College and Research Libraries Research Notes 77 2. Arthur M. McAnally and Robert B. Downs, "The Changing Role of Directors of University Li- braries," College and Research Libraries 34:103 (Mar. 1973). 3. Louis Kaplan, "Letter to the Editor: Directors of University Libraries," College and Research Li- braries 38:521 (Nov. 1977). 4. Jerry L. Parsons, "How Have They Changed? Characteristics of Research Library Directors, 1958 and 1973," Wilson Library Bulletin 50:613;617 (Apr. 1976). 5. William L. Cohn, "An Overview of ARL Directors, 1933-1973," College and Research Libraries 37:137; 143-44 (Nov. 1976). 6. Julie A. C. Virgo, Sarah How, and Annette Fern, comps., ACRL University Library Statistics, 1978-1979: A Compilation of Statistics From Ninety-Eight Non-ARL University Libraries (Chicago: American Library Association, 1980), p.1b. Six libraries with vacant or acting director positions were not included in the survey. 7. Parsons, p.613. 8. Parsons, p.614. 9. Organization and Staffing of the Libraries of Columbia University, a Case Study, prepared by Booz, Allen, and Hamilton, Inc., sponsored by the Association of Research Libraries and others (Westport, Conn.: Redgrave Information, 1973), p.45. 10. Parsons, p.616. 11. Parsons, p.614. 12. Cohn, p.141. 13. · Parsons, p.617. 14. For a brief discussion on the mobility of women directors, see WilliamS. Wong and DavidS. Zu- batsky, "The First-Time Appointed Academic Library Directors, 1970-1980, A Profile," Journal of Library Administration 4:43-44 (Spring 1983). Changes in Rank Lists of Serials Over Time: Interlending versus Citation Data Maurice B. Line The British Library Lending Division carried out three major surveys of its lending patterns in 1975, 1980, and 1983. The rank list of seri- als requested for loan showed considerable vari- ation over time. There was also low overlap in the top titles requested. A comparison was made of these rankings with the rankings from Journal Citation Reports (JCR) produced by the Institute for Scientific Information. The JCR rankings had a high degree of overlap, 95 percent for the top 100 in Science Citation In- dex, while the Lending Division had only a 57 percent overlap. The reasons for this variation are discussed. Three major surveys carried out by the British Library Lending Division in 1975, 1980, and 19831.2' 3 produced rank lists of serials in order of demand. Comparisons of tP.ese rank lists showed very consider- able changes over time, suggesting that it might be dangerous to rely unduly on a rank list relating to one year. Changes in the precise rank order would be expected; what was unexpected was the low overlap in the top titles requested. For interest, a similar, comparison was made between the rank lists produced by the Institute for Scientific Information and Maurice B. Line is director general of the British Library Lending Division, Boston Spa, Wetherby, West York- shire LS23 7BQ, United Kingdom. This note also appeared in Interlending and Document Supply 12 (Oct. 1984). I am grateful to Betty Smith, bibliographic research officer at the British Library Lending Division, for doing the comparisons between the lSI rank lists; and to Trevor Palmer, of the Lending Division's Research Section, for making comments and suggestions on a draft of this note. 78 College & Research Libraries published annually as Journal Citation Re- ports (JCR)-a volume of the annual Science Citation Index (SCI) and Social Science Cita- tion· Index (SSCI). The results are given in table 1. Nineteen eighty-three data were not available for either SCI or SSCI, and SSCI data were not available for years ear- lier than 1977; the nearest approximations were therefore used for comparison. The differences are striking. At the most extreme, there was 95 percent overlap in SCI's top one hundred in 1979 and in 1982, compared with only 57 percent in the Lending Division's top one hundred in 1980 and 1983. At the least extreme, there was a 78 percent overlap in SSCI' s top two hundred in 1977 and in 1982, <;ompared with 56 percent in the Lending Division's top two hundred in 1975 and in 1980. The high overlap in the lSI lists is remarkable, since, as Urquhart pointed out, there are good statistical reasons for expecting sub- stantial change. 4 What are the reasons for these large dif- ferences between Lending Division and lSI data? The absence of humanities titles from]CR will not have had any effect, be- cause there are very few humanities seri- als among the titles most requested from the Lending Division. One obvious expla- nation is sampling error, which would ap- ply to the Lending Division data but not to lSI's, which are compiled from a whole population of citations. A considerable difference in the precise rank order in the Lending Division lists would be expected. However, the actual numbers of requests January 1985 for the most used serials in the Lending Division samples are large, and sampling error alone is very unlikely to account for the much smaller overlap, especially as both comparisons (1975-80 and 1980"""83) yield similar results. Another explanation, which is likely to be nearer the truth, is that interlibrary loan (ILL) demand is subject to much more fluctuation than citations in journals; it is affected by local finances-for example, budgetary restrictions may favour more ILL demand because acquisitions are re- duced, or less because economies may be sought in interlibrary borrowing. The rela- tive volume of demand made by academic and industrial libraries may change sub- stantially (there was in fact a shift towards industrial library use between 1980 and 1983, though it was not very large and is unlikely to have had more than a small ef- fect on the rank order). Interests change: journals in the life sciences and related subjects and in electronic technology rose up the lists between 1975 and 1983 at the expense of such subjects as pure chemis- try and physics. This is a known element in the differences between Lending Divi- sion lists, but one might expect it also to apply to the lSI lists, if one dismisses the possibility that interests in the U.K. (from which about three-quarters of serial de- mand on the Lending Division comes) change more quickly than interests in the world at large, the U.S. in particular. However, one major difference between citations and ILL demands is that citations TABLE 1 Top x titles on lists 100 200 300 400 500 1,000 SIMILARITY OF RANK LISTS OF SERIALS AT THREE-YEAR AND FIVE-YEAR INTERVALS: ILL DATA VERSUS CITATION DATA IDData Serials Requested from Lending Division 1975/1980 1980/1983 (5 years) (3 years) 60 56 54 56 56 56 57 62 61 62 61 60 % of Titles Common to Both Lists Citation Data Science Citation Index 1975/1980 1979/1982 (5 years) (3 years) 88 83 86 88 87 83 95 93 93 91 92 Social Science Citation Index 1977/1982 1979/1982 (5 years) (3 years) 83 78 81 79 80 88 84 87 88 87 To be read as follows: Of the top 100 titles in the 1975 and 1980 ILL rank lists, 60 percent were common to both; of the top 300 titles in the 1979 and 1982 SO rank lists, 93 percent were common to both . are made mainly by authors in academic institutions, whereas ILL demands for se- rials come about equally from academic in- stitutions and from industrial and com- mercial organizations (in fact, both categories accounted for 35 percent of de- mand for serials in another survey carried out in 1983 at the Lending Division). Pos- sibly academic requests show more stabil- ity than other ILL requests: the data col- lected by the Lending Division did not include information on requesting organi- zations, so it is not possible to test this hy- pothesis. The upsurge of interest in bio- technology, bioengineering, other life science-related subjects, and electronic technology would be reflected more rap- idly in industry than in academic institu- tions, where relatively little staff move- ment would have occurred over a short period and where the existing staff would presumably have continued to write and cite as before, whether they are physicists or biochemists. There are other reasons why citation rank lists might show more stability. Serials cited most are likely to be more "academic" in nature, and these may constitute a more stable population than serials aimed at the industrial mar- ket. Self-citation (by serials and authors) would favour stability, as would the fact that some works are cited repeatedly, not Research Notes 79 necessarily because they are used very heavily but because they are standard pa- pers that must be cited or because they are 'lifted' from bibliographies in other arti- cles. These factors would not only help to explain the differences between the lSI rank list comparisons and the Lending Di- vision comparisons, but they would mean that citations, while they might reflect tol- erably well the use being made of aca- demic libraries as a whole (not in individ- ual libraries, where local factors are likely to be influential), are a poor indicator of to- tal serial uses. It may well be that the instability of the Lending Division rank lists is a little 'un- real,' in the sense that a longer survey pe- riod or a much larger sample would re- duce the differences. The stability of the lSI rank lists is likely to bear much less re- lation to reality, in that citations are much more stable than actual uses. Whatever the reasons for the differences described, they are a matter of some practical inter- est. A national core collection of serials de- signed to serve academic institutions might be identified, with more confidence that it would be reasonably stable over a period of time than a collection aiming to serve all types of organizations, let alone one designed to serve mainly industry. More research into this matter is desirable. REFERENCES 1. C. A. Bower, "Patterns of Use of the Serial Literature at the BLLD," BLL Review 4:31-36 (1976). 2. Ann Clarke, "The Use of Serials at the British Library Lending Division," Interlending Review 9:111-17 (Oct. 1981). 3. Karen Merry and Trevor Palmer, "Use of Serials at the British Library Lending Division in 1983," Interlending and Document Supply 12:53-56 (Apr. 1984). 4. J. A. Urquhart, "Has Poisson Been Kicked to Death? A Rebuttal of the British Library Lending Divi- sion's Views on the Inconsistency of Rank Lists of Serials," Interlending Review 10:97-99 (Nov. 1982). Discove•· why PAIS BULLE liN and PAIS I IIREIGN LANGUAGE INDEX a•e the sou•ces to use when the subject is •••• • Economics • Business • Finance • Public Policy. Issues • Statistics • International Relations • Trade and Commerce • Political Science • Government It's n. o mystery why thousands of people use PAIS. They know that when it comes to coverage of the social sciences, PAIS is unmatched in its breadth of coverage and in the quality of the information it contains. PAIS' coverage is more than just political science and government. Fields such as business, economics, statistics, and international trade are also covered in depth. And the PAIS source list includes more than just journals. Books, government documents, conference proceedings, statistical reports, hearings, and many other sources are carefully examined for vital new information. All the evidence points towards the use of PAIS. Examine the PAIS BULLETIN and the PAIS FOREIGN LANGUAGE INDEX closely next time you are searching for information on any important public issue. Public AHairs lll#tlrmatioll 5ert~lce 11 W. 40th Street • New York, NY 10018 • (212) 736-6629 Serving information seekers since 1914. PAIS is also available online through DIALOG, BRS, ttnd Data-Star. Ill Ill