College and Research Libraries JAMES C. BAUGHMAN Toward a Structural Approach to Collection Development A method for developing acquisition priorities based upon scholarly need is discussed, using some of the newer techniques in the area of information science. The structuralist approach requires a method interpreted through the study of behavior and properties of subject literatures. Behavior refers to temporal con- ditions while properties relate to the organizing principles of class and order. Subject literature behavior and properties are described in this paper through the techniques of citation count- ing, Bradford's law, and Coffman's indirect method. · 0 NE OF THE MOST SERIOUS PROBLEMS that -exists today in the library is collec- tion development. During the past several decades while money flowed rela- tively freely, librarians concentrated on building larger facilities, acquiring ma- terials at an exponential rate, and ex- panding staffs to control the informa- tion explosion. The 1960s stand as a testimonial to this approach. During this period, the prevailing slogan "the bigger, the better" led one to believe that excellence relates directly to the quantity of volumes that a library holds. Ironically, this slogan has forged the librarian's action into the kiva of materials administration rather than meaningful collection development- 1 ames C. Baughman is associate profes- sor, School of Library Science, Simmons College, Boston, Massachusetts. This article is an abridgement of a paper which received one of the 1976 Research Competition awards presented by the ALA Library Re- search Round Table. This research was sup- ported by a .grant from the Simmons College Fund for Research, and the com- plete paper is to be published in a collect- ed work in the near future. the raison d'etre of the library. Librarians are now beginning to real- ize that they cannot continue to operate under the assumption that _there is no limit to the amount of material that can be acquired, organized, and stored. The production rate of documents is too great and fiscal pressures too acute to allow librarians to retain their for- mer modus operandi. . It is time that li- brarians also realize that it is possible to acquire core quality collections without reaching astronomical figures.! Quality is a relative concept related to relevance, which in the library environ- ment is determined by a measure of ef- fective contact between the collection of knowledge records and the user-not by size. Comprehensiveness in the nine- teenth ·century gave way to the "ideal" of completeness; quantity in the aver- age library in the twentieth century must give way to the "ideal" of quality. While knowledge records need to be acquired on a continual basis, the im- portant question is: Can they be ac- quired effectively in terms of knowledge and use? This is the socially relevant question for librarians, and it can be I 241 242 1 College & Research Libraries • May 1977 answered only through the use of objec- tively verifiable data based on the prem- ise that acquisitions can be controlled, priorities for collection development established, and funding sought and justified. Progressing toward the 1980s, librari- ans will need to experiment with new methods for collection development, in- cluding the structuralist approach. The structural approach is discussed and il- lustrated here as one way to address the collection development problem. The structural approach seeks to find a pat- tern of relationships, since effective col- lection building is assumed to rest on identifying a structure. Once the struc- ture is determined, a plan for collection development will quickly emerge. A FRAMEWORK FOR DISCUSSION ON COLLECTION DEVELOPMENT Major Constructs in Collection Development Figure 1 graphically illustrates the re- lationships among the major constructs integral to collection development. These constructs are: ( 1 ) use-cluster of demands; ( 2) knowledge-cluster of disciplines, subjects, topics, and areas of study; and ( 3) librarianship-cluster of subject literature relationships. Collec- tion development is represented in Seg- ment D of the diagram; it is the intertwining of the concepts of plan- ning, implementation, and evaluation. Collection planning is a design for ac- cumulating documents that belong to- gether as determined by the needs, goals, objectives, and priorities of the library. Collection implementation refers to the process of making documents accessible for use. Collection evaluation involves examining and judging with respect to goals and objectives. Thus, collection development is a plan which can be implemented and evaluated and may be represented as follows: Collection Collection planning + implementation Collection COLLECTION + evaluation = -DEVELOPMENT The enlacing of these concepts into col- lection development policy leads to a system that is complementary, cyclical, and self-improving. Use Versus Use Potential Librarianship, as a professional, pur- poseful, goal-directed activity in the area of collection development, means responsibility of decision-making with reference to use and knowledge. ( See Figure 1. ) In the operational sense, use relates to demands and knowledge re- lates to needs, or use potential, consider- ing that documents are acquired on a subject basis for use. There is a difference between demand and need. A user may demand one docu- ment when in fact another might better meet his or her need. The need factor is closely associated with the concept of use potential; for if this user were to be introduced to a document which truly met the need, he or she would in fact use that document. The primary constraint here is the document (con- tent) in relation to the user's need. The interplay is among object, content, and use. 2 The demands that users make on a li- brary collection represent How from the user to the knowledge base and may be evaluated by user studies. The user study is valuable as a technique for de- termining use patterns and should be employed as a base for decision-making in the "use" area of collection develop- ment. However, user studies have several built-in limitations, two of which are: ( 1 ) they establish so-called user groups whose behavior tends to be personal, introspective, and variant; and ( 2) they measure only the materials currently held in a given library whose subject collections may range from poor to ex- cellent. Collection Development./ 243 Fig. 1 Collection Development: A Structural Approach This self-limiting method of deter- mining collection development policy does reveal special interests singular to the clientele of an individual library, but it fails to improve the quality of a collection in terms of new acquisitions of value. Depending solely upon the user study for collection development, decision-making usually places the li- brarian behind the eight ball, since de- mands are observable only after-the- fact. One can run to catch up, but it might be better to be running to keep ahead! When library collection policy is for- mulated with an eye to the structure of subject literatures, the librarian may be in a better position to cope with the voluminous production of knowledge records from which he or she must iso- late and secure those documents that have use potential. The librarian has the responsibility of building collec- tions that include items with inherent value, representing a potential need for library users. Here the flow is from the knowledge base to the user. How often has the librarian heard the patron say: "I had this in mind, but can you recom- mend something better?" Patrons in this 244 I College & Research Libraries • May 1977 situation know that they need informa- tion but are not quite sure of their in- formation needs. The interplay between demand and need must receive thoughtful attention for effective collection development. The question is: How far in each direc- tion should acquisition policy go in or- der to build the "best" library collection for users? The proper balance between use and use potential is not precise, nor can it realistically be. The relationship, never having been clearly defined, rep- resents an existential dilemma for the librarian. This is why it is important for the librarian to be aware of the cluster of demands and the cluster of disciplines within the framework of collection development. THE STRUCfURE OF SUBJECT LITERATURES Since librarians build library collec- tions for use and use potential (need), their decision-making should take into account the structure of subject litera- tures. In collection development the sub- ject (discipline) is the superordinate constraint, since users use documents in terms of a subject and its literature. Rational inquiry into problem areas produces a body of knowledge that is organized into a subject field. This sub- ject matter is publicly shared through a literature, which records, displays, stores, and transmits pertinent states of knowledge at a given point in time. From this rationale one can make the assumption that a state of knowledge is reflected in its literature. This rela- tionship is established through logical inference, i.e., if one ascertains the structure of a literature, then it may be assumed that a structure has been deter- mined on an indirect basis for a knowl- edge area, since the literature contains pertinent states of knowledge.3 The overriding viewpoint here is that subject literature structure, a way of seeking relationships, will provide un- derstanding of the literature's behavior and properties-not its intellectual con- tent per se. Thus, the structural ap- proach to subject literatures involves the study of the behavior and properties of a subject literature. The "behavior," i.e., a characteristic way of acting, may be interpreted with reference to "literature statics" (a point in time) and "literature dynamics" (a period of time). Literature statics may be investigated through several of the newer information science techniques, now referred to as bibliometrics (the measurement of bibliography) in their specific instance of use. One of these techniques is Bradford's law,4 the appli- cation of which is discussed under "Core Publishers" below. The "properties" of a subject litera- ture refer to ( 1 ) class and ( 2) order To understand knowledge it must be organized (class) and sequenced (or- der). To be useful, knowledge must be organized. Of course, no system of or- ganization can be absolutely successful, but without it no system can approach success. An innovative method for or- ganizing and sequencing knowledge is suggested by Goffman. 5 The organiza- tional aspect ( class ) of this method is discussed under "Associated Subjects" below. Investigation into the behavior and properties of a subject literature may be facilitated by dividing the literature situation into parameters. This study is a continuation of the author's previous work in sociology. The parameters of time and language in relation to sociol- ogy literature are discussed in a previous study. 6 This paper presents further dis- cussion of the parameters of ( 1 ) asso- ciated subjects and ( 2) form (objects) and introduces ( 3) publishers. Associated Subjects- Establishing Subject Relationships The grist of the librarian's mill is the subject; thus, a major concern should be to determine the inter- and multi- relationships among subject areas. Although the interdependence among the various disciplines within the social sciences has long been recognized as a condition necessary to their progress, each discipline has traditionally been treated as a separate entity. In the past several decades, there has been a shift in emphasis in the social sciences from one of isolation to one of merging in- formation to produce patterns of knowledge that overlap. More and more academic programs are reorganizing their course offerings along new inter- and multidisciplinary lines. Such cur- ricula changes will need to be reflected in the library collection. The beginning point should be the identification of subject relationships. Identifying overlapping subject areas indicates which subject areas are in com- munication. This communication needs to be determined since empirical rela- tionships among associated subjects lay the foundation for present-day library collection development. Without an un- derstanding of subject relationships, the librarian is lost in a sea of disjoint documents representing content from which selections must be made. D·eter- mining subject relationships is impor- tant, not only for the physical organiza- tion of materials, but also for the intellectual organization of knowledge for collection development purposes. Subject relationships may be deter- mined on a macro or micro level- macro referring to general areas and micro to specific topics. In approaching both the macro and micro levels, the procedure used in this study to establish subject relationships follows Coffman's indirect method. The Coffman method differentiates between the relevant and the not-so-relevant subject areas by in- vestigating the overlap of subject cover- age in individual journals. For the author's investigation into subject relationships in the social sci- Collection Development I 245 ences on the macro level, the data base, used in connection with the Coffman model, was Volume I ( 197 4-75) of the Social Sciences Index ( SSI) .7 In volume 1 of SSI there were 4,995 sub- ject headings with 30,673 entries. Since the concern here is with macro relation- ships, only the most productive subject areas-those headings with eleven or more unique entries-were included for this part of the study. The 309 subject headings with eleven or more unique en- tries were grouped into sixty-one Li- brary of Congress classification areas in order to condense the subject headings into systematized groupings that could be studied for interrelationships on the macro level. Coffman's indirect method was then applied to the sixty-one subject group- ings. The results of this procedure indi- cate that there is one major class consist- ing of eleven strongly interrelated subjects. By further applying the LC classification system~ the eleven subject areas may · be condensed into six social .science disciplines. These disciplines, presented in Figure 2, may be said to be interrelated since there are strong com- munication links among them as deter- mined by the Coffman model. The significance of the macro subject relationships is that with these findings librarians may be alerted to the spec- trum of subjects that need to be consid- ered together in building library collec- tions. The findings suggest that building strong central collections in the social sciences may be more advisable than de- partmentalizing, since developing de- partmental libraries would require extensive duplication of informational materials with resulting higher expendi- tures. The librarian needs to consider sub- ject relationships not only on the macro but also on the micro level. If a user is interested in a specific topic (micro level), such as "human ecology," he or she may be expected to become interest- 246 I College & Research Libraries • May 1977 I I History Psychology E BF Legend: BF I Economics HB, HC, HD Psychology History: America Economic theory I I I Sociology Political Education Science HM, HQ, HT, JF LB HV E HB HC HD HM HQ HT HV Economic history and conditions. National production Land. Agriculture. Industry Sociology Social groups Communities. Classes. Races Social pathology. Social and public welfare. Criminology JF LB Constitutional history and administration. General Theory and practice of education Fig. 2 Social Science Disdplines in Communication ed in related topics within the content structure of the general subject of soci- ology. The question then becomes: "If a user is studying 'human ecology,' what other topics would most likely be of in- terest because of their close alliance with the topic 'human ecology'?" To search for solutions in this prob- lem area, the author conducted an inves- tigation into the interrelationships of subject headings on the micro level within the discipline of sociology. In this case, all subject headings ( includ- ing "see also" references) under the general heading "Sociology" in Volume 1 of SSI ( 1974-75) constituted the data base. It was not necessary to group these subject headings since the interest here is in micro relationships. Therefore, all eighty subject headings were related, without interpolation, by Coffman's model. Eleven distinct classes emerged. One of these classes, for example, shows the topic "human ecology" ( pri- marily concerned with a population's collective interaction with its environ- ment) to be in communication with ( 1) cities and youth, ( 2) population, ( 3) urban sociology, and ( 4) urbanization. Form The study of subject literatures can impart information through the param- eter of form. The acquisitions officer needs to know whether to concentrate • more heavily on journals or on books in a given subject area. In order to deter- mine if there is a reliance on serial or nonserial materials in the social sciences in general, a study of five selected major social science journals was conducted. All citations from the complete issues of these journals for the year 1974 were tabulated by percentage of serial and nonserial citations. The results are pre- sented in Table 1. TABLE 1 FoRM OF THE LITERATURE CITATIONS FROM SELECTED MAJOR SociAL SciENCE JouRNALS Journal American Anthropologist American Economic Review American Historical Review American Political Science Review American Sociological Review Serial Percent 42.93 59.02 23.3 34.59 44.41 Nons erial Percent 57.07 40.97 77.6 65.36 55.59 Note: The author acknowledges the helpfulness of several of his students in preparing the data for this table. This information clearly illustrates that current social science scholars rely, as depicted in the citation patterns of these major disciplinary journals, upon nonserial literature for a major portion of their support material. It may be conjectured that within the parameter of form the literature of the social sci- ences may behave differently from the literature of the natural sciences. The natural science literature consists almost exclusively of papers published in jour- nals.8 This would seem to have signifi- cance in that it may be more necessary to build stronger monographic collec- tions in the social sciences than in the natural sciences. Core Publishers Librarians have long sensed the need to follow selected publishers' catalogs in subject areas of interest to the li- brary's users. This procedure is auto- matic and is considered fundamental to successful collection development. The operating assumption here is that a selected group of publishers accounts for a large percentage of the books be- ing published on a given subject. This everyday logic-in-use may be examined on a formal basis by using Bradford's law. 9 Originally formulated with re- spect to journal literature, this law, in general, states that a small percentage of journals accounts for a large per- centage of significant articles in a spe- cific subject area. Worthen investigated the Bradford law with respect to monographs in a specific medical subject area and deter- mined that it applied. 10 Since the social science literature relies more heavily on nonserial literature than does the nat- ural science literature, it is important to determine the applicability of the Bradford law to social science mono- graphic literature. To test Bradford's law in relation to nonserial literature in the social sci- ences, the discipline of sociology was specifically investigated. The data base for this experimental test was a· random sample of 495 monographs representing 10 percent of the 4,954 independent bibliographic units isolated through a citation count of 446 journal articles from 71 different journals in the disci- pline of sociology. 11 The B'radford law Collection Development I 247 was applied to the random sample of monographs, the results reported in Table 2. TABLE 2 MAXIMAL DIVISIONS OF MoNOGRAPHS PUBLISHERS Number of Number of Br adford Zone Monographs Publishers Multiplier (b) 1 • 0 •• • •••• 46 2 2 • • • 0 • ••• • 48 3 1.5 3 .. . . ...... . 57 4 1.3 4 0 ••••• •• • 53 5 1.3 5 • 0 ••• 0 ••• 46 7 1.4 6 . . ....... 48 10 1.4 7 . . . .. ..... .. 0 48 16 1.6 8 •• • 0. 0 ••• 46 23 1.4 9 • •• •• 0 0 •• 49 42 1.8 10 . . . ... ... 54 54 1.3 Total 0 • • • 495 166 1.50 0 Average The Bradford multiplier (b) in Table 2 is an empirical rendition of the Brad- ford series (I : n : n 2 • • • ) where the ratio ( n) is the Bradford multiplier (b). This indicates that approximately the same number of documents is produced by a number of publishers which in- creases from zone to zone such that the ratio between the number of publishers in the second and first zones is the same as between the third and second, fourth and third, etc. The minimal nucleus of publishers (Zone 1) consists of two publishers rep- resenting forty-six monographs, fol- lowed by three publishers representing forty-eight titles (Zone 2), four pub- lishers with fifty-seven titles (Zone 3), etc. Successive zones of publishers pub- lishing about the same number of monographs form the approximate geo- metric series 1 : ( 1 . 5) : ( 1 . 5 ) 2 : • • • : ( 1 . 5 ) 9 • The results also show that half of the monographs were accounted for by only twenty-one publishers or 12.6 percent of the total number of publish- ers in the sample. Thus, it may be stated that the litera- ture of sociology exhibits a Bradford- ian distribution when the publishers are ranked by the productivity of cited 248 I College & Research Libraries • May 1977 monographs. The most productive pub- lishers on the subject tend also to pub- lish the monographs that are most frequently cited. The commonsense no- tion of using core selected publishers' catalogs to assist in building collections is empirically verified. CoNCLUSION The basic underlying assumption of this study is that the structure of sub- ject literatures will provide the librarian with a fruitful guide to collection de- velopment. Collection development con- sists of: ( 1 ) collection planning (·assess- ing needs, setting goals, and establishing priorities); ( 2) collection implementa- tion (action, communication, schedul- ing, disseminating, etc. ) ; and ( 3) collec- tion evaluation (assessing and j~dging with respect to goals and objectives). The structure of subjects, as well as the structure of subject literatures, is determined by concepts and generaliza- tions intertwining to form the founda- tion for reflective inquiry. The concepts employed in this investigation center on the behavior and properties of subject literatures. Thus, the structuralist in the subject literature situation seeks out pat- terns and relationships without explicit reference to the intellectual content of the literature units per se; i.e., the struc- turalist's overriding concern is with the understanding of literature forms and processes, rather than the literature's scholarly content. Some of the major conclusions that may be drawn from this study are as follows: 1. Subject relationships, which are fundamental to the intellectual organi- zation of knowledge for collection de- velopment, may be determined by studying the communication links that exist among associated subjects. 2. Communication links may be de- termined on a macro (disciplinary) and a micro (subject, topics, etc.) level. 3. Social science scholars cite non- serial literature more frequently than serial literature; thus, collection devel- opment officers need to consider, in par- ticular, monographic literature units in social science collections. 4. The behavior of literature units in publishing, as determined by the cita- tions of scholars in journal articles, con- forms to the Bradford law, which means that a few publishers account for the bulk of the cited monographs. Half the cited books were accounted for by only 12.6 percent of the pub- lishers. REFERENCES 1. James C. Baughman, "The Library Ecology Problem," in Karl Nyren, ed., L] Special Reports # 1: Library Space Planning ( New York: Library Journal, 1976 ), p.45-47. 2. Conrad H. Rawski, "Subject Literatures and Librarianship," in Larry Earl Bone, ed., Library School Teaching Methods: Courses in the Selection of Adult Materials (Urbana, Ill.: Univ. of Illinois, 1969), p.97-103. 3. Conrad H. Rawski, "The Scientific Study of Subject Literatures," mimeographed (Cleveland, Ohio: Case Western Reserve University, School of Library Science, 1976), p.l. 4. Samuel C. Bradford, Documentation, 2d ed. (London: Crosby Lockwood & Son, Ltd., 1953), p.154. 5. William Coffman, "An Indirect Method of Information Retrieval," Information Storage and Retrieval 4 : 361-73 (Dec. 1968). 6. James C. Baughman, "A Structural Analysis of the Literature of Sociology," Library Quarterly 44:293-308 (Oct. 1974). 7. See Social Sciences Index, 1974-1975, vol. 1 (New York: H. W. Wilson, 1975). 8. Rolland E. Stevens, Characteristics of Sub- ject Literature, ACRL Monograph no. 6 (Chicago: American Library Assn., 1953 ), p.17. 9. Bradford, Documentation, p.154. 10. Dennis B. Worthen, "The Application of Bradford's Law to Monographs," Journal of Documentation 31:19-25 (March 1975). 11. Baughman, "A Structural Analysis," p.296.