Scientometrics, Vol. 17, Nos 5 - 6 (1989) 401-413
A SCIENTOMETRIC ASSESSMENT OF AGRICULTURAL RESEARCH IN SOUTH AFRICA A. POURIS
Foundation for Research Development, P.O. Box 395, Pretoria (South Africa) (Received January 4, 1989)
This article reports the results of a scientometric assessment of agricultural research in South Africa over the period 1974-1984. The Science Literature Indicators Database of CHI is used and South Africa is compared with 7 other countries spread in America, Asia, Oceania, and Africa. The criteria used for the assessment are the contribution of each country to international agricultural literature (in terms of publications) and their impact in the "Schubert-Gl~inzel-Braun Impact Scale". It was found that, although the South African contribution has improved in that period, it is comparable to that of Brazil and Argentina; that Nigeria and Israel produce 3 times more, and: that Australia and Canada contribute more than one order of magnitude of publications more than South Africa. As far as research impact is concerned "Hant Science" research in South Africa is rated "fair" in the SchubertGl~inzel-Braun scale, whilst "Dairy and Animal Science" and "Veterinary" research are rated "poor".
Introduction The usefulness o f scientometric analyses in c o m p l e m e n t i n g the more traditional decision-making processes is well established. Scientometric analyses have been used to evaluate and c o m p a r e the performance o f nations (Smith et al., 1986, Garfield, 1984; Braun et al., 1987), scientific disciplines (Crouch et al., 1986; Arunachalan et al., 1984; Pouris 1988), journals (Cawkell 1978; Pouris 1986), individual scientists
(GarfieM 1983) etc. In this article we a t t e m p t to use scientometric analysis in order to assess the p e r f o r m a n c e of agricultural research in South Africa. Figure 1 summarizes the links o f agriculture w i t h the m a c r o e c o n o m i c variables Gross Domestic Product, Exports and .Etnployment - in South Africa. The contribution o f agriculture to GDP has been reduced f r o m 21% in 1911 to 4.2% in 1984; and its contribution to e x p o r t s has b e e n reduced f r o m 20% in 1972 to 7% in 1984. As far as e m p l o y m e n t is concerned, the w o r k f o r c e e m p l o y e d in agriculture has shrunk to 13% in 1985 f r o m the 27% in 1970. Despite its diminishing importance, agriculture is one o f South Africa's key industries playing a vital role in the c o u n t r y ' s efforts for
Scientometries 17 (1989J
Elsevier, Amsterdam-Oxford-New York Akaddmiai Kiad6, Budapest
A. POURIS: AGRICULTURAL RESEARCH IN SOUTH AFRICA
Manufacturiog 5.5 Other 45.8 o)
Mining !7.6
Manufacturing 16.6 ~ M i n i n g
Other 5 1 . 6 ~ Agriculture etc 1911 21.1 1939
Manufacturing 31.2 ~Mining
Other 50.3 Agriculture etc 13.6
1984
Agriculture 4.2
Fig. la. Sector contribution to GDP. Source: Official Yearbook of SA (various issues)
K Jtherk442. ~ b)
Manufacturing 20.2 ~ I / ~ Mining ~'x'xx~'~N"~ 86 Other [ 5 .2~ 1970
Manufacturing 23
~ining
Agriculture 27
6
~ 1980
Monufacturing 23.3
~
f r~______~tMining 94 Other 1 5~ ~ g4r A -g r 7 ~
Agriculture 16
1985
ic ure 13.5
Fig. lb. Distribution of the workforce per empt0yment sector (All races. No more than 5% (80 000) of all economically active whites are now employed in agriculture, compared with 30% (166 000) in the 1920s. Source: Official Yearbook of SA (various issues) and National Manpower Commission "Annual Report 1986")
c)
YEAR Total exports (Rmil0 Total ogricuL exports Agriculture as % of total exlgorts
1972 1949 401 20.5
1982 12294 2063 10.7
1983 20653 1728 8.4
1984 25379 1842 7.3
Fig. lc. Contribution of agriculture to exports revenue. Source: SA Statistics (various issues)
self-sufficiency. In addition, agriculture is contributing to the decentralization of economic activity. Notwithstanding the needs o f a rapidly growing population, South Africa suffers from limited and vulnerable resources. Only 12% of the Country's total area is arable and only one-third o f the country receives more than 500 m m o f rain per annum. Appropriate planning and support o f agricultural research is therefore o f paramount importance. The recognitioia o f the importance o f agricultural research is mirrored in the voting for priorities of the Scientific Priorities Committee which has been responsible until recently for advice on the allocation o f research funds in the government sector. Out o f 18 research fields " f o o d production" was voted second in importance just below the need for "education and training" .(Dept of National Education 1986). Consequently, governmental R&D budgets reflect this perception. Agriculture received 402
Scientometries 17 (1989)
A. POURIS: AGRICULTURALRESEARCH IN SOUTH AFRICA 22% of the governmental R&D budget during 1985/86 while manufacturing received only 17% and energy and health from 8% each (Fig. 2 ) . I t should be mentioned that in this respect there is a big difference between South Africa and other industrialized countries which spend only 2 to 5% of their governmental R&D budget for agricultural research (e.g. USA 2%, W-Germany 2%, France 4%, UK 5%) (Lederman et al., 1986).
Agriculture 22 % IV~nufacturing 17 % Energy 8 */. Heolth Other
8% 45 %
Manufacturing 17 ~ Ensergy ~ I-leolth ~,~"~8 ~ 1 ~ ~ . ~
~
Agriculture 22
/1
Others 45
Fig. 2. Distribution of governmental R&D expenditure (South Africa 1985/86). Source: Dept of 'National Education "Resourcesfor R&'D1985/86" Pretoria, South Africa The yardstick used to assess research performance is the quantity and the impact of research publications produced by scientists residents in South Africa and published in internationally prominent scientific journals. The rationale of this gauge is that for basic research most contemporary knowledge is encapsulated in the form of research articles in scientific journals. It is apparent that the use of this yardstick restricts the assessment only to the performance of scientists working in the frontier of science and to their supporting mechanism. Performance in training and teaching, another usual preoccupation of the people undertaking research, is reflected only partially in the yardstick of aggregate publications. Good teachers will produce good researchers who in their turn will produce scientific publications. Similarly, the performance of extension services, particular to agriculture, is not reflected in the yardstick used. Extension services are used to transmit knowledge from the researcher to the practitioner and the only way to measure their effectiveness is by measuring the gap between best available technology and current practice) Next we describe the approach adopted for the assessment, we outline the results of the exercise and we conclude with a summary and conclusions section.
Methodology The policy related questions which are addressed are: 1. How the output and impact of South African agricultural research compares to that of certain other countries? 2. Has South Africa's relative position in agricultural research improved or deteriorated over time? Scientometrics 17 (1989)
403
A. POURIS: AGRICULTURAL RESEARCH IN SOUTH AFRICA 3. What is the relative performance of the specialties (e.g. animal sciences, plant sciences) that constitute agricultural research? In order to answer these questions we need to decide on the use of an operational yardstick on a set of countries to be compared with South Africa and on an approprial database containing some operational variables representing the research output of the various countries. Some of the most frequently use d yardsticks of research output and impact are: numbe o f publications as a percentage of world share, absolute number of publications, number of citations received as a percentage of world share, number of citations over a time period per article, the Relative Citation Index (citations per article over the world norm) etc. We choose to use as yardsticks: 1. the particular country's contribution to world scientific literature 2. the world share of citations to the country's articles 3. the Schubert-Gl~nzel-Braun subject field citation scale. A country's contribution to world scientific literature is the share of this country's publications per 1000 world publications per discipline. Presumably countries with high number of contributions can be considered relatively stronger than countries with small number of contributions. Similarly, countries whose research receives a proportionally larger number of citations can be considered as producing research with higher impact (at least as it is valued by contemporary peers), than countries that receive proportionally fewer citations. The Relative Citation index (and the implicit ratio of citations per paper) have been considered in this regard. The Relative Citation Index compares the citations per paper for a given country in a given field and year to the world norms for that field in that year. However, this Index is subject to wide fluctuations when it is based on a small number o f papers and citations. As agriculture and its specialties in the countries under examination are relatively unproductive, as far as research publications are concerned, we decided against the use o f this index. The "Schubert=Gl~inzel-Braun subject field citation scale" is constructed by considering the distribution of the citation frequency of the world papers published in the journals of particular subfields as are grouped in the Science Citation lndex (Schubert et al., 1987). The scale is divided into 5 segments - uncited papers, poorly cited papers, fairly cited papers, remarkably cited papers and outstandingly cited papers. The advantage of the "Schubert-GlLnzel-Braun subject field citation scale" over the "Relative Citation Index", for our investigation, is that the former examines the citation ratios over five year periods. This grouping increases the number of papers considered and t~ence a more stable assessment is possible. Seven countries were chosen to be compared with South Africa. These are: Nigeria, New Zealand, Australia, Brazil, Argentina, Israel and Canada. These countries cover 404
Scientometrics 17 {1989)
A. POURIS: AGRICULTURAL RESEARCH IN SOUTH AFRICA a wide spectrum in terms o f development, population size, size of scientific population, area, emphasis on agricultural research, etc. However, all of them have some characteristics which make them suitable to be compared with South Africa. For example, Argentina and Brazil are at similar levels of development with South Africa (development is mirrored in the Gross National Product per capita). Nigeria emphasizes agricultural research (40% of the government budget is for research in agriculture), New Zealand is an agricultural c~ Israel has scientific connections with South Africa, Australia competes in the international markets With South Africa, etc. Another factor that influenced the choice o f countries is language. As we shall discuss, databases used for scientometric analysis, are usually biased in favour of English speaking countries. 1his ,bias can influence assessments when non English speaking countries are compared with English speaking ones. Hence, the inclusion of non-english or bilingual countries (Brazil' Canada, Israel etc.) in the set of our countries would neutralise the fact that South Africa is a bilingual country. Table 1 gives some of the characteristics of these countries. Table 1 Characteristics of various countries Country
S.,Africa Nigeria N Zealand Australia Brazil Argentina Israel Canada
GNP/capita 1984 ($) 2 260 770 7 240 11 890 1 710 2 230 5 100 13 140
Area Sq km (000) 1221 923 268 7686 8511 2766 20 9976
Population (000) 31 586 92 037 3 233 15 556 132 580 30 097 4 159 25 150
Equivalent R&D no of scientists & engineers 15 348 3 545 8 080 36 215 32 508 10 484 53 735 57 020
R&D expenditure as % of GNP 0.9 0.3 0.9 1.1 0.6 0.4 2.6 1.4
(1985) (1977) (1979) (1981) (1982) (1981) (1978) (1984)
Sources: 1. "The Europe Year Book 1987", Europe Publication Ltd, London England. 2. "Statistical Yearbook 1987" UNESCO, France. 3. "Resources for R&D 1985/86" Dept of National Education, Pretoria, South Africa. The database used for our assessment is the National Science Foundation's (NSF) Science Literature Indicators Database. This database is compiled by Computer Horizon Inc (CHI) Research on behalf of NSF for use in the biennial US Science Indicators report. It is based on the Science Citation lndex (SC1) which is p r o d u c e d ' b y the Institute for Scientific Information (ISI). The primary data set covers 2000 of the World's leading scientific journals over the period 1 9 7 3 - 8 4 . The set of journals covered is kept constant over the years so longitudinal comparisons are unaffected
Scientometrics 17 (1989)
405,
A. POURIS: AGRICULTURAL RESEARCH IN SOUTH AFRICA by changes in the coverage of different countries and subfields by the Science 67ration Index. The data in the citation tape include only papers appearing in the journals in the 1973 fixed journal set and only citations for which both the cilting and cited papers were in journals in the 1973 fixed journal set. In terms of-papers and citations, 4.000.000 papers and 50.000.000 Citations are contained in the database. Papers and citations areJallocated to 176 individual countries, to 106 scientific subfields and 9 fields. The subfield definition in the database is journal based. That is, individual papers are not assigned to subfields but journals are assigned to subfields, and all papers in a given journal are assigned to that journal's subfield. Some journals contain papers from several different subfields. These journals are apportioned among those subfields according to the approximate share of the journal which is devoted to each subfield. Each paper from that journal is then given the same field assignment as the journal as a whole, so that it is split among the subfields. Citations to the papers are, similarly split. Fields and subfields are defined by the citation patterns of publications. Journals with publications which cite each other heavily are generally assigned to the same subfield. Papers and citations received are allocated to countries according to the affiliation of the publishing authors. When two or more authors from different countries coauthor a paper, this paper is apportioned among the participating countries. If, for example, three authors from three. German Institutions and one from a South African one co-author a paper, 0.75 of a paper is credited to Germany and 0.25 of a paper is credited to South Africa. Citations to that paper are similarly apportioned. While the Science Literature indicators Database is the most comprehensive source of readily accessible data on national research output, it has certain important limitations which should be mentioned. The first is the bias of the Science Otation lndex (on which the Science Literature Indicators Database is based) in favour of English language journals. Comparisons of the SC1 with the British Library Lending Division holdings (Carpenter et al., 1981) has shown that, although SC! appears to be a balanced data source for the case of the Physics and Biology disciplines, particularly for English speaking countries, there are significant differences between the SC1 and other sources in the national coverage of fields with a more dispersed literature, especially in the Case of journals from countries with non-Roman alphabets. In particular it has been concluded that the West German and French journal coverage is generally good, the Japanese and "other" country coverage seems to be adequate for most international comparisons but the USSR coverage is so deficient that international activity indicators based on this Soviet coverage would be seriously affected. The issue of whether the SC1 is representative of significant research undertaken in Less Developed Countries (LDC) creates the possibility of another deficiency. The SC1 coverage of the scientific literature is designed to encompass the most significant 406
Scientometrics 17 H989)
A. POURIS:AGRICULTURALRESEARCHIN SOUTHAFRICA research in the world. However, as LDC's research is generally believed to be peripheral to the mainstream research effort, it is possible that important research in LDCs is overlooked. Empirical findings, however, refute this notion. A comparison of the SC1 to the Index Medicus (Frame, 1979) has shown that the coverage of LDCs' biomedical papers was roughly comparable for both indexing services, even though the Index Medicus covers two and a half times as many biomedical journals as the ISI. Finally a limition particular to the Science Literature Indicators Database springs from its structure which is based on a constant set of journals established in 1973. Articles published in entirely new journals are not covered , with the result that newly emerging or fast growing areas of science may be under-represented. The CHI Research has investigated the issue by examining an alternative data series based on the 3000 journals covered by ISI in 1981. Unfortunately, it was found that the larger database was more biased in favour of English language countries (Irvine et al., 1985) than the 1973 series. Similarly, an expanding database would not be appropriate for longitudinal studies. In any case as we report in the notes section 2 the result of our assessment are robust in this respect. Assessment
The results of the publication and citation analysis of agricultural research are presented in a series of tables and graphs. Table 2 shows the respective percentage shares of all agricultural papers (i.e. general Agriculture, Dairy and Animal Sciences, and Forestry) published worldwide by authors in different countries during 19741984. For example, of the 7143 papers listed in the Science Literature Indicators Database as having been published during 1984, 26.8 (i.e. 3.75%0) were produced by authors with South African address. South Africa's contribution to world agricultural literature has increased over the decade from 0.23%o in 1974 to 3.75 in 1984, a growth of 32% per year. Some scepticism, however, is warranted due to the small size of publications in 1974. In absolute terms, South Africa's contribution is comparable only to that of Brazil's and Argentina's. Nigeria and Israel produce 3 times more publications per year while Australia's and Canada's contribution is more than one order of magnitude larger. South Africa's agricultural research is relatively weak even when is compared with the aggregate performance of the general discipline Biology in which Agricultural research belongs. South Africa's contribution to biological sciences was 5.8%o in 1984 (Pouris 1988) in comparison with the 3.7570o of agriculture. As far as the other countries are concerned;Nigeria and Argentina increased their contribution over the decade, New Zealand saw its contribution shrinking while all
Scientometrics 17 H 989J
407
A. POURIS: AGRICULTURALRESEARCH IN SOUTH AFRICA Table 2 Number of agricultural (animal and plant) publications in the SCI (1974 base) per 000 world publications Country S.-Africa Nigeria N Zealand Australia Brazil Argentina Israel Canada
1974
1976
1978
1980
1982
1984
0.23 5.18 25.43 41.90 3.47 0.99 9.44 75.35
1.31 7.02 24.71 44.55 2.59 1.10 8.76 78.80
1.78 7.65 16.53 43.83 5.24 1.66 8.87 80.97
1.80 6.80 19.90 43.20 4.!0 2.50 8.10 97.29
1.95 8.55 15.55 43.14 5.61 2.97 10.01 93.85
3.75 9.04 19.40 41.86 3.23 2.28 10.96 88.07
the other countries kept their contributions constant. The group as a whole has improved its contribution to the world agricultural literature marginally from t 6 . 2 % in 1974 to 17.8% in 1984. In order to give a disaggregate picture of performance we show the contributions of different countries to Dairy and Animal Sciences and Plant Sciences which constitute the Agricultural Science (Tables 3 and 4). The South African contribution to D.dry and Animal Sciences appears to be more erratic than the contribution to Plant Sciences where a clear upward trend emerges. Nigeria and New Zealand appear to p_~ovide much larger contributions in Agriculture (plant sciences) than in Dairy and Ardmal Sciences'while the opposite is happening in Canada. Finally, Table 5 shows the contributions of different countries to veterina~ ~e.,,~arch over the period 1 9 7 4 - 1 9 8 4 . South Africa contributes only 1.34%o to the
wc:~Id veterinary literature and has not improved its contribution over the period. New Table 3 Number of agricultural (Animal & Dairy Science) publications in the SCI (1974 base) per 000 world publications Country S.-Africa Nigeria N Zealand Australia Brazil Argentina Israel Canada
408
1974
1976
1978
1980
1982
1984
0.00 2.08 8.93 35.73 0.00 0.00 13.52 32.19
0.69 2.12 3.18 26.52 0.26 0.53 9.39 106.10
2.03 2.70 4.39 45.09 0.67 0.67 8.11 118.71
0.00 2.94 4.25 40.19 0.65 0.33 9.80 156.27
0.77 2.56 2.20 32.80 1.54 2.30 14.60 130.71
4.86 4.21 7.29 39.18 2.54 0.27 11.78 132.91
Scientometrics 17 (1989}
A. POURIS: AGRICULTURALRESEARCH IN SOUTH AFRICA Table 4 Number of agricultural (Plant Sciences) publications in the SCI (1974 base) per 000 world publications County S.-Africa Nigeria N Zealand Australia Brazil Argentina Israel Canada
1974
1976
1978
1980
1982
1984
0.28 5.94 29.45 43.40 4.32 1.23 8.45 71.25
1.49 8.40 30.81 49.63 3.25 1.26 8.58 72.39
1.72 8.81 19.37 43.54 6.31 1.89 9.05 72.14
2.22 7.68 23.46 43.93 4.90 3.07 7.74 83.90
2.34 10.50 19.91 46.51 6.94 3.19 8.51 81.82
3.36 10.73 23.63 42.80 3.47 2.98 10.67 72.41
Table 5 Number of veterinary publications in the SCI (1974 base) per 000 world publications Country S.-Africa Nigeria N Zealand Australia Brazil Argentina Israel Canada
1974
1976
1978
1980
1982
1984
1.72 5.88 4.07 60.79 .1.35 0.45 4.07 72.43
1.36 4.79 6.84 65.60 1.14 1.36 8.21 64.87
2.17 8.86 4:67 60.89 2.43 1.12 3.63 61.67
1.63 13.49 3.80 67.65 1.52 1.71 3.91 60.51
2.30 6.17 1.27 71.08 1.82 0.36 3.20 70.68
1.34 4.11 1.27 67.04 1.49 1.31 1.76 73.14
Zealand, Israel, Brazil, and Argentina contribute similarly small shares like South Africa, while Australia and Canada make much larger contributions producing 67 and 73%0 of the international literature respectively. The next question which we wish to investigate is the relative impact of South African agricultural related research on the world scientific community. Tables 6 and 7 show the shares of citations received by the agricultural and veterinary papers of individual countries per year. For example, South African papers published in 1978 received over the period 1 9 7 8 - 1 9 8 4 . 6 5 citations corresponding to 1.87% of the total citations to the world publications for this period. The obvious preblem with raw citations, however, is that they are biased in favour of groups (countries, institutes, etc.) with large research efforts. One way of off-setting, at least partially, this deficiency, is to estimate the average number of citations per paper for each group and compare it with a predetermined yardstick. Scient ometrics 17 (1989)
409
A. POURIS: AGRICULTURAL RESEARCH IN SOUTH AFRICA Table 6 Share of citations received by aggicultural papers Country S.-Africa Nigeria N Zealand Australia Brazil Argentina Israel Canada
1974
1976
1978
1980
1982
1984
0.31 2.58 17.49 44.80 1.31 0.13 12.78 78.96
1.68 4.91 20.35 40.77 1.05 0.44 8.62 90.55
1.87 3.47 13.90 32.68 1.91 0.56 5.47 68.43
1.93 2.48 12.62 40.46 2.24 2.29 9.02 113.26
3.19 2.65 13.32 41.68 20.70 2.51 8.26 176.47
1.89 6.64 35.02 79.64 2.11 2.42 7.91 69.72
Table 7 Share of citations received by veterinary papers Country S.-Africa Nigeria N Zealand Australia Brazil Argentina Israel Canada
1974
1976
1978
1,980
1982
1984
2.55 3.61 6.83 60.08 0,29 0.10 0,89 15.83
0.35 1.29 4.69 66.31 0.17 0.82 4.28 76.49
0.74 3.67 6.22 76.31 0.58 0.82 1.32 66.71
0.69 4.78 2.95 68.02 0.34 0.43 3.47 82.11
2.06 3.26 0.36 74.48 0,69 0.00 3,16 94,92
0.00 6.37 0.00 28.60 0.00 0.00 0.00 54.45
I
FAIR 1.43 POOR
0.433
RQtio: 0.1% S. Africa Nigeria 0.962 N. Zealand 1.247 Australia 0.348 Brazil 0.053 Argentina 1.266 Israel 1.602 Canada
T
T , ,-
,t..)
N
~
<
z
_ ~
z
a:l tti
<
Fig. 3. Citation performance of agricultural related research of different countries for 1980-84 (Schubert -Gl~inzel-Braun Scale)
410
Scientometrics 17 (1989)
A. POURIS: AGRICULTURALRESEARCH IN SOUTHAFRICA Figure 3 shows the citation performance of the agricultural related research of different countries in the "Schubert-Gl~inzel-Braun Scale" for the period 1980-84. South African research in Agriculture (Plant Sciences) is rated "fair" while research in Dairy and Animal Science, and Veterinary Science is rated "poor". As far as the other countries are concerned, Canada, Israel, and Australia hold the top positions in terms of impact. We should emphasise that the "Schubert-Gl~nzel-Braun Scale" evaluates the average paper. A country producing a few citation classics and a large number of low citation articles will be graded "poor", at the same level with the country that produces only low impact papers. Only detailed comparisons of the distribution of citation of the papers could provide further evidence of the relevant impact of the research of different countries. However, for countries producing similar number of papers, the "Schubert-G1Luzel-Braun Scale" is an a.dequate one.
Summary and conlusions This article is an effort to assess basic agricultural research in South Africa. Agriculture is considered an important socioeconomic sector in South Africa despite its shrinking importance in terms of contribution to exports, employment and gross domestic product. The importance of Agriculture springs form the country's need to feed the fast growing population and the political circumstances which dictate selfsufficiency as a general policy goal. The yardstick used to assess research performance was the activity and the impact of South African publications internationally. Activity is proxied by the number of publications produced by scientists with corporate address South Africa, and the impact by the number of citations received by these publications. The rationale for this approach was the fact that for basic research most contemporary knowledge is encapsulated in the form of research articles in scientific journals. The National Science Foundation's Science Literature Indicators Database is used as the main source of information. This database contains data from 176 countries and 106 specialties for the period 1973-1984. South Africa is compared with a set of 7 other countries - Nigeria, New Zealand, Australia, Brazil, Argentina, Israel, and Canada - and the world norms. During the under examination period, South African agricultural literature exhibited a considerable growth starting from almost zero to contribute 3.755700 of the world agricultural literature in 1984. The South African contribution during 1984 is comparable with those of Brazil and Argentina. However, Nigeria and Israel produce 3 times more publications than South Africa, whilst Australia and Canada contribute more than one order of magnitude more than South Africa. Agricultural research output is also Scientometrics 17 (1989)
411
A. POURIS:AGRICULTURALRESEARCHIN SOUTHAFRICA smaller in comparison with the South African biological research output in general (Biological Research contains Agricultural Research as a speciality). In 1984 south African biological disciplines contributed 5.8%o of the world literature, while agricultural research contributed only 3.75%0. The disaggregation of Agricultural research into Dairy and Animal Sciences and Plant Sciences shows that there is no considerable difference between the two specialties. However, examination of veterinary research in the country indicates that South Africa contributes only 1.34%o to the world Veterinary literature and that the size of contribution has remained constant over the 1974-1984 period. As far as research impact is concerned, South African Agricultural-Plant Science research is rated "fair" in the "Schubert-Gl~inzel-Braun Scale", whilst Dairy and Animal Scinece, and Veterinary research are rated "poor". Among the other countries, Canada, Israel, and Australia produce the bulk of publications produced by the group and their research has consistently higher impact than the rest of the countries in the group.
Notes
1. Some indication of the directions of R&D and its effects on production can be given through analysis of various productivity ratios. For example, Agricultural production in South Africa increased at a rate of 3.7% per year between 1950-1980 (Economic Advisory Council of the State President "Report of an Investigation into the Restructuring of Agriculture" Nov t986). During the same period, the cultivated area increased by 3.8% per year. Assuming that the productivity of the new cultivated areas was as good, we can conclude tentatively that R&D has not affected land productivity in South Africa. This means that R&D was either unable to assist production or that it was not transferred from the lab to the farmer. 2. In order to verify the pecking order of the countries in our group, we ranked them according to their contributions to world literature in 1984 with data, firstly from the Science Literature Indicators Database, and secondly from the computerised version of the Science Citation Index. While the first database contains papers published in 2000 journals, the second database covers more than 3600 journals. Subsequently, we estimated the Spearman Rank Correlation Coefficient for the rankings of Plant Sciences and Dairy and Animal Sciences (seperately). The correlations were found, to be. statistically significant at the 0.05 level of significance for the former and at the 0.01 level for the latter. Similarly, supportive evidence for our findings is provided by Arvanitis et al., (1988). On the basis of the literature indexed by the PASCAL database (which is 412
Scientometrics 17 (1989)
A. POURIS: AGRICULTURAL RESEARCH IN SOUTH AFRICA c o m p l e t e l y d i f f e r e n t in s t r u c t u r e f r o m t h e ISI d a t a b a s e s ) t h e y identified t h e c o n t r i b u t i o n o f d i f f e r e n t c o u n t r i e s in Tropical Soil Sciences for t h e y e a r 1983. The rating o f t h e c o u n t r i e s a c c o r d i n g t o t h e i r c o n t r i b u t i o n s is a l m o s t e x a c t l y identical t o o u r rating.
References 1. ARUNACHALAM S. RAO M. K. D., SHIRASTRAVA P. K. (1984), Physics research in Israel. A premliminary bibliometric analysis Journal of the A merican Society for Information Science, 8: 185-95. 2. ARVANITIS R. CHATELIM Y., (1988), National scientific strategies in tropical soil, Social Studies of Science, 8(1): 113-146. 3. BRAUN T., GL~_NZEL W., SCHUBERT A., (1987), One more version of the facts and figures on publication output and relative citation impact of 107 countries 1987-1980, Scientometrics, 11: 9 - 1 5 . 4. CAWKELL A. E., (1978), Evaluating scientific journals with journal Citation Reports. A case study in acoustics, Journal of the American Society for Information Science, Jan 1978, pp. 41-6. 5. CARPENTER M. P., NARIN F., (1981), The adequacy of the Science Citation Index (SCI) as an indicator of international sicentific activity, Journal o f the American Society for Information Science, Nov 1981, pp. 430-439. 6. CROUCH D., IRVINE J., MARTIN B. R., (1986), Bibliometric analysis for science policy. An evaluation of the UK's research performance in Ocean Currents and Protein Crystallography,. Scientometrics, 9 : 239-7. 7. Department of National Education (1986) "Annual Report" Government Printer, Pretoria RSA. 8. FRAME J. D., (1979) National economic resources and the production of research in lesser developed countries, Social Studies of Science, 9: 233-46. 9. GARFIELD E., (1983), How to use citation analysis for faculty evaluations, and when it is relevant, Current Contents, Oct 31, 1983. 10. GARFIELD E., (1984), Latin American research - Where it is published an how often it is cited, Current Contents, May 7, 1984. 11. IRVINE J., MARTIN B., PEACOCK T., TURNER R., (1985), Charting the decline in British science, Nature, 316: 587-90. 12. LEDERMAN L. L., LEHMING R., BOND J. S., (1986), Research policies and strategies in six countries. A comparative analysis, Science and Public Policy, 13(2): 6~/-76. 13. POURIS A., (1986), The South African Journal of Science - A bibliometric evaluation, South African Journal o f Science 82: 4 0 1 - 2 . 14. POURIS A., (1988), Assessing basic research in South Africa, South African Journal of Science, June 1988. 15. SCHUBERT A., GL~NZEL W., BRAUN T., (1987), Subject field characteristic citation scores and scales for assessing research performance, Scientometrics, 12: 267-92. 16. SMITH D. C., COLLINS P. M. D. HICKS-SWYATT D. M., (1986), National performance of basic research, Nature, 23 Oct, 1986.
Scientometrics 17 (1989)
413