«Andrés Gregor Zelman The University of Amsterdam 2002 ii Mediated Communication and the Evolving Science System: Mapping the Network Architecture of ...»
As expected there was a Policy and Informatics bias found as evidenced by the most frequently occurring journals; namely: Scientometrics and Research Policy, and by Social Science Informatics in the Journals in Citing articles. Next, we observed the
country distribution for the SOEIS member and associates publications. Figure 6.3:
SOEIS – All Publications by Country shows the most frequently published countries in the SOEIS citation environment.
98 100 90 80 70 60 50 40 30 20 10 0
It was unexpected to find Spain as the most frequently published country as it was not one of the six countries that directly participated in the SOEIS research project. Upon closer inspection it proved to be due to an unusually high number of co-authors in the relatively few Spanish contributions. The Netherlands, England, Germany, Italy, and Switzerland all appear in Figure 6.3; all countries directly involved with the SOEIS are represented, with the notable exception of Greece.4 The publications of the SOEIS core were then compared on the basis of most frequently published country; these results are shown below in: Figure 6.4: SOEIS – Core Publications by Country.
The core publications of the SOEIS members was heavily represented by the Netherlands, and was trailed by the USA, England, Italy, France and finally Switzerland. The USA and France are the only non-project countries here. It is 4 This is not to say that there were no Greek publications during this period; electronic publications were not included in this analysis as they are not included in the Web Of Science databases.
Next the co-authorship distribution between the SOEIS group, and the SOEIS members proper was compared. A final means of understanding the architecture of the publication environment of the SOEIS was obtained through the calculation of coauthorship activity in the Cited database. Clearly, pursuing the co-authorship relations in the Citing database would prove interesting for its overlap with the Cited database;
however, since the analysis is focused on the dynamics of the SOEIS communication system the co-authorship analysis was limited to just the articles cited by the SOEIS group. Table 6.2: Co-authored Publications, below, shows differences between the SOEIS-Core and the SOEIS-All.
The rate of co-authored publications per year in the cited database remained consistent with the rate of the overall publication in the cited architecture, with the exception of 1999, as outlined above. The number of co-authorships between the people in the SOEIS group was highest in 1999 and 2000. More generally, it was found that the co-authorships were proportionately higher in the SOEIS group than in the SOEIS core.
Network A means of measuring the network parameters of the SOEIS publication environment was then performed. Note the difference between the SOEIS – Core and SOEIS – All;
the Cited references were obtained by searching the citations of the individual authors in the SOEIS group and SOEIS core.5 The tabulated results are displayed below in Table 6.3: Total Cited & Citing References.
When the percentage ratio was compared between the cited references of SOEIS-Core and SOEIS-All, the core was to found fluctuate greatly over the five years, with a notable dip in the 1998 and 1999 where the SOEIS members arguably spent more effort on the project than on the publication of articles. By contrast, the cited 5 The SCI, SSCI and AHI indexes were queried for the individual authors of the SOEIS group, and their cited were references collated. It should be noted that in some cases we were unable to confirm if all publications identified with each respective name were attributable to that specific individual. All results found using the author names were used in the analysis.
100 references for the whole SOEIS group were found to remain relatively consistent over the five years.
The following three charts clearly illustrate the referencing preferences of the cited (who the SOEIS cited) and citing (who were citing SOEIS relevant publications) dimensions of the SOEIS publication environment. We begin with the sum of the most frequently cited references by the SOEIS core: Figure 6.5: Sum of Most Cited Referenced Journal Frequency (SOEIS-Core), then the sum of most frequently cited references of the whole SOEIS group: Figure 6.6: Sum of Most Cited Referenced Journal Frequency (SOEIS-All), and finally the sum of most frequently citing references: Figure 6.7: Sum of Most Citing Referenced Journal Frequency.
Figure 6.5: Sum of Most Cited Referenced Journal Frequency (SOEIS-Core) The most cited journals referenced by the SOEIS-Core reflect the bias towards Policy and Informatics Journals as described above.
The top five most frequently occurring referenced journals in the publication environment include Research Policy, Scientometrics, The Journal of the American Society for Information Science, Social Science Informatics, and the Journal of Information Science, respectively. Of these top five journals, the most frequently occurring (Research Policy) stands alone; that is, without journals with which one would expect it to be clustered (such as Science Public Policy and Research Evaluation, for example). By contrast, Scientometrics shares a relatively high frequency of being cited with those journals with which it is often clustered (The Journal of the American Society for Information Science, Social Science Informatics, and the Journal of Information Science). Collectively these journals suggest a cognitive bias of the SOIES-Core publication environment towards Policy and Informatics, with an emphasis on the Social Sciences. Next, all articles referenced by the SOEIS group and its associates were grouped and tabulated by journal. Using the aggregate publication data for the five year period between 1996 through 2000, a summation of the most cited referenced journal frequency was calculated. The results are shown below in Figure 6.6: Sum of Most Cited Referenced Journal Frequency (SOEIS-All).
101 500 450 400 350 300 250 200 150 100 50 0
Similar to the SOEIS-All, the SOEIS-Core sum of most cited referenced journal distribution (shown above in Figure 6.6) reveals a cognitive bias towards policy and informatics journals. Research Policy figures as the most frequently referenced journal by the SOEIS group, closely flowed by Scientometrics. However, neither journal is shown to cluster with similar journals as witnessed in the distribution for the SOEIS-Core.
Next, all of those articles which referenced the SOEIS related publications (the citing articles) were grouped and tabulated by journal. Using the aggregate publication data for the five year period between 1996 through 2000, a summation of the most citing journal frequency was calculated. Figure 6.7: Sum of Most Citing Referenced Journal Frequency, below, shows the results.
The sum of the most citing referenced journal frequency analysis identified Scientometrics as the main journal authors use most frequently to cite SOEIS materials, followed by Research Policy and Social Science Informatics. Similar to the results of the Most Cited Frequency analyses, described above, Research Policy 102 stands alone, without journals one would expect to be of similar citing frequency, such as Science Public Policy and Research Evaluation. Unlike the most cited analysis, here in the most citing it is found that with the notable exception of Social Science Informatics, Scientometrics does not share high citation frequency rates with any of the journals with which it clustered in the cited analysis (the Journal of the American Society for Information Science, Social Science Informatics, and the Journal of Information Science). Similar to the cited analysis, when taken together these three most frequently occurring journals (Scientometrics, Research Policy, and Social Science Informatics) suggest a cognitive bias towards Informatics and Policy of those who cite SOEIS related materials.
As illustrated above, once the sums of most frequently occurring journals in both cited and citing dimensions were compared, it was determined that Research Policy and Scientometrics were the most crucial journals for the SOEIS publication environment.
The network analysis has found that there is a direction in the information flow evident here: the SOEIS group has been shown to predominantly cite Research Policy, while it is itself cited by articles in Scientometrics. It would appear that the SOEIS group uses Research Policy to be visible in the Scientometrics, or ‘infometric’, community. In the following section: System, the publication environments of specifically these two journals on both the cited and citing dimensions are addressed.
This will aid an understanding of these environments with respect to which journals cluster with Scientometrics and Research Policy, and more importantly how these journals in turn appear to interact upon each other. The concern here is what happens at the interface between these journals, given their centrality to the SOEIS’ citation environment.
System The system oriented analysis entailed the aggregation of journal-journal citation data over the years 1996 – 2000 using Research Policy and Scientometrics as the entrance journals for the analysis. These two journals have been selected as the seed journals because they are the two with the highest sum of most cited and most citing journal frequencies, as described above. A cut-off rate of 1% has been used of the total cited and total citing of Research Policy, and a cut off rate of 0.5% for Scientometrics.6 The total is normalized over the journals contained in the Social Science Citation Index (SSCI). Examined together, the cited and citing environments of these two journals reveal the relevant relationships between these journals, and show the significance of changes in the disciplinary focus of the SOEIS group over the time dimension.
The resulting factor loadings are described below and are graphically represented using multi-dimensional scaling plots in order to illustrate the significance of changes taking place. The cited environment is examined as distinct from the citing. The cited database, as the collective sum of the texts referenced by SOEIS members and the periphery group of researchers, refers to the systemic aspect of the SOEIS that 6 Originally this research was performed using a cut-off rate of 1% for both Research Policy and Scientometrics. It was determined that a more effective means of comparing these results would be to lower the cut-off threshold for Scientometrics since so few journals were revealed using the higher cutoff rate.
Research Policy and Scientometrics have been selected as the seed journals for the analysis of the systemic dimensions of the SOEIS publication dimension precisely because they have both been identified as crucial to both the cited and citing environments. It is important to explore the journal-journal relationships for each seed journal in order to provide a basis for comparison. What concerns us here is the interface between these journals which we have identified as central to the operation of the SOEIS publication dimension. In what follows I will first review the cited environment of both Research Policy and Scientometrics chronologically for the years 1996 to 2000, and in the next section will address the citing environment for the same period.7 Commentary concerning the overlap of these two fields and their interface will be provided at the end of the section.
Cited 1996 In the Research Policy cited environment of 1996, Research Policy loaded with Scientometrics, and appeared near to the Sociology, Regional Planning, and Traditional Management clusters. Scientometrics was also near Sociology, but appeared closer to STS, Innovation, Research & Development, and Technological Management than Research Policy. There was also a notably significant Economics cluster in the environment, though distant. The SOEIS referencing behaviour in 1996, at least in the context of Research Policy, exhibits an interesting link between Research Policy and Scientometrics as they load together yet appear divided between certain literatures given their positioning.
In the Scientometrics cited environment for 1996, Scientometrics loaded with Informatics Journals and appeared near to the STS, Information Science, and Education literatures, and near Research Policy which loaded with the Management cluster. Similar to the cited environment of Research Policy for 1996, Economics and Sociology clusters were evident. Interestingly, Scientometrics and Research Policy clustered near each other – Research Policy loading with Management, and near the Economics cluster.
The factor analyses performed here are best perceived using multi-dimensional scaling plots whereby the distances between relevant clusters can be visually approximated. Only a selection of the results will be presented here to highlight the
central findings. Below, Figure 6.8: Research Policy Cited 1996 and Figure 6.9:
Scientometrics Cited 1996 reveal the approximated distances between journal clusters.
See Appendices C-II and C-III for larger versions of the plots displayed below, and for the respective factor loadings of Research Policy and Scientometrics 1996 – 2000.
7 For the full names of the abbreviated journals referred to in this analysis, see Appendix C-I.