College and Research Libraries Book Selection, Collection Development, and Bounded Rationality Charles A. Schwartz A model of how librarians may actually go about book selection is presented in three ways: bounded rationality, tacit knowledge, and symbolic content. It is based on the garbage can deci- sion process widely used in political and educational organizations, which tend to have open- ended goals, problematic preferences, hazy technology, and poor feedback. The proposed model is reconciled with the modern-day concept of intelligent choice. nterest in developing a decision-making model of the book selection process has grown because of dissatisfac- tion in some circles with the traditional se- lection literature, particularly its broad surveys of principles and criteria. 1 Over the past decade, writers have criticized this literature for consisting of'' superficial and self-evident generalities" 2 that leave the reader "wondering which criteria are most important and how to apply them. ''3 At first glance, it is not clear that the newly proposed models have much in common. Some are quantitative and would require, if feasible, computers to work, while oth- ers are interpretive and rely on intuition. A closer look, however, indicates that they are remarkably similar in scope and methodology. In scope, the proposed models cover much the same ground as the older sur- veys. The focus of inquiry, chiefly on bib- liographic factors and collection goals, has not been materially clarified or advanced beyond the point reached decades ago. In methodology, however, the models share a set of theoretical concerns unchar- acteristic of the traditional literature. One is classical rationality, defined as simply a procedure for deciding what is correct be- havior by relating means systematically to ends. 4 Closely associated is normative anal- ysis, a prescriptive approach to modeling along the lines of ''all rational actors in such-and-such a situation will believe or act in such-and-such a way.'' This article takes a different approach to modeling the selection process. Its scope is expanded, beyond the features of a book or a collection policy, to comprise human limitations and their organiza- tional consequences. The assumption of classical rationality is replaced by the con- cept of bounded rationality, formulated by Nobel Laureate Herbert Simon to denote our inability to deal with all facets of a choice situation. 5 Concomitantly, norma- tive bias about the way things ''should'' be done is superseded by behavioral analy- sis of certain fundamentally ambiguous properties of the selection process. These properties include: 1. open-ended goals for collection develop- ment but few measurable objectives or other standards for success; 2. problematic preferences owing to the ex- traordinary growth in scholarly publish- Charles A. Schwartz is Social Sciences Librarian at the Fondren Library, Rice University, Houston, Texas 77251. The author wishes to thank David Kaser, Judith Serebnick, and Herbert White for their comments on an earlier draft of this paper. 328 Book Selection and Bounded Rationality 329 ing over the past several decades, and to the broad question of assessing its intellec- tual quality; 3. hazy technology for writing or using book reviews; and 4. little or no feedback about the eventual use or value of a given selection. For collection development as a whole, the addition of a fifth property-fluid participation-provides a so-called gar- bage can process model in which several streams of problems, choices, and deci- sion makers flow through relatively inde- pendent organizational channels. The classical theory of rationality was not designed to handle such properties, and has never handled them satisfacto- rily. 6 Indeed, the issue is no longer how people can hope to relate means systemat- ically to ends, for nearly every social sci- ence recognizes that in the face of even moderate uncertainty over measures of value, striving for theoretically correct courses of action is neither feasible nor sensible. Rather, the issue is whether peo- ple operate reasonably within real-world constraints. 7 By considering book selec- tion in a framework of bounded rational- ity, this article seeks to reexamine a few presumptions of our craft and to make the use of subjectivity somewhat less of a mystery, somewhat more of a technique. "Those who want to understand book selection in behavioral terms face what seems to be an unsolvable problem: that 'selection is always a private, cognitive activity that does not submit to precise observations or delineation. 1 11 The model set forth here is developed in five main parts. Part one assesses previ- ously proposed schemes of classical ra- tionality in book selection. Part two out- lines new approaches to rational choice behavior. Part three applies a particular approach-bounded rationality in a gar- bage can decision process-to book selec- tion. Part four describes the role of tacit knowledge and the concept of intelligent choice. Finally, part five presents an over- view of theoretical prospects in this field and suggests a few areas for future re- search. CLASSICAL RATIONALITY Those who want to understand book se- lection in behavioral terms face what seems to be an unsolvable problem: that ''selection is always a private, cognitive activity that does not submit to precise ob- servation or delineation. ''8 This problem can be managed only by making some as- sumption about a hypothetical bibliogra- pher's cognitive processes. If no simplify- ing assumption is made-and not every writer who is critical of the traditional se- lection literature makes such an assumption-no behavioral model is pos- sible. For example, Hendrik Edelman in a well-known essay (on collection develop- ment by classification of library materials) takes the position that, because ''selection will be made by different people or even groups of people,'' a ''consolidated selec- tion practice statement is out of the ques- tion. " 9 The assumption of classical rationality holds that an individual identifies objec- tives (say, of collection development), searches for alternative courses of action, evaluates each alternative in comparative terms, and then chooses the best course (book). This conventional notion is as- sessed first for quantitative models of the selection process, then for interpretive schemes. Quantitative Models One type of quantitative model centers on numerical judgments. The method used is to compare books by assigning them scores for certain collection criteria. The model of John Rutledge and Luke Swindler, for example, includes six crite- ria, each with a different range of points to represent their relative importance. Thus, subject matter is worth up to 30 points, in- tellectual quality up to 25, and so on for a total possible score of 100 per book. 10 In more elaborate fashion, the John DePew model is a combination of a flowchart hav- ing dozens of decision boxes, some politi- 330 College & Research Libraries cal in nature, and an equation of fifteen variables for calculating a book's numeri- cal score. 11 The second type of quantitative model depends on probability distributions. Robert Losee proposes that books be ranked according to the statistical range of certain bibliographic features whose pa- rameter values would be computed from machine-readable cataloging (MARC) rec- ords, both for previously selected andre- jected books. Given the size of new book production in the United States (some 50,000 titles annually), this scheme would seem to require a database of overwhelm- ing proportions. Losee's hypothetical ex- ample, however, refers to a tiny, carefully defined situation in which an art history bibliographer would make selections by just three factors: a book's height, the number of times the term art appears in a review of the book, and whether a faculty member recommended the book. These factors are assumed to have normal, Pois- son, and binary distributions, respec- tively.12 Although such models purport to pro- vide rationality to the selection process, there is no evidence that they are either theoretically sound or a sensible guide for practitioners. In this field of library work where personal theory and practice are al- lied, these models generate needlessly complex analysis, entailing what econo- mists call the ''excessive costs of per snick- erty calculation,' ' 13 and they overlook or ignore basic principles for making an '' op- timally imperfect decision.'' 14 The principle of dominance pertains to an obvious choice situation, one in which a book (or other option) is easily evaluated on the basis of a few salient criteria. Be- cause of this principle, complex analysis is unnecessary except with respect to non- dominant options. 15 Although this practi- cal aspect of selection methodology is dis- regarded in quantitative models (as if all books were tough choices), it is taken for granted by some publishers. Thus, it is not uncommon for university presses to print only a few thousand copies of a book to be sold at a high price because the academic library "market is assumed to be both cap- tive and inelastic. " 16 Apparently, the May 1989 same assumption underlies inflation of in- stitutional over individual subscription rates for scholarly joumals. 17 Two additional principles can be de- scribed in terms of a trade-off between precision and relevance of selection crite- ria. On the one hand, the Rutledge- Swindler and the DePew models focus, for the most pcri"t, on relevant criteria but misrepresent them by an unwarranted emphasis on precision. Such emphasis is known as the fallacy of misplaced concrete- ness. 18 It begins with the presumption that good decisions require precise measures of alternative choices. Where the essential numbers are of doubtful validity, how- ever, there is little point in characterizing a decision process as rational in those terms. Rutledge and Swindler concede that a ''decision to assign a specific score [to a book] will be determined by subjec- tive judgments," yet they gloss over the incongruity of this numbers game by dub- bing it an art of selection. 19 In the DePew model, the numbers game is extended to political affairs, such as whether a book requester is a power base or a troublemaker. Even the judgment of the requester is scored on a scale of zero to eleven. 20 When one considers the objec- tion of many professors to having a library request turned down, one can roughly imagine the aggregated fury of an entire faculty upon discovering its members be- ing graded by librarians for knowledge of scholarly literature. On the other hand, the Losee model uses quantitatively precise but largely ir- relevant selection criteria. Substituting literature-wide probability distributions for individuals' numerical judgments does not alleviate the fallacy of misplaced concreteness but rather comEounds it by the convenience of reverse logic. 21 Such logic starts with the methodology that is most agreeable, or most in accord with one's notion of rationality, and then moves from the predetermined scheme backward to find selection criteria for which the scheme is relevant. In his discussion of normal distributions, for example, Losee cites the price, height, weight, and num- ber of pages of a book. These features are physically measurable, some are accessi- Book Selection and Bounded Rationality 331 ble from MARC records, but they are hardly logical choices as the basis for eval- uation. Certain other issues of quantifying the selection process, such as figuring the in- tellectual quality of a book, are discussed below in section three. For now, attention is turned to the practical side of selection models. Because time is a scarce resource, and the gathering and processing of infor- mation involve both economic and oppor- tunity costs, a decision maker must decide how much time and other resources to de- vote to a choice situation. The basic rule is stated succinctly by philosopher John Rawls: "We should deliberate up to the point where the likely benefits from im- proving our plan are just worth the time and effort of reflection." 22 This optimization problem of balancing the costs and benefits of complex analysis has not been addressed in any discussion of book selection. While the traditional li- brary literature can do without it, quanti- tative models of book selection sorely need a practical orientation. Only if one gathers a lot of information about biblio- graphic details, and then processes it by charts or equations, can one act in a way that the authors of those models would judge to be intelligent. But to obtain that sort of information, one would have to function quite unfittingly. One would be- come immersed in picayune matters, such as deciding which of fourteen numerical categories best suits a book review (De- Pew model) or counting the number of times a particular positive or negative word occurs in a review (Losee model). Bibliographers will find, on calculation or reflection, that this would be an unreason- able way of allocating their time, let alone making decisions. Interpretive Models While it is commonplace in the library field to regard selections more or less as the interplay of collection policy and indi- vidual judgment, Ross Atkinson is the only writer to have suggested a model of the decision-making process along such lines. Atkinson describes a hypothetical bibliographer as evaluating a book accord- ing to an internalized typology of citation (or reference) contexts. Most crucial are the contexts of resolution, categorized as archival (what is known about the collec- tion), communal (what is known about the interests of the clientele), and thematic (what is known about the subject litera- ture). Classical rationality of the interpre- tive kind is brought into play by a process Atkinson terms the ''inevitable weighting in different situations of the three con- texts'': Depending upon such factors as the goals of the library, the nature of the subject, the status of the subject within the library, and the predilec- tions of the individual selector, the three contexts will be prioritized. One context will always take precedence over the other two. 23 What is troublesome about Atkinson's approach is that contexts of resolution do not form a prospective model of the selec- tion process. Rather, they represent after- the-fact generalizations. It is tempting to evaluate decisions once their outcomes are known, yet philosophers and psychol- ogists agree that we are in constant danger of making errors whenever we try to do so. 24 This is because such contexts and other schemata-akin to what the late Mi- chael Polanyi called tacit knowledge25- generally form the background of our de- cisions but remain unspecificable, outside of conscious awareness and therefore ''in- compatible with introspective or observa- tional access to causal necessity. '' 26 Everyone uses tacit knowledge- intuition together with experience-to make sense of others' behavior and to per- form skilled tasks. The nature of tacit knowledge is not well understood, but certain of its characteristics are clear enough. First, it departs in a fundamental sense from the classical theory of rational- ity: "we can know more than we can tell. " 27 Whenever we are faced with a choice or a task of any complexity, we do not spell out the key criteria by which all other criteria are to be judged and then ap- ply them in a formal way. Instead, we have an indeterminate range of anticipa- tions and rules of application that cannot be consciously directed. Thus, in everyday life we can make fine discriminations-recognizing a friend's 332 College & Research Libraries face among a thousand, indeed among a million-without being able to put this knowledge into words. Likewise, Herbert Simon emphasizes in his discussion of bounded rationality how tacit consider- ations pervade professional life: In any field of expertise, possession of an elabo- rate discrimination net that permits recognition of any one of tens of thousands of different ob- jects or situations is one of the basic tools of the expert and the principal source of his intui- tions. 28 In this perspective, book selection has a critical tacit dimension-similar to what a British librarian terms the experienced glance 29 -that cannot be adequately ac- counted for in classical models of rational- ity or conventional notions of subjectivity. A second characteristic of tacit knowl- edge is the converse of the first: "we sometimes tell more than we can know.' ' 30 When people attempt to report on why they made a particular choice, they do not do so by consulting a memory of the cog- nitive process, for "it is the result of think- ing, not the process of thinking, that aR- pears spontaneously in consciousness.' ' 31 Instead, people simply make judgments about how plausible it is that a given factor would have influenced the decision. From the standpoint of subjective feelings of prediction and control, plausibility judg- ments are natural and disconfirming feed- back is hard to come by. Nevertheless, such judgments are not based on any true introspection, but are implicit and exist only in use . Third, tacit knowledge is personal, in the sense of involving one's personality, and also in the sense of being solitary. As a rule, no two persons share the same tacit theories. 32 The type of considerations that Atkinson focuses on-collection aims, cli- entele interests, and subject literatures- are familiar norms in the library field, yet far too general to determine particular choices of any complexity to the exclusion of other, nondominant options. Some evi- dence is provided by the Report of the Na- tional Enquiry into Scholarly Communica- tion, which found no common denomina- tor among seven case studies of the book selection process: May 1989 A variety of procedures are employed by the in- stitutions, by item selectors within the same in- stitution, and by the same selectors over time. Indeed, if there are twenty-five selectors in an institution, there are at least twenty-six proce- dures for item selection. 33 ''No theoretic ingenuity can account for the profound ways in which some people can organize their experiences and reshape their capacities for judg- ing new works.'' Finally, the concept of tacit knowledge carries with it mechanisms of learning and maturation that, being only partially translatable into words, cannot be cap- tured in models. Atkinson rightly de- scribes a bibliographer's experience as ''constantly evolving,' ' 34 yet the essential argument, made many years ago by T. S. Eliot, is that no theoretic ingenuity can ac- count for the profound ways in which some people can organize their experi- ences and reshape their capacities for judging new works .35 Perhaps we should adopt a provisional outlook about ''con- texts of resolution" and other interpreta- tions of the selection process, remember- ing with Lord Tennyson that Our little systems have their day; They have their day and cease to be. NEW APPROACHES TO RATIONAL CHOICE BEHAVIOR In the social and behavioral sciences, there has been growing interest in making sense of choice situations characterized by abstract goals, ambiguous stimuli, and hazy technology. Such interest has led to development of three interrelated con- cepts: bounded rationality, garbage can decision process, and loosely coupled sys- tem. These concepts have been applied to educational institutions, publishing houses, governmental bodies, multina- tional corporations, accounting firms, welfare agencies, appellate court systems, military bureaucracies, and research proj- ects. 36-44 The studies have little to say about libraries directly and only slightly more to Book Selection and Bounded Rationality 333 report by way of broad quantitative find- ings. But a realistic model of book selec- tion can be found in their systems of thought. Bounded Rationality In the 1950s Herbert Simon and others found that the micro-assumptions of ra- tionality in the classical theory of the firm are contrary to fact. There is no evidence that managers actually equate marginal costs and revenues, or that consumers dis- tribute their purchases in such a way as to maximize their utility. 45 From this research work arose the concept of bounded ratio- nality, which takes the limitations of the decision maker and the complexities of the environment as central concerns. Simon has demonstrated that choice be- havior does not conform to textbook de- scriptions because individuals lack the computational capacity as well as the nec- essary information to act this way. It is not a question of approximation; the shortfall in information-processing capabilities, relative to environmental complexities, is so great that teamwork and even com- puters are of minor assistance. Decision makers introduce, as reasonable re- sponses, a number of familiar procedures to simplify a choice situation. These in- clude replacing optimal goals with satisfic- ing ones; devising heuristic (rule-of- thumb) strategies, so that selections will be made after searching only a small part of the total choice situation; allocating time and other resources with an eye to the varying quality of decisions; and, with experience, adlusting aspiration levels to the attainable. Garbage Can Decision Process The garbage can analogy is not one of or- ganizational structure, but of a common type of choice situation that appears to defy conventional assumptions of ration- ality. According to such assumptions, ''choice opportunities lead first to the gen- eration of decision alternatives, then to an evaluation of their consequences, then to an evaluation of those consequences in terms of objectives, and finally to a deci- sion. " 47 There is considerable evidence that this scenario greatly oversimplifies many of the problems that practitioners face, especially in educational and public- sector organizations. Case studies of the garbage can process typically focus on singular choice in- stances for group resolution, such as the location of a new medical school or the re- organization of a university curriculum. 48 With rare exceptions, such studies have not dealt with recurrent, less burdensome decisions at the individual level, like book selection. To illustrate their general tenor and findings, consider the selection of an academic dean. In the garbage can pro- cess, goals having to do with the character of the institution-hiring women and mi- norities, upgrading neglected depart- ments, funding new programs-would typically be taken into consideration in the selection of the dean. 49 This type of deci- sion is shaped (somewhat like library col- lection development as a whole) by sev- eral relatively independent streams of problems, participants, choice opportuni- ties, and solutions flowing through orga- nizational networks. At the individual level, the research pa- per has been accurately described in terms of tacit knowledge and the garbage can process. 51 Certainly, well-ordered models of scholarly inquiry-formulate a theoreti- cal problem, select an appropriate method, analyze and interpret the results, confirm or deny the theory-are a poor ac- count of the research experience. Robert Merton, dean of Academic sociologists, explains the discrepancy: [There is a] rockbound difference between the finished versions of scientific work as they ap- pear in print and the actual course of inquiry. . . . Typically, the scientific paper or monograph presents an immaculate appearance which re- produces little or nothing of the intuitive leaps, false starts, mistakes, loose ends, and happy ac- cidents that actually cluttered up the inquiry. 52 (That sort of contextual information is screened out of scholarly publications by editorial conventions. 53) Loosely Coupled System The idea of loose coupling represents the extent of incoherence and disjuncture between organizational means and ends. Writers use it with a variety of garbage can 334 College & Research Libraries decision situations in mind: (1) a relative lack of formal standards, (2) delegation of discretion, (3) occasions in which any one of several means will produce the same end, ( 4) infrequent inspection or coordina- tion of activities, and (5) the absence of linkages for some theory to be relevant. For example, in educational organizations there is little feedback from ends to means, or feed-forward from means to ends, so the casual connections between the two are not testable. 54 Bounded Rationality in Book Selection Not all behavior within an organization conforms to a single model. In a library, catalogers tend toward a highly pro- grammed mode in the interests of biblio- graphic control, whereas administrators usually adopt a contextual approach, shifting between bureaucratic and partici- patory styles. For bibliographers, four perspectives of bounded rationality may make ordinary experience in book selec- tion more explicable. These perspec- tives-open-ended goals, problematic preferences, hazy technology, and poor feedback-suggest a necessarily more complex, loosely coupled model of means and ends in collection development than that described by conventional theories of rational choice. Open-Ended Goals Libraries have broad, idealistic goals, e.g., "to serve the academic community," but few measurable objectives, time con- straints, or other standards for success. If it were possible to change this situation without trivalizing it into a definitional problem, we would want to do so for the sake of personnel management. 55 How- ever, several recent studies of choice be- havior show that, where means-ends rela- tions are essentially ambiguous-as they are in book selection-schemes of explicit objectives are neither representative of stable preference orders nor suitable for creative interpretation of what the goal might mean. 56 About the only way of mak- ing open-ended goals tractable is to devise loosely coupled subgoals (such as collec- tion development policies broken down May 1989 by subject descriptors). The aim of sys- temizing selection goals and their criteria into some grand model appears unrealis- tic. Problematic Preferences: Book Proliferation Bibliographers seem destined to face an increasing gap between library budgets and potential acquisitions owing to the ex- traordinary growth of the scholarly enter- prise in the past several decades. The number of new books published in the United States displayed no clear trend during the first half of the century, hover- ing around 10,000 titles a year. From 1950 to 1975, however, the number of new titles published annually increased fourfold, and by the mid-1980s it surpassed the 50,000 mark (in addition to some 100,000 titles imported yearly). Moreover, in the decade after 1975 the average price of hardcover volumes just about doubled. 57 In an extensive survey of these publish- ing trends, the American Academy of Learned Societies concluded that "find- ing ways to help librarians cope intelli- gently with the flood of material engulfing them is central to any rational solution to the problems of scholarly communica- tion.' ' 58 This task must be carried out in the face of downward pressures on the to- tal number of books that libraries are will- ing to buy-given the sharp increases in the number and cost of journals, continu- ing reductions in federal grants for acqui- sition programs, and other constraints on financing higher education. ''Although evaluation of intellectual quality is an integral part of book se- lection, there has been virtually no discussion of this problem in the li- brary literature." Problematic Preferences: Intellectual Quality Although evaluation of intellectual quality is an integral part of book selec- tion, there has been virtually no discus- sion of this problem in the library litera- Book Selection and Bounded Rationality 335 ture. Perhaps a general assumption has been that ''Quality is rather like obscenity: ... defining [it] is a futile exercise, even though a group of individuals may be able to form a consensus on the quality of a par- ticular item.' ' 59 While the concept cannot be rolled into a neat definition, such a defi- nition is neither necessary nor appropri- ate. A wide range of studies show that (1) well-established criteria of intellectual cri- teria do exist but that (2) a group of indi- viduals, even subject experts, fully com- mitted to the same criteria will likely disagree on their judgments of a given work. Several large-scale surveys of journal editors and referees in the social and be- havioral sciences have found a striking consensus on the importance of a limited number of criteria for evaluating manu- scripts. The main norms indude illumina- tion of a salient problem, generalizability, comprehensiveness, practicality, original- ity, and integration of diverse theoretical perspectives into a single model. 60 One survey characterized research milestones in just three attributes: providing a differ- ent direction for researchers or practition- ers, bringing in this new "truth" from outside a sterile field, and treating a spe- cific and enduring problem. 61 Such criteria and attributes, like any val- ues, are equivocal by nature and, when employed together, are weighed differ- ently by individuals. For example, in a survey of 299 members of editorial boards for psychology journals, there was only moderate agreement on evaluations of particular manuscripts and even less cor- relation between those evaluations and subsequent judgments by other readers, as measured by citation counts. 62 More generally, Robert McC. Adams of the Na- tional Research Council has found that this absence of a clear consensus on the relative merits of scholarly works "bears little relation to field, across the entire span from the physical to the social sci- ences. " 63 The acceptance or popularity of a piece of research in the social and behavioral sci- ences has little to do with its scientific sup- port. Instead, what counts as knowledge may be only loosely coupled to empirical evidence, internal consistency, or predic- tive power. 64 Some observers suggest that pictorial models with high reliability but low precision (such as the broken window theory of neighborhood decay or the J- curve theory of political revolution) are es- pecially useful for the handling of complex concepts. 65-67 Others contend that the most interesting studies are those with the greatest distance between independent and dependent variables, e.g., the predic- tion that single vehicle automobile acci- dents are influenced by suicide stories on television soap operas. 68 Intellectual quality is often associated with iconoclasm, the breaking of estab- lished theoretical premises. In this view, the most important works are those that stand in stark relief to taken-for-granted concepts: By denying the validity of routinely held cogni- tive assumptions, a theory attracts attention, forcing its audience to re-evaluate subject mat- ter from a novel viewpoint. New propositions are interesting or uninteresting only in relation to this baseline of traditionally accepted knowl- edge.69 The corollary is that a piece of research will be considered unimportant if it merely proves, however carefully, what is already well known. This is particularly true for practitioners who need ways of reducing the uncertainty of problems where com- mon sense can go either way: Managers are accustomed to making decisions under relatively high levels of uncertainty. They have little interest in the kind of knowl- edge about causes and effects that merely in- creases the certainty of an already apparent re- lationship. They are decidedly more interested in either conceptual knowledge that helps them order their thinking about an action area or plausible hypotheses that have not previously occurred to them. 70 Iconoclasm, in the broadest sense, un- derlies Thomas Kuhn's theory of scientific revolutions as a succession of tradition- bound periods in which an older para- digm is replaced in whole or in part by an incompatible new one. The significance of paradigms is that they give rise to what Kuhn calls normal science: "law, theory, application, and instrumentation to- gether'' to form coherent traditions of re- 336 College & Research Libraries search. 71 Paradigm change depends not on the logic of experimental evidence but rather on techniques of persuasion: the paradigm's promise to solve problems, its aesthetic qualities, or something else to convince others that it is on the right track. Whereas Kuhn professes faith that para- digm choices are rational, other social phi- losophers contend that'' our standard the- ories of what constitutes rationality are not adequate to illuminate this complex process." 72 Hazy Technology: Book Reviews Technology is knowledge, not only of how to make things but also of how to make choices. Two types of studies have been done to gauge the reliability of the re- view process in scholarly communication and in book selection. One type concerns book criteria, the other reviewer bias. Some of the reports on book criteria date back several decades, but they generalize well with more recent studies that indicate reviews are overwhelmingly positive though somewhat superficial. For exam- ple, Dean Champion and Michael Morris, in an analysis of 2,378 reviews in three so- ciological journals over a 23-year period (1949-71), found only 18.2 percent of the reviews to be negative in their overall ap- praisals. "The most frequent criticisms centered upon substance (29.4 percent), while theory and methodology received less than 5 percent emphasis, respec- tively.' ' 73 The paucity of sound criticisms was attributed to an implicit norm of reci- procity in the review process and a conse- quential "fear of adverse reaction from one's colleagues.' ' 74 Similar results, though with different conclusions, were obtained by Beth Ma- cleod in an analysis of 2,600 reviews in Choice and Library Journal in 1978. Only 6 and 9 percent of the reviews in the respec- tive journals were negative recommenda- tions. 75 More often than not, reviewers made no mention of the quality of writing, depth of analysis, nature of research, or uniqueness of the book. Less than 10 per- cent of reviewers stated and expressed a judgment about the author's thesis. Ma- cleod found that "reviewer anonymity May 1989 [which was the procedure in Choice at that time] does not appear to make for more critical reviews." She concluded, instead, that the preponderance of favorable rec- ommendations was essentially a matter of initial editorial screening of doubtful ti- tles.76 This conclusion is supported by the apparent agreement on the part of editors that if a book is bad, it is not worthy of a • 77 review. Reviewer evaluation of books often is challenged as unreliable. This is hardly surprising in the social and behavioral sci- ences, given the diversity of methodologi- cal orientations and the lack of law-like generalizations. Although it is widely rec- ognized that uncertainty in the theoretical sphere sometimes is resolved by reference to social criteria, such as an author's sta- tus, reports on specific patterns of bias are largely inconclusive. Some patterns are contradictory, such as generational effect (prestigious reviewers tend to evaluate prestigious authors particularly well) as opposed to noblesse oblige (the higher the prestige of the reviewer relative to that of the author, the better the review). 78 Other patterns are marginal, e.g., Macleod's finding that "reviewers [of both sexes] were slightly more likely to criticize a book by a woman for its shallowness, and to praise a book by a man for its depth. " 79 Overall, the reports suggest an improved situation since the 1950s and 1960s, when a negative review might impugn an entire school of thought. 80 Now the greatest criti- cism tends to come from within, rather than outside of, the author's own theory group.st By and large, people recognize the hazi- ness of book reviews, as indicated by the adage that ''bad reviews are better than no reviews at all." Apparently, the tenor of reviews in the social sciences has little im- pact on book sales or library circulation. 82 Given the widespread and unavoidable differences in evaluation of scholarly works, particularly at the time of publica- tion, ''one review is very likely to give a rating far from the mean, and even three or four reviews cannot be relied upon to represent the central tendency of the total ratings. " 83 This is not to say that reviews are uninteresting or unhelpful, but Book Selection and Bounded Rationality 337 simply-with reference to quantitative models of book selection-that ''a totally inappropriate procedure is to derive a general rating from a review and to use that ratin~ mechanically in the evaluation process.'' ''Feedback is a concept that has little practical bearing on book selection as a routine process." Poor Feedback Feedback is a concept that has little prac- tical bearing on book selection as a routine process. It is treated in the library litera- ture in two distinct ways: performance ap- praisals of a general kind for individuals, and collection use studies as a measure of organizational effectiveness. We know, for example, that past book use tends to predict future use, and that the most re- cent materials likely receive the most use. 85 Such findings, as well as use statis- tics, are helpful for collection manage- ment from a retrospective standpoint, as in deciding what to discard or to send to storage. But they are not conducive to book selection in a prospective, singular sense, as in identifying specific titles to ac- quire. Indeed, it is difficult to imagine a work- able feedback mechanism for bibliogra- phers based on collection statistics when one considers the time and trouble of pre- paring such feedback studies compared to those of tacit knowledge. Further, the samples used in these studies are of lim- ited significance, inasmuch as record book use in most libraries is low, around the 50 percent mark. Also to be considered is the ''heavy measure of conservatism'' en- tailed in providing for fresh patron inter- ests. 86 A MODEL OF BOOK SELECTION A model of book selection -drawing in part on the previous text-is described be- low in terms of three main aspects: bounded rationality in a garbage can deci- sion process, tacit knowledge, and sym- bolic content. Bounded Rationality There is no reason to suppose that there is a single decision process to which all in- stances of book selection (or rational choice generally) should conform. Collec- tion goals are varied and imperfectly un- derstood. Choice alternatives are some- times dominant, sometimes problematic. Technology is hazy. Performance mea- sures are vague. Decision quality cannot be evaluated autonomously but rather re- quires a long period to establish and de- pends to some extent on others' decisions. Even with careful analysis, feedback is ambiguous. As a rule, it is virtually impos- sible to define a criterion of rationality for this type of situation (or, what amounts to the same thing, to provide a solution to the garbage can decision process). A partial exception seems to establish this rule. Mention was made above of the heuristic maxim articulated by Rawls, that we should deliberate on a choice up to the point where the likely benefits of analysis are just worth the time and trouble. This would appear to be unassailable logic. Yet, oecause decision makers acting under uncertainty routinely ignore it for intelli- gent reasons (see discussion below on symbolic content), one can plausibly ar- gue that it is deficient not only as a de- scriptor of behavior but also as a criterion of rationality. The general point to be emphasized is that the assumption of classical rationality is not appropriate for book selection and should be replaced by the concept of bounded rationality in a garbage can deci- sion context, given our inability to deal with all facets of a choice situation. Tacit Knowledge The ideal of a known, systematic, and justifiable cognitive process, as suggested in Atkinson's interpretive model, does not exist. Instead, reliance must be placed on tacit knowledge, whose basic premises differ sharply from such an ideal. Tacit knowledge accredits the human capacity to acquire knowledge by steps one cannot specify, it allows for choices to be made 338 College & Research Libraries within an intuitive framework that is largely personal, and it accepts the fact that such intuitions are incapable of proof or disproof. 87 An interpretive model, by contrast, ex- presses more the human ability to form post factum theories of one's own behavior than an ability to relate means to ends in a prospective manner. 88 We have already considered the meth- odological problems and excessive costs of quantitative models of classical rationality in book selection. There remains an issue of decision logic. The common presump- tion of the quantitative schemes is that once one has minutely examined and nu- merically scored different features of a book, one has mastered the whole. Un- derlying this notion is the logic of positiv- ism: If you cannot measure, your knowl- edge is meager and unsatisfactory. With computers called in to handle all the num- bers, book selection becomes viewed as an engineering problem; in Losee's model, if a change in collection development policy requires computer reprogramming, "hu- man selection of new books will be neces- sary for a time.' ' 89 However, any attempt to break a book down into numerous variables that can be analyzed separately misses the essence of knowledge and literature: the whole is greater than the sum of the parts. 90 In this perspective of evaluating a work as a co- herent whole, we return to a critical tacit dimension of the selection process-the experienced glance-that cannot be estab- lished by argument, but is rather some- thing known directly to the practitioner. "Imitation offers guidance to it, but in the last resort . . . we alone can catch the knack of it; no teacher [or model] can do this for us.' ' 91 Symbolic Content In at least one respect, book selection (along with certain other kinds of organi- zational behavior) confounds both the classical and the bounded models of ra- tionality. This has to do with the symbolic requirements and signaling opportunities o£ choice situations in a garbage can deci- sion process. As we have seen, classical rationality is outcome oriented. Primary attention is May 1989 placed either on systematic analysis of ends and means to find the best possible decision (quantitative models), or on care- ful reflection to find the most likely reason for a given decision (interpretive models). Bounded rationality, on the other hand, takes an opposite approach. It attributes special value to decision efficiency. Heu- ristic devices and minimum, or satisficing, criteria are emphasized as reasonable re- sponses to limitations of time and information-processing capabilities. It makes little sense to focus on out- comes in the book selection process, as though selections were discrete or salient events, given the stream of selections year after year and the lack of feedback. Nor is it advisable to stress the practical side of decision making by showing how the problematic character of selections might be decreased by weakening the require- ments for rational choice. This is because the central purpose of book selection for the practitioner may be neither outcomes nor efficiency; rather, the main point may be the process itself: Where the substantive outcome has relatively low salience for many participants or where it is difficult to establish decision efficacy by observ- ing outcome effects, we would expect process pleasures to become particularly relevant. 92 In such a view, the way in which a bibliog- rapher goes about his or her work may be explicable only if we recognize that a deci- sion process is an arena for exercising so- cial values, for displaying competence, and for exhibiting appropriate behavior with respect to a core ideological construct of modern life: the concept of intelligent choice. 93 Where knowledge of inputs is incom- plete and, at the same time, outcomes are ambiguous, conspicuous consumption of information may be a sensible strategy. For a bibliographer, reading book reviews serves, in part, to provide a ritualistic as- surance that proper attitudes about collec- tion development exist. For an adminis- trator, gathering information, asking for information, and justifying decisions in terms of information have come to be sig- nificant ways of showing that the organi- zation is well managed. The symbolic value of information is Book Selection and Bounded Rationality 339 probably more common to garbage can decision processes (administration and collection development) than to biblio- graphic functions (cataloging and circula- tion). It is probably more common at the top of the hierarchy than at the bottom; and, for bibliographers, it is probably more common in subject literatures lack- ing paradigms than in scientific fields. However, this signaling strategy may be compromised-that is, information seek- ing may be regarded as a sign of indeci- siveness or lack of faith-if the organiza- tion acknowledges interpretations of decision making that emphasize limited rationality, a garbage can process, and loosely coupled structure. Not surpris- ingly, symbolic investments in informa- tion depend upon conventional concep- tions of intelligent choice. CONCLUSIONS Two things can be said on behalf of the foregoing model. It treats book selection as a rather distinct category of a class of decision-making situations well-known in social science literature; and, consistent with that body of research, it does not at- tempt to explain too much. There are about sixty studies of the gar- bage can decision process. Nearly all of them discuss choice situations as dissimi- lar, nonrecurrent events for group resolu- tion. A few studies, chiefly in public ad- ministration, address the garbage can decision process at the level of the individ- ual policymaker, but they maintain the as- sumption of dissimilar types of choice sit- uations and are content to rely on Charles Lindblom's classic notion-a precursor to bounded rationality-of muddling through. 94 Until now, no one has consid- ered the garbage can decision process from the perspective of a practitioner, act- ing alone, who faces the same type of choice situation on a daily basis. Hence, the important role of tacit knowledge to account for intuition together with experi- ence in a book selection model. It must be realized that the model pre- sented here is an example of non theory. There is no strict proof of the existence or extent of tacit knowledge; its acceptance II depends on the observation of complex performances which not only do not re- quire conscious direction but which defy attempts to codify rules for success.''95 At the same time, conventional models of book selection, with lists of choice varia- bles overburdened with box scores and process arrows, are not a theoretical expla- nation of anything. Nor do such schemes, as Simon noted in his Nobel address, II even remotely describe the processes that human beings use for making deci- sions in complex situations. " 96 Rather, when individuals and organizations are faced with a profusion of goals and crite- ria, they behave much like a system with almost no goals at all. Also, the model in this paper avoids pre- scriptions. The importance of careful anal- ysis or reflection is itself subject to ration- ale appraisal and will depend, like so much else, on the individual and the con- tingencies of his or her situation. 97 Model- ing cannot fruitfully involve the pursuit of such contextual detail. Moreover, a nor- mative approach would require bibliogra- phers' acceptance of fairly rigid norms about what is rational and acceptable be- havior. What is so vital to the develop- ment of either a normative or an explana- tory theory-a paradigm 11 signaling the gestalt in which the situation is to be seen"98 -would be calamitously narrow for book selection. It may be useful to suggest possible areas of future research. One such area is prompted by the thought that some of the bounded rationality properties described primarily with reference to social and be- havioral science books may not fully per- tain (1) to collection development in the physical sciences, given the role of normal paradigms for gauging intellectual quality in those fields, or (2) to collection develop- ment in the humanities, in view of various cultural establishments of extremely influ- ential reviewing sources. 99 Another area for reconsideration might be serials management. Some writers have proposed methods of classical ra- tionality for journal evaluation that would require, for example, the development of lists of citation rankings or productivity/ cost ratings for scores of fields. 100' 101 No one has really addressed the optimization problem of balancing the costs and bene- fits of such complex and time-consuming 340 College & Research Libraries analyses. Perhaps the conclusions drawn in this article with reference to book selec- tion do not fully apply to serials manage- ment, since journal acquisitions and de- acquisitions are more discrete and probably more important decisions. Further research in the area of collection development might benefit from the adoption of a particular point of view- namely, that collection development can be described as a garbage can decision process if one adds the dimension of fluid May 1989 participation. Thinking about collection development in that framework would at least promote frank recognition that there is no general way of evaluating the organi- zational effectiveness of libraries. Karl Weick, who conceived the idea of a loosely coupled system, once wrote of the need for redirecting our concern to one of effec- tive organizing of the decision process, since the doin~ of the process it.self is what really counts. 2 REFERENCES AND NOTES 1. Examples of this literature include Robert N. Broadus, Selecting Materials for Libraries, 2d ed. (New York: Wilson, 1981); Arthur Curley and Dorothy Broderick, Building Library Collections, 6th ed . (Metuchen, N.J.: Scarecrow, 1985); William A. Katz, Collection Development: The Selection of Mate- rials for Libraries (New York: Holt, 1980); and G. Edward Evans, Developing Library and Information Center Collections, 2d ed . (Littleton, Colo.: Libraries Unlimited, 1987). 2. Ross Atkinson, "The Citation as lntertext: Toward a Theory of the Selection Process," Library Resources & Technical Services 28:109 (Apr.-June 1984) . See also Hendrik Edelman, "Redefining the Academic Library," Library Journal101:55 (Jan . 1, 1976); and Hendrik Edelman, "Selection Methodology in Academic Libraries," Library Resources & Technical Services 23:33 (Winter 1979). 3. John Rutledge and Luke Swindler, "The Selection Decision: Defining Criteria and Establishing Priorities," College & Research Libraries 48:124 (Mar. 1987). 4. James G. March, "The Technology of Foolishness," in Ambiguity and Choice in Organizations, 2d ed., ed . James G. March and Johan P. Olsen (Bergen, Norway: Universitetsforlaget, 1979), p.70. 5. Herbert A. Simon, "Rational Decision Making in Business Organizations," in Advances in Behav- ioral Economics, ed. Leonard Green and John H . Kagel, v.1 (Norwood, N.J.: Ablex, 1987), p.18-47. This article is the lecture Herbert Simon delivered in Stockholm, Sweden, December 8, 1978, when he received the Nobel Prize in Economic Science. For a broader discussion, see Herbert A. Simon, Administrative Behavior: A Study of Decision Making Processes in Administrative Organization, 3d ed . (New York: Free Press, 1976). 6. Simon, "Rational Decision Making," p.25. 7. Helmut Jungerman, "The Two Camps of Rationality," in Models of Thought, ed. Herbert A. Si- mon (New Haven, Conn.: Yale Univ. Pr ., 1979), p.659. 8. Atkinson, "Citation as Intertext," p.109. 9. Edelman, "Selection Methodology," p.35. 10. Rutledge and Swindler, "Selection Decision," p.123-31. 11. John N. DePew, "An Acquisitions Decision Model for Academic Libraries," Journal of the Ameri- can Society for Information Science 25:237-46 (July-Aug. 1975). 12. Robert M. Losee, "A Decision Theoretic Model of Materials Selection for Acquisitions," Library Quarterly 57:269-83 (July 1987) . 13. William J. Baumol and Richard E. Quandt, "Rules of Thumb and Optimally Imperfect Deci- sions," American Economic Review 54:24 (Mar. 1964) . 14. Ibid., p .23. 15. Detlof von Winterfeldt and Ward Edwards, Decision Analysis and Behavioral Research (New York: Cambridge Univ. Pr., 1986), p.388. I 16. Walter W. Powell, "Publishers' Decision-Making: What Criteria Do They Use in Deciding Which Books to Publish?" Social Research 45 :230 (Summer 1978). 17. Richard De Gennaro, ''Escalating Journal Prices : Time to Fight Back," in Libraries, Technology, and the Information Marketplace: Selected Papers, ed . Richard De Gennaro (Boston : Hall, 1987), p.103-13. 18. Alfred North Whitehead, Science and the Modern World (Middlesex, England : Penguin, 1938), p .74. 19. Rutledge and Swindler, "Selection Decision," p .128. Book Selection and Bounded Rationality 341 20. DePew, "Acquisitions Decision Model," p.243-44. 21. John Kenneth Galbraith, A View From the Stands: Of People, Politics, Military Power and the Arts, ed. Andrea D. Williams (Boston: Houghton, 1986), p .34-38. 22 . John Rawls, A Theory of Justice (Cambridge, Mass.: Harvard Univ. Pr., 1971), p.418 . 23. Atkinson, "Citation as Intertext," p.llS. Emphasis added. 24 . For an overview of this literature, see Richard E. Nisbett and Timothy DeCamp Wilson, "Telling More Than We Can Know : Verbal Reports on Mental Processes," Psychological Review84:231-59 (May 1977). 25. Michael Polanyi, The Tacit Dimension (Garden City, N.Y. : Doubleday, 1966). 26 . Kenneth S. Bower, "Intuition and Discovery," in Theories of the Unconscious and Theories of the Self, ed. Raphael Stern (Hillsdale, N.J.: Analytic Press, 1987), p.79. 27. Polanyi, Tacit Dimension, p.4. 28. Herbert A . Simon, "Alternative Visions of Rationality," in Models of Thought, p.108. 29 . Kenneth Whittaker, Systematic Evaluation: Methods and Sources for Assessing Books (London: Clive Bingley, 1982), p.49. 30. Nisbett and Wilson, "Telling More," p.247. 31. George A. Miller, Psychology: The Science of Mental Life (New York: Harper, 1962), p.56; cited in Nisbett and Wilson, "Telling More," p.25. 32. Polanyi, Tacit Dimension, p .25. 33. National Enquiry into Scholarly Communication, Scholarly Communication: The Report of the Na- tional Enquiry (Baltimore: Johns Hopkins Univ. Pr., 1979), p.143. 34. Atkinson, "Citation as Intertext," p.114. 35. T.S. Eliot, The Use of Poetry and the Use of Criticism (Cambridge, Mass.: Harvard Univ. Pr., 1933), p.6- 9. 36. See, for example, Michael D. Cohen, James G . March, and Johan P. Olsen, "A Garbage Can Model of Organizational Choice," Administrative Science Quarterly 17:1-25 (Mar. 1972); Karl E. Weick, "Educational Organizations as Loosely Coupled Systems," Administrative Science Quar- terly 21:1-19 (Mar. 1976 ); March and Olsen, Ambiguity and Choice; Frank W. Lutz, ''Tightening Up Loose Coupling in Organizations of Higher Education," Administrative Science Quarterly 27:653-69 (Dec. 1982); and Michael D. Cohen and James G. March, Leadership and Ambiguity: The American College President (Boston: Harvard Business School Press, 1986). 37. Walter W . Powell, Getting Into Print: The Decision-Making Process in Scholarly Publishing (Chicago: Univ . of Chicago Pr., 1985) . 38. See, for example, James G . March and Johan P. Olsen, "The Uncertainty of the Past: Organiza- tional Learning Under Ambiguity," European Journal of Political Research 3:147-71 (June 1975); James G. March and Johan P. Olsen, "What Administrative Reorganization Tells Us About Gov- erning," American Political Science Review 77:281-96 (June 1983); John S. Dryzek, "Don't Toss Coins in Garbage Cans: A Prologue to Policy Design," Journal of Public Policy 3:345-67 (Oct. 1983); Jeffrey L. Pressman and Aaron Wildavsky, Implementation, 3d ed. (Berkeley: Univ. of California Pr., 1984); and James G. March and Johan P. Olsen, "The New Institutionalism: Organizational Factors in Political Life," American Political Science Review 78:734-49 (Sept. 1984) . 39. Geert Hofstede," Angola Coffee-Or the Confrontation of an Organization with Changing Val- ues in Its Environment," Organization Studies 1:21-40 (1980) . 40. D. J. Cooper, D. Hayes, and F. Wolf," Accounting in Organized Anarchies: Understanding and Designing Accounting Systems in Ambiguous Situations,'' Accounting, Organizations, and Society 6:175-91 (1981). 41. B. Gummer, "Organized Anarchies, Loosely Coupled Systems, and Adhocracies: Current Per- spectives on Organizational Design," Administration in Social Work 6:77-84 (1982); and D . Bran- non, "Decision-Making in Public-Welfare: Scientific Management Meets Organized Anarchy," Administration in Social Work 9:23-33 (1985). 42 . Lawrence B. Mohr, "Organizations, Decisions, and Courts," Law and Society Review 10:625-42 (Summer 1976). 43. Alan Ned Sabrosky, James Clay Thompson, and Karen A. McPherson, "Organized Anarchies: Military Bureaucracies in the 1980s,'' The Jou rna[ of Applied Behavioral Science 18:137-53 (1982); and Ambiguity and Command: Organizational Perspectives on Military Decision Making, ed. James G. March and Roger Weissinger-Baylon (Marshfield, Mass.: Pitman, 1986) . 44. Joanne Martin, "A Garbage Can Model of the Research Process," in Judgment Calls in Research, ed. Joseph E. McGrath, Joanne Martin, and Richard A . Kulka (Beverly Hills, Calif.: Sage, 1982), p.17-39. 45. Herbert A . Simon, "Theories of Decision Making in Economics and Behavioral Science," The 342 College & Research Libraries May 1989 American Economic Review 49: 253-83 (June 1959); and Simon, "Rational Decision Making," p.42-43. 46. Gary S. Beck," A Theory of the Allocation of Time," The Economic ]ournal75:493-517 (Sept. 1965); Steffan Burenstam Linder, The Harried Leisure Class (New York: Columbia Univ. Pr., 1970), ch.6; and James G. March, "Bounded Rationality, Ambiguity, and the Engineering of Choice," Bell journal of Economics 9:587-608 (1978). 47. Cohen, March, and Olsen, "Garbage Can Model," p.2. 48. Kare Rommetveit, "Decision Making Under Changing Norms, 11 in Ambiguity and Choice, p.140-55; and James G. March and Pierre J. Romelaer, "Position and Presence in the Drift of Decisions," in Ambiguity and Choice, p.251-76. 49. Johan P. Olsen, "Choice in an Organized Anarchy," in Ambiguity and Choice, p.82-139. 50. J. M. Ziman, Knowing Everything About Nothing: Specialization and Change in Scientific Careers (New York: Cambridge Univ. Pr., 1987), p.73. 51. Martin, "Garbage Can Model of the Research Process." 52. Robert K. Merton, On Theoretical Sociology (New York: Free Press, 1967), p.4 . See also James H. Austin, Chase, Chance, and Creativity: The Lucky Art of Novelty (New York: Columbia Univ. Pr., 1978), p.xii. 53 . Contemporary Classics in the Social and Behavioral Sciences, ed. Neil J. Smelser (Philadelphia: lSI Press, 1978), p.vii-viii. 54. Weick, "Educational Organizations," p.5; and Kim Cameron, "Critical Questions in Assessing Organizational Effectiveness," Organizational Dynamics 9:70-71 (Autumn 1980). 55. For a general discussion, see Herbert S. White, Library Personnel Management (White Plains, N.Y.: Knowledge Industry Publications, 1985). 56. March, "Bounded Rationality," p .595. 57. Oscar Handlin, "Libraries and Learning," The American Scholar 56:205 (Summer 1987); and Na- tional Enquiry, Scholarly Communication, p.85-87,149. 58. National Enquiry, Scholarly Communication, p.13. 59. Bert R. Boyce and Danny P . Wallace, ''In Defense of 'Ignorant Drivel,' "American Libraries 18:654 (Sept. 1987). 60. Janet M. Chase, "Normative Criteria for Scientific Publishing," American Sociologist 5:262-65 (May 1970); Duncan Lindsey, The Scientific Publication System in Social Science (San Francisco: Jossey-Bass, 1978); Wirt M. Wolff, "A Study of Criteria for Journal Manuscripts," American Psy- chologist 25:636-39 (July 1970); and Stephen D. Gottfredson, "Evaluating Psychological Research Reports: Dimensions, Reliability, and Correlates of Quality Judgments," American Psychologist 33:920-34 (Oct. 1978). 61. John P. Campbell, Richard L. Daft, and Charles L. Hulin, What to Study : Generating and Developing Research Questions (Beverly Hills, Calif.: Sage, 1982), p.91-114. 62 . Gottfredson, "Evaluating Psychological Research," p . 930-33. 63. Robert McC. Adams, "Rationales and Strategies for Social Science Research, II Bulletin, American Academy of Arts and Sciences 34:15 (1981). 64. For an overview of this literature, see W. Graham Astley, "Administrative Science as Socially Constructed Truth," Administrative Science Quarterly 30:502-504 (Dec. 1985). 65. Kurt Lewin, "Formalization and Progress in Psychology," in Studies in Topological and Vector Psy- chology, ed. Kurt Lewin, Ronald Lippitt, and Sibylle Korsch Escalona (Iowa City: Univ. of Iowa Pr., 1940), p .9-42. 66 . James C. Davies, "Revolution and the J-Curve," in Violence in America: Historical and Comparative Perspectives, A Report Submitted to the National Commission on the Causes and Prevention of Violence, v .2 (Washington: Govt. Print. Off., 1969), p.547-77. 67 . Chris Argyris, "Research as Action: Useable Knowledge for Understanding and Changing the Status Quo," in The Theory and Practice of Organizational Psychology: A Collection of Original Essays, ed. Nigel Nicholson and Toby Wall (New York: Academic, 1982), p.205. 68 . David P . Phillips, "The Impact of Fictional Television Stories on U.S . Adult Fatalities : New Evi- dence on the Effect of Mass Media on Violence," American journal of Sociology 87:1340-59 (May 1982); cited in Barry M . Staw, "Repairs on the Road to Relevance and Rigor : Some Unexplored Issues in Publishing Organizational Research," in Publishing in the Organizational Sciences, ed. by L. L. Cummings and Peter J. Frost (Homewood, Ill.: Irwin, 1985), p.98 . 69. Astley, "Administrative Science," p.503. 70 . Richard E. Walton, "Strategies With Dual Relevance, II in Doing Research That Is Useful for Theory and Practice, ed. Edward E. Lawler III et. al. (San Francisco : Jossey-Bass, 1985), p.197-98 . 71. Thomas S. Kuhn, The Structure of Scientific Revolutions, 2d ed., enlarged (Chicago: Univ. of Chi- cago Pr., 1970) . 72. Richard J. Bernstein, The Restructuring of Social and Political Theory (New York: Harcourt, 1976), p.93 . Book Selection and Bounded Rationality 343 73 . Dean J. Champion and Michael F. Morris, "A Content Analysis of Book Reviews in AJS, ASR, and Social Forces," American journal of Sociology 78:1261 (Mar. 1973). 74. Ibid., p.1257, 1264. 75. Beth Macleod, "Library journal and Choice: A Review of Reviews," The Journal of Academic Librari- anship 7:27 (Mar. 1981) . 76. Ibid., p.28. 77 . Lewis A. Coser, Charles Kadushin, and Walter W . Powell, Books: The Culture and Commerce of Publishing (Chicago : Univ. of Chicago Pr., 1985), p.309 . 78. William E. Snizek and E. R. Fuhrman, "Some Factors Affecting the Evaluative Content of Book Reviews in Sociology," The American Sociologist 14:108-14 (May 1979). 79 . Macleod, "Library journal and Choice," p.28. 80. Lawrence E. Riley and Elmer A. Spreitzer, "Book Reviewing in the Social Sciences," The Ameri- can Sociologist 5:358-63 (Nov. 1970); and Norval Glenn, "On the Misuse of Book Reviews," Con- temporary Sociology 7:254-55 (May 1978). 81 . William E. Snizek, E. R. Fuhrman, and Michael R. Wood, "The Effect of Theory Group Associa- tion on theE valuative Content of Book Reviews in Sociology,'' The American Sociologist 16:185-95 (Aug. 1981). 82. On the tenor of reviews, see Judith Serebnick, "Book Reviews and the Selection of Potentially Controversial Books in Public Libraries," Library Quarterly 51:390-409 (Oct. 1981); and Coser, Ka- dushin, and Powell, Books, p.323, 329- 31 . On library circulation, see John P . Schmitt and Steward Saunders, "An Assessment of Choice as a Tool for Selection," College & Research Libraries 44:375-80 (Sept. 1983) . 83. Glenn, "On the Misuse," p .254. 84. Ibid ., p.255. 85. For reviews of this literature, see Robert N. Broadus, "Use Studies of Library Collections," Li- brary Resources & Technical Services 24:317-24 (Fall1980); and Robert W. Burns, Jr ., "Library Use as a Performance Measure: Its Background and Rationale," The journal of Academic Librarianship 4:4-11 (Mar. 1978). 86. Broadus, "Use Studies," p.323. 87. Knowing and Being: Essays by Michael Polanyi, ed. Marjorie Grene (Chicago: Univ. of Chicago Pr., 1969), p .133-34; and Alasdaire Macintyre, After Virtue: A Study of Moral Theory, 2d ed. (Notre Dame, Ind. : Univ. of Notre Dame Pr ., 1984), p.15 . 88 . Olsen, "Choice in an Organized Anarchy," p .83; and March, "Bounded Rationality," p.593. 89. Losee, "Decision Theoretic Model," p.281. 90. Clifford Geertz, "Blurred Genres: the Reconfiguration of Social Thought," American Scholar 49:165-79 (Spring 1980); and Jon Huer, Art, Beauty, and Pornography: A Journey Through American Culture (Buffalo: Prometheus Books, 1987), p .72. 91. Knowing and Being, p .126. 92. James G. March and Johan P. Olsen," Attention and the Ambiguity of Self-Interest," in Ambigu- ity and Choice, p.47. See also Powell, Getting Into Print, p .98-100; Brown and Wildavsky, Imple- mentation, p.247-48; and Lawrence Mohr, Explaining Organizational Behavior (San Francisco: Jossey-Bass, 1982), p .158. 93 . Martha S. Feldman and James G. March, ''Information in Organizations as Signal and Symbol," Administrative Science Quarterly 26:177-84 (June 1981). 94. Charles Lindblom, "The Science of Muddling Through," Public Administration Review 19:79-88 (Spring 1959). 95 . Martin X. Moleski, "The Role of Retortion in the Cognitional Analyses of Lonergan and Pola- nyi," in Self-Reference: Reflections on Reflexivity, ed. Steven J. Bartlett and Peter Suber (Boston: Martinus Nijhoff, 1987), p.220. 96. Simon, "Rational Decision Making," p.42 . 97. Rawls, Theory of justice, p.418. 98. Kuhn, Structure of Scientific Revolutions, p .189. 99. See Richard Ohmann, Politics of Letters (Middletown, Conn.: Wesleyan Univ. Pr., 1987). 100. An excellent source is Robert N. Broadus, ''The Applications of Citation Analyses to Library Col- lection Building,'' in Advances in Librarianship 7, ed. Melvin J. Voigt and Michael H. Harris (New York: Academic, 1977), p .299-335. 101. See, for example, Donald A. Windsor, "De-Acquisitioning Journals Using Productivity/Cost Rankings," De-Acquisitions Librarian Newsletter 1 (Spring 1976), p .8-10; and Richard De Gennaro, Libraries, Technology, and the Information Marketplace: Selected Papers (Boston: Hall, 1987), p.103-13. 102. Karl Weick, "Re-Punctuating the Problem," in New Perspectives on Organizational Effectiveness, ed. PaulS. Goodman and Johannes M. Pennings (San Francisco: Jossey-Bass, 1977), p.216-17.