Science in Africa : A bibliometric approach.
N. Narvaez-Berthelemot, J. M. Russell 
, R. Arvanitis, R. Waast, J. Gaillard 
Publié dans Scientometrics 54 (2) : 229-241, 2002
- Voir aussi : Arvanitis, R., Roland Waast et Jacques Gaillard (2000). "Science in Africa : A bibliometric panorama using PASCAL database." Scientometrics 47 (3) : 457-473.
- écouter aussi l’émission sur France Inter (Février 2007)
What is the state of art with respect to science in Africa ? Is scientific research in the continent focussed on issues related to development ? What is the role of basic research ? Is scientific capability in decline or is it taking on new directions ?
Regional scientific activity is a popular subject for bibliometric analysis. Nonetheless, the African region has received little attention from scientometric perspective, perhaps due to the backward nature of science in a continent that encompasses some of the poorest countries in the world. The globalisation of science has resulted, among other things, in a generalised increment in international scientific collaboration making necessary updated information on scientific co-operation, co-authorship and influence. This is important not only for the scientifically advanced countries but, perhaps, more so in the case of the developing world whose contribution to scientific achievement and legacy does not receive the same level of attention and reflection.
The purpose of this paper is to present an overview of the scientific output of the different countries that make up the African continent and to compare their evolution with the rest of the world. Data are also presented and analysed on their affinity with specific fields, institutional participation and production in international collaboration.
Two multi-disciplinary databases in the sciences were used in the present study : the French PASCAL database for the main analysis and the Science Citation Index (SCI), produced in the United States by the Institute for Scientific Information, for comparative purposes. The PASCAL database produced by the Institut National de Information Scientifique et Technique (INIST), is the only multidisciplinary database edited in Europe, covering essential world literature in the sciences, technology and medicine. Priority is given to French and European literature representing 40% of indexed documents. In 1994 the database covered 4,356 scientific journals edited in 150 countries. Twelve of these are African nations.
Our period of study covered the seven years from 1991 to 1997. The output of 49 African countries was recorded in the PASCAL database, but due to the scant publication of some, detailed bibliometric analysis was carried out only with respect to the high producers, such as South Africa, Egypt, Nigeria, Tunisia, Morocco, Kenya, Algeria, Cameroon, Zimbabwe, Senegal, Cote d’Ivoire, Tanzania, Ethiopia, Ghana, and Sudan. Only journal articles were taken into consideration. Main fields and sub fields were correlated with the countries to establish research affinities. The journal classification scheme developed by Computer Horizons Inc. (CHI), New Jersey, USA, was used to denote fields and sub fields for the SCI data.
The data presented compares the output of the different countries, the participation of the principal institutions, the distribution of output by fields of research, the evolution of research over the period analysed, and the international collaboration.
The production of the African continent in the PASCAL database in all fields corresponded to 4% of the European production at the beginning of 1991. By the end of the year 1997 this had declined to 3%.
In 1996 the principal producer, South Africa, had a publication volume comparable to Greece, while that of the second most productive African country, Egypt, was similar to that of Portugal. No other African countries reached publication levels comparable to European countries. This comparison brings to light the remarkable scientific growth of these two "small" European countries as a result of their affiliation to the European Union. Our first table analyses the 42,227 articles from 1991 to 1997 for 49 African countries, by total production and by country rankings for 1991 and 1997. Output for the individual countries is compared with the population statistics for 1995 and the GNP/inhabitant.
Production trends vary depending on the database used, especially over long periods. Nonetheless, in our study we found a consensus that for the five years from 1991to 1996, the African continent compared to Europe or to the rest of the world, lost 15 to 20% of its relative capacity to contribute to scientific advancement. The number of articles written by African scientists in the PASCAL database diminished 5% during this period. In contrast SCI registered a slight growth for the same period (+9% for Africa in general and +13% for the high producing countries). Nonetheless, this represents an overall loss in their percentage of the total world production, since the growth in output of the European countries was greater (+35%), as was total world production (+27%). The 15 most productive African countries had 32.5% more articles in SCI than in PASCAL for the period 1991-1997.
The recent production of three different regions : North Africa (except Libya), Central Africa and South Africa, is described. Scientific capacities are unequally distributed in the continent and are not necessarily related to the economic status or populations of the individual countries. This will be described in the full paper divided into five hierarchical groups.
The trends for the 15 highest producing countries have differed between decades. While regular growth was established in the 70s and the 80s in certain key countries in the continent, the 90s appeared to change the situation. We can summarise the events as follows. The difficulty experienced by the two “giants” of the continent, Egypt and South Africa to maintain their position ; the upsurge of the Maghreb countries (Algeria, Morocco and Tunisia) ; the reduction in the publication output of Nigeria ; and the emergence of Cameroon, Tanzania and Senegal.
Five relatively homogenous geo-linguistics zones will be detailed : North Africa, Republic of South Africa, the rest of Anglophone Africa, the rest of Francophone Africa and the remaining countries (mainly Portuguese speaking). This division is useful because it differentiates the regions by language, by scientific institutions and traditions, by the strength of their co-operation links, by scientific resources, and by their distinct development problems. The data for the five zones are related to fields of specialisation resulting in distinctive profiles in fields such as agricultural sciences, health, and other sciences. The linguistic and regional bias of the two databases used in the present study plays a role in presenting contrasting pictures. For instance, the weight given to tropical agricultural in the PASCAL database presents a panorama closer to the realities of North and Central Africa, while SCI’s superior coverage of subjects in clinical medicine and chemistry gives a more accurate picture for East and South Africa.
Other tables will show the total production and annual growth rates of articles in 10 fields, the comparison between the PASCAL and SCI data for 1991 and 1996, as well as data on the output by country and by discipline. Seventy sub fields are analysed. In this way we hope to identify the relative strengths in different fields of the African countries.
In developing countries, participation of various institutions in a research project is indicative not only of internal scientific collaboration but also shows the relative scientific capabilities of individual national institutions. The so-called coverage index for developing countries has already been used to analysis institutional participation in Latin America. In the present study countries like South Africa, Sudan and Kenya, for example, were found to have a higher number of participating institutions than other African countries.
The international collaboration of the 15 most productive countries in Africa was also studied. Approximately 38% of articles in SCI proved to be in international collaboration. The levels of international collaboration vary from one field to another, depending of the country.
United States is the foreign country collaborating most with these African countries (24% of articles), followed by France with 21%. However, when analysing international collaboration by regions, Europe is top of the list with 60% followed by the Americas with 29%. The full paper will also describe intra-African collaboration.
All bibliometric databases suffer from certain bias. In spite of the inconveniences of using the French database PASCAL, conceived as a bibliographic retrieval database and not for bibliometric purposes, it is useful in regard to the francophone countries. It was useful also for example because of its coverage of certain continuing education journal titles for medical practitioners as well as network bulletins in agriculture in which results from Africa are highlighted. SCI, on the other hand, is focussed more on the fundamental aspects of applied science. The tendency towards the “privatisation” of science (contract research, the race to get out proposals) in the African continent is detrimental to the generation of fundamental research and lessens its contribution to mainstream scientific knowledge. Such is the case of global interest in the environmental sciences, especially in the minor African countries.
Specific results and percentages vary from one base to another. However, by using two different databases, analysis allows the identification of the most consistently influential locations and “actors” of research.