8
Jointly published by AkadØmiai Kiad, Budapest Scientometrics, and Kluwer Academic Publishers, Dordrecht Vol. 59, No. 3 (2004) 337344 Received December 9, 2003 Address for correspondence: MASASHI SHIRABE Graduate School of Science and Engineering, Tokyo Institute of Technology Ookayama 2-12-1, Meguro, Tokyo, 152-8550 Japan E-mail: [email protected] 01389130/2004/US $ 20.00 Copyright ' 2004 AkadØmiai Kiad, Budapest All rights reserved Likelihood of inbound/outbound access to co-authorship MASASHI SHIRABE, a,b HIROYUKI TOMIZAWA b a Graduate School of Science and Engineering, Tokyo Institute of Technology, Tokyo (Japan) b National Institute of Science and Technology Policy, Ministry of Education, Culture, Sports, Science and Technology, Tokyo (Japan) We shall generalize the concept of our previous paper (SHIRABE & TOMIZAWA, 2002), which proposed an index for international scientific co-authorship. Based on a simple model of domestic and international co-authorships, we focused on likelihood of overseas access to co-authorships in the paper. Here, in consideration of bidirectionality of international co-authorship, we shall extend our previous index to two symmetrical indices. The indices can draw a reasonably clear picture of international co-authorship, with regard to difference in patterns of international co-authorship among countries. Introduction In this paper, we shall generalize the concept of our previous paper (SHIRABE & TOMIZAWA, 2002). Taking into consideration of bidirectionality of international co- authorship, we shall extend the previous index to two symmetrical indices. The purpose of the present paper is to suggest indices drawing a clear picture of international co- authorship concerning difference in patterns of international co-authorship among countries. Among many fields where globalization is in progress, developments in science and technology are most exceptional, especially in post-Cold War period since 1990s (NISTEP, 2000). This situation makes all the more important quantitative studies of international co-authorship in scientific papers (e.g., NARIN & WHITLOW, 1990; NARIN et al., 1991; LUUKKONEN et al., 1993; GL˜NZEL & DE LANGE, 1997a, 1997b; MELIN & PERSSON, 1998; MELIN, 1999; GL˜NZEL & SCHUBERT, 2001; GL˜NZEL, 2001). The reason why studies in this area are very active is that international co-authorship itself is a remarkable phenomenon, but we should not miss the fact that its quantitative indicators raise interesting mathematical issues, such as these scale dependency.

Likelihood of inbound/outbound access to co-authorship

Embed Size (px)

Citation preview

Page 1: Likelihood of inbound/outbound access to co-authorship

Jointly published by Akadémiai Kiadó, Budapest Scientometrics,and Kluwer Academic Publishers, Dordrecht Vol. 59, No. 3 (2004) 337–344

Received December 9, 2003Address for correspondence:MASASHI SHIRABEGraduate School of Science and Engineering, Tokyo Institute of TechnologyOokayama 2-12-1, Meguro, Tokyo, 152-8550 JapanE-mail: [email protected]

0138–9130/2004/US $ 20.00Copyright © 2004 Akadémiai Kiadó, BudapestAll rights reserved

Likelihood of inbound/outbound access to co-authorshipMASASHI SHIRABE,a,b HIROYUKI TOMIZAWAb

a Graduate School of Science and Engineering, Tokyo Institute of Technology, Tokyo (Japan)b National Institute of Science and Technology Policy, Ministry of Education, Culture, Sports,

Science and Technology, Tokyo (Japan)

We shall generalize the concept of our previous paper (SHIRABE & TOMIZAWA, 2002), whichproposed an index for international scientific co-authorship. Based on a simple model of domesticand international co-authorships, we focused on likelihood of overseas access to co-authorships inthe paper. Here, in consideration of bidirectionality of international co-authorship, we shall extendour previous index to two symmetrical indices. The indices can draw a reasonably clear picture ofinternational co-authorship, with regard to difference in patterns of international co-authorshipamong countries.

Introduction

In this paper, we shall generalize the concept of our previous paper (SHIRABE &TOMIZAWA, 2002). Taking into consideration of bidirectionality of international co-authorship, we shall extend the previous index to two symmetrical indices. The purposeof the present paper is to suggest indices drawing a clear picture of international co-authorship concerning difference in patterns of international co-authorship amongcountries.

Among many fields where globalization is in progress, developments in science andtechnology are most exceptional, especially in post-Cold War period since 1990s(NISTEP, 2000). This situation makes all the more important quantitative studies ofinternational co-authorship in scientific papers (e.g., NARIN & WHITLOW, 1990; NARINet al., 1991; LUUKKONEN et al., 1993; GLÄNZEL & DE LANGE, 1997a, 1997b; MELIN &PERSSON, 1998; MELIN, 1999; GLÄNZEL & SCHUBERT, 2001; GLÄNZEL, 2001).

The reason why studies in this area are very active is that international co-authorshipitself is a remarkable phenomenon, but we should not miss the fact that its quantitativeindicators raise interesting mathematical issues, such as these scale dependency.

Page 2: Likelihood of inbound/outbound access to co-authorship

M. SHIRABE, H. TOMIZAWA: Inbound/outbound access to co-authorship

338 Scientometrics 59 (2004)

The relationship between international co-authorship and national size has already beenanalyzed in several empirical studies (e.g., MELIN, 1999). But its mathematicalproperties are still ambiguous and need to be examined in depth.

It is necessary to note the meaning of the “national size” to discuss this issue.Actually the national size means the size of science-related activities. Many pastbibliometric studies have used the number of scientific papers as index of this sizeexplicitly or implicitly, but considering the purpose of dealing with the size of scientificactivities, it is possible to employ other indices, for example, the number of researchers(e.g., full-time equivalent as OECD defines). This number can be regarded as the figureof actual and potential authors of academic papers. From this point of view, the numberof researchers can be associated with bibliometric quantities. We shall take the view inthis paper.

While introducing these non-bibliometric quantities might give an impressionleading to more complicated discussion, but in fact, it could be rather useful in somecases. Arguably, it helps when building a model to study a phenomenon with suspectedrelationship like “scientists in larger countries more easily can find partners within theircountry than scientists from smaller countries” (MELIN, 1999), because it is hard to lookclosely into phenomena with possible network effects in human resources without anydirect/indirect consideration for the number of nodes (i.e., researchers).

Our previous paper proposed a new index for international scientific co-authorship,in order to exclude the influence of a country’s size on international co-authorship. Theindex, which can exclude such influence in part, is developed in consideration ofpublication activities at the level of researchers to some extent. Nevertheless,bidirectionality of international co-authorship cannot be considered in it. Thus, here weshall extend the index to two indices, which reflect bidirectionality of international co-authorship.

Indices

The basic concept of our indices is the adjustment of the share of international co-authored papers, which is executed by the calculation of likelihood of inbound/outbound access to co-authorships at the level of researchers.

In our previous paper (SHIRABE & TOMIZAWA, 2002), we introduced an index,which is defined as the likelihood that a researcher outside a specific country seizes anopportunity for publishing a paper with one or more researchers in the countryon one occasion. Considering bilateral character of international co-authorship (i.e.,overseas researchers’ access to researchers in a specific country for co-authorship

Page 3: Likelihood of inbound/outbound access to co-authorship

M. SHIRABE, H. TOMIZAWA: Inbound/outbound access to co-authorship

Scientometrics 59 (2004) 339

and vice versa), this index can be seen as likelihood of inbound access to co-authorships(LIA) in the specific country. The same index was named as “likelihood of overseasaccess to international co-authorship” in the previous paper, though. As induced in it,

kSe )1(logLIA −−= where k is the number of researchers outside A and S is the share

of internationally co-authored papers in total in A.At the same time, we can focus on A’s researchers’ access to overseas researchers

for co-authorship. That will lead to the other index, likelihood of outbound access to co-authorships (LOA), which is defined as the likelihood that a researcher in a specificcountry seizes an opportunity for publishing a paper with one or more researchersoutside the country on one occasion.

By definition, LOA is symmetric to LIA. Thus, we can induce LOA in the same wayas LIA was induced in our previous paper. That is to say, l

Se )'1(logLOA −−= , wherel is the number of researchers in A and S’ is the share of international papers co-authored by at least one researcher from A in papers authored/co-authored byresearchers outside A.

Figure 1 shows intuitive interpretations of LOA and LIA. Potential occasions tointernational co-authorship for researchers in A increases as research activities outsideA become active. A proxy variable of such activeness is the number of papersauthored/co-authored by researchers outside A. As the number of researcher in A (i.e.,actual/potential actors of international co-authorship in A) increases, the absolutenumber of A’s international co-authored papers is expected to increase. However, theshare (i.e., the relative number) of international co-authored papers decreases as shownin empirical studies (e.g., MELIN, 1999). The model LOA is based on reflects suchrelations. On the other hand, LIA represents the flip side of international co-authorship.LIA reflects an aspect of A’s international co-authorship in the opposite direction to thatof LOA, from the outside of A. Therefore, LOA and LIA are symmetric to each other,and yet they signify different but mutually related aspects of international co-authorshipas show in the next section.

A numerical example

Table 1 and Figure 2 show the results of our calculations. Due to our limitation inresources, we used data from an NSF report (1998). Consequently, likelihood ofinbound access to international co-authorship (LIA) in the table is not accurate. Inparticular, as the figure for total number of researchers is not that for the world but that

Page 4: Likelihood of inbound/outbound access to co-authorship

M. SHIRABE, H. TOMIZAWA: Inbound/outbound access to co-authorship

340 Scientometrics 59 (2004)

for OECD countries, each LIA is consistently bigger than its real figure. We should,therefore, view the result broadly on the basis of such inaccuracy and consistency. Thisis the reason we present normalized data in Table 1 and Figure 2.

While the correlation between number of researchers and share of internationallyco-authored paper (% of co-authored) is –0.70, the correlations of LIA and LOA tonumber of researchers are –0.49 and –0.43, respectively. Still, the correlations of LIAand LOA to shares of internationally co-authored paper remain high (i.e., at 0.90 and0.42 respectively). Thus, we can safely say that LIA and LOA are adequate indices forinternational co-authorship.

Figure 1. Intuitive interpretations of LIA and LOA

Page 5: Likelihood of inbound/outbound access to co-authorship

M. SHIRABE, H. TOMIZAWA: Inbound/outbound access to co-authorship

Scientometrics 59 (2004) 341

Table 1. Likelihood of inbound/outbound access to co-authorships1, and share of internationallyco-authored papers in 19952

Country LIA LIA LOA LOA SIC(×108) (ratio to an average (×108) (ratio to an average (% of

of 25 countries) of 25 countries) co-authored)Japan 9 0.45 2 0.14 27US* 14 0.71 9 0.61 32Australia+ 13 0.66 11 0.72 49Austria* 22 0.11 21 1.43 64Belgium 25 0.13 18 1.25 68Canada 15 0.80 16 1.07 52Denmark 23 0.12 22 1.51 63Finland 18 0.91 15 1.06 50France 18 0.93 13 0.88 53Germany 18 0.93 11 0.73 61Greece* 20 1.02 18 1.23 62Hungary 29 1.51 16 1.12 73Ireland 21 1.08 14 0.99 66Italy 18 0.93 15 1.05 49Mexico 24 1.24 7 0.50 67Netherlands 18 0.92 22 1.52 54New Zealand 15 0.79 21 1.48 58Norway 21 1.06 12 0.85 58Poland 26 1.32 8 0.56 78Portugal 29 1.48 8 0.52 72Spain 16 0.81 14 0.93 55Sweden 20 1.05 19 1.32 59Switzerland# 27 1.38 30 2.07 76Turkey 12 0.63 5 0.35 46UK 15 0.77 16 1.10 53

LIA: likelihood of inbound access to co-authorshipLOA: likelihood of outbound access to co-authorship (×108)SIC: share of international co-authorship 1 Numbers of researchers, which are used in calculation of LIA/LOA, are not those in 1995 (* 1993, + 1994, #1996).2 The data were collected by NSF (1998), and originally from OECD, EAS (MSTI database), ISI, and NSF.

Page 6: Likelihood of inbound/outbound access to co-authorship

M. SHIRABE, H. TOMIZAWA: Inbound/outbound access to co-authorship

342 Scientometrics 59 (2004)

Figure 2. LIA and LOA in 1995 (ratio to the average of 25 countries)Produced from NSF (1998).

Judging from Table 1 and Figure 2, the picture of international scientific co-authorship can be seen in detail with the indices. Therefore, it seems reasonable tosuppose that the introduction of this pair of indices is useful. For example, the indicescannot only suggest that the position of Japan is low in terms of international co-authorship as existing indices like share of international co-authored papers in nationalpublication show, but the present indices can also suggest that Japanese researchers failto seize occasions for international publication in comparison with the extent to whichother countries’ researchers seize occasions for international publication with Japaneseresearchers. That is to say, we can interpret that Japanese researchers are likely to bepassive in international co-authorship. Moreover, we can say that Mexico, Poland,

Page 7: Likelihood of inbound/outbound access to co-authorship

M. SHIRABE, H. TOMIZAWA: Inbound/outbound access to co-authorship

Scientometrics 59 (2004) 343

Portugal, and Turkey are among relatively passive countries in internationalco-authorship, although they differ in shares of international co-authored papers innational publication; they are in a severe excess of “imports” of internationalco-authorship.

But it should be paid attention to the point that this result is based upon a simplemodel, not discriminating between first author and other author(s) of co-authored paper.Different trend might be observed when calculation is based upon another model withmeasured data, which can be obtained without tremendous work. This type of analysisis a very interesting issue to be studied hereafter.

Discussion

In this paper, we proposed a pair of indices for international scientific co-authorship(i.e., LIA and LOA), which are based on a simple model of domestic and internationalco-authorships. These indices draw a reasonably clear picture of internationalcollaboration, concerning difference in patterns of international co-authorship amongcountries. For example, we can minutely understand the situations of countries likeJapan, Mexico, Poland, Portugal, and Turkey; they are passive countries with regard tointernational co-authorship.

Here we apply LIA and LOA for scientific and engineering fields as a whole, partlybecause of lack of resources. However, the indices can be calculated field by field, andsuch application is expected to show the detail of international co-authorship. Furtherresearch is required to clarify the applicable area of these indices.

References

FRAME, J. D., CARPENTER, M. P. (1979), International research collaboration, Social Studies of Science,9 : 481–497.

GLÄNZEL, W. (2001), National characteristics in international scientific co-authorship relations,Scientometrics, 51 : 69–115.

GLÄNZEL, W., DE LANGE, D. (1997a), Modelling and measuring multilateral co-authorship in internationalscientific collaboration. Part I, Scientometrics, 40 : 593–604.

GLÄNZEL, W., DE LANGE, D. (1997b), Modelling and measuring multilateral co-authorship in internationalscientific collaboration. Part II, Scientometrics, 40 : 605–626.

GLÄNZEL, W., SCHUBERT, A. (2001), Double effort = Double impact?, Scientometrics, 50 : 190–214.INSTITUTE FOR SCIENTIFIC INFORMATION (1996), Multinational coauthorship: Has collaboration peaked?

Science Watch, 7 : 1–2.LUUKKONEN, T., TIJSEEN, R. J. W., PERSSON, O., SIVERTSEN, G. (1993), The measurement of international

scientific collaboration, Scientometrics, 28 : 15–36.

Page 8: Likelihood of inbound/outbound access to co-authorship

M. SHIRABE, H. TOMIZAWA: Inbound/outbound access to co-authorship

344 Scientometrics 59 (2004)

MELIN, G. (1999), Impact of national size on research collaboration, Scientometrics, 46 : 161–170.MELIN, G., PERSSON, O. (1998), A bibliometric study of collaboration at some European universities, Journal

of the American Society for Information Science, 49 : 43–48.NARIN, F., WHITLOW, E. S. (1990), Measurement of Scientific Cooperation and Co-authorship in EC-related

Areas of Science, EC-Report EUR 12900.NARIN, F., STEVENS, K., WHITLOW, E. S. (1991), Scientific co-operation in Europe and the citation of multi-

nationally authored papers, Scientometrics, 21 : 313–323.NATIONAL SCIENCE FOUNDATION (NSF) (1998), Science and Engineering Indicators, NSF, Arlington.NISTEP: National Institute of Science and Technology Policy (2000), Science and Technology Indicators,

NISTEP Report No.66, http://www.nistep.go.jp/index-e.html, (NISTEP, Tokyo).SHIRABE, M. TOMIZAWA, H. (2002), Likelihood of overseas access to international co-authorships,

Scientometrics, 53 : 123–129.VAN RAAN, A. F. J. (1997), Science as an international enterprise, Science and Public Policy, 24 : 290–300.VAN RAAN, A. F. J. (1998), The influence of international collaboration on the impact of research results,

Scientometrics, 42 : 423–428.