College and Research Libraries W. M. SHAW, JR. A Practical Journal Usage Technique A practical journal usage technique is currently being employed at the Case Western Reserve University Libraries. This technique can be used in a wide variety of libraries, with extensive collections, for an indefinite period of time and with minimal expense. The associated study yields relevant man- agement information from a straightforward analysis. The fundamental re- sult of the study is a stable segregation of the collection into two compo- nents, those volumes and titles that are used and those that are not used. A technique for rating the relative liability of the unused titles is also pro- vided. IN RECENT YEARS the increasing cost of journal subscriptions and the rapid growth of the journal literature have had a sig- nificant impact on the stable or declining budgets and fixed space of many academic libraries. 1 These circumstances have created the need for more cost-conscious manage- ment of the journal literature collection in major academic libraries. Improved man- agement of the collection can be facilitated by knowledge of how the use of the collec- tion is distributed over the individual titles. Unfortunately, it is difficult and expensive to monitor the use of large collections on an ongoing basis, particularly when no record of use is generated by standard procedures, such as circulation records. The technique described below provides a simple , inex- pensive way to obtain the necessary man- agement information for an indefinite period of time. 1 W. M. Shaw, Jr. , is assistant professor, School of Library Science, Case Western Reserve Uni- versity (CWRU), Cleveland, Ohio. This article is a revised and updated version of a paper origi- nally presented to the Library Research Round Table at the Annual Conference of the American Library Association in June 1977. The data for this study were collected by several CWRU li- brary school interns: Katherine Wilson, Karen Croneis, Jane Hodges, and James Arifayan. Their efforts are gratefully acknowledged. This study was conducted by the CWRU Libraries office of research and systems, coordinated by Esther Greenberg. BACKGROUND There have been numerous journal usage studies reported in the literature. These studies have addressed such issues as: 1. which titles should be added to the col- lection; 2. which titles might be canceled, dis- carded, or stored; 3. which titles might be bound in a less expensive manner, or not at all; and _4 . which titles might be converted to mi- croform. The data required to support decisions in the above areas have been taken from a va- riety of sources including: citation analy- ses, 2 - 5 interlibrary loan requests, 6 photocopy requests, 7 ,s questionnaires, 9 loan slips, 10 shelving statistics, n- 13 and other internal measures. While the origin of the data varies, there are several characteristics that are fre- quently observed in these studies. 1. The data are collected during a limited time period, e.g . , four months or one semester. 2. The studies frequently occur in special libraries or otherwise limited collections, especially scientific and medical collections. 3. The results of the studies are customar- ily expressed as a ranking of titles based on the relative, absolute (not normalized) fre- quency of use. 14 The same measure is gen- erally employed to show how the use varies as a function of volume age. I 479 480 I College & Research Libraries • November 1978 Studies in which the data are derived and analyzed in the general manner described above are defined here as utilizing the tra- ditional journal usage technique. DISCUSSION OF THE TRADITIONAL JOURNAL USAGE TECHNIQUE The three characteristics that the tra- ditional studies frequently share possess in- herent disadvantages. First, it is difficult to determine how long such a study should be conducted before re- liable results can be obtained. While a re- cent study has shown that the majority of low-use titles can be identified in a rela- tively short period of time, 15 such results do not ·diminish the need for long-term studies. There are two reasons for this: 1. It cannot be assumed that the same limited time period will identifY the same percentage of those titles with a given use characteristic in all libraries. Consequently, each library will have to ·determine this time period independently until or unless it can be shown repeatedly that the same time period has the same meaning in a large number of libraries. 2. Neither the library nor the environ- ment in which it operates is static. Man- agement decisions can affect the usage pat- tern in such a way that other studies might be required. Clearly, new titles are added to collections (even these days), and it would be desirable to begin assessing their use immediately. Moreover, changes in the curriculum or the clientele can influence the demands placed on an academic library. In addition to these uncertainties, it is not clear how far into the future the results of a study conducted over a limited time period can be meaningfully used as a man- agement tool. Second, the results derived from small, specialized collections do not lend them- selves to generalizations that may be adopted by large academic and research li- braries. The attention to scientific and med- ical collections provides little data applicable to diverse collections which include hold- ings in the humanities and social sciences. The issue here is actually more funda- mental. Library managers must make deci- sions that require title-by-title distinctions. It is unlikely that decisions of this type, however scientifically motivated, could be universally adopted, even by similar librar- ies. The fact that a title is not used in li- brary A is no guarantee that it is of no use to library B. Since generalizable results are not likely to be produced, a generalizable technique is a more reasonable objective. Finally, the data generation, collection, and analysis phases of the traditional frequency-of-use study are · typically cum- bersome, labor-intensive, and expensive exercises. The associated ranking analysis seeks to distinguish between the relative number of uses experienced by the indi- vidual titles but provides no simple measure for the performance of the collection. The most common summary statistic reported in the literature gives the percentage of the collection that generates some percentage of the uses, the 20/80 syndrome. This is an interesting statistic, but it does not allow a library to monitor its journal performance as a function of time; nor does it allow two libraries to compare their rela- tive performance. Moreover, it is hardly useful for the library manager to know that title A has been used 150 times and title B has been used 15 times, during some lim- ited time period, when a substantial propor- tion of the collection generates no use dur- ing an extended time period. The traditional journal study may generate substantial quantities of data that are of doubtful practi- cal utility. 16 It is clear that it would be desirable to have a technique which could be used in large collections, for an indefinite period of time, with minimal expense, and which would yield relevant management informa- tion from a straightforward analysis. A prac- tical journal usage technique is currently being employed at the Case Western Re- serve University (CWRU) Libraries that possesses these useful characteristics. DESCRIPTION OF THE CWRU STUDY The CWR U study is restricted to the bound journal volumes which the users are encouraged not to reshelve. The use data are generated by simply applying a small, pressure-sensitive label to the spine of a volume the first time it is reshelved by li- library personnel. A tagged volume is re- A Practical journal Usage Technique I 481 shelved with no further action. Thus at any given time the tagged volumes represent those that have been used at least once since the study began . Current issues may be included if the tag is affixed adjacent to the title label on the shelf, i.e. , by tagging the shelf not the jour- nal issue. These tags indicate whether or not the current issues of journal titles are used. The data are collected periodically by counting the total number (or an appro- priate sample) of volumes and titles and those that have been tagged. The associated analysis simply requires computing the cumulative fraction of vol- umes and titles that have been used during progressively expanding time periods . A similar technique has been used as a weed- ing criterion for monographs in a public library. 17 The cumulative fraction of bound volumes and/or titles used in expanding time periods can serve as a simple measure of perfor- mance for the journal collection. This meas- ure would be expected to increase rapidly at first, then more slowly , approaching a con- stant fraction as time progresses. When the constant fraction condition is reached, there is a high probability that those volumes or titles that have not been used will not ex- perience significant use in the foreseeable future . The fundamental result of this study is, then, a stable segregation of the journal lit- erature collection into two categories: (1) those volumes and titles that are used and (2) those that are not used. Thus decisions regarding· the disposition of volumes or ti- tles can be made with confidence that they will stand the test of time. Moreover, when the constant fraction condition is reached, the shelvers would be applying labels in- frequently. Consequently, the procedural costs, in terms of the time required to up- date the data, diminish as time progresses. RESULTS OF THE CWRU STUDY The CWR U Libraries are composed of two major libraries (Freiberger and Sears) and their satellites. The Freiberger Library contains materials in the humanities, social sciences, and behavioral sciences. The Sears Library contains materials in the sciences, engineering, management, and economics. The results reported here were derived from the public access areas (not storage) , which for Freiberger include approximately 2,700 titles and 59,000 volumes and which for Sears include approximately 1,600 titles and 24,000 volumes. The current study of the journal literature collection of the two libraries was begun in October 1975. Data have been collected and analyzed periodically since that time. Figure 1 gives the cumulative fraction of .volumes and titles used as a function of time for the two libraries. The results associated with the Sears collection are different from those associated with the Freiberger collection in terms of overall use and in terms of the tendency toward the constant fraction condi- tion. The lower overall ut-ilization of the Freiberger collection has been influenced by the fact that a substantial portion of the journal titles were classified and shelved among the monographs. The classified titles showed relatively little use compared to the majority of the collection, which is arranged alphabetically by title . The existence and lo- cation of these classified titles were proba- bly not apparent to all users. Many of these titles were declassified and integrated into the main collection during the course of this study. Currently, the use of these titles is comparable to that of the rest of the collec- tion. This event demonstrates how a man- agement decision can affect the use of the collection. The tendency toward the constant fraction condition has been distorted by an unusual event at the Sears Library. The Sears curve for titles shows an anomalous increase be- tween the eighth and the twelfth month. This increase is due indirectly to a flood of the first floor of that library in August 1975. The first floor contained many older vol- umes of important titles. In the aftermath of this event, it was determined that no jour- nal titles would be housed in this area and that a selected group would be shelved in the main collection on the second floor. In order to provide space for the selected items and other additions to the collection some volumes and titles had to be moved to storage. Those items moved to storage showed little or no use in a previous tra- ditional study and the current study. Dur- 482 I College & Research Libraries • November 1978 100 - - 80 - 6 6 6 6 6 6 Q -LLJ V) :::::1 I- 60 - • z: ~ LLJ • u c::: 6 • LLJ -c... LLJ 6 • • > ;:: 40 • • :5 • 0 0 :::::1 :E: 0 :::::1 . u 20 0 • • - 0 • • • I I I I I I I I I I I I I I I 4 8 12 16 20 24 28 32 Tir-E