Braun Tibor, Glänzel Wolfgang, Schubert András: Országok, szakterületek, folyóiratok tudománymetriai mutatószámai 1981-1985 (A MTAK Informatikai És Tudományelemzési Sorozata 6., 1992)
Methodology
A. SCHUBERT, W. GLÄNZEL, T. BRAUN : SCIENTOMETRIC DATAF1LES Ad 3. Country assignment of papers covered by the SCI database is generally based on the country field of the corporate address record (which, in turn, follows the byline of the original publication). Decision is to be made on how coauthors are to be taken into account. In the case of all author count, all papers having at least one coauthor from a given country counts one full paper for the country in question. In the case of fractional counting each paper is divided among the contributing countries (various kinds of weighting may come into question). First author count is, actually, a special case of fractional counting, when weight of the country of the first named author is 1, while all other contributing countries have a weight of 0. All author count has several drawbacks: besides of the danger of counting equally unequal contributions, because of duplicate counts, the counts of a group of countries don't add up to a group total. Beyond doubt, the optimal method would be a fractional counting with proper weights, but how such weights could be determined? First author count, although apparently somewhat biassed, offers the unquestionable advantage of technical simplicity and, surprisingly, approximates any kind of reasonable fractional count within 10% in case of samples larger than 100 papers, and within 1% in case of samples larger than 1000 papers. Based on these arguments the method of first author count has been adopted in this study. Ad 4. The field/subfield classification of papers is a neuralgic point of all kind of scientometric evaluations. Even if each single paper would be classified separately (which is obviously impossible in case of analyzing millions of papers), justifiable objections could be raised in at least half of the cases. Without having much more arguments than that of authority, the method developed and successfully applied by the CHI/NSF database 1 0 was adopted here. Journals were clustered into subfields, subfields into fields, and each paper has been classified into the field/subfield of the journal in which it was published. The subfields and the constituent journals were extracted from the original SCI database". The 108 subfields were aggregated into 5 major fields: Life Sciences, Physical Sciences, Chemistry, Engineering, Mathematics. Some journals could not be categorized into any of the subfields, but were assigned directly to one of the major fields. Accordingly, papers published, say, in the Journal of the American Chemical Society have not been counted in any of the chemistry subfields, but appeared in the Chemistry major field counts. 126 journals have not been categorized into any of the fields/subfields; papers published in them appear only in the All Fields Combined aggregate counts. Scientometrics 16 (1989) 7
