Following the methodology established byPrice, this paper analyzes the empirical evidence of citation matrices. Using the data cleaned and tabulated by Computer Horizons, Inc. from the Science Citation Index data banks, it is shown that the non-diagonal elements of the square citation matrices can be accounted for very satisfactorily by assigning each nation a characteristic output and input coefficient in each field measured; the ratio of these coefficients provides a measure of quality. Deviations from this simple model give measures of particular linkage strengths between nations showing some evidence of preferences and avoidances that exist for reason of language, social structure, etc. It is also shown that the diagonal data can be accounted for by the measurable phenomenon that each nation seems to publish partly for the international knowledge system and party for its own domestic purposes. Thus, three parameters and a cluster map can parsimoniously describe the citation data within the limits of random error.
If the inline PDF is not rendering correctly, you can download the PDF file here.