator and the controversy surrounding it, Leta and Brito Cruz (2003) have discussed the growth in impact of Brazilian scientific publications. According to these authors, the impact factor of Brazilian publications, measured by the ratio citations (three years‐window): publications, increased from 1.0 in 1981 to 1.9 in 1998.

For the present study, the impact of publications in the three fields was analysed (Figure 6). Even if we neglect papers from 2000, it is clear that the normalised average of citations per paper is increasing in all three fields. It increased from 1.2 to 1.9 in astronomy, from 1.2 to 2.6 in immunology, and from 1.1 to 2.4 in oceanography. Although impact data were normalised (Methodology, v.s.), only in astronomy was the average for 2000 similar to that found for 1996. This suggests that publications from this field are more rapidly cited then those from the other two fields. Despite this difference, analysis of the uncited publications (Table VI) published in the years 1988, 1992, 1996 and 2000 shows that this variable represents a very similar fraction of the whole publications from the three fields: 26 per cent in astronomy, 22 per cent in immunology and 24 per cent in oceanography.

Details of the most highly cited publications from each of the three fields are presented in Table VII. The cumulative number of citations of these publications differs among the fields. In immunology, the range of citations of the top‐cited publications was higher than that found for the top‐cited publications from astronomy which were in turn more highly cited than those from oceanography.

The type of publication also varies. In astronomy and oceanography, there are “reviews” and “notes” or “letters” among the most cited publications, while in immunology there are only “articles”. All the top‐cited publications were co‐authored with an international partner, especially from the US. This is, however, not true for national collaboration. Among the 15 publications listed in Table VI only three have more than one Brazilian institution involved. Curiously, although USP was responsible for the largest number of publications from the three fields, the university was responsible for only two of the top‐cited publications.

As for the subject matter of the top‐cited publications, it seems that most are oriented toward international interests. This is clear for all the publications in astronomy and three of the five publications in immunology. In this field, the most cited publication is related to a tropical disease that is endemic in some Brazilian regions, Chagas disease. In contrast, in oceanography the most cited publications are oriented toward national interests: four are related to the Amazon River, the largest river in the country, and one publication that reviews concepts for the understanding of deep‐water, the focus of PETROBRAS research.