The Association of Vision Science Librarians' citation analysis of Duane's Clinical Ophthalmology

INTRODUCTION

The Association of Vision Science Librarians (AVSL) is always looking for tools to help members evaluate their collections so that they can determine if their acquisitions programs are well directed and well funded. Methods of collection evaluation can be broadly grouped into collection-centered and use-centered methods [1]. Although use statistics can be helpful guides for librarians in knowing what books to keep, it is hard to evaluate the quality of a collection through use alone. Therefore, a collection-centered method was sought as an additional way to evaluate the collections.

In collection-centered evaluations, collections are often compared to an authoritative bibliography such as the “Brandon/Hill List of Print Books and Journals for the Small Medical Library” [2]. Although the AVSL has an Opening Day Book, Journal and A-V Collection: Vision Science [3] bibliographic tool, it is meant as a guide for a beginning collection and only contains materials that are currently available. Also, because the list is relatively small, a library could own most of the materials on the list yet still have an underfunded, underdeveloped library collection without a lot of depth. For these reasons, another evaluation tool in addition to the Opening Day Book, Journal and A-V Collection: Vision Science would be beneficial. It would also be useful to have a tool that would let libraries similar in size and scope compare themselves to each other.

Nisonger suggested using citation-checking techniques as an evaluation tool for subject-specific libraries [4]. In his 1983 article, he used two groups of citations to compare for reliability. The first group of citations was selected from three years of a leading journal in the field with a random citation picked from each article. This sampling generated a group of 150 citations. The second group was selected from one volume each of five separate journal titles covering approximately the same year. A citation was selected randomly from each article in each journal title. This selection generated a group of 142 citations. Two samples from each source were taken to help test reliability. Each of the random lists was given to five libraries to see if the number of materials they held from the lists would rank them in similar order as to percentage of materials held. In his study, this citation method proved reliable, in both cases ranking the five libraries in the same order.

Heidenwolf used Nisonger's technique to evaluate the Public Health Library collection at the University of Michigan [5]. Her sample size was 623 citations. She also cites Broadus and Mosher as proponents of citation checking as a means of collection evaluation. Citation checking is also described as one of the major methods of collection evaluation in the American Library Association Guidelines for Collection Development [6].

Although Nisonger and Heidenwolf used random citations from selected journal titles, Currie used bibliographies from textbooks to do his citation analysis [7]. The textbooks he used were required texts used by the faculty.

In vision science, there is a comprehensive reference book called Duane's Clinical Ophthalmology [8]. It has seven volumes, including an index, and is updated yearly. Therefore, this reference book has a broader scope over many subfields of vision science and is more up-to-date than a regular textbook. A random list of citations from the end of each chapter in Duane's Clinical Opthalmology was made. The participating vision science libraries compared their collections to the list. A second random list was generated from the same book to see if the libraries would come out in relatively the same order for both comparisons. This second list helped determine consistency and increased the sample size.

METHODS

One reference was selected randomly from the end of each chapter in the 1998 edition of Duane's Clinical Ophthalmology, edited by William Tasman, M.D., and Edward Jaeger, M.D. This made a total of 355 references on citation analysis list number one. Then another reference was selected randomly from each chapter for an additional list of 355 references, and this became citation analysis list number two. Selection of all citations was made using a random number table. Both lists, a summary sheet, and directions were emailed or mailed to selected AVSL members for them to check their collections. Each library then returned the summary sheet containing the percentages of books and journals from each list. Percentages were then compared. Participation was voluntary, and twenty-one out of thirty-six invited libraries from the United States, Puerto Rico, and Canada returned their forms.

There were a few duplicate citations between the two lists, and these were included. Only book and journal titles were counted. If, for example, the picked citation represented unpublished literature, it was disqualified and a new reference picked randomly from the chapter. Each list had a journal section, book section, and a combined number, which were all converted to percentages. A library could claim to hold a journal article if it owned the title, year, and issue that the article was in. Only paper copies were counted, although if the study is repeated, electronic formats will have to be considered. When comparing the book section, the library had to have the exact edition of a book to count it. If it was a different edition it counted as half of a book on their total. A different publisher would not matter.

The combined percentage of book and journal titles held was the one used for comparison of the libraries. If the library was part of a campus, then the entire campus collection could be used for the checklist. A branch library in another city could not be used. For example, University of Michigan would be able to say it held any title kept in any library on its main campus but not at its Flint branch. University of California, Berkeley, could use its campus collection but not the books kept at other University of California branches.

One purpose of the study was to see what percentage of each citation list each vision science library held. Libraries were compared to see if they fell on the list where common sense would place them. For example, large university libraries should be ranked near each other, as well as private optometric universities. Also newer libraries would not have as much of the older material as long-established ones. The library with the most citations was not necessarily the “best” library, nor was it the purpose of the study to find out who had the best collection. It was helpful to see how each library compared to others with similar scopes, ages, and collection goals. This would provide a broader standard with which to measure the collections, because the citation analysis would contain older literature as well as peripheral literature that was not necessarily vision science related. Also, if libraries were rated in approximately the same order on both citation lists, this would help establish the usefulness of Duane's Clinical Ophthalmology as an assessment tool.

RESULTS

The results of the study are found in Table 1 and Table 2. The libraries were ranked according to the number of books and journals on the list the libraries owned. The libraries were ranked in the same relative order on both citation analysis list one and list two. If there was some variation, it was usually only within one or two percentage points.

Table 1 Results from citation analysis list number one

Table 2 Results from citation analysis list number 2

All of the libraries in the study owned a higher percentage of journal titles than book titles. The libraries lower on the list had closer percentages of book and journal ownership.

Also, most of the libraries of similar scope and age seemed to be ranked near each other. Although there was some overlap, the top few libraries on each list were part of large university campuses. The next tier was independent optometry schools and other stand-alone collections. The last tier tended to be smaller public universities with newer collections. Budget information was not gathered.

CONCLUSIONS

The citation analysis from the reference book, Duane's Clinical Ophthalmology, provides the AVSL libraries with useful information for collection evaluation. Not only do the two lists rank the libraries similarly, they also rank libraries near other libraries of similar scope. This comparison gives a much broader view of individual collections. It can be used in addition to the Opening Day Book, Journal and A-V Collection: Vision Science for collection evaluation. With this method, a broader sample with older and peripheral literature is included. Hopefully, this information will be useful to the individual libraries for funding purposes as well as quality collection management.

The AVSL plans to repeat this analysis again in four or five year cycles. Next time the study is done, the AVSL will have to decide whether to include electronic copies of journals and books.