bell.p65 Making the Library Management Systems Acquisition 347 Making the Library Management Systems Acquisition: Achieving Resolution of a Tough Decision Steven J. Bell and Cynthia Cronin-Kardon At some point in their careers, library administrators will likely be in­ volved in the acquisition of a new library automation system. Whether it is a first-time acquisition or a migration from old to new, the decision is perhaps the most challenging the administrator will ever have to make. Despite an abundance of information in the library literature on the me­ chanical and managerial aspects of acquiring a new automated system, there is scant information on, or investigation into, the decision-making process that leads to the selection of an automation vendor. Based on the premise that the automation decision is both complex and risky, it is a decision fraught with ambiguity, uncertainty, and conflict. This study offers exploratory research on the automation decision process and those factors that lead to decision difficulty. Using the Garbage Can Model of Organizational Choice as a theoretical framework, this study uses deci­ sion resolution as a criterion of decision success. Based on research to identify factors that have enabled decision makers to achieve resolu­ tion, the authors of this study seek to make recommendations that will enable administrators to better manage a tough decision. t a regional conference, the di­ rector of a research library, in discussing his library’s migra­ tion to a new library manage­ ment system (LMS), succinctly summed up the attitude of LMS decision process participants. “It was exhilarating, but I wouldn’t want to do it again for ten years.” The decision to acquire a new li­ brary automation system, referred to here as an LMS, is straightforward. Whether acquiring a system for the first time or migrating to a new system, libraries make this decision for clear reasons. New sys­ tems take advantage of the latest and emerging computer technology in creat­ ing a better search and work environment for staff and patrons. What is extremely difficult, and usually a source of tremen­ dous uncertainty, is the selection of an au­ tomation vendor. A central thesis of this article is that the decision to acquire new information tech­ nology is, as described by Paul C. Nutt, “a tough decision.”1 Tough decisions are filled with ambiguity, uncertainty, and conflict. In selecting an LMS, the library organization must analyze its automation Steven J. Bell is Director of the Gutman Library at the Philadelphia College of Textiles & Science; e-mail: BellS@philacol.edu. Cynthia Cronin-Kardon is a Reference Librarian in the Lippincott Library of the Wharton School at the University of Pennsylvania; e-mail: croninkc@wharton.upenn.edu. 347 mailto:croninkc@wharton.upenn.edu mailto:BellS@philacol.edu 348 College & Research Libraries July 1998 needs carefully and then determine which LMS product matches them. Rarely does an automation vendor offer a product with all the features a library requires, making the LMS decision one of trade­ offs. Should the library sacrifice its exist­ ing highly functional OPAC for vastly improved technical processing modules? Should the library align itself with a new vendor ’s untested, but promising, prod­ uct or the less innovative, but market- proven, product of an established ven­ dor? The decision is ripe for conflict among the decision participants. LMS decisions are characterized by their toughness. To understand why this is, it is impor­ tant to consider the following characteris­ tics of the typical LMS decision situation: � The technology changes rapidly so that information gathered last month may be totally irrelevant in light of today’s de­ velopments. � The information required to make this strategic decision is based on what vendors choose to disseminate. Is it wiser to go with vendor A, who promises to de­ liver features in the next version six months from now, or vendor B, who can deliver today? � The nature of the LMS technology marketplace is itself conducive to uncer­ tainty. Owing to the small size and capi­ talization of many automation vendors, it is understandable that LMS decision makers are asking, “Will this company be here to support this product a year from now?” Though many vendors are reliable, the library systems industry has a num­ ber of elements that contribute to its unpredictability. Nutt found that decision makers often ignored the uncertainty and conflict in tough decisions, which resulted in unsuc­ cessful decision-making processes. 2 Though the LMS acquisition is likely the riskiest and most crucial strategic decision that library administrators make, little is known about the decision-making pro­ cess they use. As the literature review in this article will show, the study of deci­ sion-making for large-scale information technology in academic libraries is largely unexplored territory. This article’s goal is to initiate an investigative study of auto­ mation acquisition and related decision- making behavior in academic libraries. In it is presented the results of a survey of libraries either migrating to a new LMS or making a first-time acquisition. The survey results will provide descriptive information about the decision structure and participants, and insights into how the decision process takes place. The study used the Garbage Can Theory of Organizational Choice as a framework for its analysis of LMS decision-making. A key premise of this article is that the A key premise of this article is that the LMS decision process is complex and therefore subject to dysfunction. LMS decision process is complex and therefore subject to dysfunction. If the decision process lacks proper planning and management, the chances of decision failure increase. An objective in undertak­ ing this research is to provide library ad­ ministrators with a set of recommenda­ tions they can use to improve the LMS decision process. One thing is clear: The LMS decision is unlikely to become less ambiguous. Library administrators need information they can use to better man­ age decisions made under ambiguous conditions. A Theoretical Framework for Analysis One of the challenges of any analysis of decision making is in determining the cri­ teria for judging successful decisions. From among different models used for decision-making analysis (e.g., rational– analytical, bureaucratic, behavioral, po­ litical, etc.), the authors selected Michael D. Cohen, James G. March, and Johan P. Olsen’s Garbage Can Model of Organi­ zational Choice.3 This model describes a Making the Library Management Systems Acquisition 349 decision-making process that occurs in or­ ganized anarchies. The central features of an organized anarchy are problematic preferences, unclear technology, and fluid participation. As described by Cohen and March, an organized anarchy “does not know what it is doing. Its goals are either vague or in dispute [problematic prefer­ ences]. Its technology is familiar but not understood [unclear technology]. Its ma­ jor participants wander in and out of the organization [fluid participation].”4 The model describes a dysfunctional decision- making process and suggests ways in which decision makers might better man­ age the process. It is often used to ana­ lyze decision-making in higher educa­ tion. Cohen and March called the American college or university the prototypic organized anarchy. With their greater staff homogeneity, hierarchical structures, and goal com­ monality, academic libraries tend to defy characterization as organized anarchies. This may be why the model rarely is used to describe academic libraries or to study library decision making. A notable excep­ tion is Joan Giesecke’s study of a decision to choose the best system for governing a university library.5 This article, in compar­ ing the political and garbage can models, concluded that library managers need to understand decision-making models to successfully manage decision processes. Selecting an appropriate strategy can in­ fluence decision outcomes. Giesecke found that the Garbage Can Model ex­ plained the diverse debates of the deci­ sion makers on a stream of unrelated top­ ics and their inability to achieve resolu­ tion with unanimous support of the li­ brary staff. The authors of this study think that the Garbage Can Model is particularly rel­ evant for LMS decision-making. Owing to high-risk and ambiguity factors, these decisions tend to exhibit the dysfunc­ tional behaviors common to decision- making in chaotic organizations. Cohen, March, and Olsen identified a series of properties that characterized an organization’s garbage can decision pro­ cess. Among these characteristics were the level of organizational energy ex­ pended on the decision process (high), the time required for the decision (lengthy), and the actions of the participants (fluid). The garbage can process describes a de­ cision environment that results in deci­ sion resolution failure. Decision resolution is one of three de- Oversight means that a decision, rather than being resolved, is essentially ignored. cision styles discussed by Cohen, March, and Olsen. It is an important concept in this analysis of LMS decision making. The anarchic organization makes decisions in one of three ways. Resolution is the opti­ mal and normative manner of choice de­ termination in organizations. In resolu­ tion, an appropriate choice matches a de­ cision situation after a period of work by decision makers. It is the least common style in the garbage can decision process. What occurs most often is described as either oversight or flight. Oversight means that a decision, rather than being re­ solved, is essentially ignored. Decision makers do not decide but, rather, shift their energies to some other decision without concern for the original problem. Flight means that decision makers shift their energies to a new potential solution instead of resolving the decision based on existing and readily available solutions. How do Cohen, March, and Olsen’s decision styles relate to LMS decision- making? In analyzing LMS decision out­ comes, resolution represents a primary criterion of success. If the decision mak­ ers resolve the decision, it indicates that the process is less characteristic of the Garbage Can Model than those decisions resulting in oversight or flight. In this study, the authors have tried to identify those decision process characteristics that 350 College & Research Libraries July 1998 are most conducive to resolution and those that contribute to oversight or flight. The authors seek to identify the behav­ iors most conducive to resolution in or­ der to build a set of recommendations decision makers can follow in large-scale information technology acquisitions. Often success is linked to decision out­ comes. A successful decision outcome would be judged by criteria such as cost savings, improved system efficiencies, better service delivery, or more satisfied users. Outcomes can be measured, but this requires sufficient time after imple­ mentation to determine the level of suc­ cess. In the case of an LMS, it may take a year or more, plus the collection of com­ parable data, to determine if system ex­ penditures are lower, if search failure rates are lower, and if other measures of success are present. For purposes of this article, a successful decision is one that achieves resolution, regardless of the out­ come, although the authors recognize that the two are strongly connected. Recom­ mendations the authors make are linked primarily to identifying strategies that ad­ ministrators should use for decision reso­ lution. Successfully resolved decisions are completed more quickly, are less confus­ ing to decision participants, and gener­ ally result in faster implementation Review of the Literature The LMS migration is a relevant topic for a study of decision-making behavior. In the past few years, many libraries either have acquired a new system or are plan­ ning to do so. The most comprehensive source of data on library automation trends is the “Automated System Market­ place” report. Published each April in Li­ brary Journal, this survey reports informa­ tion on system installations, vendor revenues, market growth, and other LMS data. The authors examined this annual re­ port for the period between 1994 and 1997. In 1994, system migrations were not reported. However, vendors did report that new installations increased by 69 percent between 1992 and 1993.6 The 1995 survey was the first to report migrations, which numbered 151.7 The 1996 survey reported a dramatic 176 percent increase from 1994 to 1995, with the number of mi­ grations rising from 151 to 418.8 For 1997, the report authors did not report migra­ tion data, stating that “the issue is no longer whether or not to automate, but which system provides the best gateway to all electronic information.”9 The 1997 report stated only that the automation marketplace remained stable. To determine what beyond automation data is reported in the library literature about LMS decisions, the authors searched ERIC, LISA, and Information Science Abstracts. In addition, ABI/In­ form was searched for articles on infor­ mation technology decision-making. Books in Print, LC MARC Records, and local OPACs were searched for mono­ graphs on the topic. A considerable amount of information on library auto­ mation exists, but within that literature virtually nothing was found on the deci­ sion-making process for selecting an LMS. The automation literature is largely directed to the practical aspects of acquir­ ing a new LMS, from developing system selection criteria to procedures for a suc­ cessful implementation. For example, John W. Head and Gerard B. McCabe’s Insider’s Guide to Library Au­ tomation contains a multitude of articles that provide practical advice on nearly every phase of the automation process, from developing local automation sys­ tems to selecting, buying, and installing them.10 Most of the authors discuss auto­ mation at their own sites. This book and others like it, such as John Corbin’s Man­ aging the Library Automation Project, which provides a request for proposal (RFP) sample, costing worksheets, and other practical tools for LMS selection, offer valuable reading for those embarking on an LMS acquisition.11 The journal literature contains a signifi­ http:acquisition.11 Making the Library Management Systems Acquisition 351 cant body of information on designating selection criteria, designing RFPs, and dealing with installation and hardware issues. For example, Edward R. Johnson identified pitfalls in selecting library au­ tomation systems.12 Based on his experi­ ence at Oklahoma State University, he concluded that automation selection must be particularly sensitive to rapid technol­ ogy change, the psychological impact of delays, and the inclusion of library users in the selection process. Mona Couts and colleagues promoted the use of a concept paper as an alternative to the RFP. RFPs are typically time-con­ suming to produce and reflect current needs only. The concept paper provides v e n d o r s w i t h a d e s c r i p t i o n o f t h e library’s vision of its future and re­ quires the vendor to describe how its system’s qualities match that vision.13 Their article is a good example of those that focus on a single facet of the selec­ tion process. Others focus on the experi­ ence of their own institution. In an article providing a comprehen­ sive, detailed look at a single institution’s selection process, Philip Schwarz de­ scribed everything from the factors lead­ ing to the system selection decision to RFP development and the structure of the de­ cision committees.14 Although he con­ cluded that the project was successful, there is no actual analysis of how this organization’s approach to the decision process contributed to that success or how this process was successful owing to its differentiation from the process of librar­ ies experiencing decision failure. Surveys of groups of libraries help identify common elements of the selec­ tion process and provide practical advice on conducting the automation selection. Russell T. Clement surveyed twenty-one small institutions, focusing on factors used in selection decisions.15 He found that cost and software issues were most significant. Julie Hallmark and C. Rebecca Garcia asked the automation administra­ tors of thirty-three libraries what they would and would not do again in repeat­ ing their automation selection process.16 Here, again, the focus is on practical ad­ vice. These, and other articles of this type, largely ignore the structure of the deci­ sion process and the decision participants. They also fail to develop strategies that administrators might use to increase the likelihood of the process achieving reso­ lution. Discovering this lack of information on library automation decision processes, the authors identified additional articles on the general topic of decision making. Some, the authors postulated, may con­ tain discussions of automation system selection. Douglas G. Birdsall and Oliver D. Hensley briefly mentioned the impor­ tance of decision committee structure in discussing library strategic planning.17 Others, including Robert S. Runyan and Carl H. Losse and Arlyle Mansfield Losse, discussed individual or group decision- making in libraries but fell short of pro­ viding concrete insight into decision- making for the acquisition of library tech­ nology.18-19 Although the literature search for this article indicated that automation decision makers can find an abundance of advice that may help with the mechan­ ics of the selection process, a gap was found in the exploration of how a deci­ sion process structure can contribute to decision resolution. Methodology The intent of the design of this research instrument was twofold: first, to col­ lect data about the structure of the LMS decision-making process; and second, to identify elements of the process that contribute to, or detract from, decision resolution. To achieve this, the authors mailed a survey questionnaire to a group of 142 academic libraries identified as hav­ ing acquired or migrated to a new LMS in the past eighteen months. Several methods were used to identify survey participants, including: � news from colleagues and list http:planning.17 http:process.16 http:decisions.15 http:committees.14 http:vision.13 http:systems.12 352 College & Research Libraries July 1998 TABLE 1 Respondents by Collection Size (N = 101) Collection Size Frequency Percent Up to 100,000 100,000 to 500,000 500,000 to 1 million 1 million and greater Total 26 48 8 19 101 25.7% 47.5 07.9 18.8 100.0 servers; � automation columns in library journals; � customer notes and lists on vendor Web sites; � solicitation of automation vendors for names of recent customers. The questionnaires, with a cover letter explaining the research project, were mailed to each library’s system director, if his or her name was known, or to the library director. The cover letter directed the recipient to have the chair of the se­ lection committee complete and return the questionnaire. Most often, the chair was the library director. The first mail­ ing resulted in the return of 71 ques­ tionnaires and a follow-up mailing brought in another 51, bringing the to­ tal number of questionnaires returned to 112. The 112 completed surveys rep­ resented a response rate of 78 percent. Analysis of the returned surveys indi­ cated that eleven were unusable, which adjusted the response rate of useable surveys to 71 percent. Initially, the pri­ mary criterion for participation in the study was migration to a new LMS within the last eighteen to twenty-four months. However, because 46 percent of the re­ spondents were automating for the first time, it was decided to include these first- time acquirers along with those migrat­ ing to new systems. The authors thought this might produce some interesting com­ parisons between first-time acquirers and those migrating. Following are two levels of analysis of the collected data. First, a series of sum­ mary statistics examines the survey popu­ lation. Of particular interest are the char­ acteristics of the LMS selection process (length, deadlines, use of RFPs and con­ sultants, etc.) and the structure of the se­ lection committee (how organized, chairs, reporting, etc.). The nature of the selec­ tion process and its structuring are di­ rectly related to decision resolution. Sec­ ond, a series of cross-tabulations was run to determine how strongly different de­ cision process events correlated with de­ cision difficulty. The results of these cross- tabulations are used to test the following hypotheses: � Migrating libraries experience more difficulty in resolving the decision than first-time acquirers do. � Use of either an RFP or a consult­ ant will result in less difficulty in resolv­ ing the decision. � Limiting the number of vendors/ systems in the selection process will re­ sult in less difficulty in resolving the de­ cision. � Larger, more collaborative commit­ tee structures will result in less difficulty in resolving the decision. � Decision processes that are lengthy result in difficulty in resolving the deci­ sion. � Decision processes with specific deadlines are more likely to result in de­ cision resolution. Survey Population TABLE 2 Respondents by Institution Type (N = 99) Institution Type Frequency Percent Community college 12 12.1% 4-Year college 40 40.4 University - no Ph.D. 15 15.2 Research university 32 32.3 Total 99 100.0 Frequency missing = 2 Making the Library Management Systems Acquisition 353 TABLE 3 Reasons for Migration (N = 101) Reason for Migration Frequency Percent System not adequate 22 28.6 Reduce computing costs 3 03.9 Mandate: Univ. admin. 10 13.0 Mandate: Computer center 2 02.6 Failure of existing system 1 01.3 New technology/feature 21 27.3 Other 18 23.4 Total 77 100.0 Frequency missing = 24 Of the libraries responding to the ques­ tionnaire, 42 percent were public and 58 percent private. The majority were midsized, four-year colleges. Tables 1 and 2 identify the respondents by collection size and institution type. The larger rep­ resentation of smaller libraries may ac­ count for the higher-than-expected re­ sponse from first-time acquirers as opposed to migrating libraries. Whereas larger libraries take an automated envi­ ronment for granted, smaller institutions with limited resources are still acquiring for the first time. Of those libraries that were migrating from an existing system, 13 percent were moving from a locally developed system to one purchased from an automation vendor. Survey respondents were asked to give their reasons for choosing to migrate or acquire an LMS. The most frequently cited reasons were inadequate existing systems and desire for new technology and fea­ tures (see table 3). Surprisingly, few indi­ cated that the impetus for migration was cost reduction. Cost is considered a pri­ mary reason to migrate from an older, mainframe system to newer, client-server technology. The move to the new LMS was initiated by the library director in 46 percent of the cases. There was a fairly even distribution among three other sources of initiation (library strategic plan, joint library and computing strate­ gic plan, committee action). In a few cases, just three percent, an automation crisis forced an LMS acquisition. Length of the decision process, defined as starting when a selection committee is officially formed and ending when an official recommendation or selection de­ cision is made, may indicate the level of difficulty encountered in the decision pro­ cess. The authors wanted to know how long it took respondents to reach their LMS decision. Table 4 shows this distri­ bution. The majority of the libraries com­ pleted their selection process in a year or less. Respondents’ comments indicated that a number of the libraries were con­ sortia members and that the LMS ac­ quired would be the one that is compat­ ible with those of other members. In these cases, the decision process will move faster because the choice options are de­ fined more rigidly. Decision process length can depend on whether the decision makers must meet a specific deadline. The imposition of a deadline on decision makers should cause them to reach a decision on time, whereas the lack of one will allow for procrastina­ tion. Although most of the respondents (42%) had no deadline imposed on them, 29 percent did have a specific deadline and another 29 percent knew they needed to complete the decision within a specific three- to six-month time frame. Of those with a deadline or quasi-deadline, the vast majority (92%) met it. TABLE 4 Length of Decision Process (N = 101) Time Frame Frequency Percent 1-6 months 28 28.0 6-12 months 34 34.0 13-18 months 19 19.0 19-24 months 7 07.0 More than 24 months 11 11.0 Total 99 100.0 Frequency missing = 2 354 College & Research Libraries July 1998 The authors wanted to know more about the respondents’ decision commit­ tees and the impact of committee struc­ ture on decision resolution. The majority of the committees (78%) consisted of fewer than ten members. Among those, committee size most frequently ranged from five to nine members. The tendency was for single decision committees, but in 13 percent of the cases, libraries re­ ported using multiple committees. Mul­ tiple committees generally involved more than ten individuals. Committee mem- Cost is considered a primary reason to migrate from an older, mainframe system to newer, client-server technology. bership was nearly evenly divided be­ tween those organizations that involve only library staff in the decision (44%) and those that create a committee composed of library and nonlibrary staff (56%). Nonlibrary members were either faculty or computing staff. Students rarely served on LMS decision committees. Other ad­ ministrators, such as business adminis­ trators or deans, were identified infre­ quently as committee participants. The person most often responsible for assign­ ing participants to the LMS committee was the library director (77%). A primary characteristic of the orga­ nized anarchy is problematic preferences. By this, Cohen, March, and Olsen meant an organization without clear, under­ standable goals. In deciding which LMS to select, problematic preferences can be a source of difficulty. Within the library, participants may be unclear about the goals the organization is trying to achieve by acquiring a new system. Is the prior­ ity cost savings, a better OPAC, smoother- running technical processing modules, or the need to establish computing indepen­ dence from the university’s central com­ puter operation? Under these conditions, the participants develop greater ambigu­ ity and confusion, resulting in the inabil­ ity to reach resolution or to do so in a rea­ sonable amount of time. Two techniques available to libraries seeking to clarify their preferences are the use of RFPs and automation consultants. Although rigorous and time-consuming to produce, the RFP forces the decision committee to focus the intent of the deci­ sion process and should serve to elimi­ nate or reduce problematic preferences. A consultant can help the decision com­ mittee sharpen its goals or aid in the RFP production. Whatever benefits a consult­ ant may provide are certainly offset by the cost and extra time required for the consultant to study and understand the needs of the organization. An RFP was used by 46 percent of the respondents to select the LMS; only 28 percent used a consultant in the selection process. LMS selection is complicated by the availability of dozens of system vendors. Libraries can narrow the field by target­ ing vendors whose products are geared to libraries of a particular size and type. For example, some vendors cater to the needs of large, academic research librar­ ies whereas others focus on the needs of multiple-branch public libraries. The au­ thors wanted to know how many vendors typically are involved in a library’s choice process. It is important to note that the responses reflect the overall decision pro­ cess and not necessarily the number of vendors on the respondent’s “short list.” The majority of the respondents (73%) examined the systems of three to five ven­ dors. Twelve percent included only one to two vendors, and 15 percent included more than five. Those including just one or two vendors in their selection process typically were consortia members. Although, anecdotally, administrators often describe the LMS selection process as among the most difficult decisions of their professional careers, the survey re­ sults do not reflect this. When asked about the difficulty of the process, 61 percent of the respondents chose the description, Making the Library Management Systems Acquisition 355 TABLE 5 Relation of Decision Process Variables with Decision Difficulty “the decision process was straightforward, but not easy.” Only 19 percent described the process as “difficult,” and only a combined ten percent chose to describe it as “very difficult” or “the most difficult decision of my career.” Respondents also were asked about the length of time needed to reach a selection deci­ sion. When asked if that time was less than needed, about right, or not enough, 69 percent indicated the length of the process was Variable Chi-Square Phi Probability Committee size* 10.225 .320 0.001 Library size 12.668 .356 0.001 RFP used .054 .02 .816 Consultants used .347 -.059 .556 Number of vendors 3.29 .069 .18 Migrate or new 1.99 .158 .142 Length of process 3.265 .071 .181 * Smaller committee (10 members or fewer) right. That 22 percent found the length of the process insufficient could indicate that although in those cases a decision was reached, more time may have allowed for a better resolution. Overwhelmingly, respondents re­ ported that they were “satisfied” (40%) or “very satisfied” (51%) with their selec­ tion. That question was framed as “how satisfied was the committee” with the se­ lection. The answer may be biased com­ ing from the committee chair. The chair may take the committee’s ability to reach a decision as an indicator of satisfaction with the process. Committee participants can and do disagree with the chair about the selection. Participants who dissent with the selection decision may feel pres­ sure to resolve the decision, which leads them to agree to a selection choice. How­ ever, in only ten percent of cases did re­ spondents report anything other than a unanimous decision. Also, anything other than a complete decision disaster or total failure to reach a decision could be per­ ceived as satisfaction with the process, even if certain elements of the process left participants unsatisfied with the out­ come. Testing the Hypotheses and Discussion of Results To test the relationship between decision process variables and decision resolution, the authors used the chi-square test to determine the degree of independence or dependence between selected variables and decision difficulty. When necessary, the responses were collapsed into two categories (e.g., large/small) to create 2 x 2 tables. This increased cell size and per­ mitted use of the same measure of corre­ lation (Phi) for all cross-classifications. Phi varies from -1 (100% negative correlation) to +1 (100% positive correlation). The key dependent variable was the degree of dif­ ficulty of the decision process. Difficulty to complete and inability to complete the process are indicators of decision resolu­ tion failure. The five possible responses to the questions (easy, straightforward, difficult, very difficult, and most difficult) were recoded into two categories (easy/ straightforward and difficult). Of the six hypotheses tested, few are supported by the statistical tests. Table 5 shows the results of the analysis. Only two variables, committee structure and library size, had significant positive cor­ relation to decision difficulty. Larger or more complex committees had more dif­ ficulty with decisions. Committees with more than ten members or libraries us­ ing multiple committees reported diffi­ culty in decision making 57 percent of the time. Institutions using committees of ten or fewer people reported difficulty in de- cision-making 21 percent of the time. This suggests that smaller committees encoun­ ter less difficulty in achieving resolution, which refutes the authors’ hypothesis that 356 College & Research Libraries July 1998 larger committees should encounter less difficulty in resolving decisions. The Gar­ bage Can theory suggests larger commit­ tees bring more energy to the process, contributing to resolution. The lack of correlation between several variables, including first or subsequent In retrospect, the process may seem easier and more satisfactory than it actually was. migration, use of RFP or consultant, num­ ber of vendors, or decision length of time and the degree of decision difficulty, sug­ gests that in the LMS decision process, academic libraries cannot be said to re­ flect the Garbage Can Model of Organi­ zational Choice. One exception is the positive correlation between library size and difficulty. Larger institutions had more difficulty than smaller institutions in the decision process. This provides some support to the authors’ hypothesis that larger libraries experience more dif­ ficulty achieving resolution than small li­ braries. Larger libraries, with more com­ plex issues to confront, are more likely to have problematic preferences. A number of factors may account for both the weakness of the correlations and the authors’ inability to convincingly prove the hypotheses. First, and perhaps most significant, is the research method itself. The survey method seems inad­ equate for analysis of a complex decision process. Although the survey proved use­ ful for gathering statistical data on the respondents, it was less successful in de­ termining what aspects of the process contributed to decision-making difficul­ ties. The authors now believe that the case study method, using interviews with de­ cision participants, would be more effec­ tive in allowing participants to discuss and elaborate on decision difficulty. The discrepancies between the anecdotal evi­ dence and the survey results leads to the suspicion that a form of cognitive disso­ nance explains why many respondents report little or no difficulty with the LMS selection process. Having expended great effort in ac­ quiring the LMS, the respondents, in re­ calling the process, now may perceive it quite differently. In retrospect, the pro­ cess may seem easier and more satis­ factory than it actually was. Whatever their beliefs about the difficulty of the process were at that time, the cognitive dissonance causes a different response to the survey questions. The authors think that the case study method would elimi­ nate or minimize the effect of cognitive dissonance. When in-depth interviews are conducted, participants are more likely to provide greater detail and in­ sights into the decision process that can allow the interviewer to more fully un­ derstand where difficulties occurred in the process. The interviewer also has the ability to use probing questions to help the respondent reflect on how the struc­ ture of the decision committees, other decision makers, the vendors, and the organization contributed to the chal­ lenges of acquiring the new LMS. The survey population may be the source of additional problems. A larger number of respondents were first-time LMS acquirers than initially anticipated. The authors believe that migrating to a new vendor makes for a more difficult decision and that the inclusion of too many first-time acquirers may skew the results toward suggesting that the process is easy. Migrating to a new system in­ volves more risk because the challenge is to acquire a system that loses none of the functionality of the existing system but, rather, brings improvement in all areas while decreasing computing costs and in­ creasing processing speed. Moving from no automation to an automated system is much easier. No matter what one ac­ quires, it is sure to be an improvement. Those migrating have more to lose. No library director wants to hear campus constituencies complaining that the new Making the Library Management Systems Acquisition 357 system, despite the cost of the acquisition, just is not as good as the old one. The survey population also included many small libraries. In those organiza­ tions, the decision process may indeed be simpler. The decision committees are smaller and typically include the library director (who frequently makes the final decision), and lack of funds may inhibit the small library’s ability to conduct an extensive decision-making process. In smaller libraries, a quick, simple decision process may be the norm. A closer inspec­ tion of the responses of the few large re­ search university libraries in the survey population indicated that they were more likely to characterize the process as some­ what difficult, very difficult, or among the most difficult decisions made. Larger re­ search libraries have a much broader con­ stituency to serve, have the resources to consider multiple vendors, and are likely to have larger, more complex decision committee structures. Their need to in­ volve nonlibrary personnel in the deci­ sion process and to gain support, feed­ back, and buy-in from a more diverse group of constituents also contributes to greater decision complexity. It may be that large libraries are experiencing the great­ est difficulty in achieving decision reso­ lution. However, when looking at just the large libraries, the sample size became too small to test the hypotheses reliably. Future Research Although the statistical analysis of the survey data failed to prove the majority of their major hypotheses, the authors maintain that the Garbage Can Model of Organizational Choice is a viable model for describing the LMS decision process. The authors know that many participants in the LMS decision acknowledge the dif­ ficulty in making this strategic, costly choice. The authors also know there are factors that can impede the decision maker ’s ability to achieve decision reso­ lution. The challenge is to determine the best method to discover what those fac­ tors are and then to develop recommen­ dations to help LMS decision makers achieve resolution more easily. The authors see a need for further LMS decision-making research and recom­ mend that it utilize the case study research method. The research should target librar­ ies just embarking on the process of mi­ grating from an old LMS. The research­ ers should cover the entire decision pro­ cess and include participant observation and interviews. This will be more time- consuming and costly, but it is a proven method used by decision-making schol­ ars to study all types of strategic deci­ sions. Future research should involve a better mix of libraries or examine specific segments of the library community, such as those at the research university level. More difficulty is likely to be experienced at larger, more complex library organiza­ tions. If possible, comparative research on different-size organizations should be considered. What benefits can be gained from this research? First, there is a need to deter­ mine under what conditions and for what libraries the LMS decision process is apt to produce the dysfunctional behaviors that cause flight and oversight. Not all li­ braries will face great ambiguity in acquir­ ing a new LMS. For example, a consor­ tium member acquiring the system used by all consortia members. However, should the consortium leadership choose to migrate to a new LMS, that would be a far riskier venture and one more suscep­ tible to those conditions conducive to decision failure. Next, a set of prescrip­ tive measures should be identified that allows administrators and decision mak­ ers to create an improved decision pro­ cess that either eliminates garbage can properties or minimizes their impact. The critical areas for study of the LMS decision process are those behaviors that contribute to, or detract from, resolution. Using the Garbage Can Model of Orga­ nizational Choice as a framework for analysis suggests focusing on two areas: 358 College & Research Libraries July 1998 � Decision structures: Who should have access to the decision process? Does the decision process work best when only a few top administrators are involved, or is it best to invite individuals from throughout the organization to partici­ pate? Do collaborative structures (those involving a large cross-section of library and nonlibrary personnel) work better than the limited structures? � Problematic preferences: How well do the decision makers understand the goals of the decision process? Are their expec­ tations of their ability to identify and ac­ quire the best LMS for the institution re­ alistic, and are they prepared to anticipate and manage the inherent risks and con­ straints of complex technology acquisi­ tions? Does use of the RFP help decision makers better define their decision pro­ cess goals? In addition, researchers should exam­ ine other factors that may lead to flight Under increased conditions of constrained resources and complex and technically confusing options, acquiring information technology is not likely to get easier or less risky. or oversight. These include the number of vendors that decision makers choose to include in the process, the length of time the process takes and whether a deadline is imposed on the decision mak­ ers, and how the degrees of complexity and ambiguity differ in first versus sub­ sequent LMS acquisitions because pre­ scriptive measures may vary in each case (although automated systems are soon likely to be so ubiquitous that first-time acquirers will have all but disappeared). The goal of future research should be to identify specific actions that decision makers can take to overcome the chal­ lenges of making choices in an environ­ ment characterized by many unknown factors. Such recommendations may take the form of guidelines or tasks that ad­ ministrators can follow to achieve deci­ sion resolution. For example: � Form committees that provide wide campus representation. � Limit the number of vendors to no more than three. � Begin actively examining vendor products only when the goals for the ac­ quisition are clearly defined. � Use an RFP to help focus goals if appropriate. � Impose a deadline of one year from the day vendors are identified to actual resolution. Whatever form future research takes, it will help for the profession to define measurable criteria for evaluating suc­ cessful LMS acquisitions. Although this study has concerned itself primarily with the decision process, what comes after the LMS decision is far more important. No matter what recommendations future re­ searchers make to decision makers, their efforts will be in vain if they fail to con­ nect decision-making behavior to effec­ tive decision outcomes. Longitudinal studies of libraries are needed to deter­ mine whether long-term factors such as savings to libraries, system performance improvements, or increases in patron sat­ isfaction levels are actually achieved. If the case study method is used, participat­ ing libraries could remain under obser­ vation for a two- to five-year period after system implementation. This would pro­ vide researchers with the long-term data needed to assess the factors that lead to successful decisions and outcomes—and the relationship between the two. Conclusions The LMS decision maker already has an abundance of advice on various mechani­ cal and procedural aspects of acquiring an automated system. The decision-mak­ ing dimensions of the LMS selection pro­ cess should be explored further so that administrators can better manage these decision situations. If they fail to manage LMS and other technology choice oppor­ Making the Library Management Systems Acquisition 359 tunities effectively, the decision situation becomes dysfunctional. Choice opportu­ nities become garbage can decisions, and the process loses inertia as it grinds down and becomes mired in conflict and con­ fusion, leaving decisions unresolved. Guided by recommendations to help them acknowledge the inherent ambi­ guity of this process and to give atten­ tion to structuring the decision process, decision makers can manage the dys­ functional behaviors. Proper manage­ ment of the decision process will lead to good information technology deci­ sions.20 In an era of technological uncertainty, one sure thing is that administrators will be faced with more complex strategic technology acquisition decisions. The re­ sults of further study into LMS decision- making may have applications as a gen­ eral set of guidelines that help library administrators confront the challenges of choosing the technologies needed to build the libraries of the next century. Under increased conditions of constrained re­ sources and complex and technically con­ fusing options, acquiring information technology is not likely to get easier or less risky. Thinking back to the library director whose last LMS acquisition cre­ ated mixed emotions of excitement and dread, it is hoped that by the next time this director enters the LMS market, our understanding of the acquisition process allows far more control and manageabil­ ity of this strategic decision. Acquiring an LMS will still be a tough decision, but not so tough that library directors might opt to retire rather than face it again. The authors would like to acknowledge Dr. Michael Halperin for his assistance with the instrument design, data collection, analysis, and interpretation of the data; and Jane Bryan for her helpful review and comments. Notes 1. Paul C. Nutt, “Preventing Decision Debacles,” Technological Forecasting and Social Change 38, no. 2 (1990): 159–74. 2. Ibid. 3. Michael D. Cohen, James G. March, and Johan P. Olsen, “Garbage Can Model of Organi­ zational Choice,” Administrative Science Quarterly 17, no. 2 (1972): 1–25. 4. Michael D. Cohen and James G. March, Leadership and Ambiguity: The American College President (New York: McGraw-Hill, 1974), 3. 5. Joan Giesecke, “Recognizing Multiple Decision-making Models: A Guide for Managers,” College & Research Libraries 54 (Mar. 1993): 103–13. 6. Jose-Marie Griffiths and Kimberly Kertis, “Automated System Marketplace: 1994,” Li­ brary Journal 119 (Apr. 1994): 50–59. 7. Jose-Marie Griffiths and Gerald Lundeen, “The Changing Face of Automation: Automated System Marketplace 1995,” Library Journal 120 (Apr. 1995): 44–54. 8. Jeff Barry, Jose-Marie Griffiths, and Peiling Wang, “Jockeying for Supremacy in a Net- worked World: Automated System Marketplace 1996,” Library Journal 121 (Apr. 1996): 40–51. 9. Theresa Pepin, Jeff Barry, and W. David Penniman, “The Competitive Edge: Expanded Access Drives Vendors,” Library Journal 122 (Apr. 1997): 47–56. 10. John W. Head and Gerald B. McCabe, eds., Insider’s Guide to Library Automation: Essays of Practical Experience (Westport, Conn.: Greenwood Pr., 1993). 11. John Corbin, Managing the Library Automation Project (Phoenix, Ariz.: Oryx Pr., 1985). 12. Edward R. Johnson, “Lessons Learned from Unsuccessful Experiences,” Journal of Aca­ demic Librarianship 19 (Nov. 1993): 306–7. 13. Mona Couts, Charles Gilreath, Joe A. Hewitt, and John Ulmshneider, “Use of a General Concept Paper As RFP for a Library System: A New Model for Library System Procurement,” Advances in Library Automation and Networking 5 (1994): 177–202. 14. Philip Schwarz, “Selection of an Automated Library System for the University of Wiscon­ sin Cluster Libraries,” Information Technology and Libraries 6 (Mar. 1987): 40–56. 15. Russell T. Clement, “Cost Is Not Everything,” Library Journal 110 (Oct. 1985): 52–55. http:sions.20 360 College & Research Libraries July 1998 16. Julie Hallmark and C. Rebecca Garcia, “System Migration: Experiences from the Field,” Information Technology and Libraries 11 (Dec. 1992): 149–59. 17. Douglas G. Birdsall and Oliver D. Hensley, ”A New Strategic Planning Model for Aca­ demic Libraries,” College & Research Libraries 55 (Mar. 1994): 149–59. 18. Robert S. Runyon, “Some Principles of Effective Decision Making in American Libraries,” Journal of Academic Librarianship 8 (July 1982): 144–150. 19. Carl H. Losse and Arlyle Mansfield Losse, “Creative Thinking in Decision Making: A Bibliography,” College & Research Libraries 48 (July 1987): 297–301. 20. Steven J. Bell, A Critical Analysis of the Strategic Decision Making Process for Information Technology in Academic Libraries (Ed.D. diss., University of Pennsylvania, 1997).