Analysis of library collections is very expensive, cumbersome, labor intensive, and time consuming. Collection development librarians sometimes neglect this activity because of the size of the collections, magnitude of the analysis, and lack of an automated method to analyze the entire library collection or any part of it. It was unthinkable until 1989 to try to compare library collections quantitatively and automationally against national standards or prominent institutions. Similarly it was very difficult to evaluate specific areas or an entire collection. However, with the cooperation of OCLC and AMIGOS Bibliographic Council, Inc., it became a reality with the new product called OCLC/AMIGOS Collection Analysis CD.

2. DESCRIPTION OF THE CD

The Collection Analysis CD allows the user to compare the participating libraries' hold- ings that are entered into the OCLC Online Union Catalog against peer institutions. As shown in Table 1, there are 14 standard peer groups based on collection size and academic degree program.

In addition the user may define a 15th group based on its own specification.

NAME                                                                                                        NUMBER OF INSTITUTIONS

• Association of Research Libraries (ARL)

- All ARL member in OCLC                                                                               72

- ARL libraries 1st quartile (largest)                                                                 17

- ARL libraries 2nd quartile (next largest)                                                       18

• Academic libraries with 700,000+ volumes

- Large academic libraries                                                                                    86

- Collections with 1 million volumes                                                                  37

- Collections with 700,000-999,999 volumes                                                      49

• Academic libraries with 300,000-699,999 volumes                                         209

- Supporting doctoral programs                                                                          67

- Supporting masters and bachelors programs                                                113

• All academic libraries with 100,000-299,999 volumes                                    473

- Supporting graduate degrees                                                                           266

- Supporting undergraduate degrees only                                                       146

• All academic libraries with 50,000-99,999 volumes                                        134

• The entire 1.7 million record database                                                            974

________________________________________________________________
 
 

The Collectrion Analysis CD has the following product capabilities:

• Compares the user's library materials collection against national standards,larger institutions or smaller libraries

2.1. Hardware

• OCLC M310 or M38 Workstation or IBM AT with:

The database of the OCLC/AMIGOS Collection Analysis CD is extracted from the OCLC Online Union Catalog for mostly monographic titles (excluding serials, government docu-ments, and dissertations) published within a defined decade. The 1990 annual edition of the database includes titles published between 1978 and 1988 and it contains approximately 1.7 million abbreviated bibliographic records. Each record in the database must have a Library of Congress (LC) classification number and be held by at least one Association of Research Library (ARL) or academic library.

Figure 1 shows that each record contains author, title, publisher, year of publication, OCLC control number, and International Standard Book Number (ISBN). The records contain up to 100 characters of author/title information, of which 20 are extracted from MARC fields 100, 110, and 111, and the remaining ones are taken from MARC field 245. Since both holdings information and the comparative database were based on the OCLC Online Union Catalog, the Collection Development CD is available only to OCLC member institutions.

The Collection Analysis CD is available to purchase in two options: One contains only the user's holdings data and the standard 14 peer groups' data. The other has data of the user's holdings data, user's specified peer group (up to 99 OCLC member institutions) and the standard 14 peer groups' data.
 
 

4. USE OF OCLC/AMIGOS COLLECTION ANALYSIS CD AT STERLING C. EVANS LIBRARY AT TEXAS A&M UNIVERSITY

The library purchased the Collection Analysis CD and the needed hardware in the spring of 1990. It allowed the collection development librarians to evaluate parts of the Sterling C. Evans Library collection and prepare for collection development at several levels. Comparison of the Sterling C. Evans Library with ten peer libraries, as shown in Table 2, could be made to measure collections.

Prior to the collection evaluation, the researcher sent a questionnaire to department heads and departmental library representatives in an attempt to gather information related to current and projected curriculum and research directions of the departments. The current and fore-casted areas of emphasis of each academic department were converted into Library of Congress call number ranges and were analyzed by the Collection Analysis CD.

An example of the current and projected curriculum and research, their matching Library of Congress Subject Headings, and their related LC classification numbers of the Department of Finance is listed in Figure 2:

The software allowed the use of the following main menu: Collection metrics, subcollection metrics, and bibliographic lists. The collection and subcollection metrics menu included: Counts, proportions, overlap, holdings distribution, gap, and uniqueness. Sub-collection metrics had the additional ability to view the more specific call number ranges. The bibliographic list menu covered overlap, gap, unique evaluator and peer groups, evaluator list, and peer group list. All bibliographic lists could be saved in an ASCII file or printed out as needed. As can be seen in Figure 3, the software also allowed the use of the established
 

parameters which were consistent throughout the entire project.

The analysis started with the general collection counts where the basic call number ranges were adopted from the National Shelf List Count. Table 3 shows the summary of the collection counts for the entire collection.
 
 

It is important to note that the records contained on the Collection Analysis CD are only titles which were published during 1978-1988; therefore their numbers are much smaller than the total shelflist count of the Sterling C. Evans library and its selected ten peer groups. Table 4 includes three main groups:

1. Peer group (Selected ten institutions)

2. Evaluator (Sterling C. Evans Library's titles) and

3. Overlap (Overlap of titles between the Sterling C. Evans collection and the ten peer institutions).

The peer group heading is further divided into number of individual titles held by any member of peer group. The collective holdings are the cumulation of the peer group titles, and the unique titles were held by only one library.

The subcollection counts (Table 4) subdivides the H-HX call number ranges into further categories like HG1-HG9999, which represents mainly the subjects financing and banking. Evans Library has 2682 titles in the call number range, of which 2572 overlap with the ten selected peer groups. Therefore, Texas A&M University has only 110 titles which are not owned by the peer group libraries.

The Collection Analysis CD is also able to generate gap lists at different percentage levels, as shown in Figure 4, which help the librarians to prioritize materials for purchase.

Titles which were missing from the evaluator's institution but found in 90 percent of the peer libraries generally were purchased before materials found in 70 or 80 percent of peer libraries. Such comparison made possible one method of refining and prioritizing collection needs. In a time of tighter budget and greater users' demand, prioritizing is indispensable.
 
 

The generated gap titles, such as those in Figure 5, were then used for retrospective collection development purposes.

5. CONCLUSION

The OCLC/AMIGOS Collection Analysis CD worked very well in making broad com-parisons with peer and other groups of libraries and in generating selective lists of acquisition possibilities of retrospective materials. However, raw data comparisons could not be fine tuned, and only the 1978-1988 decade of monographic publications were possible for com-parison. Therefore, making an evaluation of the entire library collection was impossible.

The new information provided by the Collection Analysis CD allowed collection develop-ment librarians to prioritize needs, to modify collection development policies, and to decide in which areas the library will excel in the future.

REFERENCES

Dillon, Marin and Kevin Flash, "Design issues for microcomputer-based collection analysis system," Microcomputers for Information Management, 5 (4): 263-273 (December 1988).

Harrell, Jeanne, "Use of the OCLC/AMIGOS Collection Analysis CD to determine comparative collection strength in English and American literature: A case study," Technical Services Quarterly, 1992 (forthcoming).

Jarmillo, George R. "Computer Technology and its Impact on Collection Development." Collection Management 10 (1-2): 1-13 (1988)

Kreyche, Michael, "BCL3 and NOTIS: An automated collection analysis project," Library Acquisitions: Practice and Theory, 13 (4): 323-328 (1989).

OCLC Online Computer Library Center. OCLC/AMIGOS Collection Analysis CD User Guide. Dublin, OH: OCLC, 1990.

Sowell, Steven L., "Expanding horizons in collection development with expert systems: Development and testing of a demonstration prototype," Special Libraries 80 (1): 45-50 (Winter 1989).