An Informetric analysis of publication and research collaboration patterns in natural and applied sciences in South Africa

This study is the part of an ongoing research which presents the mainstream scientific output and collaboration of five research universities of South Africa over a 9 year period between 1995 and 2003. Since a part of this research has already been published in 2006 where the main emphasis was on publication output, this paper concentrates more on collaboration. The paper discusses the distribution of publications by institutions, index of specialization, collaboration and patterns of co-authorship. The results show that South African authors collaborated more frequently with international authors (73.99%) than with national authors (26.01%). This was confirmed statistically at a confidence level of p-value <0.025. A further non-parametric chi-square statistical analysis illustrated that there are significant differences in the proportion of co-authorship amongst the five institutions (p-value<0.005). The results obtained shows that there is a sharp decline in publication output from 1995 until the end of 1998 and then again from 2003. The decrease in publication output is also an indication of the lack of collaborative research by South African scientists.


Introduction
A primary feature of modern science is that researchers collaborate in teams.Research teams appeared sporadically already during the 19 th century (Havemann, 2001), and are now so common that on various occasions scholars in science studies have stipulated using teams as basic units of investigation as opposed to individual researchers.Very often, in many fields, researchers without teams cannot keep up with developments on the research front.
According to Haiyan, Hildrun and Zeyuan (2004), there have been varying manifestations of collaborative work.And since the pioneering work of Price (1963), and Beaver and Rosen (1978, 1979, 1979), a large number of scholars have stressed different forms and roles of scientific collaboration in different scientific fields.An investigation into these researchers can be made by analysis at micro level (individuals), meso level (institutions), or macro level (countries) (Glanzel 2002, Kretschmer 2004).
The last few decades have witnessed a restructuring in scientific research where the pattern of scientists working alone gradually giving way to collaborating with colleagues from the same institution or from other institutions.In this way, it has increasingly become a collaborative endeavour (Subramanyam 1983).There is also a strong trend towards borrowing from, and interpenetration across disciplines.As a result of these trends, there is a rise in collaboration both within interdisciplinary research and an increasing interest in collaboration amongst both researchers and science policy makers.The global tendency towards more collaboration in scientific research -often crossing the borders of institutes and countries -has been demonstrated in many scientific fields (Havemann, 2001a).For researchers to work together in order to solve problems is so common now in many branches of science and technology, that in many cases it makes more sense to consider groups as the basic units of research rather than individual scientists (Seglen and Aksenes, 2000).To establish and maintain collaboratiive links can be more or less easy -depending on the scientific, cultural, political, and geographical barriers that must be overcome.Havemann (2001) is of the opinion that in recent years, some of these barriers have been lowered, especially in transnational collaboration.
How research collaboration and productivity are correlated has been studied by many scholars.A recent study by Seglen and Aksnes (2000) delves in to scientific productivity and group size and the result of the study carried out by Gupta and Karisiddapa (1998) shows that there has been a systematic increase over time in the number of papers per author in the subset of collaborative researchers.
There were three comprehensive bibliometric studies carried out by Jacobs and Ingwersen(2000); Ingwersen and Jacobs (2003); and Jacobs (2006) which measured the trend in publication of South African scientists.First of these studies carried out a research on the development and research of scientists for a sixteen year period starting from 1981-1996.The five standard fields chosen were animal and plant sciences, and veterinary medicine.In the second study, where the period of study extended from 1981-2000, the authors used plant and animal science, chemistry, physics, general and internal medicine and the stress was on the citation impact.They compared South African research with New Zealand and Mexico and compared the publication volume with world shares.The study investigated the productivity of scientists for the entire period 1981-2000 and brought out the strengths and weaknesses of different areas of science in the country.The study also came to the conclusion that the citation impact of the sciences only started picking up after 1994 after the embargo on the country had been lifted.

Purpose
The present study has investigated the position of South Africa in relation to collaboration both inter-institutional and international.International scientific collaboration has generated increased interest in recent years due to the higher quality of collaborative papers as shown by higher average impacts compared to solely national publications (Van Raan, 1998) and the benefit gained by peripheral countries from international collaboration which integrates their national publications in the international scientific network (Russell, 1995).
The purpose of this paper was to explore mainstream scientific research and collaboration of the authors both nationally and internationally as well as to illustrate the extent of scientific development in South Africa.Although there has been extensive studies on collaboration in other countries no such comprehensive studies have been carried out in South Africa, hence this study.
• To identify the main South African institutions that are actively involved in the production of the main disciplinary fields • To investigate the growth and development of South African scientific publications, 1995-2003 among the selected universities • To examine the extent of production in the seven selected fields such as physical sciences, plant and animal sciences, engineering and clinical medicine.• To determine the extent of international collaboration among the faculties of these five institutions.
• To study the specialization index for selected fields using publication data for the given period

Methodology
The data for study has been collected from CD-ROM and Web versions of the Science Citation Index (SCI) and social Sciences citation Index (SSCI) which is published by the Institute of Scientific Information (ISI), presently the Thomson Scientific.All papers were recorded in the annual volumes of the Science Citation Index (SCI) of the Institute for Scientific Information (ISI) for the period of 1995-2003.The data for each document included author names, title, abstract, date, document type, and cited references.Author names were standardised as some authors may report their names differently in different papers.Each author was identified by his/her surname and first initial only (Newman, 2001).
The following five research institutions of South Africa were selected for the study.They were the Universities of Pretoria (UP), Cape Town (UCT), Natal2 (NATAL), Stellenbosch (STELL) and Witwatersrand (WITS).The five institutions were considered as the research universities of the country as they seem to produce the bulk of scientific publication.This was proved by a research done by the author in her doctoral research where all twenty one universities were included.The research output of the five selected universities were 51% of the total and all the other universities put together only produced about 49% of the output.
The performance of the five institutions was judged and compared on the basis of the following qualitative and quantitative indicators, (a) size of scientific activity measured by volume of production during the period of study (b) each institution's fields of specializations using the specialization index(SI) formula (c) and the collaboration of scientific activity measured by co-authorship.Once retrieved, records were analyzed using Microsoft Excel in order to identify the production distribution through out the period of study, distribution of publications by fields and institutions, and distribution by type of document, in order to determine whether one institution is more or less active in a specific field in comparison to other institutions.The specialization index (SI) formula was used (Godin, Robitaille, & Côté, 2001).It was calculated as the share (%) of publications of institution X in field Y divided by the share (%) of publications of all the institutions in field Y.
The study was also analytical in nature, with the application of suitable statistical tools to strengthen the empirical validity.The computer software, SPSS was used for processing.A non-parametric chi-square test was applied in the analysis of the differences of co-authorship amongst institutions, and a t-test was used to the rate of co-authorship rate.A further ANOVA analysis was conducted to study whether or not there was significant inter-institutional variation in their total publication out put in the mainstream exists.

Results and discussions
The results are represented under items 3.1 to 3.4 and in table one and figures 1 to 3 below.

The disciplinary distribution of South African publications
Using the information available in the address field of the articles, the number of articles produced from the most productive South African institutions was calculated.In this study, only articles from the 9 year period of 1995-2003 were included.19399 articles were identified in total from 7 fields of study amongst the 5 institutions.An analysis of the total out put presented in Table 1 indicates that the University of Cape Town (UCT) accounts for the largest share of South African publications (26.80%) followed by UP (19.84%).The University of WITS and STELL are in the range of 18-19%.The University of Natal with a share of 16.28% accounts for the least of all the other institutions.A further statistical analysis was employed to indicate if there was a significant level of inter-institutional differences identified in the total out put of publications.However, the result of a statistical analysis at p-value >0.10 does not reveal significant interinstitutional variation in their total publication output.
In terms of disciplinary scientific publications for the period of study, out of the total of 19399, 5725 or 29.51% came from the field of Clinical Medicine, 4044(20.85% from Plant and animal ( PLT&ANM) science.BIO&BICH counted for 935(4.82%)and Physics, 2693 (13.88%) and Engineering had a total of 2520 which was equivalent to13.00%respectively of the institutions' output for the period.The other disciplinary fields, in order of contribution, were ENV&ECO 1746 (9.00%), and CHEM, 1736 or (8.94%).University of Pretoria had 560 publications out of the total output of 5725 publications by the five institutions in clinical medicine which was only 14.55% and was considered the lowest among the contributing institutions.However, notable differences were found with respect to the contribution made by individual institutions to different fields of study.The University of Pretoria, for example, had the highest percentage of contribution in PLT&ANM science, 1436 (37.31%),PHY 602 (15.64%) and, the least was BIO&BICH with only 102 (2.62%) out of the total production.The University of WITS, on the other hand, has got the highest publication share in clinical medicine 1516 (42.90%) and the lowest in BIO&BICH 158 (4.50%).The University of Cape Town and Natal were strong in the fields of Clinical medicine and PLT&ANM science while the University of Stellenbosch showed its highest production in the fields of Clinical medicine 970(26.51%)and Engineering 751 (20.52%).
In general, coverage varies across different fields.The research and publications in Clinical medicine and, PLT&ANM science were very encouraging.In Engineering and Physics, the coverage was somewhat lower.In the CHEM and ENV&ECO the coverage tended to be poorer.The least contribution was from BIO&BICH.These disciplinary differences have been proven using the appropriate statistical analysis and it was found that scientific publication differences amongst the seven fields of study were justified at a confidence level of p <0.005.

Table 1
Distribution of publications according to subject field and institutions