HIGHLY VISIBLE SCIENCE: A LOOK AT THREE DECADES OF RESEARCH FROM ARGENTINA, BRAZIL, MEXICO AND SPAIN
Jane M. Russell, J. Antonio del Río and Héctor D. Cortés
Jane M. Russell. Ph.D. in Information Science, City University, London, UK. Senior Researcher, Universidad Nacional Autónoma de México (UNAM), Address: Centro Universitario de Investigaciones Bibliotecológicas, UNAM. Ciudad Universitaria, 04510 México DF, México. e-mail: jrussell@servidor.unam.mx
J. Antonio del Río. Ph.D. in Physics, UNAM. Senior Researcher, Centro Morelense de Innovación y Transferencia Tecnológica, CCYTEM, Mexico. e-mail: antonio@servidor.unam.mx
Héctor D. Cortés. Computer Systems Engineer, Centro de Investigación en Energía, UNAM, Temixco, Morelos, Mexico. e-mail: hdcg@cie.unam.mx
SUMMARY
Since the international visibility of scientific research is especially important for developing countries, the multidisciplinary journals Nature and Science were analyzed for the papers published from 1973 to 2005 by Latin America’s three most productive countries, Argentina, Brazil and Mexico, and these compared with those from Spain. The total numbers of publications were: Spain, 696; Brazil, 411; Mexico, 227; and Argentina, 127. Both Spain and Brazil published over 65% of the total papers in Nature, in contrast to Mexico and Argentina where production was more evenly distributed between the two titles. An increasing presence of all countries over the 33 year span was found, more so for Spain and from 1993 onwards. A high level of international coauthorship was similar in all cases (72-75%). The most visible institutions in the Latin American countries were the large national universities. From 12-18% of publications remained uncited while three on the genome received more than 2000 citations. Analysis of the most highly cited journals indicated a concentration of publications in physics, astronomy and astrophysics, geosciences and molecular biology, findings which were reflected in the results of a text mining analysis of abstracts.
CIENCIA ALTAMENTE VISIBLE: UNA VISTA A TRES DÉCADAS DE INVESTIGACIÓN EN ARGENTINA, BRASIL, MÉXICO Y ESPAÑA
RESUMEN
]]> Dado que la visibilidad internacional de la investigación científica es de especial importancia para los países en desarrollo, fueron analizados los trabajos publicados entre 1973 y 2005 por los tres países más productivos de Latinoamérica, Argentina, Brasil y México, en las revistas multidisciplinarias Nature y Science, comparándolos con los de España. El número total de las contribuciones fue de 696 para España, 411 para Brasil, 227 para México y 127 para Argentina. Tanto España como Brasil publicaron más del 65% del respectivo total en Nature, mientras que en el caso de México y Argentina, la producción entre las dos revistas fue más equilibrada. Se notó una creciente presencia de los cuatro países a lo largo de los 33 años estudiados, especialmente con respecto a España a partir de 1993. Todos los países mostraron un alto nivel de colaboración internacional (72-75%). Las instituciones latinoamericanas con más presencia fueron las grandes universidades nacionales. Entre 12 y 18% de las contribuciones por país no fueron citados y tres trabajos sobre el genoma recibieron más de 2000 citas. Un análisis de las revistas más citadas mostró la concentración de títulos en Física, Astronomía y Astrofísica, Geociencias y Biología Molecular, resultados que se reflejaron de igual forma en los temas resultantes de un proceso de minería de textos en los resúmenes.CIÊNCIA ALTAMENTE VISÍVEL: UMA VISTA A TRÊS DÉCADAS DE INVESTIGAÇÃO NA ARGENTINA, BRASIL, MÉXICO E ESPANHA.
Resumo
Devido a que a visibilidade internacional da investigação científica é de especial importância para os países em desenvolvimento, foram analisados os trabalhos publicados entre 1973 e 2005 nos três países mais produtivos da América Latina, Argentina, Brasil e México, nas revistas multidisciplinares Nature e Science, e comparados aos de Espanha. O número total das contribuições foi de 696 para Espanha, 411 para Brasil, 227 para México e 127 para Argentina. Tanto na Espanha como no Brasil foram publicados mais de 65% do respectivo total em Nature, enquanto que no caso do México e da Argentina, a produção entre as duas revistas foi mais equilibrada. Notou-se uma crescente presença nos quatro países ao longo dos 33 anos estudados, especialmente em relação à Espanha a partir de 1993. Todos os países mostraram um alto nível de colaboração internacional (72-75%). As instituições latino-americanas com mais presença foram as grandes universidades nacionais. Entre 12 e 18% das contribuições por país não foram citadas e três trabalhos sobre o genoma receberam mais de 2000 citações. Uma análise das revistas mais citadas mostrou a concentração de títulos em Física, Astronomia e Astrofísica, Geociências e Biologia Molecular, resultados que foram refletidos nos temas resultantes de um processo de busca de textos de resumos.
KEYWORDS / Collaboration / Iberamerica / Mainstream Science / Nature / Science /
Received: 02/27/2007. Modified: 08/07/2007. Accepted: 08/08/2007.
Introduction
Of special concern to developing countries is the extent to which their science is visible to the international scientific community and, especially, how much of their contribution to global science can be considered at the cutting edge of international research. A great deal of the science that is carried out in lesser developed countries and regions is virtually invisible to the global research population (Gibbs, 1995), due to its widespread publication in a series of national journals which are, by and large, not picked up by the leading bibliographic services (Cetto and Alonso-Gamboa, 1998). Nonetheless, Latin American countries have shown a steady increase over the last two decades in the number of papers they publish in mainstream journals (Albornoz, 2004), indicating a desire for greater international recognition.
Of all mainstream journals, the two multidisciplinary, weekly journals, Nature (published commercially in the UK) and Science (published by the American Association for the Advancement of Science) are the best-known and most widely read, with a publishing tradition dating back to the latter half of the 19th century (Braun et al., 1989). Contributions are short thus inviting readership even by the busiest of scientists and the inclusion of a broad range of contributions encourages readership among a wide variety of disciplinary specialties. Both journals are genuinely cosmopolitan and publication of research work is open to authors from all countries (Stankus, 1999).
In 2003 Nature and Science published 859 and 926 articles, respectively. In that same year, papers from these two journals attracted a total of 343528 and 311593 citations, respectively (ISI, 2003). The fact that they are consistently among the top group of ten or so journals with the highest impact factors of a total of almost 6000 also speaks for the quality, visibility and significance of these unique titles for the advancement of scientific knowledge. For these same reasons scientists compete for publication in their pages, producing ever decreasing circles of competition, selection, and quality. As Stankus (2002) points out the world’s scientific elite recurringly publishes in a journal like Nature, enhancing their professional reputations and that of the journal itself. Such is the status of publishing articles in these two prestigious titles that a ranking weight of 20% was given to this activity by the Shanghai Jiao Tong University in their 2006 academic ranking of universities worldwide (IHE, 2006).
]]> The specific criteria for publication of scientific papers in Nature are two-fold, that they be of "outstanding scientific performance" and "reach a conclusion of interest to an interdisciplinary readership". Only about 10% of papers received are published; hence selection criteria are rigorous (Nature, 2006). In the case of Science, the information for authors state that "priority is given to papers that reveal novel concepts of broad interest" and that "competition for space in Science is keen" (Science, 2006). Of particular interest to both journals are controversial issues and high-risk papers. Both journals have a policy of publishing the best research papers, paying little heed to the symmetry of research topics in any particular issue, such that a new hot topic might dominate the pages of these journals before the natural subject balance is eventually restored. Decisions on submitted manuscripts are expedient and this coupled with the fact that papers are generally short, encourages authors, in a race for priority, to send their initial findings to Nature or Science, with a follow-up paper in the leading journals of the discipline to which the discovery belongs (Stankus, 1999). Consequently, papers published in both Nature and Science have an exceptionally wide impact, both among scientists and, frequently, among the general public, and are among the most highly cited research reported in the international literature.It can therefore be considered that papers published in these two journals reach a far broader audience, both within and outside the international scientific community, than those published in specialized journals. Taking into account that a high international profile is of particular importance for the advancement of science in peripheral nations, especially those whose main communication language is not English, publication in these two widely-read journals can be taken as one of the measures of the international visibility of the research performed by countries such as those from Latin America, as well as identifying research work that is at the forefront of its field.
Earlier studies on Latin American mainstream science have concentrated on the visibility of specific research areas (e.g., Collazo Reyes et al., 2004), certain institutions (e.g., Pereira-Friedich and Dos Santos-Rodríguez, 1998), defined aspects of science activity (e.g., Sancho et al., 2006) or the overall production in the mainstream literature (e.g., De Moya-Anegón and Herrero-Solano, 1999). In the present study, focus is placed on the characteristics of research published in the two most highly-cited and most highly-read mainstream journals and a description given of a simple, direct method for measuring a country’s high profile research production. This methodology was previously applied to 10 years of Mexican science published in Nature and Science (Del Río and Cortés, 2007).
Method
Taking four Spanish or Portuguese-speaking countries, Spain, Argentina, Brazil and Mexico, we analyzed all publications, regardless of type, in the multidisciplinary journals Nature and Science, from 1973 to 2005. Full records, including abstracts, references and citation data, were downloaded individually for each country from the Science Citation Index Expanded using the Web of Science, adopting the following strategy: source (Nature AND Science) AND address, either Spain, Argentina, Brazil or Mexico NOT New. Data were worked using software developed for bibliometric analysis at the Center for Energy Research (CIE) of the Universidad Autónoma de México.
Groups of records for each country were analyzed using the following variables: Total production and its distribution over time, patterns of international collaboration, most productive institutions, citation frequencies, and cited journals.
A text mining technique was used (Ortuno et al., 2002) to find the most relevant words occurring in the abstracts of papers of all four countries. The distance between a particular word occurring in the text of an abstract was compared to the standard deviation of all words in all abstracts. A normalized standard deviation higher than 1 indicates that the distribution of the word within a particular abstract is not random, allowing to determine which words or string of words can be considered relevant for that particular text (Cortés et al., 2006). The reasoning behind this assumption is that the standard deviation is an indicator analogous to enthropy and can sometimes play a role as a measure of order, or disorder (Reiss et al., 1986). The advantage of this particular technique is that it does not require a labor-intensive revision of individual words to extract the keywords from a text, but rather provides a ready-made list of the most frequently occurring words and strings of words whose distribution within a text is not random and, therefore, likely to be significant. Contributions such as letters, editorial material and book reviews, do not have abstracts and, therefore, are not included in the text mining analysis.
Results
The total numbers of papers in both journals found for the four countries were as follows: Spain, 696, Brazil, 411, Mexico, 227 and Argentina, 127. The relative percentages of papers in Nature and Science can be seen in Figure 1. Both Spain and Brazil published over 65% of their total papers in Nature, while the production of Mexico and Argentina were more evenly distributed between the two titles.
All four countries showed an increasing aggregate presence in the two journals over the 33 years analyzed, as can be seen in Figure 2. Argentina, Brazil and Mexico first got into double figures from 1995 onwards. Argentina reached its maximum level of publications (11) in 2001-2003 and again in 2005. Brazil published 36 papers in 1995 and since then has shown a steady but fluctuating increase in the annual production of papers, reaching an all time maximum of 55 in 2005, a figure comparable to the 58 papers produced by Spain in that same year. Mexico had 17 publications in both 2002 and 2003, with levels dropping slightly for 2004 and 2005. Spain showed a notable rise in publications from 1993 onwards, even though a steady rise was apparent from the mid 80s, coinciding with her entry into the European Union.
All four countries had similar percentages of papers published with international collaboration: Spain 74% (515), Mexico 74% (166), Argentina 72.5% (92) and Brazil 72.3% (297). Figure 3 shows the bilateral collaboration patterns of the four countries in the two titles, specifically with the USA, with England, and with France. Brazil, Mexico and Argentina show similar percentages of approximately 23-25% in bilateral collaborations with the US. Spain showed the lowest percentage with the US (13.2%) and a higher percentage of bilateral collaborations with countries other than the US, France or England, than did the three Latin American countries. In particular, Spain published 9 contributions together with Germany, but less than 5 with other nations. The three Latin American countries showed fewer than 3 bilateral collaborations with any other country outside the Latin American region, with the exception of four Brazilian papers published with Switzerland. Brazil published 8 papers in bilateral collaboration with Panama, but coauthorships with Spanish-speaking countries did not figure prominently in the bilateral collaborations of the other two Latin American countries.
The most visible institutions in all three Latin American countries (Table I) were by far the large national universities, which dominated the scene particularly in the case of Mexico, where the UNAM authored or coauthored 48% of the grand total of the country’s publications. National research councils occupied first place in the Spanish papers and second slot in those from Argentina.
When looking at the most frequently occurring authorships for Argentina, Brazil and Mexico, it is found that the same institutions publishing as sole authors are at the top of the list (Table II). Only with regard to Spain is there a change in the order, with the Autonomous University of Madrid in first position followed by the Spanish Higher Council for Scientific Research (CSIC). In the case of Brazil, the impetus given to research by state funding in the province of São Paulo could explain the commanding position of its main university. With regard to Mexico, the UNAM is known to be responsible for almost half of the country’s international research papers.
An analysis of the number of times the publications from each country were cited in subsequent years is given in Figure 4. No adjustment was made for the 30 year range of cited years. Over 18% of Mexican papers remained uncited; the corresponding figure for Argentina was 14%; for Brazil, 12% and for Spain, 13%. A quarter of the publications from the Argentine, 29% from Brazil, 22% from both Mexico and Spain received 1 to 9 citations. The most frequent citation levels greater than 9 were 30-49 for Argentina and Mexico, and 50-99 for both Brazil and Spain. The highest level of citation, between 200 and 3576, was achieved by almost 8% of papers from Spain and by 3-4% of those from Argentina, Brazil and Mexico.
]]>
The three most highly cited articles, receiving more than 2000 citations, are on the subject of the genome, one on the human genome which included an author from Spain (3576 citations), one on E. coli which included a Mexican author (2774 citations) and the other, on drosophila with a Spanish coauthor (2186 citations). The most highly cited publication in the group of Brazilian publications on carbon nanotubes received 1104 citations and that of Argentina on ecological systems, 762. Four Spanish articles obtained 1000-2000 citations.
An analysis of the journal titles most frequently cited in the reference lists of a group of papers gives an indication of the main subject areas present in these publications. In the present study, both Nature and Science are, perhaps not surprisingly, the most frequently cited journals, Nature more so than Science (Figure 5). The results show general agreement with the distribution patterns of the publications of the four countries between these two journals. Frequently cited titles were principally in the areas of physics, astronomy and astrophysics (principally Spain and to a lesser extent, Mexico), geophysics (predominantly Mexico but also Brazil) and in molecular biology (mainly Spain) and biochemistry. Papers from all four countries cited the multidisciplinary proceedings of the US National Academy of Sciences.
The most significant single words and phrases of 2-3 words found in the abstracts from the different countries are presented in Table III. For Argentina the words indicate the presence of papers in physics, biology, and in the geosciences, while the analysis of strings of two or three words revealed important occurrences of terms and phrases showing the concentration of papers in geosciences of the southern hemisphere. In the Brazilian papers, relevant phrases were few and revealed no specific research topics. However, findings suggest that leprosy is a big concern for the Brazilians, as could be the subject of forest fires. A general interest in astronomy is also apparent. In the case of Mexico, words that showed standard deviations >1 indicate three main research thrusts: astronomy, molecular biology, and geosciences. When looking at the phrases, related themes became apparent. Spanish research published in the two multidisciplinary journals was in the general areas of environmental sciences, biological sciences, and astronomy. However, no particular research theme emerged from the present analysis, suggesting that Spanish research in many branches of science is present in these mainstream journals.
Discussion
According to Stankus (1999), both Nature and Science look favorably on work that is sanctioned by prominent scientists via letters of endorsement. In this context, it would appear more difficult for less well-connected scientists, such as the great majority from developing countries, to get published in these prestigious journals. Nonetheless, the present study shows that at least a small minority of Latin American and Spanish scientists do realize what Stankus (2002) describes as every scientist’s dream since its launch in 1869, getting published in Nature. He also hazards a guess that less than 0.1% of the world’s scientists realize this goal and, then, just once in a lifetime.
The notable increase in publications in these two journals over recent years found in the present study is real in the sense that the numbers of papers published between 1981 and 2004 in both Science and Nature have remained relatively stable and if anything, have decreased slightly. An increased presence in these two multidisciplinary journals is consistent with a general increase in the number of papers published by Latin American countries in the mainstream scientific literature. Mexico, for instance, almost doubled its production in ISI journals in a period of less than 10 years, from 2,916 in 1995 to 5,783 in 2003 (CONACYT, 2004). In the Big Science research area of high energy physics: particles and fields, Mexican production in the SLAC-SPIRES-HEP specialized database, increased almost tenfold from 31 papers in 1980 to 292 in the year 2000 (Collazo Reyes AND Luna Morales, 2002).
Notwithstanding, our findings for the four countries also show that relatively small proportions of publications are authored by institutions from only one country. In accordance with world trends, the figure for Mexico of 74% of papers in international collaboration found in the present study is much greater than the 34% found for the 1990 mainstream Mexican papers (Russell, 1995). The impetus given to science by the incorporation of Spain into the European Union is also apparent from the present results, as is the high level of intercountry collaboration in the Spanish group of papers.
]]> A comparative study of Nature and Science using randomly selected articles or reports for 1981, 1982 and 1983, showed the presence of authors from the USA in almost 90% of papers in Science while the UK, West Germany, Japan, Canada and other industrialised nations were well-represented in Nature (Kaneiwa et al., 1988). Mexico was the only Spanish-speaking country in the sample of papers by the most frequently authoring countries in Science, and none were found in Nature. Over 50% of the papers, in both journals, were either in the area of medicine or biology, traditional research areas for Latin American countries. The authors conclude that, in general terms, Nature has a stronger international character than Science. Our results concur with this conclusion, given that in the case of the four Spanish speaking countries selected for the present analysis, there was a tendency towards a higher percentage of publications in Nature than in Science.The levels of uncited papers from both journals found for the four countries in the present study are higher than those previously reported for papers published in Nature as a whole. A study on the citation patterns from 1965-1972 of a small sample of 327 papers published in Nature during 1965 (Ghosh, 1975) found that 7.39% were never cited in the eight-year period while 10.4% remained uncited during the first five years after publication. A smaller percentage (0.9) of longer papers or leading articles remained uncited as compared to the percentage (6.4) of short papers or letters.
A keyword analysis of reports published in Nature from 1971-1998 (Arkhipov, 1999) showed that the approximate percentages of papers in the main scientific fields were: Biology and Medicine, 50%; Physics and Astronomy, 20%; both Earth Sciences and Chemistry, 12%; Human Society Sciences, 4%; and both in Mathematics and Mechanics, and in High Technology, 1%. Although Nature has a bias towards the life sciences, in keeping with the high quality general science journals, nonetheless, it remains open to work in the physical sciences, particularly in astrophysics, geophysics, and high energy physics but with few in chemistry and mathematics (Stankus, 2002). The results of the present study also show a predominance of papers in physics, astronomy, geosciences and biology in keeping with the general pattern of the two journals.
When we compare the number of publications in the two journals with the expenditure in R&D as a percentage of GDP for the year 2000 for the four countries: Spain, 0.94%; Brazil, 1.04%; Mexico, 0.37%; and Argentina, 0.44% (Albornoz, 2004), we find that only Mexico approaches Spain with regard to "value for money". If Mexico had invested the same percentage of its GDP in R&D as Spain did, a production of approximately 580 papers could have been "expected", compared to the 227 that were actually published.
The present study demonstrates the usefulness of the software employed for analyzing publications in the two influential, multidisciplinary journals, Nature and Science, to determine a country’s high international scientific profile. Using this method it was possible to identify trends in publications, collaborations and citations, as well as prominent institutions and actors, and main research thrusts. While the presence of Latin American and Spanish papers is important in mainstream journals as a whole, publication in the two multidisciplinary journals Nature and Science gives another dimension to the concept of international visibility.
Acknowledgements
The authors wish to acknowledge Yoscelina Hernández García for the literature searches and the partial financing from the Mexican Consejo Nacional de Ciencia y Tecnología and the Government of the State of Morelos Project FOMIX 9250.
References
1. Albornoz M (Coord.) (2004) El Estado de la Ciencia. Principales Indicadores de Ciencia y Tecnología Iberoamericanos/Interamericanos 2003. RICYT / CYTED / OEA / REDES. Buenos Aires, Argentina. 286 pp. [ Links ]
2. Arkhipov DB (1999) Scientometric analysis of Nature, the journal. Scientometrics 46: 51-72. [ Links ]
3. Braun T, Glanzel W, Schubert A. (1989) National publication patterns and citation impact in the multidisciplinary journals Nature and Science. Scientometrics 17: 11-14. [ Links ]
4. Cetto AM, Alonso-Gamboa O (1998) Scientific periodicals in Latin America and the Caribbean: A global perspective. Interciencia 23: 84-93. [ Links ]
5. Collazo Reyes F, Luna Morales ME (2002) Física mexicana de partículas elementales: Organización, Producción Científica y Crecimiento. Interciencia 27: 347-353. [ Links ]
6. Collazo Reyes F, Luna Morales ME, Russell JM (2004) Publication and citation patterns of the Mexican contribution to a Big Science discipline: elementary particle physics. Scientometrics 60: 130-143. [ Links ]
7. CONACYT (2004) Indicadores de actividades científicas y tecnológicas. www.siicyt.gob.mx (Retrieved 01/20/2005). [ Links ]
8. Cortés HD, del Río JA, García EO, Robles M (2006) Web application to profiling scientific institutions through citation mining. Enformatika 14: 419-423. [ Links ]
9. De Moya-Anegón F, Herrero-Solano V (1999) Science in America Latina: A comparison of bibliometric and scientific-technical indicators. Scientometrics 46: 299-320. [ Links ]
10. Del Río JA, Cortés HD (2007) La ciencia mexicana en las revistas Nature and Science: la última década. Ciencia (Revista de la AcademiaMexicana de Ciencias) 58: 61-68. [ Links ]
11. Ghosh JS (1975) Uncitedness of articles in Nature, a multidisciplinary scientific journal. Inf. Proc. Manag. 11: 165-169. [ Links ]
12. Gibbs WW (1995) Lost science in the Third World. Scientific Am. 273: 76-83. [ Links ]
13. IHE (2006) Academic Ranking of World Universities 2006. Institute for Higher Education. Shanghai Jiao Tong University. http://ed.sjtu.edu.cn/rank/2006/ARWU2006Methodology.htm. (Retrieved 02/08/2007). [ Links ]
14. ISI (2003) 2003 Journal Citation Reports. Science Edition. Institute for Scientific Information www.dgbiblio.unam.mx/bases.html. (Retrieved 12/10/2004). [ Links ]
15. Kaneiwa K, Adachi J, Aoki M, Masuda T, Midorikawa N, Tanimura A, Yamazaki S (1988) A comparison between the journals Nature and Science. Scientometrics 13: 125-133. [ Links ]
16. Nature (2006) Getting published in Nature. www.nature.com/nature/submit/get_published/ index.html. (Retrieved 12/23/2006) [ Links ]
17. Ortuno M, Carpena P, Bernaola-Galvan P, Muñoz E, Somoza AM (2002) Keyword detection in natural languages and DNA. Europhys. Lett. 57: 759-764. [ Links ]
18. Pereira-Friedich M, Dos Santos-Rodriguez P (1998) Looking at science in Brazilian universities: The case of the Instituto de Biofísica Carlos Chagas Filho. Scientometrics 42: 247-258. [ Links ]
19. Reiss H, Hammerich AD, Montroll EW (1986) Thermodynamics of non-physical systems: formalism and an example (single-lane traffic). J. Stat. Phys. 42: 647-687. [ Links ]
20. Russell JM (1995) The increasing role of international cooperation in science and technology research in Mexico. Scientometrics 34: 45-61. [ Links ]
21. Sancho R, Morillo F, de Felippo D, GÓmez I, FernÁndez MT (2006) Indicadores de colaboración científica inter-centros en los países de América Latina. Interciencia 31: 284-292. [ Links ]
22. Science (2006) Information for authors. www.sciencemag.org/about/authors/. (Retrieved 12/23/2006). [ Links ]
23. Stankus T (1999) Why reading Science has been second Nature for over a hundred years. Technicalities 19: 4-6. [ Links ]
24. Stankus T (2002) Nature: The next generation. Technicalities 22: 4-7. [ Links ] ]]>