College and Research Libraries HARRY M. KRIZ Subscriptions vs. Books m a Constant Dollar Budget Citation analysis was used as an aid in collection development in the field of engineering. It was found that for graduate students books are of greater importance than journals. Substantial reductions in the number of subscrip- tions released funds for the purchase of books within the limitations of a constant dollar budget. Library usefulness appears to have been increased. INFLATION HAS LED MANY LIBRARIANS to examine critically their collection build- ing practices. Particular attention has been given to rapidly rising subscription prices, in part because subscriptions consume the major portion of the acquisitions budget in many university libraries. Another factor contributing to the emphasis on subscrip- tions has been the availability of rank-order listings of useful and/or important peri- odicals as derived from citation analysis. 1 While attention has been focused on the usefulness of some periodicals relative .to other periodicals, much less attention has been given to the usefulness of periodicals. relative to other forms of library materials. For example, one might ask whether it is better to continue a subscription to a sel- dom used translation journal or to purchase duplicate copies of a few heavily used mon- ographs. Traditionally, academic librarians have re- sponded to funding shortages by maintain- ing periodical subscriptions in preference to buying new books. Thus one reads from time to time of libraries that have sus- pended book buying. 2 Less anecdotal evi- dence of the tendency to stop buying books in order to continue subscriptions is pro- vided by a recent report, which states that academic libraries increased expenditures Harry M. Kriz is agriculture-engineering li- brarian, West Virginia University, Morgantown. for periodicals by 36 percent in the two-year period ending in 1975, though prices in- creased by only 34 percent. In contrast, book expenditures increased only 2.8 per- cent during these two years, although book prices increased by 8 percent. 3 These figures should not be surprising in light of White's discussion of the dramatic changes in library materials budgets that occurred in 1969-73. The transfer of funds from the book to the serials budget resulted in a "drastic" reduction in the acquisition of books during that period. 4 The tendency to buy subscriptions rather than books probably is based on the belief of librarians and faculty that periodicals are the fundamental tools of research. It is rea- soned that one of the characteristics of a good research library is a large number of currently received periodicals. The extraor- dinary increases in the prices of periodicals during the past several years make it neces- sary to examine in more detail the assump- tion that periodicals deserve precedence when a choice must be made between books and subscriptions. A study of the library materials used by students writing master's theses in the West Virginia University (WVU) College of Engi- neering suggests that, at least in engineer- ing, a large number of subscriptions is less important than an adequate book collection. This conclusion is based on the measure- ment of the use of library materials in terms of citation frequency. As with any quantita- I 105 106 I CoUege & Research Libraries • March 1978 tive measure of library use, there are limita- tions inherent in the use of citation data. However, the results have provided useful guidance in budget decisions at West Virginia University, and similar considera- tions may prove useful to librarians in other subject areas at other academic institutions. BACKGROUND Incentive for a study of the use of peri- odicals relative to books was provided by several published reports showing that there are profound differences in the citation pat- terns in different fields. Jones and his asso- ciates compared results of several studies and found that the percentage of references that cited periodicals ranged from a low of 21.5 percent in English history to a high of 92.7 percent in chemistry. In general sci- ence, 82 percent of the references cited pe- riodicals . 5 Fussier's study of the literature of chemis- try and physics showed the remarkable sta- bility of citation patterns in these two fields. The percentage of references that cited journals had remained virtually unchanged for nearly fifty years. 6 Garfield has commented frequently on the subject-dependent nature of citation patterns. For example, only 20 percent of references processed for the Science Cita- tion Index in 1975 cited nonjournal items, but 62 percent of the references in the So- cial Sciences Citation Index cited such items. 7 The relatively low incidence of jour- mil citation in the social sciences is shared by some fields of applied science ~ In 1972 it was found that 45 percent of references in applied chemistry journals cited nonjournal items, hut only 9.6 percent of references in chemistry journals cited such items. 8 One might expect a similar emphasis on nonjournal citations in the field of engineer- ing. Indeed, highly cited works in engineer- ing tend to be books rather than journal ar- ticles, a characteristic of the literature which distinguishes engineering from other scien- tific and technical fields. 9 •1o These studies suggest that decisions to continue or cancel subscriptions in an academic library must be made on a subject-by-subject basis . When budget limi- tations require a choice to be made between the purchase of books and subscriptions, it seems that subscriptions would be the bet- ter investment in the field of chemistry. In applied chemistry, however, lesser used subscriptions might have to be canceled in order to maintain an adequate book collec- tion. Although studies of citation patterns in the literature of a given subject field can be useful, they do not provide an adequate base upbh which to make decisions in a given library. A study of the citation pat- terns of the individual library's particular group of users should be made before mak- ing a decision concerning the purchase of librarY materials to support that user group. There are several reasons for making a study of the local users: 1. A library must serve the local group of authors, not a subject field. Thus a librarian needs to know what is being used and cited by those who use the library, not what is being cited by those who publish in a par- ticular set of journals. In this regard it should be noted that the literature of a spe- cialty must be distinguished from the litera- ture of interest to workers in that specialty. Garfield pointed out that the literature used by agricultural scientists is not the same as the literature produced by agricultural sci- entists.11 Further, a listing of the most cited physics journals is quite different from a list- ing of the journals most cited by physi- cists.12 Finally, a study by Scales13 indicates that rank-order lists of journals for a field as a whole do not necessarily correspond to the particular journal needs even in large libraries with a large and diverse group of users. 2. Studies of the literature as a whole provide only an average measure of the use of library materials. Local conditions may lead to wide departures from this average, but there are no data available to indicate the extent of variation among different user groups in the same general subject field. 3. The use of library materials by stu- dents may differ considerably from the use by professionals publishing in a field. Studies based only on references in journals do not provide adequate data on the educa- tional needs served by an academic library. 4. Published studies do not usually con- sider use variations by subfield, an impor- tant consideration for the academic librarian r who must balance the conflicting needs of different departments. 5. A study of the local user group can provide a time-averaged measure of the use of library materials. The ranking of a journal on a list can be strongly dependent on the time period under study. 14 RESULTS AND DISCUSSION Use of engineering library materials was measured by analysis of references in mas- ter's theses accepted by the WVU College of Engineering during the four years 1971- 74. As the thesis is not a degree require- ment, only 126 students out of the 296 graduates wrote a thesis during this period. A complete report on the results of this study is available on request from the au- thor. An analysis of the data on journal cita- tions has been published elsewhere. 15 The present discussion is concerned only with the relative use of journals and books. Table 1 presents a summary of the data on this topic. It can be seen that only one-third of the references in the WVU engineering master's theses cited journals and that fully two- thirds of all references cited nonjournal items. This is the same percentage found by Garfield in his study of highly cited engi- neering works. 16 The percentage of non- journal citations varied from a low of about 50 percent in chemical engineering to a high of almost 90 percent in aerospace en: gineering. The figure for chemical engineer- ing is very close to the 45 percent quoted above for applied chemistry. These results indicated that information sources other than journals were of primary Subscriptions vs. Books I 107 importance in library support of the engi- neering master's program at WVU. In addi- tion, it was recognized that relatively mas- sive reductions in expenditures for subscrip- tions to engineering journals would have virtually no measurable effect on the availability of useful journals to support en- .gineering students. 17 (Holland recently re- ported that major budget cuts in the engi- neering library at the University of Michi- gan would have little effect on service to users. 18) Thus it was decided that when a choice had to be made between buying books or subscriptions, the subscriptions would be canceled. The decision to buy books rather than subscriptions was reinforced by several con- siderations. First, the space problem at WVU is severe. The engineering book col- lection has been heavily weeded to the ex- tent that the entire collection increased by only 10 percent in five years despite an an- nual acquisitions rate of 10 to 12 percent throughout this period. Second, bound vol- umes of engineering journals occupy 55 percent of available space in the engineer- ing collection, but they account for only 13 percent of circulation despite a liberal policy that allows journals to circulate for two weeks just as any other book. Finally, in fiscal 1973, when budget re- strictions began to significantly affect the li- brary's capacity to continue subscriptions, expenditures for subscriptions were 13 per- cent higher· than book expenditures. In sum, the bulk of library resources in terms of money and space was being devoted to that part of the collection that received the least use. TABLE 1 CITATIONS TO JOURNALS IN WEST VIRGINIA UNIVERSITY ENGINEERING MASTER'S THESES 1971-1974 Number of Number of Citations to journals Department Theses References Number Percent of References Aerospace 14 105 12 11.4% Chemical 16 516 255 49.4% Civil 30 888 199 22.4% Electrical 28 543 219 40.3% Industrial 18 640 201 31.4% Mechanical 20 310 114 36.8% -- -- Totals 126 3002 1000 33.3% 108 I College & Research Libraries • March 1978 As a result of these considerations, ex- penditures for engineering subscriptions at WVU were reduced from $25,000 in 1973 to $23,000 in 1976. This reduction was achieved despite the fact that the average price of engineering subscriptions rose by about 50 percent during this time19 and de- spite an increase of 34 percent in the total subscription expenditures by the WVU Li- brary system. As a result of the cancellation of subscriptions, the library was able to con- tinue spending about $20,000 annually on engineering books. Had subscriptions not been canceled, expenditures for books would now be less than $8,000, a figure that could not begin t9_ support a quality collec- tion in engineering. The library user's perception of library service often takes precedence over strictly budgetary considerations or objective meas- ures of library use. Therefore, faculty and student feedback is sought constantly con- cerning the adequacy of the WVU engineer- ing collection. Faculty were at first surprised to learn that journals were so little cited by their students. Some thought that this repre- sented a deficiency in WVU's programs, until citation data from the published litera- ture was prtsented. Others felt the low use of journals represented a deficiency on the part of engineers in general. These indi- viduals seemed to feel that the use of jour- nals by chemists and physicists represented an ideal that should characterize all of sci- ence and technology. Perhaps the differences in literature use between engineering and the "pure" sci- ences reflect the deeper differences be- tween the two fields. Engineers frequently use. well-known principles to design a de- vice for a new application, in which case standard reference books might be of greatest value. Unstable government fund- ing and interdisciplinary studies may re- quire reference to basic works in many dif- ferent fields as engineers seek to extend their expertise to many different specialties. In the pure sciences there is a greater tend- ency toward long-term studies of highly specialized topics on the edge of what is known, and such work cannot be presented in books in a timely manner. The faculty's uneasiness over the rela- tively low use of journals did not extend to the selection of particular titles to be can- celed. Apparently the list of titles selected for cancellation, prepared with the aid of ci- tation data, coincided with the faculty's per- ception of what is useful and desirable. The entire list of canceled subscriptions was re- viewed by the faculty a second time in the fall of 1976, two to three years after most of the subscriptions had been stopped. There were no requests for reinstatement of any title. Funds released by canceling subscriptions have been used to continue efforts at build- ing an extensive and timely book collection representing the full breadth of engineering interests at WVU. As a result, annual circu- lation is now 62 percent higher than it was four years ago, although student enrollment has increased by only 2 percent. Despite the massive cancellations and the rapidly in- creasing use of the library accompanying the growth in quality of the book collection, there has been no increase in the number of interlibrary loan requests. CONCLUSIONS Satisfactory accommodations to library budget limitations have been made in the WVU engineering collection with the aid of citation analysis. Objective data relating to the use of all portions of the library collec- tion aided in eliminating the less useful journals in favor of more useful books. While long-term effects of inflation may yet be serious if costs continue to rise, short- term effects may be reduced in many librar- ies by using similar techniques. In particu- lar it is important to examine those areas where service far exceeds need and to em- phasize areas of service of greatest impor- tance to the local group of library users. REFERENCES 1. Eugene Garfield, "Significant Journals of Sci- ence," Nature 264:609-15 (Dec. 16, 1976). 2. See, for instance, Library journal 100:358 (Feb. 15, 1975); 100:2092 (Nov. 15, 1975); 101:652 (March 1, 1976). 3. "Academic Library Statistics Released for 1974-75," College & Research Libraries News 37:57 (March 1976). 4 . Herbert S. White, "Publishers, Libraries, and Costs of Journal Subscriptions in Times of Funding Retrenchment," Library Quar- terly 46:359-77 (Oct. 1976). 5. Clyve Jones, Michael Chapman, and Pamela Carr Woods, "The Characteristics of the Lit- erature Used by Historians," journal of Li- brarianship 4:137-56 (July 1972). 6. Herman H . Fussier, "Characteristics of the Research Literature Used by Chemists and Physicists in the United States. Part II," Li- brary Quarterly 12:119-43 (April1949) . 7. Eugene Garfield, "Anonymity in Refereeing? Maybe-But Anonymity in Authorship? No!" Current Contents: Engineering, Technology , & Applied Sciences 7 (11) :5-7 (March 15, 1976). 8. Eugene Garfield, "Some Comments on Pure and Applied Research-Stimulated by a List of Works Cited by Applied Chemists," Cur- rent Cqntents: Engineering, Technology, and Applied Sciences 5 (49):5--8 (Dec. 4, 1974). 9. Eugene Garfield, "Journal Citation Studies. 21. Engineering Journals," Current Contents: Engineering, Technology, & Applied Sciences 6 (27):5-10 (July 7, 1975). 10. Eugene Garfield, "Characteristics of Highly Cited Publications in the Engineering Sci- ences," Current Contents: Engineering , Technology, & Applied Sciences 7 (12) :5-10 (March 22, 1976). 11. Eugene Garfield, "Journal Citation Studies. Subscriptions vs. Books I 109 20. Agriculture Journals and the Agricultural Literature," Current Contents: Engineering, Technology, & Applied Sciences 6 (20):5-11 (May 19, 1975). 12. Eugene Garfield, "Journal Citation Studies. XIV. Wherein We Observe that Physicists Cite Different Physics Journals Than Other People," Current Contents : Engineering , Technology & Applied Sciences 4 (40) :5-8 (Oct. 2, 1974). 13. Pauline A. Scales, "Citation Analyses as In- dicators of the Use of Serials: A Comparison of Ranked Title Lists Produced by Citation Counting and From Use Data," journal of Documentation 32:17-25 (March 1976). 14. Garfield, "Significant Journals of Science." Garfield, "Characteristics of Highly Cited Publications in the Engineering Sciences." 15. Harry M. Kriz, "Citation Counting and the Future of Engineering Libraries," Engineer- ing Education 67:707-10 (April 1977). 16. Garfield, "Journal Citation Studies. 21. Engi- neering Journals." 17. Kriz, "Citation Counting." 18. Maurita Peterson Holland, "Serial Cuts vs. Public Service: A Formula," College & Re- search Libraries 37:543-54 (Nov. 1976). 19. F . F . Clasquin, "Periodical Prices : 74-76 Update," Library journal 101:2015-19 (Oct. 1, 1976).