Visualizing a Discipline: An Author Co-Citation Analysis of Information Science, 1972-1995

This study presents an extensive domain analysis of a in articles, regardless of which of their works are cited. discipline—information science—in terms of its auACA synthesizes many such counts. Now that 15 years thors. Names of those most frequently cited in 12 key have passed since it was introduced by White and Griffith journals from 1972 through 1995 were retrieved from So(1981), the present writers wish to explore this literaturecial Scisearch via DIALOG. The top 120 were submitted based technique as a means for contributing to intellectual to author co-citation analyses, yielding automatic classifications relevant to histories of the field. Tables and history. As in that earlier article, we shall use authors graphics reveal: (1) The disciplinary and institutional affrom information science to illustrate that, although ACA filiations of contributors to information science; (2) the is applicable in any discipline (Bayer, Smart, & Mcspecialty structure of the discipline over 24 years; (3) Laughlin, 1990; Eom, 1996; Hoffman & Holbrook, 1993; authors’ memberships in 1 or more specialties; (4) inertia McCain, 1986), readers of this journal will best be able and change in authors’ positions on 2-dimensional subject maps over 3 8-year subperiods, 1972–1979, 1980– to judge its validity when it is applied to their own field. 1987, 1988–1995; (5) the 2 major subdisciplines of inforUltimately, we are interested in ACA as a way of visualizmation science and their evolving memberships; (6) ‘‘caing a field through a representative slice of its literature, nonical’’ authors who are in the top 100 in all three and we shall develop evidence from it for those who subperiods; (7) changes in authors’ eminence and influence over the subperiods, as shown by mean co-citation would define information science in terms of its specialcounts; (8) authors with marked changes in their ties. Unlike McCain (1990) or White (1990a, 1990b), mapped positions over the subperiods; (9) the axes on the present article is not an introduction to ACA but an which authors are mapped, with interpretations; (10) eviexercise in domain analysis, in the sense developed by dence of a paradigm shift in information science in the Hjørland and Albrechtsen (1995). On one level, it relates 1980s; and (11) evidence on the general nature and state of integration of information science. Statistical routines to the creation of an intellectual framework for informainclude ALSCAL, INDSCAL, factor analysis, and cluster tion science (Buckland & Liu, 1995, p. 389); on another analysis with SPSS; maps and other graphics were made level, it relates to the visualization of linguistic data (Wilwith DeltaGraph. Theory and methodology are suffiliams, Sochats, & Morse, 1995). ciently detailed to be usable by other researchers. Because the data of ACA are merely noun phrases and associated citation counts, they produce history of the Introduction cliometric sort, which leaves out almost all the good parts, such as who had shouting matches, who slept with whom, Co-citation analysis shows that literatures cohere and change in intelligible ways over time, whether one defines and what actually gave rise to the most significant work. them in terms of individual articles and books, whole All ACA can do, for the historian of ideas or any other oeuvres, or journals. Author co-citation analysis (ACA) party, is to identify influential authors and display their is the subcategory that maps oeuvres, and, by implication, interrelationships from the citation record. It is no substithe people who produce them. The raw data are counts tute for extensive reading and fine-grained content analyof the times that selected author pairs are cited together sis, if someone is truly interested in the intellectual history of a field. But with Social Scisearch data, which start in 1972, it is now possible to show author relationships over Received December 3, 1996; revised March 6, 1997; accepted March several multi-year periods—in the present study, three 6, 1997. periods of 8 years each, overlapping most of the existence of the American Society for Information Science, which q 1998 John Wiley & Sons, Inc.