Citation Analysis: Queries and Caveats

Study of citation patterns has become a leading analytical tool in the sociology of science, and offers an enticing way to evaluate the research productivity of individual scientists. However, there remain some basic conceptual and nmethodological concerns. Debate touches on the validity of citation counts as an indicator of 'scientific quality' (of papers or authors), of recognition, of scientific diffusion, and of utilization (who cites whom).I To Menard, citations are the only enduring record that a paper is used,2 and Cole and Cole take citations to be a roughly valid indicator of influence.3 But Kaplan wonders what citations mean, in the absence of know'ledge about actual norms and practices.4 Plausible reasons for citation include selectively supporting one's contentions (i.e. 'persuasion'), indicating knowledge of the literature, providing clues about content to citation index scanners -as well as acknowledging genuine communication.5 Citation does not actually prove a reference has been read (or even seen),6 yet journal citations appear to be the best practical indicator of the worth of research.7 Citations may be used to evaluate papers (either individually or in groups e.g. experimental vs. theoretical physics; or short notes vs. full reports); or, similarly, to evaluate journals; or scientists. This paper concerns citation analysis of scientist-authors (both individuals and groups), and discusses the methodological problems involved. The problems to be faced depend on the uses to which the measures are to be put. The problems and opportunities of a supervisor evaluating the citation profile of a young scientist (an arguable practice) are poles apart from those of a sociologist studying groups of scientists in an international specialty, or of a historian studying its development. However, there are some generalizable concerns: these include the possibility that radical work will go unrecognized (and uncited) by contemporaries: that outstanding ideas become widely accepted and not formally cited (e.g. the Mossbauer effect);8 and that citation counts do not