Scientometrics, VoL 5. No. 3 (1983) 163-175

INSTITUTIONAL SECTORS OF 'MAINSTREAM' SCIENCE PRODUCTION IN SUBSAHARAN AFRICA, 1970-1979: A QUANTITATIVE ANALYSIS C. H. DAVIS

Institute d'histoire et de sociopolitique des sciences, Universit~ de Montrdal (Canada) (Received October 14, 1982)

The place of production of 'mainstream' scientific authors in Subsaharan Africa is examined in terms of institutional sectors for the period 1970 through 1979. Patterns of production of 'mainstream' scientific literature and the citation visibility of this literature are also examined, for a shorter period of time, in terms of institutional sectors. It is shown that the university and public sectors predominate in the production of 'mainstream' authors. These same sectors also assure more consistent intra-African visibility of research results than do other sectors. However, the growth of the university and public sectors appears to have slowed considerably since the mid-1970s. Research conducted within regional and subregional cooperative organizations declined dramatically during the decade. A growing emphasis on external interventions under multilateral (rather than bilateral) auspices is noted.

Introduction Quantitative data on scientific activity in Subsaharan Africa are rare, and this is particularly true of time-series of data. It is often not possible to construct diachronic statistical pictures of scientific activity in Africa using conventional scientometric data such as those published by international organizations. The use of counts of scientific authors, publications or citations provides a partial, though biased, solution to this problem. The Science Citation lndex (SCI) and Who is Publishing in Science (WlPIS, superseded by the Current Bibliographical Directory o f the Arts and Sciences) have provided convenient data bases for a n u m b e r of studies of aspects of the participation of Subsaharan Africa in the production and circulation of 'mainstream' scientific infoimation I . The present study uses SCI and WlPIS as data bases to develop a quantitative analysis of patterns and trends in the production of 'mainstream' science by institutional sector in 36 Subsaharan African countries for the period 1970 through 1979. While national differences in

Scientometrics 5 (1983)

163

c. H. DAVIS: SCIENCE IN SUBSAHARANAFRICA 'mainstream' science production are readily calculable using the annual national listings provided by WlPIS, a reclassification of the data by institutional sector is of interest because it makes possible art intra-regional comparative analysis of those African science producing institutions that are visible via the SCI-WlPIS information system. We know that African universities predominate in the production of 'mainstream' science in Africa, but to what extent do they predominate? What proportion of African '~ainstream' science is produced in other kinds of institutions? How do patterns of institutional production vary among African states, how has the production of each sector varied throughout the 1970s, what fields of science are produced by each sector, and what differences may be noted in the international and intra-African visibility of the various sectors? The present study provides some answers to these questions. In so doing, the quantitative picture we construct confirms certain of our intuitive expectations about recent institutional trends in African scientific development: the rise and levelling off of university research, the decline in regional and subregional cooperative research, and the relatively important role played by organizations of non-African origin, with multilateral research organizations tending to overshadow bilateral research implantations in Africa. Methods and limitations

For the first part of this study, all 'mainstream' scientific authors listed for 36 Black African countries 2 in WIPIS from 1970 to 1979 inclusive were classified according to sector of institutional affiliation. The sectoral classificatory scheme used here was designed to take into account groups of institutions of historical and contemporary importance to the support of scientific research in Africa; the classificatory schema is not meant to suggest distinctions of ontological significance. 3 The institutional sectors are as follows: A. The university sector, including university and teaching hospitals, training centers, and all other schools. B. The public sector, which includes all governmental research organizations, ministries, parastatal organizations, and hospitals not directly affiliated with teaching establishments. C. The global international organizations sector. In the case at hand, institutions in this sector are primarily the World Health Organization, the Food and Agriculture Organization, the United Nations Economic Commission for Africa, and diverse projects identified by articles' authors as being affiliated with the above or other universal intergovernmental organizations. 164

Scientometrics 5 (1983)

c. H. DAVIS:SCIENCEIN SUBSAHARANAFRICA D. The 'hybrid' internationalresearch organizations sector. This category is designed to permit a distinction between the organizations of the United Nations family per se and organizations which, although sponsored in part by one or more U.N. bodies, are autonomous. For Africa, institutions in this sector include three international agricultural research centers under the auspices of the Consultative Group on International Agricultural Research (CGIAR): the International Institute of Tropical Agriculture (IITA) in Nigeria, the International Livestock Center for Africa (IICA) in Ethiopia, and the International Laboratory for Research on Animal Diseases (ILRAD) in Kenya. We have also included in this sector the International Centre of Insect Physiology and Ecology (ICIPE) in Kenya which, although an indigenous African research institution, entertained in the 1970s a close association with the CGIAR. E. The African regional and subregional intergovernmental research institutions sector, which includes, notably, the several research organizations of the former East African Community, the West Africa Rice Development Association (WARDA), the Institut du Sahel, the Organisation de Coordination et de Coop6ration pour la lutte contre les Grandes Epid6mies (OCCGE), and the Organization of African Unity. F. The French overseas research sector, which is composed primarily of the Office de la Recherche Scientifique et Technique Outre-Mer (ORSTOM) and the research institutes and services under the auspices of the Groupment d'Etudes et de Recherches pour le D6veloppment de l'Agronomie Tropicale (GERDAT). Both ORSTOM and the GERDAT institutes are specialized tropical research organizations maintained largely by the French government for use in bilateral research cooperation with client states. G. All other bilaterally-maintained institutions, including Britain's Medical Research Council laboratories in the Gambia, all embassies, all philanthropic foundations, and all non-African governmental affiliations with addresses in Africa. H. The commercial organizations sector, consisting for the most part of research laboratories in the agricultural and mining industries. I. Residual category: religious missions, unidentifiable organizations, unaffiliated authors. For the second part of this study, a population of 'mainstream' scientific articles produced in the 36 countries was assembled, consisting of all articles which could be found in the 1975 SCI for authors in these countries listed in WIPIS for 1975. Citations to these articles were collected from the SCI for the period 1976-1979 inclusive, and the geographical location of each citing author was retrieved from WIPIS4. Before discussing the results of this research, a word is in order about what 'mainstream' science is, and what measurement of it is not likely to demonstrate. "Mainstream' science is usually def'med as the world's most heavily utilized sciences, where international visibility and citation frequency are partial determinants of the choice of the set of periodicals defining the parameters of 'mainstream' science. $cientometric s 5 {1983}

165

C. H. DAVIS: SCIENCE IN SUBSAHARANAFRICA There is serious question as to the extent to which measurement of 'mainstream' science produced in Africa can be used as an index of the general development of research in Africa. It is of course possible to find strong correlation between the 'mainstream' science production of African countries and a numbe of standard indicators of industrialization (consumption of energy, national production of wealth, and so forth), just as it is possible to establish such correlations for most of the world's countries. Such empirical correlations suggest that international comparisons of national scientific "sizes' using data on 'mainstream' scientific production are feasible in some respects. On the other hand, the vast majority of African scientific publications are not indexed by the Science Citation Index. The distinction between 'mainstream' science and (presumably) 'non-mainstream' science, a distinction operationalized by the choice of scientific periodicals indexed by international scientific information services, would seem to marginalize a good deal of respectable science produced in developing areas. For instance, it has been found that Nigerian entomological articles published in foreign journals received 1,74 times more citations than articles published in Nigerian or West African journals 6. A study of three major agricultural research information services (AGRIS, AGRICOLA, and CAB) found that AGRIS, which relies on nationallycollected data rather than on criteria of journal use for informational input, provides by far the best developing country coverage of the three services 7. A study of social science research produced at the University of Ibadan has shown that the bulk of this research is internationally invisible because the most important national social science publishing outlets are unindexed by international information services s. Examples such as these could easily be multiplied. Our point is that to equate 'non-mainstream' research with inferior research is to sanction criteria of scientific achievement which are not entirely fair to newly developing scientific communities. As Jean-Louis Boursin has pointed out, that the Institute for Scientific Information "has succeeded in persuading the scientists of the world that it is the infallible barometer of their research level is a feat that has been surpassed only by the Swedish Academy thard~ to the bequest of Alfred Nobel ''9. In the present article no prejudice is intended as to. the quality of 'invisible' scientific literature in Subsaharan Africa.

Institutional Sectors of 'Mainstream' Scientific Activity Figure 1 shows the percentage of each institutional sector in the 1970 to 1979 cumulative annual total of 'mainstream' scientific authors in 36 subsaharan African countries. Nearly two-thirds of these authors were to be found in the university sector, while another sixth were to be found in the public sector. Sectors of undoubtable external origin (United Nations, hybrid international, French, and other bilateral) ac166

Scientometrics 5 (1983)

C. H. DAVIS: SCIENCEtN SUBSAHARANAFRICA / ~

/

~

\

/ F (6.7'1,) / ~ C(1.8"/.) 16.5%)/ ' ~ f " / / D ( 1 . 9 * / . ) / / , , , ~ % c (3.o'~.)

~

Fig. 1. Sectoral Composition of the (Slraulative Annual Production of 'Mainstream"Scientific Authors of 36 Subs~aran African Countries, 1970-1979. A: Universities and other teaching establishments, B: Public sector, C: United Nations Organizations, D: 'Hybrid' international researeh organizations, E: African Regional and Subregional research organizations, F: French overseasresearch organizations, G: Other bilateral organizations, H: commercial organizations, I: miscellaneous and unidentified institutions N= 12481

counted for about one-eighth o f Africa's scientific authors in the 1970s. About three percent of all authors were from regional or subregional cooperative research organizations, and about two percent were from commercial organizations 1o. Ratios of expectation are used in Table 1 to describe the relative strength of each sector in the total cumulative production of 'mainsteram' authors in each state or group of states. A ratio of expectation is the ratio between observed and expected value, where it is assumed that each value in a matrix is determined by the proportion of its row and column to the whole. The ratio of expectation is then the observed divided by the expected value 1i. The primary advantage to the presentation of matrix data in this form is that anomalies to the assumed distribution of values are immediately apparent. A disadvantage is that small variations in value can give rise to gross singularities when rows or column contain disproportionately large or small values, as is the case here, since the university column contains more values than all other columntogether. Thus the university sector is the implicit norm. It may be seen in Table 1 that there are few extremely high or low ratios of expectation in national production of 'mainstream' authors in the university and public sectors. About a third more than expected of Nigeria's production of mainstream authors came from the university sector, and in Kenya about a third less. In general, former French colonies in Africa produce fewer authors in the university sector than expected. The remaining sectors are less evenly distributed among African states. The concentration of 'mainstream' authors from regional and subregional research organizations in states of the former East African Community is evident, as is the concentration of authors from French science assistance organizations in former French colonies, $cientometrics 5 (1983j

167

C. H. DAVIS: SCIENCE IN SUBSAHARAN AFRICA

r O

0.~

Eo ~ ~

9

.

..~.~.

0

o

o~ 0e-,

.~ ~

o ~

e

o.

=g O

0

I ~'~

~

Cq *-~ r

~

~"

e~ c~ 0o 9

o

.

~

O

g~

e-, 0

94 c5 ,4 c~ c5 ~

c~.4 .,,.,.~

~o

e c e 4 ~

~
m~ e.

168

Scientometrics 5 (1983)

C. H. DAVIS: SCIENCE IN SUBSAHARAN AFRICA n o t a b l y the I v o r y Coast. There are three patterns m a i n f e s t e d o f u n e x p e c t e d l y high p r o p o r t i o n s o f external-sector research in total national o u t p u t o f ' m a i n s t r e a m ' authors: a c o s m o p o l i t a n pattern comprising high ratios o f e x p e c t a t i o n o f research by international organizations ( E t h i o p i a and Kenya), a bilateral pattern based on research by F r e n c h organizations ( t h e Ivory Coast, Senegal), and a m i x e d pattern o f bilateral and multilateral research (the smaller f r a n c o p h o n e countries)~ Performance

of Institutional

Sectors over Time

P r o d u c t i o n o f "mainstream" authors in Subsaharan Africa increased at an overall annual rate o f a b o u t nine percent in the 1970s (see Table 2). The increase was loglinear (r = 0.96) and was due to regular o f fairly regular increases in p r o d u c t i o n Table 2 Growth Rates of 'Mainstream' Author Production in Subsaharan Africa, 1970-1979, by Sector. 70-74

75 - 7 9

70-79

slope r

0.14 0.98

0.04 0.98

0.09 0.96

university sector slope N = 8108 r

0.17 0.99

0.05 0.96

0.11 0.96

0.17 0.92

0.02 0.29

0.06 0.72

United Nations organizations slope 0.04 N = 230 r 0.44

0.14 0.86

0.10 0.90

"Hybrid' international organizations slope 0.36 N = 239 r 0.50

0.27 0.86

0.37 0.90

African regional and subregional organizations slope - 0.04 - 0.09 N = 380 r - 0.52 - 0.81

- 0.07 - 0.89

French overseas research organizations slope 0.04 - 0.02 N = 833 r 0.57 - 0.38

0.02 0.55

other bilateral organizations slope N = 225 r

0.12 0.85

- 0.03 - 0.37

0.04 0.52

Commercial organizations slope N = 221 r

0.05 0.48

0.06 0.82

0.02 0.41

All sectors N = 12481

public sector N = 2064

Scientometrics 5 (1983)

slope r

169

C. H. DAVIS:SCIENCEIN SUBSAHARANAFRICA throughout the decade by about one-third of the countries in the area accounting for about two-thirds of Africa's annual production of 'mainstream' authors. Among these countries were the largest science producers of Africa: Nigeria, Kenya, the Ivory Coast, Zambia, and Tanzania. Three countries (Chad, Uganda, and the Malgasy Republic) exhibited regular decline in production during the decade. The predominance of Nigeria and Kenya increased throughout the period: their combined production of "mainstream' authors rose from 42 percent of the Subsaharan African total in 1970 to 59 percent in 1979, Slightly more than 75 percent of the cumulative total of authors was produced in former British colonies, and about 15 percent in former French colonies. These figures may reflect a linguistic bias in WlPIS/SCI journal selection, but they also probably reflect the more rapid university development in anglophone areas of Africa. Table 2 presents data on the behavior of the various science-producing institutional sectors of Africa during the 1970s: data are provided about the slopes of the best lines of regression of graphed natural logarithms of annual production of 'mainstream' authors for the periods 1970-1974, 1975-1979, and 1970-1979. Coefficients of correlation are also provided. It may be seen that the growth of the two largest sectors (university and public) was much more rapid during the first half of the decade than during the second half. Thus the two sectors most dependent on African policies and resources show signs of saturation. Furthermore, both bilateral sectors show negative growth for the second half of the decade. The sector dependent on collective African policies and resources (African regional and subregional research organizations) shows an overall regular decrease in numbers of annual 'mainstream' authors throughout the decade. This is a measure of the declining fortune of regional and subregional intergovernmental cooperative research institutions in Africa. It is to be noted that the decline precedes the collapse of the East African Community in 1977, but is accelerated byit. The most spectacular increases in the numbers of 'mainstream' authors have come from international organizations, anct mis especially in the second half of the decade. The 'hybrid' international agricultural research centers, the oldest of which (IITA in Nigeria) dates only from 1968, show an overall annual increase of about 37 percent for the decade. Scientific Fields of 'Mainstream' Activity by Sector Table 3 presents expectation ratios of production of scientific articles and citations to these articles for 1975 and 1976-1979 respectively. Percentages of production of 'mainstream' articles per institutional sector correspond only approximately with percentages of production of 'mainstream' authors per sector because of varying rates of productivity of articles by scientists among sectors, and because only a portion of 170

Scientometrics 5 f1983)

C. H. DAVIS: SCIENCE IN S U B S A H A R A N AFRICA

0

r.-.

11 II ZZ

oG

,-~.~

.,...

,~- r

e-. o

e4~,

O

o O

.fi

e.,

O

Z~

e4~

O

G

GG

,~.G e~e-i

GG

OG

0

"~.o

~.~

~

dd

dd

~d

dd

~

~o~ ~ "~ o ~'~

,-

e'L

Scientometrics 5 (1983)

171

C. H. DAVIS: SCIENCE IN SUBSAHARAN AFRICA

oS~ I I

II

~~

~ I ~ ~ _~ _ . _ _I_ _ _ _

~

~

.

. ~

~

. ~

~

9

-~.~

E ......

)

~o~ " I 0

~0

~

~.~.~

I 9

.

o

I .mg ~

~.~ ~

0

~~ z.~ ._~ .~. ~0 "~ .a 2 0

~ .~ 0 ~ 0

172

Scientometrics 5 (1983)

C. H. DAVIS: SCIENCE IN SUBSAHARAN AFRICA

'mainstream' authors listed in WIPIS produced articles in journals indexed by the SCI. In Tables 3 and 4 a residual category of miscellaneous publications not belonging to any of the five principal scientific fields, was eliminated. 12 It is immediately apparent from Table 3 that the university sector is the only physical science-producing sector in Africa, and that ratios of citation visibility of articles in all fields of science produced by the university sector correspond closely to the ratios of article production. It is especially in the smaller sectors that may be seen disproportions between production and visibility. In the public sector, institutions produced articles for the most part in the fields of agriculture, biology and medicine, but it is to be noted that African public-sector biology is disproportionately invisible, while public-sector environmental science literature is disproportionately visible. While the French bilateral and African regional sectors produce articles in the same fields, patterns of production and visibility are quite different. African regional and subregional research organizations produce research in agriculture and the environmental sciences that is disproportionately invisible. On the other hand, the biological information produced by these organizations is disproportionately visible. French overseas research organizations are disproportionately visible in agricultural research and very active in' environmental research (a reflection of the oceanographic research conducted by ORSTOM). In Table 4, expectation ratios of citation visibility by institutional sector and consumer state are presented. Of interest are the almost uniformly "normal' rates of expectation of citation of articles produced in the university sector. The only consumers citing university-produced 'mainstream' research less frequently than expected are the African francophone countries (excluding national self-citations), for whom African university research is only half as visible as expected. The implication is that non-national university research is underutilized in francophone Africa. On the other hand, public-sector research is cited more frequently than expected in Africa, and this is especially true of non-national public sector research. The very high ratio of expectation of UN research visibility in francophone Africa is due to citation among researchers in UN-sponsored medical research projects. "Hybrid' international research organizations are strongly cited in their host countries and are fairly visible in anglophone Africa, but appear to have been invisible in francophone Africa in the 1970s. These institutions have, however, greater than expected visibility in tropical areas outside Africa (Latin America, Australia, Asia). Another surprise concerns the visibility of 'mainstream' research produced in African regional and subregional organizations: while expected visibility obtains in aggregated anglophone Africa, visibility is only half that expected in host coun$cientometrics 5 (1983) 3

173

C. H. DAVIS: SCIENCE IN SUBSAHARANAFRICA tries and nil in francophone African countries. Research emanating from this sector appears to have been particularly visible in Latin American and the United Kingdom. As for research produced in the French overseas research institutions, it is invisible in angiophone Africa but is more visible than expected in host countries and in other francphone areas of Africa. R e , a r c h produced in nonFrench bilateral organizations is less visible than expected everywhere in Black Africa. Concluding Remarks Despite the previously-mentioned biases in the data bases used in this study, the results confirm certain intuitive expectations about the institutional loci of ',mainstream' science production in Subsaharan African during the past decade. The data solidly support the inference that African universities play a primordial role in the production of 'mainstream' research, University-produced articles are cited with expected frequency all over the world (given the definition of expectation that we have used), with the exception of the francophone African areas noted earlier. National public-sector insti. tutions, while not enjoing as wide an international visiblity as African universities, produce research with unexpectedly high intra-African visibility. All other institutional sectors, with the qualified exception of non.French bilateral research, display patterns of visibility which are deeply marked by the' linguistic cleavages between a r e a s in which English or French is the usual international language of science. Of all institutional sectors, it has been the university and public sectors which appear to bridge intra-African distances with the most consistency. Patterns of production 'mainstream' science by the university and public sectors indicate a considerably decrease in rate of growth in the second half of the 1970s. Are we here faced with a methodological artefact? The development of unindexed publishing outlets could explain the apparent decline in the rate of growth of African production of 'mainstream' science in the two largest institutional sectors. An alternative explanation might consider the diminishing resources which are available for research in many African states with the worsening of the world economy.

Gratitude is expressed to Paule Laberge for assistance in the collection of the data used in this study.

174

Scientometrics 5 (1983J

C. H. DAVIS: SCIENCE IN SUBSAHARAN AFRICA

References 1. For instance, E. G. KOVACH, Country Trends in Scientific Productivity, Who Is Publishing In Science 1978, 33-40; J. D. FRAME, Francis NARIN and M. P. CARPENTER, The Distribution of World Science, Social Studies o f Science 7 (1977) 501-516; I. ADAMSON, The Size of Science in Old Nigerian Uaiversities, Scientometrics 3 (1981) 317-324; and Y. RABKIN et al, Citation Visibility of Africa's Science, Social Studies o f Science 9 (1979) 499-506. 2. These countries, in descending order of scientific !size', are Nigeria, Kenya, Ghana, Uganda, Zambia, Tanzania, Ivory Coast, Senegal, Ethiopia, Cameroon, Zaire, Malagasy Republic, Sierra Leone, Malawi, Botswana, Chad, Lesotho, Upper Volta, Mozambique, the Gambia, Togo, Mauritius, Congo, Liberia, Mali, Benin, Rwanda, Central African Republic, Niger, Gabon, Swaziland, Angola, Mauritania, Somalia, Burundi, and Guinea. 3. For purposes of classification of institutions the following texts were helpful: The World o f Learning Europa, London, 1978; Mary and Chen CHIMUTENGWENDE, Eds, Guide to World Science, v. 19: Central Africa, Hodgson, Britain, 1975~ and UNESCO, National Science Policies in Africa Paris, UNESCO, 1974. 4. Self-citations were eliminated from the population of citing authors. 5. This def'mition is taken from FRAME et al, op. cit. p. 502. 6. S . M . LAWANI, Citation Analysis and the Quality of Scientific Productivity, Bioscience 27 (1977) 31. 7. J. E. WOOLSTON, Information in the Service of Development, paper presented at the Symposium on Science and Technology in Development Planning, Mexico, 1979. 8. W. O. AIYEPEKU, The Periodical Literature Component of Social Science Research in Ibadan,Nigerian dournal o f Economic and Social Studies 17 (1975) 41-59. 9. Jean-Louis BOURSIN, Le Franfais, Langue Scientifique, in Journal Officiel de la Rdpublique Francaise, Document de rAssembl6e Nationale No. 2311, 1981,285. My translation. 10. The sectoral distribution of 'mainstream' authors in aggregate Subsaharan Africa is not radically different from the distribution in Qu6bec. In both cases from two-thirds to threequarters of all 'mainstream' scientific authors are in universities, from one-sixth to one-fourth are in the public sector, and two or three percent are affiliated with commercial organizations. The principle difference in sectoral distribution of "mainstream" authors is that in Africa, at least one-seventh of all authors belong to international organizations, whereas this sector hardly exists in Qu6bec. 11. For a presentation and discussion of this methodology see D. de SOLLA PRICE, The Analysis of Scientometric Matrices for Policy Implications, Scientometrics 3 (1981) 47-54. 12. This residual cate$ory accounted for about three percent of the population of publications. 13. For a description of how an adverse economic environment has affected research in one African state, see J. HANLON, Ghana: Science Hangs on Amid Chaos, Nature 279 (17 May 1979) 182-184.

Scientometrics 5 (1983) 3*

175