Aparna Basu

International Collaboration in Science in India and its Impact on Institutional Performance

In this paper, our objective is to delineate some of the problems that could arise in using research output for performance evaluation. Research performance in terms of the Impact Factor (IF) of papers, say of scientific institutions in a country, could depend critically on coauthored papers in a situation where internationally co-authored papers are known to have significantly different (higher) impact factors as compared to purely indigenous papers. Thus, international collaboration not only serves to increase the overall output of research papers of an institution, the contribution of such papers to the average Impact Factor of the institutional output could also be disproportionately high. To quantify this effect, an index of gain in impact through foreign collaboration (GIFCOL) is defined such that it ensures comparability between institutions with differing proportions of collaborative output. A case study of major Indian institutions is undertaken, where Cluster Analysis is used to distinguish between intrinsically high performance institutions and those that gain disproportionately in terms of perceived quality of their output as a result of international collaboration.

International Collaboration in Indian Scientific Papers

Internationally co-authored publications may be regarded as an indicator of scientific co-operation between countries and is of interest in science policy. In this study, the extent of international collaboration in Indian science has been estimated from SCI data in 1990 and 1994. We find an increase in collaboration both in terms of output and the extent of the network and significantly higher impact (IF) associated with internationally co-authored papers in several disciplines. However, there was no significant increase in IF of collaborative papers over time, whereas Indian papers in general showed a statistically significant, though small, increase in average impact from 1990 to 1994. The bulk of Indian scientific co-operation was with the developed Western nations and Japan, but it was often the smaller countries with a few co-authored papers which showed higher average impact. Co-operation with South Asian countries, initially low, has doubled in four years. By a combination of multivariate data analysis techniques the relative positions of Indias partners in scientific collaboration have been mapped with respect to the fields of co-operation.

Using ISI's 'Highly Cited Researchers' to obtain a country level indicator of citation excellence

SummaryA high level of citation to an authors work is, in general, a testimony to the fact that the authors work has been noted and used by his peers. High citation is seen to be correlated with other forms of recognition and rewards, and is a key indicator of research performance, among other bibliometric indicators. The Institute for Scientific Information (ISI) defines a highly cited researcher (HCR) as one of 250 most cited authors of journal papers in any discipline. Citation data for 20 years (1981-1999) is used to calculate the share of HCRs for countries in 21 subject areas. We find that the US dominates in all subject areas (US share ~ 40-90%). Based on the number of highly cited researchers in a country, an index of citation excellence is proposed. We find that rank order of countries based on this index is in conformity with our general understanding of research excellence, whereas the more frequently used indicator, citations per paper, gave an unacceptable rank order due to an inherent bias toward very small countries. Additionally, a high value of the index of citation excellence was found to be associated with higher concentration of highly cited researchers in affiliating organizations.

Does a country’s scientific ‘productivity’ depend critically on the number of country journals indexed?

In this paper, we examine the question whether it is meaningful to talk about the scientific productivity of nations based on indexes like the Science Citation Index or Scopus, when the journal set covered by them keeps changing with time. We hypothesize from the illustrative case of India’s declining productivity in the 1980s which correlated with a fall in its journals indexed, that an apparent increase/decrease in productivity for any country, based on observed change in its share of papers could, in fact, be an effect resulting from the inclusion of more/less journals from the country. To verify our hypothesis we have used SCIMAGO data. We found that for a set of 90 countries, the share of journals regressed on the share of papers gave a linear relationship that explained 80% of the variance. However, we also show that in the case of China’s unusual rise in world scientific productivity (to second rank crossing several other countries), there is yet another factor that needs to be taken into account. We define a new indicator—the JOURNAL PACKING DENSITY (JPD) or average number of papers in journals from a given country. We show that the packing density of Chinese journals has steadily increased over the last few years. Currently, Chinese journals have the highest ‘packing density’ in the world, almost twice the world average which is about 100 papers per journal per annum. The deviation of the JPD from the world average is another indicator which will affect so called ‘national productivities’ in addition to the number of national journals indexed. We conclude that in the context of a five fold increase in the number of journals indexed over 20xa0years, the simplistic notion of ‘scientific productivity’ as equivalent to papers indexed needs to be re-examined.

Science publication indicators for India: Questions of interpretation

We comment on a letter toNature in 1996 on the long term decline of Indian science pointing out methodological reasons why the (SCI) data used by the authors do not unambiguously lead to their stated conclusions. Our arguments are based on the contention that no valid statement on change in a countrys output may be made for a period in which the journal coverage from that country in SCI has changed significantly. We have suggested that for longitudinal comparisons of country level performance, it should be verified that the journals from that country in SCI remained constant within the period. This could be ensured if the country of publication of journals could be included as a field in the SCI database.We define a Visibility Index as the cumulated impact and derive a relation to estimate change in visibility combining changes in output and average impact. In the period during which Indian journal coverage remained unchanged, a detailed analysis of output for two years (1990–94) leads us to conclude that, with the exception of Agriculture, there has been an increase in publication in virtually every field, with significant increase in the overall mean Impact Factor. At least 25 subfields have been identified with statistically significant increase in mean Impact Factor and Visibility. The impact of foreign collaboration on visibility has also been considered. In conclusion we touch upon the question of citation as a performance indicator for Third World countries as high citation and relevance may be in conflict as objectives.

Social Network Analysis: A Methodology for Studying Terrorism

This chapter aims to bring to the reader an overview of the work done since the 9/11 terrorist attack, in the field of Social Network Analysis as a tool for understanding the underlying pattern /dynamics of terrorism and terrorist networks. SNA is particularly suitable for analyzing terrorist networks as it takes relationships into account rather than merely attributes, which are difficult to obtain for covert networks. Using graph theoretic methods and measures and open source data it has been possible to map terrorist networks and examine roles of different actors, as well as identify groups and structures within the network. The methodology is illustrated by reviewing two case studies: the 9/11 terrorist network study by Krebs, that used data from a single terrorist attack, and a study by Basu that used data from about 200 terrorist incidents in India to create a network of terrorist organizations for predictive purposes.

Designing a Composite Index for research performance evaluation at the national or regional level: ranking Central Universities in India

It is now generally accepted that institutions of higher education and research, largely publicly funded, need to be subjected to some benchmarking process or performance evaluation. Currently there are several international ranking exercises that rank institutions at the global level, using a variety of performance criteria such as research publication data, citations, awards and reputation surveys etc. In these ranking exercises, the data are combined in specified ways to create an index which is then used to rank the institutions. These lists are generally limited to the top 500–1000 institutions in the world. Further, some criteria (e.g., the Nobel Prize), used in some of the ranking exercises, are not relevant for the large number of institutions that are in the medium range. In this paper we propose a multidimensional ‘Quality–Quantity’ Composite Index for a group of institutions using bibliometric data, that can be used for ranking and for decision making or policy purposes at the national or regional level. The index is applied here to rank Central Universities in India. The ranks obtained compare well with those obtained with the h-index and partially with the size-dependent Leiden ranking and University Ranking by Academic Performance. A generalized model for the index using other variables and variable weights is proposed.

Some differences in research publications of Indian scientists in India and the diaspora, 1986---2010

The diaspora of a less developed country, who reside outside their country of origin, can contribute to the parent country through financial or knowledge transfers, connections, or on return of talented persons. The knowledgebase of the diasporas is therefore of interest to the parent country. Scientific publications of the Indian diaspora are an indicator of the existing knowledge base of Indians overseas. Samples drawn from Web of Science (1986–2010), using a selected list of unique Indian names, are analyzed with the objective of comparing and identifying distinguishing features of the diaspora. While both Indian and diaspora samples have increased over time, publication output from Indians overseas has increased more rapidly. English was by far the most frequently used language. A major difference was found in the type of publication with many more proceedings papers and meeting abstracts by the diaspora, showing increasing importance of rapid publication of novel results. Number of articles was about the same in both samples, but a more detailed look at the top 100 journals qualifies the nature of the journal space used, which again shows major differences. Articles in Nature and Science confirm the differences in the high impact range. We end with a discussion of limitations which includes effects of changing database coverage with time.

`Cognitive mobility' or migration of authors between fields used in mapping a network of mathematics

In this paper we have looked at a new measure of connectedness between research areas, namely, the migration of authors between subfields as seen from their contributions to different areas. Migration may be considered as an embodied knowledge flow that bridges some part of the cognitive gap between fields. Our hypothesis is that the rate of author migration will reflect cognitive similarity or affinity between disciplines. This is graphically shown to be reasonable, but only above certain levels of migration for our data from mathematical reviews spanning 17xa0years (1959–1975). The inter-related structure of Mathematics is then mapped using migration data in the appropriate range. We find the resulting map to be a good reflection of the disciplinary variation in the field of Mathematics.

The Albuquerque model and efficiency indicators in national scientific productivity with respect to manpower and funding in science

Due to recession in the world economy there is a trend towards a reduction in growth of R&D expenditure in the G7 countries. At the same time countries like China and Korea are investing more in scientific research. We compare the differences in the inputs to science for different countries and explore the level of efficiency in the output of scientific papers with respect to inputs such as manpower and investment. We find that the EU countries are relatively more efficient than Japan, the USA and also China and Korea so far as the production of papers is concerned. However, if efficiency is considered in terms of patents, Japan Korea and the USA are ahead. We compare our results with Albuquerque’s model linking patent to paper ratios and development, and find significant deviations for some countries. We deduce that there has been a shift from publishing towards patenting in certain countries and link it to high contribution from the business sector to R&D expenditure. Preliminary results of this analysis have been presented in Basu (In Proceedings of the 14th International Society for Scientometrics and Informetrics (ISSI) Conference, 2013).