College and Research Libraries Research Notes Reference Service and Bounded Rationality: Helping Students with Research Felix T. Chu In university library reference service librarians often get ambiguous questions to which they try to give appropriate answers. Because of limitations on resources, time, mental capability for information processing, and other factors, the decision-making process involved in answering a reference question be- comes bounded by the rationality of these constraints. Entering into this process is the ambiguous nature of good and acceptable answers according to students. This paper is based on Herbert Simon's ideas on bounded rationality and fuzzy sets as discussed by L. A .. Zadeh. ·u· he quality of reference service has been a concern of librari- ,. ans for many years. Carefully crafted studies can help prac- ticing librarians improve services and help administrators in allocating funds and justifying budget requests to uni- versity administration. To that end, there has been a preference for objective studies. In examining the accuracy of reference service, Kenneth Crews has given an overview of existing studies.1 In calling for further studies to examine the effects of using more variables, it may become possible to predict more accurately the quality of service.2 How- ever, another recent study by Jo Bell Whitlatch has raised questions about the scope of existing studies such as those based on unobtrusive testing.3 She says unobtrusive testing seems too narrowly focused to measure the range of refer- ence services and cautions librarians not to let available methodology dictate the types of evaluation. Whitlatch urges in- trospection of reference librarians about the reference process so that they may examine and identify variables that help explain quality service while they wait for more sophisticated studies that use a greater variety of evaluation methods.4 One area that merits closer examina- tion involves the assumptions of refer- ence service. An existing premise is that there is a single correct answer for a reference question. The quality of refer- ence service as measured by unobtru- sive testing studies, for example, is based on the supposed accuracy of the answer. However, there may be multiple Felix T. Chu is Reference/Catalog Librarian at Western Illinois University, Macomb, Illinois 61455. 457 458 College & Research Libraries answers that are acceptable and equally accurate. In order to measure the quality of reference service, one must find out what to measure and how to measure. Part of the what involves a better under- standing of the process of decision mak- ing involved in how to answer reference questions. It is only when the process (doing reference) is improved that the final product (the answer) can be im- proved. Some of the variables in the process that may influence the quality of reference service are explored below. THE REFERENCE SETIING One of the jobs for a reference librar- ian at an academic library is to help stu- dents find information for their course work. This may be as simple as finding the mailing address in Washington, D.C., of United States Senators from Dlinois or as complex as finding research materials for a paper on variables that influence the success of United States exports in South America. The former case is a clearly de- fined question for which only one an- swer will suffice. In the latter case, several factors influence the decisions as to what constitutes usable materials. Relevant materials may span several disci- plines such as economics, political science, and history. They may also be written in Spanish or Portuguese and assume an un- derstanding of local cultures and practices in specific countries such as Argentina or Brazil. Other factors include ambiguous understanding of terms such as success. Thus, relevancy of many of the variables is problematic. ISSUES OF RATIONALITY Herbert Simon has used the phrase bounded rationality to describe the limita- tions on rationality.5 In explaining his ideas through playing chess, he said that a player never really considers every possible move before choosing the next one. Given the time and human capacity, it is simply not possible, especially if one tries to extrapolate to future moves. A beginning player may only see a few moves and choose on that basis. A more experienced player may not consider many more moves, but in the process September 1994 may visualize several moves ahead based on the pattern of the pieces on the board. Instead of looking at individual pieces, such a player looks at the makeup of the board after successive moves. The choice is not based on all possible moves but only on a subset of possible moves being considered. The choice will then be a satisficing-a Scottish term meaning "good enough"-move.6 Thus the alter- native chosen is satisfactory based on the limited rationality of the player. One must also keep in mind that not all relevant materials are usable for a particular student. In the lecture delivered when he ac- cepted the Nobel Prize in Economic Sci- ence in 1978, Simon further postulated that the satisficing decision is based on the aspiration level of the decision maker, given the limitations on rational- ity.7 It is not that one should not aspire to a higher level. But in a given circum- stance, librarians, limited by their ra- tionality such as constraints of time and memory, provide answers that are suffi- ciently good in those instances for them and the students. Librarians often must make decisions based on incomplete information about issues that are not well understood. Robert Taylor describes this as one per- son (the librarian) trying to find out what another (the patron) wants to know but cannot describe precisely.8 Ac- cording to the classic understanding of rationality in decision making, one would consider all possibilities, calcu- late the cost of using each choice and the benefits to be gained in each case, and then choose the best one. Calculations of cost and benefit should be taken in a broader sense to mean more than expen- ditures of money and time, and knowl- edge gained in this instance. Benefits should also include understanding the process through which knowledge and experience are increased and broaden- ing of one's knowledge base so that these gains may be used later. Reference Service and Bounded Rationality 459 However, there are many limitations on rationality. Foremost are the limita- tions on the resources at a given library and the time available for searching for relevant materials. One must also keep in mind that not all relevant materials are usable for a particular student. Those materials that assume a solid under- standing of the basics in a field or that use advanced research methodology or statistics are not useful to the freshman though they may be helpful to the graduate student.9 Nor is it possible to calculate accurately the cost associated with making some of the choices to- gether with the return on making those choices since one's mental capability is limited.lt is hard to predict whether the value of the information that one finds will outweigh the time and energy spent tracking it down; nor is it possible to know the existence of all relevant mate- rials that have been published or even all materials at one's own library. Thus, one has incomplete knowledge of not only available information but alternative methods of access. Sometimes, when librarians and students don't understand each other, it becomes a process of expanding the bounds of rationality until the librarian's fuzzy set of satisfactory answers and the student's fuzzy set of acceptable answers overlap. Reference librarians answer their in- terpretation of the question based on a context bounded by their knowledge and experiences. Similar to the chess ex- ample above, an experienced reference librarian does not consider each journal index as a separate item but may con- sider classes of indexes as patterns to answer certain types of questions. The number of items or chunks that can be remembered is relatively constant among inexperienced persons and ex- perts.10 A student may attribute one title to a chunk whereas an experienced li- brarian may attribute a group name rep- resenting many related titles to one chunk. Thus, instead of considering Edu- cation Index, Current Index to Journals in Education and Resources in Education as three distinct titles, librarians may take them as one chunk to answer questions dealing with education. This also means that a satisficing an- swer to a freshman may not be good enough for a graduate student working on a thesis. It is possible for two students in the same class to need two different satisfactory answers. In one instance in- volving the author, for example, one stu- dent asked for material on the yearly percentage increase in a certain category of the population. All that could readily be found at the time were yearly totals. That was minimally satisfactory for the author because the end product in- volved performing some calculations, whereas the actual percentages are available in other sources. But to that student, the ariswer was good enough. He sat down with his calculator and did the necessary work to arrive at the per- centages. For another student working on the same assignment, that was not good enough because he could not visu- alize the process. This is knowledge gained from talking to the student. The "good enough" answers cover a whole range of alternatives whose quality is situation-dependent. It also becomes ob- vious that there is a whole set of accept- able answers for students. As with good chess players, librarians see patterns in information that may not be apparent to a student. With an understanding of the infrastructure of information, librarians act on patterns that students may not be able to follow. In order to arrive at an appropriate answer, sometimes librari- ans need to negotiate with the student. To summarize, the student does not go through all possible ways to phrase a question rationally nor does the librar- ian rationally pick the best answer. In the reference process, the two participants try to elucidate their frameworks and try to merge the two. A successful answer is when the two come to a common under- standing. In the example above involv- ing population increase, the author was able to explain the process of calculation --------------------------------------~---- 460 College & Research Libraries to the second student, which took far less time than it would have taken to trace the necessary steps in locating the actual percentages. FUZZY SETS It is apparent from the scenario exam- ined above that there is not a single an- swer nor a precise number of correct answers. The number of "good enough" answers really depends on the rational- ity of an individual. 11 At this time it is appropriate to bring up the concept of fuzzy sets introduced by Zadeh. 12 Fuzzy sets are those sets with ambigu- ous criteria of membership. In a nor- mal or crisp set, the membership is bivalent. One element is either a mem- ber of the set or not. For example, if the set is comprised of students registered for a particular class, then any student who is registered is a member of the set, and any student who is not does not belong to that set. When a student asks for factual information such as addresses of state senators from Illi- nois, only those addresses would be satisfactory, although sources used may vary for different librarians. How- ever, when the set is comprised of tall librarians, without an accompanying definition of tall, then membership be- comes problematic, depending on the understanding of the observer. In this case, membership may range from "0" to "1" on a continuum with "0" mean- ing not in and "1" meaning in, depend- ing on the likelihood of a librarian being considered tall. This is the nature of the question when a student needs to find background information leading to the passage of the Civil Rights Act of 1964. There is no crisp set of documents or sources that will address the question because students will approach the ·question with a set of assumptions based on their previous know ledge and feelings. There is only a set of satisficing materials that will help in answering the question. Again, the set of relevant documents for one person may not be the same as the set of relevant docu- ments for another person. As Bart Kosko said, "The world is gray but sci- September 1994 ence is black and white."13 Thus, rele- vance in the real world is gray and de- pends on the individual case when we do not know the user's assumptions. Similar examples may be a set of useful or extremely useful information sources to address a need. The word extremely in the above example is what Zadeh calls a lin- guistic hedge.14 The degree to which it enhances the accompanying adjective depends on interpretation of the person using it. In this sense, "good enough" may range from minimally useful to a perfect fit to the reference question. As illus- trated above in the reference setting, reference librarians are dealing with a set consisting of the range of satisfac- tory. answers that can be offered by a librarian, and a set of acceptable solutionS according to a student. It is only when these two fuzzy sets-bounded by their respective rationalities-intersect that there is a mutually satisfactory solution to the problem. CONCLUSION The examples presented above consti- tute a brief look at problems encoun- tered in helping students do research. In most cases, reference librarians know what the students want and can help thein find the information. Sometimes, when librarians and students don't un- derstand each other, it becomes a proc- ess of expanding the bounds of rationality until the librarian's fuzzy set of satisfactory answers and the student's fuzzy set of acceptable answers overlap. Then there is a solution. We must realize that, depending on the reference ques- tion, there may be more than one answer considered satisficing. The usefulness of the answer also depends on the rational- ity of the student. ISSUES FOR FURTHER RESEARCH In examining the quality of reference service, it is not enough to improve the accuracy in measuring the measurable with the basic assumption that there is a best or only answer. In view of the salience of bounded rationality in de- cision making, it seems advisable that librarians approach measurements care- Reference Service and Bounded Rationality 461 fully in evaluating the quality of refer- ence service. It is the process that must be better understood in order to improve the quality of services. Data for further research can be gathered through field observation, recordings of reference ne- gotiations on audiotape or videotape, and focused interviews of librarians and patrons. The data may then be analyzed to generate grounded theories to better understand the process. With better understanding, librarians may find more relevant variables that can be measured. REFERENCES AND NOTES 1. Kenneth D. Crews, "The Accuracy of Reference Service: Variables for Research and Imple- mentation," Library and Information Science Research 10 Quly /Sept. 1988): 331-55. 2. Ibid., 349-50. 3. Jo Bell Whitlatch, "Unobtrusive Studies and the Quality of Academic Library Reference Services," College & Research Libraries 50 (Mar. 1989): 189. 4. Ibid., 191. 5. Herbert A. Simon, "Theories of Bounded Rationality," in Decision and Organization, ed. C. B. McGuire and Roy Radner (Amsterdam: North-Holland, 1972), 161-76. 6. Ibid., 168. 7. Herbert A. Simon, "Rational Decision Making in Business Organizations," American Eco- nomic Review 69 (Sept. 1979): 503. 8. Robert H. Taylor, "Question Negotiation and Information Seeking in Libraries," College & Research Libraries 29 (May 1968): 179. 9. See my letter to the editor "'Useful' Isn't Always 'Usable'," Library fournal115 Quly 1990}: 8 and "Bibliographic Instruction and the Scholarship of Integration," Research Strategies 11 (Spring 1993): 66-72 for a discussion of usefulness versus usability. 10. Herbert A. Simon, "How Big Is a Chunk?" Science 183 (Feb. 8, 1974): 482-88. See also K. Anders Ericsson and James J. Staszewski, "Skilled Memory and Expertise: Mechanisms of Exceptional Performance," in Complex Information Processing: The Impact of Herbert A. Simon, ed. David Klahr and Kenneth Kotousky (Hillsdale, N.J.: Lawrence Erlbaum Associ- ates, 1989) for a discussion on differences in memory management between novices and experts. 11. See Taylor, "Question Negotiation," especially pp.187-88. Even though this study deals with special libraries where users are more critical of and want more substantive answers, the process of negotiation between the librarian and the user brings out ambiguities and assumptions that must be filtered to reach an acceptable answer. 12. L.A. Zadeh, "Fuzzy Sets," Information and ControlS Qune 1965): 338-53. 13. Bart Kosko, Fuzzy Thinking: The New Science of Fuzzy Logic (New York: Hyperion, 1993), 8. 14. L. A. Zadeh, "The Concept of a Linguistic Variable and Its Application to Approximate Reasoning-!," Information Sciences 8 (1975): 204. .----------------------------------------------------------------~----- - New Titles from Ae Discovering Librarians: Profiles of a Profession Mary jane Sherdin, editor Results of national studies of vocational interests of library and information professionals. The librarian profile is presented from the ACT, the Strong Interest Inventories, the Meyers-Briggs Type Indicator, and the SIGI PLUS computer- aided career guidance tools, along with analysis of demographic data. $35.95,ACRL member $31.95, 220 p. 0-8389-7753-7, 1994 Managing Student Workers in College Libraries, CLIP Note #20 Michael Kathman, jane McGurn Kathman, camps. Comprehensive guidance for managing student employees. Includes examples of policies and procedures for employment, dismissal, orientation, training, supervision, and performance review. $29.95,ACRL member $25.95, 140p. 0-8389-7752-9, 1994 ACRL University Library Statistics, 1992-93 Library Research Center, Graduate School of Library & Information Science, University of Illinois at Urbana-Champaign, comp. Data from over 109 participating libraries. Library categories include collections, personnel, expenditures, and interlibrary loan. Institutional categories include degrees offered, enrollment size, and faculty size. $69.95;ACRL member $39.95, 80p. 0-8389-7738-3, 1994 Formal Planning in College Libraries, CLIP Note # 19 Sarah Watstein, Pamela Wonsek, and Paula Matthews, camps. A survey of formal planning procedures at academic libraries. Includes examples of worksheets and checklists, vision statements, and formal planning documents. $27.95;ACRLmember$24.95, 120p. 0-8389-7743-X, 1994 To Order: Write: ALA Order Fulfillment 520 N. Dearborn St. Chicago, IL 6061 0 Call toll free: 1-800-545-2433 (press 7) Fax: 312-836-9958 AssociATION OF COLLEGE & RESEARCH LIBRARIES A DIVISION OF THE AMERICAN LIBRARY ASSOCIATION