cook.p65 Users’ Hierarchical Perspectives on Library Service Quality 147 Users’ Hierarchical Perspectives on Library Service Quality: A “LibQUAL+” Study Colleen Cook, Fred Heath, and Bruce Thompson This study confirms that a single second-order factor is associated with the delivery of high-quality library services in a research university envi­ ronment. However, a hierarchical factor analysis also demonstrated that research library users simultaneously think about library quality at mul­ tiple levels. The LibQUAL+ diagnostic tool, a product of the ARL’s New Measures Initiative, shows that although a single factor dominates user thinking about library service quality, all of the items used in the survey suffuse this factor. Nevertheless, several first-order factors contribute important unique information to the notion of service quality. As different types of users place varying degrees of importance on the first-order factors, the utility of the hierarchical model is demonstrated. n the history of many profes- baseball parlance. BA and ERA suffice as sions, careful research and rig- measures of excellence in the sport. orous design have produced Human health also has its overarching standards of measurement that higher-order measures. Since Stephen permit specialist and layman alike to achieve a perspective on performance. In baseball, the most statistically driven of all professional sports, batting average (BA) and earned run average (ERA) pro­ vide a benchmarking overview on all bat­ ters and pitchers in the history of the game. A .400 batting average is extraor­ dinary; a career ERA below 2.00 will cer­ tainly earn the achiever a niche in the Hall of Fame. And although underlying vari­ ables contribute to performance measures (bat speed, perhaps, and pitch velocity), those factors are not part of the common Hale’s publication in 1733, blood pressure measurement has been a fundamental in­ dicator of human health. Although meth­ ods have improved in the aftermath of Hale’s rather invasive methods of mea­ surement, pressure scores expressed as millimeters of mercury have become rec­ ognizable standards. A score of 120/70 expresses a desirable human condition the world over. Conversely, a score of 190/100 would be a measurement of considerable concern, although the search for causal factors could span a broad spectrum of possible contributors.1 As is the case with Colleen Cook is Professor of Library Science and Executive Associate Dean in the Sterling C. Evans Library at Texas A&M University; e-mail: ccook@tamu.edu. Fred Heath is Dean of Library Services and holder of the Sterling Evans Chair at Texas A&M University; e-mail: fheath@tamu.edu. Bruce Thompson is Professor of Educational Psychology and Distinguished Research Scholar at Texas A&M University and Adjunct Professor of Community Medicine at Baylor College of Medicine; email: bbt6147@acs.tamu.edu. 147 mailto:bbt6147@acs.tamu.edu mailto:fheath@tamu.edu mailto:ccook@tamu.edu 148 College & Research Libraries March 2001 baseball, many underlying lower-order health factors may contribute to an individual’s blood pressure reading, but from a health maintenance view, the single higher-order factor often serves as a deci­ sion point as to whether more serious in­ tervention may be required. One plausible model of user perceptions of library service quality posits that users may simultaneously think about quality at multiple levels within a hierarchical model. Just as scientific measurement is used to benchmark observable phenomena such as batting averages and blood pressures, careful statistical procedures are utilized to measure and rank based on passive human activities such as “watching” or as impre­ cise a precept as “perceiving.” For example, Nielsen Media Research, through a process refined over half a century, determines the rankings of television programs in the United States based on the viewing habits of only thirteen thousand people in only five thousands households. This, in a na­ tion in which ninety-nine million house­ holds have television sets!2 Similarly, in the state of Texas, consumers are asked by the Office of Public Insurance Counsel to rate health maintenance organizations (HMOs) on specific areas of care and physician ser­ vices as well as to provide an overall rank­ ing. Those perceptions of consumers are analyzed to create a report card in which the state’s thirty-four HMOs with the larg­ est market share are rated.3 SERVQUAL Not Enough And so it is across services and industry sectors. Developing the concept of Gap Theory, Leonard L. Berry, A. Parasuraman, and Valarie A. Zeithaml have offered mea­ sures of customer perception of service quality to establish benchmarks across a broad array of public sectors, from health service providers to airlines, to restaurants. Only customer perceptions matter, accord­ ing to the authors, whose SERVQUAL in­ strument has become a standard of mea­ surement in the private sector. Various SERVQUAL studies have shown that, indeed, it is not only possible, but also necessary to benchmark percep­ tions. The fulfillment of customer expec­ tations is key to the success of every res­ taurant. A high quality score depends on many things: a solicitous and knowledge­ able staff, menus that meet expectations for preparation and presentation, and, importantly, reliability from visit to visit. Over twelve years of study across a wide range of industry sectors, the authors found that five first-order factors contrib­ ute to an overall measure of service qual­ ity in a nonlibrary environment. These include:4 Reliability: Ability to perform the prom­ ised service dependably and accurately; Assurance: Knowledge and courtesy of employees and their ability to convey trust and confidence; Empathy: The caring, individualized attention the firm provides to its custom­ ers; Responsiveness: Willingness to help cus­ tomers and provide prompt service; and Tangibles: Appearance of physical facili­ ties, equipment, personnel, and commu­ nications materials. However, various studies have clearly demonstrated that (1) the five SERVQUAL dimensions are not recoverable in the li­ brary context, and (2) additional dimen­ sions of quality not measured by SERVQUAL are necessary. 5, 6 Library Context What has all this to do with libraries? Li­ braries have long recognized that their metrics were out of phase with the rising demands for accountability in higher edu­ cation. Among the member libraries of the ARL, for example, measurement has not yet progressed appreciably beyond basic input metrics. Despite the cautions of the ARL to the contrary, the ARL Index, based solely on expenditure metrics, have come to be widely accepted as a measure of li­ brary quality.7 With hopes of improving the situation, researchers and practitio­ ners alike have begun to explore a num­ ber of assessment alternatives. Users’ Hierarchical Perspectives on Library Service Quality 149 And so within the library service com­ munity there has been “increasing pres­ sure on libraries to assess the degree to which their services demonstrate criteria of ‘quality’ … [and] not to equate ‘qual­ ity’ merely with collection size.”8 This movement beyond sole reliance on col­ lection counts as indices of quality seems eminently reasonable. As Danuta A. Nitecki recently observed, “Flying across the Atlantic, are you more likely to judge the quality of the airline you use by the number of planes it operates or by the reliability of its schedules of departures and arrivals and the attention its staff gives you?”9 Library users appear to in­ voke similar criteria when they evaluate the services that libraries provide. Responding to this movement within the field, the ARL has sponsored a num­ ber of New Measures initiatives. In Octo­ ber 2000, ARL organized a two-day con­ ference bringing library directors and staff together with internationally recognized experts on measuring perceptions of ser­ vice quality. One of the New Measures ini­ tiatives is the LibQUAL+ study being con­ ducted by ARL and the Texas A&M Uni­ versity libraries.10 Continuing phases of the LibQUAL+ study are being supported in part by the Fund for the Improvement of Post-Secondary Education (FIPSE). However, a fundamental question that must be addressed in these initiatives in­ volves how user perceptions of library service quality should be modeled. Of course, as Clyde Hendrick and Susan Hendrick noted, in the behavioral sci­ ences “theory building and construct measurement are joint bootstrap opera­ tions.”11 That is, we progress in an itera­ tive manner by tentatively formulating a theory, developing a measure of that theory, evaluating the measure, revising the theory, and then proceeding cyclically back through this bootstrap process. Hierarchical Models One plausible model of user perceptions of library service quality posits that users may simultaneously think about quality at multiple levels within a hierarchical model. An illustrative model—in this example a model of cognitive abilities—is presented in figure 1. This actually is an approxima­ tion of the model commonly used to mea­ sure intelligence. However, similar hierar­ chical models also can be used in understanding attitudes or perceptions. Hierarchical models are useful because perspective taking at different levels al­ lows us to see different dynamics. Bruce Thompson offered the following analogy: The first-order analysis is a close-up view that focuses on the details of the valleys and the peaks in mountains. The second-order analysis is like looking at the mountains at a greater distance, and yields a potentially dif­ ferent perspective on the mountains as constituents of a range.12 At the first-order lower level, we may see trees and streams but not recognize the mountain range. At the second-order FIGURE 1 Illustrative Hierarchical (Third-order) Structure Cognitive Abilities Verbal Quantitative Comprehension Grammar Computation Equations http:range.12 http:libraries.10 150 College & Research Libraries March 2001 level, we lose sight of details but gain the global perspective of the range. Purpose of the Study The purpose of the present study was to investigate what hierarchical structure underlay library service perceptions of 3,987 participants. These data were pro­ vided from users at eleven ARL institu­ tions participating in the LibQUAL+ study. Method Participants The 3,987 participants in this phase of the LibQUAL+ study represented the range of library user groups at the eleven ARL member institutions. Included were un­ dergraduate students (n U = 998), gradu­ ate students (n G = 1,281), faculty (n F = 1,022), and staff and ancillary professional personnel (e.g., research scientists not in tenure accruing positions, n O = 686). Half the participants were males and half females. The ages of the 3,987 par­ ticipants were: (a) < 22 years, 17.8 percent; (b) 22–30 years, 29.7 percent; (c) 31–45 years, 25.3 percent, and (d) > 45 years, 27.2 percent. The disciplines of the partici­ pants were: (a) science, 17.1 percent; (b) social science, 16.6 percent; (c) libraries, 12.0 percent; (d) humanities, 11.3 percent; (e) engineering, 10.9 percent; (f) health sciences, 9.4 percent; (g) business, 8.1 per­ cent; (h) education, 6.5 percent; (i) fine arts, 3.8 percent; (j) law, 1.7 percent; (k) architecture, 1.5 percent; and (l) veteri­ nary medicine, 1.0 percent. Thus, the sample was both large and diverse. Instrumentation An instrument used with some frequency to measure perceptions of service quality is the 22-item protocol called SERVQUAL. The SERVQUAL protocol ostensibly mea­ sures perceptions of service in terms of tangibles, reliability, responsiveness, assur­ ance, and empathy.13 Within this model, “only customers judge quality; all other judgments are essentially irrelevant.”14 Based on the authors’ qualitative re­ search, the twenty-two generic SERVQUAL items do not appear to capture all the rel­ evant aspects of users’ perceptions of library service quality.15 Therefore, the authors also administered an additional nineteen items that emerged out of qualitative interviews at nine institutions. These forty-one items were administered on the Web using browser software.16 The 3,987 participants rated their perceptions of library services using 1-to-9 Likert scales. Results All five principal components with eigen­ values greater than 1.0 were extracted and rotated to the promax criterion. Promax rotation results in correlated factors. In the authors’ study, the correlations of the five factors ranged from .183 to .641. As Rich­ ard L. Gorsuch has emphasized: Rotating obliquely in factor analy­ sis implies that the factors do over­ lap and that there are, therefore, broader areas of generalizability than just a primary factor. Implicit in all oblique rotations are higher- order factors. It is recommended that these [always] be extracted and examined so that the investigator may gain the fullest possible under­ standing of the data.17 The authors then factor-analyzed the first-order factor correlation matrix. One second-order factor had an eigenvalue greater than 1.0 (ë = 2.78). Finally, to com­ plete the analysis, the authors invoked a useful interpretation aid proposed by John Schmid and John M. Leiman and also explained by Gorsuch.18, 19 This solu­ tion “orthogonalizes” the two levels of analysis to each other by removing from the first-order factors any information that also is present at the second-order level. This solution also allows interpre­ tation of both levels of analysis in terms of the observed variables. This solution is presented in table 1. Figure 2 is a graphic representation of the results. The sizes of the objects in figure 2 reflect the amount of information resid­ ing in both levels of the analysis after the http:Gorsuch.18 http:software.16 http:quality.15 http:empathy.13 Users’ Hierarchical Perspectives on Library Service Quality 151 TABLE 1 Factor Pattern Coefficients from Schmid-Leiman Solution Residual First-order Factors Item Item Core Topic A I II III IV V 19 Responsive Willingness to help users .696 .499 -.050 -.065 .026 .083 24 Empathy Deal with users in caring fashion .724 .476 .056 -.013 .059 .070 18 Responsive Readiness to respond to user .708 .460 -.049 -.041 .065 .044 34 Assurance Employees who are courteous .685 .453 .033 .003 -.010 -.073 20 Assurance Employees have knowledge .708 .447 -.020 -.003 .005 .079 9 Assurance Employees instill confidence .673 .422 -.014 -.050 -.003 .240 11 Empathy Employees understand needs .734 .387 -.021 -.004 .032 .251 41 Empathy Giving users individual attention .704 .355 .033 .117 -.042 .028 15 CollAccess Instruction in use, when needed .646 .332 -.007 -.011 .033 .231 33 Reliable Handle users' service problems .734 .293 -.017 .150 .087 -.079 13 Empathy Users' best interests at heart .738 .274 .044 .000 .102 .256 28 Reliable Performing services right .756 .246 .011 .101 .188 -.096 26 Responsive Prompt service to users .778 .239 -.013 .139 .150 .021 38 Tangibles Employees have neat appearance .580 .231 .158 .110 -.049 -.072 39 LibAsPlace A meditative place .588 -.004 .601 .033 -.051 -.059 30 LibAsPlace A haven for quiet and solitude .604 -.023 .596 .017 .007 -.078 40 LibAsPlace Space that facilitates quiet .617 -.006 .583 .048 -.012 -.113 12 LibAsPlace A contemplative environment .645 .010 .533 -.042 .005 .223 4 LibAsPlace A place for reflection .571 -.031 .474 -.092 .075 .276 14 LibAsPlace Comfortable and inviting location .665 .037 .444 -.047 .033 .311 29 LibAsPlace Space group/individual study .601 -.033 .423 .112 -.004 .081 22 LibAsPlace Center intellectual interaction .581 .023 .382 .134 -.122 .234 21 LibAsPlace A secure and safe place .577 .097 .230 -.018 .182 -.110 23 Tangibles Visual appeal materials .689 .103 .185 .132 .039 .171 37 CollAccess Complete runs of journal titles .609 -.062 -.012 .490 -.035 .121 27 CollAccess Comprehensive print collection .646 -.052 .029 .429 -.005 .163 36 CollAccess Interdisciplinary needs addressed .635 .001 .006 .406 .014 .038 32 CollAccess Library materials in the stacks .620 .022 .044 .254 .119 .004 35 Tangibles Modemn equipment .638 .037 .072 .248 .042 .173 31 Assurance Assuring accuracy/confidentiality .620 .173 .053 .176 .060 -.086 2 Reliable Providing services as promised .692 .041 -.031 -.015 .468 .038 5 Reliable Service at promised time .707 .036 .012 -.025 .458 .050 3 Responsive Keep users informed .605 .061 -.013 -.061 .388 .131 1 CollAccess Convenient access collections .640 -.046 -.005 .080 .337 .250 17 CollAccess Timely document delivery .609 .048 -.058 .167 .235 .046 16 Reliable Maintain error-free records .639 .113 .036 .096 .223 -.060 10 CollAccess Resources added to collection .589 .019 -.106 .295 .037 .403 8 Empathy Convenient business hours .546 .022 .017 .127 .088 .378 6 Tangibles Visually appealing facilities .606 -.014 .370 -.058 .092 .371 25 CollAccess Full text delivered electronically .524 .074 -.005 .287 -.126 .370 7 CollAccess Access to archives .621 -.016 .047 .100 .195 .354 Trace/Information 17.34 2.08 2.38 1.14 1.04 1.51 Note. The irst column represents the second-order actor. The next 5 columns represent the irst- order solution, based on variance orthogonal to the second order (Richard L. Gorsuch, Factor Analysis, 2nd ed. [Hillsdale, N.J.: Erlbaum, 1983J, 248 54). 152 College & Research Libraries March 2001 Schmid–Leiman solution was invoked. What is noteworthy here is that both lev­ els of analysis contain unique information on user perceptions. Discussion The SERVQUAL measure has proved use­ ful in measuring library users’ percep­ tions of service quality.20 However, the five dimensions presumed by the SERVQUAL developers have typically not been recovered in these applications.21 And the authors’ qualitative research has indicated that additional items must be added to the measure to represent some important dimensions of users’ percep­ tions of library service quality.22 The results presented in table 1 and fig­ ure 2 do indicate that a single dimension does dominate user thinking about li­ brary service quality. As indicated by the coefficients for the second-order factor presented in table 1, basically all forty-one items that the authors used saturate this factor. The result is also consistent with related findings.23 However, it is noteworthy that consid­ erable information on users’ perceptions is not present in this single, overarching second-order perspective. This was true even though the Schmid–Leiman solution vests in the higher-order factor any infor­ mation existing at both levels. Thus, the result is compelling. The constructs still operating at the first-order level, even in the presence of the general second-order service construct, involve the Library as a Place, Empathy with User Needs, Ac­ cess, Collections, and Reliability. The study results suggest that users perceive library service at a global level but also simultaneously invoke a more nuanced view involving these specific elements. The results mean that ongoing efforts to evaluate library service quality would do well to invoke both levels of characteriz­ ing service. As the field moves beyond collection counts in measuring service quality, it will be critical that the final mea­ sures be ecologically grounded in ways that honor the users’ frames of mind when they think about library services. FIGURE 2 Two Levels of User Perceptions Note. The sizes of the objects represent the amount of unique information present in each construct at both levels of the model. LibQUAL+ Empathy Place Collections Reliability Access http:findings.23 http:quality.22 http:applications.21 http:quality.20 Users’ Hierarchical Perspectives on Library Service Quality 153 Notes 1. Nassim H. Naqui and M. Donald Blaufox, Blood Pressure Measurement: An Illustrated His­ tory (New York: Parthenon, 1998). 2. Nielsen Media Research, “What TV Ratings Really Mean—And Other Frequently Asked Questions,” 2000. Available online at . (31 July 2000). 3. Tricia L. Silva, “Most HMO Members Rate Plans High, Survey Shows,” San Antonio Business Journal, 3 Jan. 2000. Available online at http://www.bizjournals.com/sanantonio/stories/2000/01/03/story1.html. (31 July 2000). 4. A. Parasuraman, Valarie A. Zeithaml, and Leonard L. Berry, “A Conceptual Model of Ser­ vice Quality and Its Implications for Future Research.” Journal of Marketing 49 (fall 1985): 41–50. 5. Colleen Cook and Bruce Thompson, “Higher-order Factor Analytic Perspectives on Us­ ers’ Perceptions of Library Service Quality,” Library Information Service Research (in press); Danuta A. Nitecki, “An Assessment of the Applicability of SERVQUAL Dimensions: A Customer-based Criteria for Evaluating Quality of Services in an Academic Library” (Ph.D. diss., University of Maryland, 1995), abstract in Dissertation Abstracts International 56: 2918A. 6. Syed S. Andaleeb and Patience L. Simmonds, “Explaining User Satisfaction with Aca­ demic Libraries,” College & Research Libraries 59 (Mar. 1998): 156–67; Colleen Cook and Fred Heath, “Users’ Perceptions of Library Service Quality: A ‘LibQUAL+’ Qualitative Interview Study” (pa­ per presented at the Association of Research Libraries Measuring Service Quality Symposium on the New Culture of Assessment: Measuring Service Quality, Washington, D.C., Oct. 20, 2000). 7. Association of Research Libraries. ARL Membership Criteria Index, 1998–99 (2000). Memo­ randum to Directors of ARL Libraries from Martha Kyrillidou, senior program officer for statis­ tics and measurement, and Julia Blixrud, director of information services, ARL, Mar. 8, 2000. 8. Peter Hernon and Charles R. McClure, Evaluation and Library Decision-Making (New Jer­ sey: Ablex, 1990), xv. 9. Nitecki, “Assessment of Service Quality in Academic Libraries: Focus on the Applicabil­ ity of the SERVQUAL,” in Proceedings of the Second Northumbria International Conference on Perfor­ mance Measurement in Libraries and Information Services (Newcastle upon Tyne, England: Univ. of Northumbria at Newcastle, 1998), 181. 10. Cook and Heath, “The Association of Research Libraries LibQUAL+ Project: An Update,” ARL Newsletter: A Bimonthly Report on Research Library Issues and Actions from ARL, CNI and SPARC (July 2000); Colleen Cook, Fred Heath, Bruce Thompson, and Russell L. Thompson, “The Search for New Measures: The ARL ‘LibQUAL+’—A Preliminary Report,” Portal: Libraries and the Acad­ emy (under review). 11. Clyde Hendrick and Susan Hendrick, “A Theory and Method of Love,” Journal of Person­ ality and Social Psychology 50 (Feb. 1986): 393. 12. Bruce Thompson, “SECONDOR: A Program That Computes a Second-order Principal Components Analysis and Various Interpretation Aids,” Educational and Psychological Measure­ ment 50 (1990): 579. 13. Parasuraman, Berry, and Zeithaml, “Refinement and Reassessment of the SERVQUAL Scale,” Journal of Retailing 67 (winter 1991): 420–50; ———, “A Conceptual Model”; —, “Alterna­ tive Scales for Measuring Service Quality: A Comparative Assessment Based on Psychometric and Diagnostic Criteria,” Journal of Retailing 49 (fall 1994): 201–30. 14. ———, Delivering Quality Service: Balancing Customer Perceptions and Expectations (New York: Free Pr., 1990), 16. 15. Cook and Heath, “Users’ Perceptions.” 16. Cook, Heath, Thompson, and Thompson, “Score Reliabilities in Web- or Internet-based Surveys: Unnumbered Graphic Rating Scales Versus Likert Scales” (paper presented at the Asso­ ciation of Research Libraries Measuring Service Quality Symposium on the New Culture of As­ sessment: Measuring Service Quality, Washington, D.C., Oct. 20, 2000). 17. Richard L. Gorsuch, Factor Analysis, 2nd ed. (Hillsdale, N.J.: Erlbaum, 1983), 255. 18. John Schmid and John M. Leiman, “The Development of Hierarchical Factor Solutions,” Psychometrika 22 (Mar. 1960): 53–61. 19. Gorsuch, Factor Analysis, 248–54. 20. Colleen Cook and Bruce Thompson, “Reliability and Validity of SERVQUAL Scores Used to Evaluate Perceptions of Library Service Quality,” Journal of Academic Librarianship (July 2000): 248–52. 21. ———, “Higher-order Factor; Nitecki, “An Assessment of the Applicability of SERVQUAL Dimensions.” 22. Cook and Heath, “Users’ Perceptions.” 23. Bruce Thompson, Colleen Cook, and Fred Heath, “How Many Dimensions Does It Take to Measure Users’ Perceptions of Libraries?: A ‘LibQUAL+’ Study,” Portal: Libraries and the Acad­ emy (under review). http://www.bizjournals.com/sanantonio/stories/2000/01/03/story1.html