The Profession Literature Loss in International Relations Charles A. Schwartz, Rice University Literature loss has two levels of meaning. In the research process, it refers to the difference between what is "out there" in some published form and what is likely to be found in a literature search. On an institu- tional level, it refers to the declining ability of research libraries to main- tain comprehensive collections in the wake of the extraordinary growth and price inflation of scholarly litera- ture since the 1970s. Recent technological innovations in the library field provide the means not only to measure literature loss on an institutional level, but also to alleviate it in one's own research work. An application of this technol- ogy to the area of international rela- tions is presented here. Part one of the paper is a brief history of the commentary on literature loss in the social sciences. Part two provides an analysis of the growing gap in the international relations area between book publication output and the aggregate holdings of 70 institutions in the Association of Research Libraries (ARL).' Part three describes the relevance of the new technology in greater detail, with an illustration for peace and conflict studies. Finally, part four draws some conclusions on the problem of literature loss at both the research and institutional levels, as well as on the closely related question of infor- mation retrieval. Commentary on Literature Loss The problem of literature loss first came to the fore 20 years ago when Robert Lane observed, in his presi- dential address to the American Political Science Association, that publication of political science arti- cles in general journals was hindering development of a cumulative body of research in the political science field itself (1972, 181). During the 1970s, a few scholars prodded the association to sponsor the organization of a more comprehensive bibliographic database. However, as Carl Beck reported toward the end of the decade, such efforts had to be aban- doned for lack of colleague support: The need for information retrieval systems in the social sdences is both real and apparent but, given the ability of many researchers to gather idiosyn- cratic research support, the need is not perceived as acute. For most of us, our information search behaviors were shaped by payoffs in the random walk that we undertook as students in col- lege and graduate schools. Although little has been done to analyze the nature of this walk in the social sci- ences, it seems a tenable hypothesis that our information behavior was shaped by "authorities": the authority of either a journal or a publisher who had attained high status in the profes- sion. The authority-conscious nature of our information usage is reinforced by the information explosion; to undertake that walk again with all the many information sources available is a traumatic experience. . . . The infor- mation milieu in the social sciences gets further compounded by the changing nature of the social sciences. Under these conditions we have a seri- ous problem of information loss (1977, 428-29). A similar effort to organize a com- prehensive database for peace research under the auspices of the United Nations also failed, although the UNESCO Yearbook of Peace and Conflict Studies was launched as a compromise measure in 1980 (Hoivik 1980; Beck 1980; Chatfield 1980). Discussion of the problem of information loss lapsed in the 1980s.2 However, a large-scale survey in 1986 of scholars' views on libraries and databases tends to corroborate Beck's concern about the relatively narrow, authority-conscious nature of literature-searching behavior. This survey found that, among 3,835 social scientists, 77% rely primarily on personal libraries for doing research; only 18% attach much importance to computerized card catalogs; and just 6% use online databases of periodical literature with any regularity (Morton and Price 1989). Evidently, social scientists tend to confront the expanding infor- mation environment by limiting their searches to relatively small, familiar parts of it. Measurement of Literature Loss Among Research Libraries Underlying any analysis of litera- ture loss is the problem of assessing the size of the universal stock of aca- demic books in a given field. That problem has no definitive solution, nor any prospect of one. Moreover, even if accurate book publication figures could be compiled, it would hardly make sense to consider them as realistic or worthy goals for research library collection develop- ment. The ancient ideal of the Library of Alexandria—to hold everything that was ever published— has been considered impossible and undesirable since the early part of the twentieth century. For the purpose of the analysis presented here—to assess the finan- cial performance of the nation's research libraries within reasonable bounds—book publication output is derived from the Online Computer Library Center, Inc. (OCLC) Union Catalog (OLUC) bibliographic data- base. The decision to use this data- base recognized the utter infeasibility of attempting to locate the publica- tion lists of hundreds of university and trade publishers, scholarly socie- ties, and the like (in peace studies alone, UNESCO lists 313 research bodies); and involved the simplifying assumption that just about any book that appeared over the past decade in the Americas or in Europe would have shown up in the OCLC-OLUC system, which contains the shared 720 PS: Political Science & Politics Literature Loss in International Relations FIGURE 1 Book Output on International Relations 1000 800 600 400 200 • (A) Total bookioutput Literature loss (B) Aggregate ARL holdings Mainstream literature (C) Subset of rare ARL holdings i Gray literature 1978 1979 1900 1981 1982 1983 1984 1986 1986 1987 OCLC-OLUC and OCLC/AMIQOS CACD databases cataloging of over 4,800 libraries, including the Library of Congress, in 26 countries. Figure 1 compares the aggregate holdings of 70 ARL institutions with total book output on international relations for the period 1978-87.3 Line A shows that output fluctuated from a low of 601 titles in 1978 to a high of 904 titles in 1984, with 867 titles in 1987. On average, 808 titles were published annually, with a stan- dard deviation of 89. Line B—covering that part of book publication output that is held in at least one of 70 ARL institutions —is derived from a different data- base, OCLC/AMIGOS Collection Analysis CD (CACD). It shows aggregate ARL holdings by year of publication (not by year of acquisi- tion). Such holdings in international relations fluctuated from a low of 525 for titles published in 1987 to a high of 662 for titles published in 1983. On average, 570 titles were acquired per year of publication, with a standard deviation of 38. Overall, ARL holdings dropped from 75% of book output in the late 1970s to 65% in the late 1980s. Line C demarcates the mainstream literature from the gray literature. The former includes books held by at least 10% of ARL institutions and comprises 55% of ARL aggregate holdings. The gray literature is held in less than seven institutions and mainly consists of small press publi- cations, foreign-language materials, conference proceedings, and legal texts. Table 1 clarifies the breakdown between the mainstream and gray lit- eratures on international relations by showing their relative distribution within ranges of ARL holdings. The distribution of mainstream literature is covered in the top nine ranges, starting with the 115 titles held in 90%-100% of ARL institutions and descending to the 10%-19% range with holdings of 838 titles. Strikingly different is the distribution of the gray literature; as many as 1,817 titles are held in only a few of the ARL institutions, with another 1,216 titles being in a "unique" category. Thus, the main point brought out by the table is that the holdings of books on international relations among ARL institutions are highly skewed, or disparate. Among the largest ARL institu- tions, the average collection holds 60% of the mainstream literature and 18% of total output. Among smaller academic libraries (with about 1 million volumes), the average collection holds 23% of the main- stream literature and 12% of total output. Research Applications of the New Technology The prospect of finding a relatively small and diminishing part of the book literature in a major research library reflects a substantial literature loss and presumably arouses interest in the OCLC-OLUC database and its scope for information retrieval. OCLC was established in 1967 to facilitate shared cataloging opera- tions. Books could be searched by author and/or title, but not by sub- ject. The OLUC subsystem, incor- porating subject and keyword retrieval capabilities, began in 1990. By then, the OCLC database con- tained over 20 million records, with 2 million being added each year. Searches can be directed to records in one or more of 365 languages, and to particular formats—books, serials, musical scores, sound recordings, machine-readable files, maps, and media (films, videos). Retrieval can be formatted to show which libraries own an item. Online charges are $24.00 per hour with an additional telecommunications charge of $8.40 per hour, but most searches can be TABLE 1 Distribution of Holdings Association of Research Libraries Books* Published 1978-87 on International Relations Percentage of 70 ARL Institutions 90-100 80-89 70-79 60-69 50-59 40-48 30-39 20-29 10-19 3-9 Unique Toial 1-100 Number of Titles per Range 115 191 184 154 143 182 225 364 838 1,817 1,216 5,429 •Excluding dissertations and U.S. govern- ment reports. December 1992 721 The Profession FIGURE 2 Book Output: Nuclear War, Peace Studies Thousands 14 1 2 10 M M - — 1 i 1 j : _ j I 1 1 I JJJ- Pi m I Sfuaies 1 i i .. i irwiaf / I j li il > / /_ 0 1800 1860 1900 OCLC-OLUC and OCLG/AMIQOS CACD databases 1960 2000 done by a trained librarian in a few minutes. OCLC-OLUC is particularly useful for interdisciplinary and historical research. Figure 2, for example, compares the growth of two book lit- eratures: nuclear war and peace studies. The number of titles on nuclear war has roughly doubled every five years since 1945, with the greatest growth in the early 1980s. The number of titles on peace studies, by contrast, displayed no pronounced trend during the nine- teenth century, doubled in size dur- ing the first 40 years of this century, doubled again during the next 30 years, and nearly again in the 1980s. Proliferation of peace studies reflects a broadening of the research agenda from "negative peace"—an absence of warfare—to "positive peace"—the presence of social and economic justice, with all that implies for radical transformation of national and international arrange- ments. There is strong debate whether this expanding agenda is leading to a loss of focus and coher- ence in the field: [As Hylke Tromp has said], "peace research has become what a black hole is in astronomy. There seems to be no social problem which in the final analysis does not have its legitimate place within peace research, and there- fore is absorbed by the definitional processes in peace research." In peace research, as elsewhere, rapid growth causes its own problems, e.g., in terms of integration. Extremely few, if any, of those who define themselves as peace researchers today have an over- view of the entire field, and hardly any peace research institute has more than some parts of it represented. To many peace researchers, it is a source of embarrassment that a field which prides itself on its transdisciplinarity should become so compartmentalized (Wiberg 1988, 45; Tromp 1980, xxvii). Some scholars have urged a return in peace research to its traditional focus on international security issues (Boulding 1989; Quester 1989; Soroos 1990). However, no such trend is evident; the amount of peace research that is cross-referenced in OCLC-OLUC with such "negative peace" descriptors as arms control or nuclear weapons has not changed much over the past decade, hovering around 10%. OCLC-OLUC is also useful for checking perceived gaps in the litera- ture. For example, Kenneth Boulding proposed a number of topics for peace historians (1989). Below are a few of these topics, along with the number of probably relevant records retrieved in a quick search for pur- poses of this paper: • Peace or war in Paleolithic, Neo- lithic, or hunting societies—153; • Relation of demography to war or peace—21; • Relation of agriculture to war or peace—1,040; • Relation of religion in the Middle Ages to peace—8. Conclusions At the institutional level, the mag- ' nitude of literature loss in the field of international relations is estimated as follows: • Thirty percent of total book out- put for 1978-87 is not held in any of 70 ARL institutions. • Another 40% of book output, being held in just one or a few ARL institutions, is also virtually lost except through EPIC. • Of the remaining 30% that com- prises the mainstream literature, the average ARL institution holds about a third of it, or roughly 13% of the books published on international relations during the period under review. • In the largest ARL institutions, a researcher will find about 60% of the mainstream literature and up to 18% of total book output for the period. As Figure 1 shows, literature loss grows rather steadily. Underlying the deteriorating position of the nation's research libraries are certain infla- tionary tendencies built into the scholarly communications system. During the 1978-87 period, the average price of a hardbound book on political science increased 180% (Bentley 1991, 404); and the cumula- tive inflation rate for institutional subscriptions to political and social science journals increased 270%, ris- ing to 356% by 1991 (Carpenter and Alexander 1991, 59). In any particu- lar field, new journals (13 in political science since 1978) also tend to "crowd out" book acquisitions. However, with the advent of OCLC-OLUC, researchers are no longer tied to their institutional card catalog systems. The importance of comprehensive retrieval, of course, 722 PS: Political Science & Politics Literature Loss in International Relations depends on the nature of the research problem at hand. OCLC- OLUC is especially useful for his- torical, interdisciplinary, or foreign- language searches. For example, 29% of the records in OCLC-OLUC on peace studies are non-English in lan- guage, as opposed to only 10% of the mainstream social-science period- ical literature (Gareau 1984). OCLC- OLUC should also help to alleviate what is called the "problem of unnecessary originality": [In some quarters, it is held,] all that is worth knowing is contained in the very latest journals or books. . . . Unfortunately, this view has led to an antihistorical bias and caused a great deal of unnecessary originality in our discipline. . . . Reincarnations of con- temporary ideas and supposedly new ideas might be greatly improved if ideational antecedents or previous investigations were sought out and incorporated into the " n e w " produc- tion of knowledge (Ault and Ekelund 1987, 652). More directly, an OCLC-OLUC search may have value in demon- strating to editors that a research proposal does indeed fill a gap in the literature. As a final note, while the OCLC- OLUC system for comprehensive book retrieval now extends online to some 4,500 institutions, the CACD system for collection analysis has only been acquired by 44 institutions. Combining the two systems to measure literature loss in a given field has not been done before, so there are no- other studies against which to compare the results of the analysis presented here. Notes The author would like to express his thanks to Helen Hughes of the AMIGOS Biblio- graphic Council for technical advice; and to Stephen P. Harter, David Kaser, and Herbert S. White for their helpful comments on an earlier draft of this article. 1. ARL currently comprises 117 institu- tions. For the period 1978-87 under review, 70 institutions are included in the OCLC/ AMIGOS Collection Analysis CD. The par- ticular holdings of two research libraries not included—Stanford University and the New York Public Library—may be of interest to peace historians, since those institutions col- lect comprehensively in the area of inter- national relations. Such holdings, however, would not materially affect the results of this study, which is based on the contemporary literature. 2. A few tangential studies were done in the library field. Two citation analyses of political science journals—one for the period 1910-60 (Robinson 1973), the other for 1968-70 (Stewart 1970)—showed the same dis- tribution of sources used in research: 30% from within the discipline, 40% from other social sciences, and 30% from outside the social sciences. Drawing on those studies, Elliot Palais (1976) assessed the coverage of various periodical indexes for 179 journals commonly cited in political science. He found the range of coverage to vary between 40% and 70% of the journals. Thus, reliance on any one index to search a broad topic means information loss in the range of 30%-60%. 3. The literature on international relations comprises the " J X " range of the Library of Congress classification scheme and falls under three primary subject headings: inter- national relations, world politics, and national security. In this hierarchical scheme, titles falling under the scores of more specific subject headings (e.g., peace, nuclear arms) are also retrieved and counted. However, the numbers are not precise owing to inevitable inconsistencies, duplications, and errors in any shared-cataloging system as OCLC. (For example, the "first" book on nuclear warfare in the database is the 1932 edition of Aldous Huxley's Brave New World, which actually dealt with chemical warfare; this record was created, apparently on the basis of a cata- loger's faulty recollection of the novel's theme, sometime in the 1970s.) Also, while researchers in the area of international rela- tions are obviously interested in far more of the literature than is classified in the " J X " range, inclusion of non-"JX" titles in any general analysis would be an arbitrary and endless process. For the analysis presented here, it was necessary to structure and bound the literature on international relations. The set of 5,500 titles, together with the set of 70 ARL research collections, are adequate to draw reasonably accurate patterns of litera- ture loss over the past decade. References Ault, Richard W. and Robert B. Ekelund. 1987. "The Problem of Unnecessary Originality in Economics." Southern Eco- nomic Journal 53(1): 650-61. Beck, Carl. 1977. "Information Systems and Social Sciences." American Behavioral Scientist 20 (January): 427-48. Beck, Carl. 1981. "Peace Research and Information Systems." In UNESCO Year- book on Peace and Conflict Studies 1980. Westport, CT: Greenwood Press, pp. 14-24. Bentley, Stella. 1991. "Prices of U.S. and Foreign Published Materials." In The Bowker Annual Library and Trade Book Almanac, ed. Filomena Simora. New Providence, NJ: R. R. Bowker, pp. 399-442. Boulding, Elise. 1989. "Introduction." In Peace and World Order Studies: A Cur- riculum Guide, ed. Daniel C. Palmer and Michael T. Kare. Boulder, CO: Westview Press, pp. 3-4. Boulding, Kenneth E. 1989. "A Proposal for a Research Program in the History of Peace." Peace & Change 14 (October): 461-69. Carpenter, Kathryn and Adrian W. Alexan- der. 1991. "Price Index for U.S. Period- icals." Library Journal 116 (April 15): 52-59. Chatfield, Charles. 1981. "The Dissemination of Peace Research Through Periodicals: A Tentative Review and Recommenda- tions." In UNESCO Yearbook on Peace and Conflict Studies 1980. Westport, CT: Greenwood Press, pp. 25-56. Gareau, Frederick H. 1984. "An Empirical Analysis of the International Structure of American Social Science." The Social Sci- ence Journal 21 (July): 23-36. Hoivik, Tord. 1981. "New Developments in Information and Documentation on Peace and Conflict Studies." In UNESCO Year- book on Peace and Conflict Studies 1980. Westport, CT: Greenwood Press, pp. 3-13. Lane, Robert E. 1972. "To Nurture a Disci- pline." American Political Science Review 66 (March): 164-82. Morton, Herbert C. and Anne J. Price. 1989. The ACLS Survey of Scholars: Views on Publications, Computers, and Libraries. Washington, DC: American Council of Learned Societies. Palais, Elliot S. 1976. "The Significance of Subject Dispersion for the Indexing of Political Science Journals." The Journal of Academic Librarianship 2(2): 72-76. Quester, George H. 1989. "International Security Criticisms of Peace Research." In Peace Studies: Past and Future, ed. George A. Lopez. The Annals of the American Academy of Political and Social Science 504 (July): 98-105. Robinson, William C. 1973. "Subject Disper- sion in Political Science: An Analysis of References Appearing in the Journal Lit- erature, 1910-1960." Ph.D. diss., Univer- sity of Illinois. Soroos, Marvin S. 1990. "Global Policy Studies and Peace Research." Journal of Peace Research 27(2): 117-25. Stewart, June L. 1970. "The Literature of Politics: A Citation Analysis.'1 Interna- tional Library Review 2(4): 329-53. Wiberg, Hakan. 1988. "The Peace Research Movement." In Peace and Research: Achievements and Challenges, ed. Peter Wallensteen. Boulder, CO: Westview Press, pp. 30-53. About the Author Charles A. Schwartz is the social sciences bibliographer at the Fondren Library, Rice University, Houston, TX 77251, and he has a Ph.D. in foreign affairs from the University of Virginia. December 1992 723