Shill.p65 Does the Building Still Matter? 123 Does the Building Still Matter? Usage Patterns in New, Expanded, and Renovated Libraries, 1995–2002 Harold B. Shill and Shawn Tonner Since the mid-1990s, more than 390 academic institutions have con­ structed new libraries or have expanded, renovated, or reconfigured an existing library. Given current concern about the future of libraries and the nature of space needs, it is important to know what impact these improvements have had on use of the physical library. Using responses from a 68-item Web survey conducted in early 2003, this article exam­ ines the impact of building improvements on usage of the physical facil­ ity. Recommendations for facility planning are offered, and implications for the “library as place” debate are explored. This report is a companion piece to an article published in the November 2003 issue. ore than 390 library building projects were completed in American and Canadian insti­ tutions of higher education be­ tween January 1995 and December 2002. The number of projects completed annu­ ally varied between eighteen and twenty- six in most years, with a “low” of eleven completions in 2001 and a “peak” of thirty-nine finished projects in 2002. Many new libraries opened in 2003, in­ cluding major facilities at Marquette Uni­ versity, San Jose State University, Arcadia University, Columbia College, and the Uni­ versity of Georgia. More than fifty other projects are currently under construction, including new or expanded facilities at the University of Houston, Harvard University, Middlebury College, Santa Monica College, Indiana University-Southeast, Appalachian State University, and Cuesta College. Design, fund-raising, and visioning activities are in progress at other institu­ tions. Princeton University has engaged prominent architect Frank Gehry to de­ sign a $60 million science library. Another renowned architect, Lord Norman Foster, is designing a $46 million library for Cali­ fornia State University-Channel Islands. The new University of California-Merced campus will have a 178,800-square-foot library when it opens in fall 2005. Cali­ fornia State University-Monterey Bay is completing schematic design for a 200,000-square-foot library scheduled to open in June 2007.1 Despite this impressive level of build­ ing activity, however, librarians face a va­ riety of challenges in making the case for physical improvements. These challenges, including growing student usage of Internet resources, off-site availability of Harold B. Shill is Director of Capital College Libraries at Penn State Harrisburg and Penn State Schuylhill; e-mail: hbs2@psu.edu. Shawn Tonner is Director of the Hill Freeman Library & Spruill Learning Center at Reinhardt College; e-mail: sct@reinhardt.edu. 123 mailto:sct@reinhardt.edu mailto:hbs2@psu.edu 124 College & Research Libraries March 2004 electronic resources, institutional needs for technology upgrades in classroom build­ ings, and declining usage statistics, are examined in detail in the companion ar­ ticle, “Creating a Better Place: Physical Improvements in Academic Libraries, 1995–2002.”2 Scott Carlson’s November 2001 Chronicle of Higher Education article, “The Deserted Library,”3 suggested to aca­ demic administrators that libraries are be­ coming marginalized in the lives of some college students, while also noting cases of significant usage increases. Many variables, including broader trends and local factors, contribute to us­ age levels within academic library build­ ings. Surprisingly, given the large cost of new and improved facilities, the relation­ ship between facility quality and library usage has not been subjected to system­ atic, empirical analysis. The lack of evi­ dence about this relationship is a serious gap in the profession’s knowledge base. As a result, librarians are forced to rely on anecdotal evidence alone in making the case that major facility improvements will have a positive impact on library usage. As reported in the companion article, the authors sent a 68-question survey to academic institutions that had recently built new libraries or significantly en­ hanced existing facilities in January 2003. The previous article discussed the nature of these physical improvements and iden­ tified trends in library design. The cur­ rent study uses data from the same ques­ tionnaire to describe the impact of these improvements on usage of the physical facility. Its conclusions should enable li­ brarians and facility planners to (1) project the likely impact of physical improve­ ments on library usage and (2) determine the types of facility features most likely to contribute to major increases in build­ ing use. An extensive literature review was provided in the earlier article and requires only one addition. Scott Bennett’s Librar­ ies Designed for Learning, published in late 2003, reported findings about space conceptualization and planning processes from a Web survey of library construction and renovation projects undertaken be­ tween 1992 and 2001.4 Design and Methodology The earlier article describes the method­ ology of the study. Respondents were asked questions about: • institutional characteristics; • project-specific features (comple­ tion year, type of project, etc.); • nature and extent of improvements made (seating, wiring, studies, etc.); • presence of nonlibrary facilities; • collection provisions; • before-and-after quality changes (seating, layout, lighting, HVAC, etc.); • usage before and after project completion. The reader is referred to the compan­ ion article for details about the first six types of variables. Because the current article focuses on usage, the development and application of the “usage change” variable are described below. In the survey, data on four usage vari­ ables—exit gate count, total circulation, in-house collection use, and reference transactions—were requested to deter­ mine (1) the extent to which facility im­ provements affected postproject usage and (2) the relationships between specific facility/institutional variables and changes in library use following project completion. The variables listed earlier were used as independent variables to determine their possible effects on postproject usage levels. The current study focuses on exit count as the key measure of facility use. Usage changes were calculated by sub­ tracting preproject totals for each usage indicator from the most recent postproject figures available. The remainder then was divided by the total from the preproject year, and a percentage of change was cal­ culated. For example, if Hypothetical University opened a new Library X dur­ ing the 2000–2001 year, the following exit count calculation might occur: 150,000 exits recorded in 2001–2002 - 96,000 exits recorded in 1999–2000 54,000 difference Does the Building Still Matter? 125 TABLE 1 Characteristics of Library Projects Included in Study (n = 182) Total Projects with Percent of Percent of Projects Exit Counts Projects in Projects with Variable (n = 182) (n = 90) Study Exit Counts Institution Type Public 118 61 64.8% 67.8% Private 64 29 35.2 32.2 Carnegie Classification Associate 21 15 11.4% 16.7% Baccalaureate 27 10 14.8 11.1 Master's 51 29 27.8 32.2 Doctoral 42 22 23.3 24.4 Branch/undergraduate 20 5 10.8 5.6 Medical 12 6 6.8 6.7 Law 9 3 5.1 3.3 Project Type New 53 30 29.1% 33.3% New, multipurpose 31 12 17.0 13.3 Addition 5 5 2.9 5.6 Addition/renovation 73 31 40.0 34.4 Other 20 12 11.0 13.3* Completion Year 1995 23 9 12.6% 10.0% 1996 19 11 10.4 12.2 1997 18 7 9.9 7.8 1998 23 16 12.6 17.8 1999 21 11 11.5 12.2 2000 27 18 14.8 20.0 2001 12 3 6.6 3.3 2002 39 15 21.4* 16.7 *Total does not equal 100% due to rounding error. 54,000 difference / 96,000 preproject total = 56.3% increase In the case of new libraries, the most recent full preproject year before the new library was occupied was used as the baseline period for this calculation. Ac­ tual “move years” were avoided to elimi­ nate data contamination from the mixing of old and new library figures. For addition and renovation projects, usage is normally depressed during the construction period as users seek a qui­ eter working environment. For these li­ braries, the full year two years prior to project completion was used as the baseline year. Postproject usage changes would be exaggerated if a midconstruction baseline year were used, so earlier years were se­ lected to elicit a more normal measure of preproject usage. Study Findings Even though 171 respondents com­ pleted most survey questions, the num­ ber of usable responses for the four fa­ cility usage questions ranged from 106 libraries (circulation) to 45 libraries (in­ house collection use). Usable exit count data were supplied by ninety libraries (25.3% of the overall survey population and 49.5% of the participating libraries). Several reasons for nonresponses on the usage questions were identified, includ­ ing: 126 College & Research Libraries March 2004 ing exit count data, theTABLE 2 major dependent vari-Postproject Changes in Facility Use able used in this report, by Key Use Measures are described in table 1. Table 1 provides a Use Measure % Increase Highest Lowest Median general breakdown of Exit Count 80.0% +1012.0% -40.8% +37.4% responding libraries by Circulation 44.8 +3350.6 -69.3 -4.1 institution type, modi-Reference 40.1 +162.4 -73.3 -21.4 fied Carnegie classifica­In-house Use 26.7 +565.5 -63.7 -23.0 • lack of complete historical data on these variables, particularly from the preproject period; • data collection gaps (malfunction­ ing exit gate counter, etc.) that made com­ parison of recent and older data unreli­ able; • changes in data-recording prac­ tices; • recent completion dates (fall 2002, etc.) that precluded the comparison of pre- and postproject data during the spring 2003 data analysis period; • respondent data collection burden. Key characteristics of the overall respon­ dent population and those libraries provid­ tion, project type, and completion year. It con­ firms that the percentage of libraries pro­ viding exit count data approximates the overall distribution of library projects within these categories. The resulting data were exported ini­ tially from the SSRI database into an Ex­ cel spreadsheet for preliminary analysis and assessment of frequency distribu­ tions, then later exported into SPSS (Sta­ tistical Package for the Social Sciences) for significance testing. The statistical mea­ sures used to determine the existence and strength of relationships are described in the findings section. Pre- and postproject facility usage (exit count) findings are reported below in the FIGURE 1 Facility Use Change in Libraries Completing Building Projects 1995-2002 (n = 90) 25.6% 21.1% 14.4% 15.6% 4.4% 18.9% (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Percent of Exit Count Change Does the Building Still Matter? 127 TABLE 3 Libraries Reporting 200% Postproject Increases in Facility Use Library Project Type Completion Date Exit Count Change (%) Univ. of San Francisco-Law Arkansas Tech University Williams College-Science Trevecca Nazarene College Green River Comm. College Mount Union College Front Range Comm. College St. Martin's College New New multipurpose New multipurpose New New multipurpose Addition & renovation New, oint use facility New 2000 1999 2000 2000 1996 2000 1998 2000 1012.0 525.4 386.5 381.7 268.6 251.7 232.5 203.9 *Note: Respondents have agreed to the publication of these data following six sections: general project char­ acteristics; technology; user space; collec­ tion provisions; nonlibrary facilities; and facility quality. Relationships between spe­ cific features and postoccupancy usage are explored. Correlations between specific characteristics and changes in postproject usage patterns are then discussed. A final section summarizes the major findings about the impact of facility improvements on library usage, indicates their signifi­ cance for academic library planning, and suggests directions for future research. Summary data for overall facility use (exit count) and the three other usage vari­ ables are reported in table 2. Although this article focuses primarily on exit count, overall usage change findings for circu­ lation, in-house use, and reference activ­ ity are included here to allow preliminary comparisons of project impact on several measures of facility use. The data in table 2 confirm that 80 per­ cent (n = 72) of the ninety libraries sup­ plying usable before-and-after exit count data increased their facility use following project completion, whereas nearly one- half of those libraries reported circulation growth. Reference and in-house collection use declines were proportionately less than those reported for the whole popu­ lation of ACRL, ARL, and AAHSL librar­ ies during the study period. The findings in table 2 indicate that, in general, building improvements had a greater overall impact on basic facility use (gate count) than on circu­ lation, reference transaction volume, and in­ house collection use. As table 2 confirms, actual changes in exit count ranged from a 40.8 percent drop in usage to an increase slightly greater than 1000 percent. The median increase across all facilities providing usable pre- and postproject exit count data was 37.4 percent. The overall exit count data indi­ cate that, in general, students are using new and improved libraries at levels greater than their use of preproject library facilities. Because this article focuses primarily on use of the physical facility, rather than on collection use or services (circulation, reference, etc.), primary attention will be focused on the before-and-after facility usage (exit or gate count) data in this analysis. Figure 1 provides a more de­ tailed breakdown of postproject exit count changes. Although 80 percent of the respond­ ing libraries experienced some degree of postoccupancy usage growth, figure 1 shows that usage change varied widely across six level-of-change categories. In some cases, that usage increase has been profound. Whereas 25.6 percent (n = 23) of the postproject libraries experienced usage increases of 100 percent of more, 8.9 percent (n = 8) of the responding li­ braries (not shown in figure 1) reported usage increases actually exceeding 200 percent. 128 College & Research Libraries March 2004 The most dramatic exit count increase was reported by the University of San Francisco’s Dorraine Zief Law Library, a new facility opened in 2000. That facility’s 1012.0 percent increase has been con­ firmed with library officials.5 The data re­ flect a change from 30,809 annual users in the former library to more than 342,593 users in the new building in 2001–2002. Given an enrollment of 670 FTE students in the law school, this usage level is par­ ticularly impressive. Other usage in­ creases exceeding 200 percent are re­ ported in table 3. Table 3 lists the libraries reporting the greatest levels of postproject exit count change. However, although the preced­ ing tables and figures confirm the gen­ eral tendency toward postproject usage increases and provide some evidence of major usage growth, they do not identify either the types of projects or the specific enhancements most likely to stimulate building use. In the following sections, the investiga­ tors will examine the relationship between specific facility features and postproject usage more closely. Chi-square tests are used to indicate where a relationship ex­ ists between a specific project attribute and postproject usage. Other statistical tech­ niques are employed later in this analysis to determine the strength of any existing relationships between specific attributes and postoccupancy usage levels. Because this study explores new ground and has practical implications, some usage-related findings for specific variables will be presented even when they are not statistically significant. The characteristics of libraries reporting the greatest increases (100%+) in postproject usage also will be examined. Findings: General Characteristics Respondents were asked questions about the type of project done at their libraries, completion dates, public/private affilia­ tion, facility size, the library’s campus location, and shared usage with other campus operations. The investigators also determined the Carnegie classification for each library’s host institution. Postproject usage change findings for these questions are presented below. Project Types Responses from the 182 libraries partici­ pating in the survey were placed into five “type of project” categories: new, stand­ alone libraries (29.8%); new libraries in multipurpose facilities (13.3%); addition and renovation projects (42.0%); additions only (2.8%); and “other” project types (12.1%). The “other” category consists mainly of renovation-only projects. Changes in postproject facility usage from the ninety libraries providing usable exit count data are shown in table 4. It was anticipated that new libraries and libraries in new, multipurpose facili­ ties would show greater increases in us­ age than addition, renovation, and “other” project types. In fact, thirty-five of the forty-two new and new, multipur­ pose facilities (83.3%) did report ex- TABLE 4 Project Type and Changes in Postproject Facility Use (n = 90) Type of Facility Exit Count Change Project (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total New library 2 3 5 6 5 9 30 New, multipurpose 0 2 2 3 1 4 12 Addition only 0 0 3 2 0 0 5 Addition/renovation 0 8 5 5 5 8 31 Other 2 1 4 1 2 2 12 Total 4 14 19 17 13 23 90 Does the Building Still Matter? 129 TABLE 5 Completion Date and Changes in Postproject Facility Use (n = 90) Completion Year Exit Count Change (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total 1995 1996 1997 1998 1999 2000 2001 2002 2 0 0 1 1 0 0 0 3 0 2 3 2 1 0 3 0 2 1 4 2 7 0 3 4 2 2 1 2 3 1 2 0 3 2 4 1 0 0 3 0 4 0 3 3 7 2 4 9 11 7 16 11 18 3 15 Total 4 14 19 17 13 23 90 panded usage, with 31.0 percent (n = 13) of them exceeding the 100 percent in­ crease level. Addition/ renovation projects did not lag far behind, with 74.2 percent (n = 23) experiencing growth and 25.8 percent (n = 8) attaining the 100 per­ cent increase level. Although projects in existing facilities were slightly more likely to record postproject usage declines, there was no statistically significant difference in exit count changes between these li­ braries and new facilities. In short, al­ though the construction of a new building provides a “fresh start” unencumbered by obsolete features, significant usage increases also can result from space improvements in existing facilities. Completion Pattern Positive changes in postproject usage were expected to be greatest in library projects completed during the latter years of the study period. The number of reporting projects remained relatively stable from 1995 through 2000 but declined signifi­ cantly in 2001. However, the number of project completions then rebounded to its highest level during the entire study pe­ riod in 2002, with 21.4 percent of the 182 responding libraries completed in that year. Findings for the ninety libraries pro­ viding meaningful before-and-after exit count data are summarized in table 5. As table 5 indicates, thirty-two of the thirty-six projects (88.9%) completed dur- FIGURE 2 Institution Type and Facility Use Change (n = 90) 130 College & Research Libraries March 2004 ing the 2000–2002 period, and providing usable before-and-after exit count data, showed usage increases. Most projects completed before 2000 still showed sig­ nificant usage gains, with forty of the fifty-four pre-2000 libraries (74.0%) re­ porting expanded use during the postproject measurement year. Interest­ ingly, thirteen of the thirty-six projects (36.1%) completed in 2000 or later re­ ported 100%+ increases in postproject usage. In contrast, five of the nine facili­ ties completed in 1995 reported that their 2001–2002 usage level was less than the level reported for the last precompletion reporting period. As expected, there was a slight ten­ dency for projects completed more re­ cently to show greater usage increases than projects completed earlier in the study period. However, the recency–us­ age relationship was not statistically sig­ nificant (p = .160 in a chi-square test). On balance, usage levels for more recently completed libraries are higher, but a clear majority of the libraries completed in the first half of the study period still showed 2001–2002 usage levels greater than those found for their preproject baseline year. In short, although newer libraries displayed slightly greater usage growth than those com­ pleted earlier in the study period, nearly 75 percent of the improved libraries are still ex­ periencing usage levels exceeding those found before project initiation. Public and Private Projects The Carnegie Foundation for the Ad­ vancement of Teaching lists classifications for 3,941 higher education institutions on its Web site (www.carnegie.org). Public colleges and universities comprised 41.7 percent of the institutions listed, and pri­ vate nonprofit and private for-profit in­ stitutions accounted for 42.7 and 15.6 per­ cent, respectively, of the institutional population.6 Although public and private, non-profit institutions are represented almost equally in higher education, 64 percent of the responding libraries were in public institutions. In short, actual li­ brary building activity was proportion­ ately greater in public institutions than in their private colleges and universities. No initial assumptions were made about possible differences in postproject usage between public and private insti­ tutions. However, as figure 2 indicates, usage increases were found more consis­ tently in private institutions than in pub­ lic colleges and universities, with twenty- seven of the twenty-nine improved librar­ ies in private institutions (93.1%) record­ ing positive changes in postproject usage. However, 73.8 percent (n = 45) of the pub­ lic institution libraries also experienced usage growth in the most recent measure­ ment year. This finding is significant at the p = .010 level in a chi-square test. In addition to showing a great prob­ ability of postproject usage increases, pri­ vate institutions accounted for a some­ what disproportionate number of the “100 percent increasers.” As the right- hand bar in figure 2 confirms, thirteen of the twenty-nine private institution librar­ ies (44.8%) reported positive usage changes at this level. In contrast, just ten of the sixty-one public institutions (16.4%) saw exit count increases of 100 percent, while 19.7 percent reported usage reduc­ tions of 0–24 percent and 6.6 percent ex­ perienced usage declines of 25–49 percent. Although most public institution projects reported postproject usage in­ creases, this significant difference be­ tween public and private institution li­ braries was not expected. One might speculate that liberal arts college librar­ ies may require greater use of print library resources, but this finding requires fur­ ther explanation. In summary, although both public and private institution libraries received heavier usage after project comple­ tion, nearly all facilities in private institutions received greater postproject use and almost half of them experienced a profound growth of 100 percent or more. Carnegie Class The investigators also wished to deter­ mine whether there were any significant differences in postproject facility usage among institutions in different Carnegie http:www.carnegie.org Does the Building Still Matter? 131 TABLE 6 Postproject Changes in Facility Use by Modified Carnegie Classification (n = 90) Modified Carnegie Exit Count Change Classification (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total Associate Baccalaureate Master's I or II Doctoral Branch/undergrad. Medical Law 1 1 1 0 0 1 0 1 0 4 2 3 2 2 4 2 6 7 0 0 0 3 3 6 5 0 0 0 2 0 7 3 0 1 0 4 4 5 5 2 2 1 15 10 29 22 5 6 3 Total 4 14 19 17 13 23 90 classes, regardless of public or private affiliation. In order to facilitate statistical analysis, they collapsed similar Carnegie classes (Baccalaureate-Liberal Arts with Baccalaureate-General, Master ’s I with Master’s II, etc.). In addition, they devised a separate category of “branch or under­ graduate library” to allow for possible differences in usage patterns within sub­ ject- or clientele-focused facilities on multilibrary campuses. Doctoral and master’s institutions com­ pleted proportionately more projects dur­ ing the study period, in relation to their numbers in higher education, than other types of institutions did. Although com­ prising just 6.6 and 15.5 percent, respec­ tively, of all colleges and universities listed by the Carnegie Foundation for the Ad­ vancement of Teaching in 2000, doctoral and master ’s institutions accounted for 23.5 and 27.4 percent, respectively, of the 182 libraries responding to the survey. Table 6 shows the distribution, by modified Carnegie class, of participating libraries that were able to provide com­ parable before-and-after exit count data. Most libraries in the four larger Carnegie classes reported usage increases, with the “increasers” ranging from twenty-four of twenty-nine libraries (82.8%) for master ’s institutions to twenty of twenty-two libraries (90.9%) for doc­ toral institutions. Declining usage was concentrated among branch/under­ graduate, medical, and law libraries, but their numbers are too small to permit any inferences. In fact, the professional school libraries displayed a slight tendency to cluster toward the extremes of large in­ creases or actual declines. As a result of these findings, the investigators have con- TABLE 7 Building Size and Changes in Postproject Facility Use (n = 90) Square Footage Exit Count Change in Facility (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total < 20,000 1 3 2 1 0 1 8 20,000-49,999 0 1 2 4 2 6 15 50,000-74,999 2 2 0 4 2 6 16 75,000-99,999 0 1 3 1 3 4 12 100,000-149,999 1 1 3 4 1 0 10 150,000-199,999 0 1 1 1 2 1 6 > 200,000 0 5 8 2 3 5 23 Total 4 14 19 17 13 23 90 132 College & Research Libraries March 2004 TABLE 8 Campus Location and Postproject Facility Use (n = 90) Library Location Exit Count Change on Campus (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Central 2 9 13 10 8 12 Neutral 1 0 3 6 2 5 Peripheral 1 5 3 1 3 6 Total 4 14 19 17 13 23 Total 54 17 19 90 cluded that there is no apparent relationship between the level of degrees offered and postproject usage patterns. Library Size There was a significant overall increase in the size of physical library facilities as a result of building projects. Prior to com­ pleting their facility projects, 53.8 percent (n = 98) of the 182 responding libraries re­ ported having less than 50,000 square feet of space, whereas only 28.0 percent (n = 51) fit into this size category afterward. In contrast, the number of libraries with space exceeding 200,000 square feet increased from 10.9 percent before the project to 25.3 percent following completion. The impact of facility size on postproject building us­ age is described in table 7. The investigators made no assump­ tions about the possible impact of facility size on usage of postproject libraries. Be­ cause twelve of the thirty-one libraries in the 20,000–74,999 square foot range (38.7%) experienced 100% increases, there may be a slight tendency toward greater usage of smaller facilities, but it is not sta­ tistically significant. Moreover, this find­ ing could reflect a larger proportional in­ crease in library space resulting from the replacement or expansion of very small facilities. However, facility size alone does not appear to be a significant factor affecting postproject usage patterns. Campus Location Most (62.1%) of the 182 responding librar­ ians indicated that their libraries occupied a central location on campus, whereas 18.7 percent reported a neutral location and 19.2 percent a peripheral location. Facility siting can be a major concern on campuses where existing buildings occupy all of the prime locations. However, as table 8 indi­ cates, there was no statistically significant relationship between library siting and postproject usage patterns among the ninety libraries providing before-and-af­ ter exit count data. Curiously, libraries with a peripheral location had the highest percentages of both 100 percent increases in use and postproject declines, with six of the nineteen periph­ eral projects (31.6% each) falling into the extreme-usage categories. Because re­ sponses to other proximity questions con­ firmed that most libraries are located within a quarter mile of parking lots, classroom buildings, and student centers, the investi­ gators concluded that physical location did not have a significant overall impact on library building usage. Proximity to dormitories and other student residences appears to have had a slight positive influence, but it was not statistically significant. Library and Nonlibrary Uses Historically, library buildings have been built primarily to accommodate library collections, services, and operations, rather than to provide shared space for the library and other, nonlibrary units. The major ex­ ception to that tendency has been the lo­ cation of subject-specialized branch librar­ ies (music, science, engineering, etc.) in the same building with faculty offices and classrooms supporting those disciplines. In recent years, however, many aca­ demic institutions have combined new or expanded library buildings with space for other campus operations. Because the presence of nonlibrary facilities could af­ Does the Building Still Matter? 133 TABLE 9 Percent of Building Allocated for Library Functions and Changes in Postproject Facility Use (n = 90) Percent for Library Use Exit Count Change (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total > 25% 25--9% 50-44% 75-70% 90-99% 1009 1 0 0 0 3 0 2 1 2 2 5 2 0 1 1 2 5 10 0 0 2 2 5 8 0 0 1 2 5 5 1 0 2 7 8 5 4 2 8 15 31 30 Total 4 14 19 17 13 23 90 fect postproject usage levels, the investi­ gators were interested in seeing whether postproject usage was affected by (1) the relative percentage of the facility allocated for library functions and (2) the presence of specific types of nonlibrary facilities. The following nonlibrary facilities were most frequently included in the 173 libraries answering this question: • conference rooms (82.7%; n = 143); • general computer labs (69.9%; n = 121); • seminar rooms (53.2%; n = 92); • multimedia production centers (45.1%; n = 78); • snack bars or cafes (32.4%; n = 56); • general use classrooms (31.8%; n = 55); • educational technology centers (26.0%; n = 45); • art galleries (24.9%; n = 43); • auditoriums (20.2%; n = 35); • writing labs (16.8%; n = 29). It was hypothesized that usage within new, multipurpose buildings would in­ crease more significantly than usage in other types of library projects because they would attract many users with nonlibrary needs in addition to those coming to use library space and resources. As the data in table 9 confirm, there is no significant relationship between the proportion of building space allocated for library functions and postproject usage levels. In fact, there appears to be a ten­ dency toward diminished usage among facilities allocating less than 25 percent of their space for library purposes, but the number of cases is too small to permit generalizations from this limited finding. In contrast, usage increases were found in 93.3 percent of the facilities devoted entirely to library use. However, the great­ est concentration of 100 percent usage increases occurred in buildings assigning TABLE 10 Premises Wiring Type and Postproject Changes in Facility Use (n = 90) Project Writing Type Exit Count Change (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total CAT-2 CAT-5 CAT-5E CAT 5 Premium CAT-6 Fiber-Optic 1 1 0 0 0 2 2 10 1 1 0 0 2 9 1 1 1 5 2 5 4 1 1 4 0 7 2 0 1 3 2 7 5 2 3 4 9 39 13 5 6 18 Total 4 14 19 17 13 23 90 134 College & Research Libraries March 2004 75 to 90 percent of their space for the li­ brary. It appears that there is no linear rela­ tionship between the proportion of a facility devoted to library functions and postproject usage, but usage growth occurs most fre­ quently in all-library buildings and most dra­ matically in buildings where other units oc­ cupy 10 to 25 percent of the available space. Findings: Technology As student ownership of portable com­ puting devices has grown, librarians have recognized the need to provide facilitywide network access for mobile users. According to a recent survey, 96 percent of all full-time Penn State under­ graduates owned a computer in fall 2002, laptop ownership was increasing, and 15 percent owned a personal digital assis­ tant (PDA) device.7 Although only 6.5 per­ cent of the 182 responding libraries re­ ported an institutional requirement for student computer ownership, the diffu­ sion of computing devices is likely to ac­ celerate as their use becomes more com­ monplace and prices trend downward. This section examines the impact of various technology solutions—premises wiring systems, data ports, wireless sys­ tems, public access workstations, and in­ struction labs—on postproject usage. Premises Wiring Systems More than 83 percent (143 of 171) of the libraries responding to this question now have, at minimum, Unshielded Twisted Pair (UTP) Category 5 (CAT-5) copper wiring throughout the building. In addi­ tion, 21.0 percent (n = 36) of the partici­ pating libraries reported that they now deliver “fiber to the desktop.” These types of wiring infrastructure provide band­ width sufficient to support high-speed downloads of textual, video, audio, and data files. It was assumed that libraries provid­ ing higher-end wiring solutions would receive greater postoccupancy usage in­ creases than those with less-robust net­ work access. Possible associations be­ tween the type of wiring used through­ out the facility (“premises wiring”) and postproject usage are depicted in table 10. Somewhat unexpectedly, the investiga­ tors found no relationship between the use of more robust types of wiring (fiber-op­ tic, UTP Category 6, etc.) and postproject usage levels. Although twenty-two of the twenty-four libraries (91.7%) deploying fiber-optic or CAT-6 cabling reported us­ age increases, their levels of usage change were not significantly different from those found for wiring categories with a lesser bandwidth. Although 33.3 percent (n = 3) of the CAT-2 libraries and 28.2 percent (n = 11) of the CAT-5 facilities reported postproject declines, the majority of librar­ ies with those wiring types still reported usage increases, with several exceeding 100 percent. Although technology capabili­ ties were expected to contribute to in­ creased usage, it appears that wiring type TABLE 11 Number of Data Ports and Changes in Postproject Facility Use (n = 90) Number Data Ports > 50 50-99 100-249 250-499 500-999 1000 + Exit Count Change (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% 2 2 2 0 0 1 2 2 3 2 1 0 4 4 7 3 2 0 2 1 4 1 10 1 2 4 2 3 4 0 2 6 0 4 6 Total 7 11 20 18 16 18 Total 4 14 19 17 13 23 90 Chi-square = 39.390 df = 25 p = .034 Does the Building Still Matter? 135 TABLE 12 Extent of Wired Seating and Postproject Changes in Facility Use (n = 90) Percent Wired Exit Count Change Seats (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total 0% 2 0 2 3 2 2 11 1-10% 0 4 4 1 1 0 10 11-24% 0 2 4 2 1 4 13 25--9% 1 1 3 1 3 1 10 50-44% 1 0 1 1 2 4 9 75-99% 0 6 2 8 3 4 23 1009 0 1 3 1 1 8 14 Total 4 14 19 17 13 23 90 Chi-square = 43.317 df = 30 p = .055 alone does not explain much of the change in postproject facility use. Data Ports and Wired Seating As higher-capacity wiring alone was not associated with increased usage, the inves­ tigators were interested to know whether the relative availability of network access points would influence postproject usage. Both wired and wireless access solutions were viewed as possible technology fac­ tors affecting facility usage. Table 11 depicts the relationship be­ tween the number of wired data ports and postproject usage patterns. Prior to project completion, 107 of the 172 responding li­ braries (62.2%) contained fewer than fifty network connections of any kind, includ­ ing dedicated ports for office and OPAC computers. Following project completion, 51.7 percent of all responding libraries reported having 250 or more data ports. As table 11 confirms, there is a statisti­ cally significant relationship (chi-square = 39.390, p = .034) between the number of data ports in a library facility and postproject us­ age. Of the libraries offering 250 to 499 data ports, 55.6 percent (n = 10) experi­ enced usage increases exceeding 100 per­ cent; 29.4 percent (n = 10) of the facilities with 500 or more data ports also sur­ passed the 100 percent benchmark. In con­ trast, only 7.9 percent (n = 3) of the librar­ ies offering 249 or fewer data ports at­ tained the 100 percent increase level. Interestingly, the relative absence of data ports also appears to discourage li­ brary usage. Seven of the eighteen librar­ ies (38.9%) reporting fewer than 100 data ports experienced postproject declines in usage, whereas only 20 percent of the re­ sponding libraries reported drops in use. Clearly, the presence of abundant network access points is significantly related to postoccupancy usage patterns. The number of data ports in a build­ ing also includes dedicated ports for pub­ lic access workstations, service points, offices, instruction labs, and computer labs. In order to consider the impact of mobile, end-user ports separately from fixed network access points, the investi­ gators asked that respondents indicate the percentage of wired public seats in the facility. The findings in this section in­ clude both open area and group study seating. Table 12 documents the relation­ ship between wired public seating and fa­ cility use. Because 25.6 percent (n = 23) of the re­ sponding libraries experienced a 100 per­ cent increase in usage, one would expect most of the highly “wired” libraries to report major increases in postproject us­ age. As expected, eight (57.1%) of the four­ teen libraries with wired network access at all seating locations reported usage in­ creases of 100 percent or more. Surpris­ ingly, however, six (26.1%) of the twenty- three facilities in the 75 to 99 percent wired 136 College & Research Libraries March 2004 TABLE 13 Wireless Network Access and Changes in Postproject Facility Use (n = 90) Wireless Access Exit Count Change Available? (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total Yes 0 7 12 12 10 9 50 No 4 7 7 5 3 14 40 Total 4 14 19 17 13 23 90 Chi-square = 12.093 df = 5 p = .034 seating range encountered declines in postproject usage. The chi-square finding of p = .055 in table 12 does not quite attain the p < .05 level indicating the presence of a statisti­ cally significant relationship between the extent of wired seating and postproject usage. However, it is significant, for plan­ ning purposes, that the majority of the completely wired libraries recorded postoccupancy usage gains exceeding 100 percent. It appears that an investment in ubiquitous or near ubiquitous wired network access in public seating areas contributes to increased levels of student facility use. Wireless Systems Wireless systems were rare in academic libraries before project completion, with 149 of the 171 responding libraries (87.1%) lacking any wireless installation. In con­ trast, 57.9 percent of the participating li­ braries reported that wireless connectiv­ ity was available in their facility, to at least some degree, by early 2003. In many cases, wireless systems were installed af­ ter the building project was completed to complement an existing wired infrastruc­ ture. Wireless systems are now common­ place, but they are not found in all librar­ ies undergoing a building improvement project since 1994. The presence of wireless systems was expected to be associated with increased facility usage as students increasingly acquire wireless devices and demand wireless access. A chi-square test indicates that a significant relationship exists at the p = .034 level, but it is an inverse relation­ ship. Unexpectedly, table 13 shows that 100 percent increases were found more com­ monly in libraries without a wireless net­ work, with only 18.0 percent (n = 9) of the wireless facilities and 35.0 percent (n = 14) of the nonwireless libraries showing in­ creases of this magnitude. In contrast, table 13 also indicates that postproject usage TABLE 14 Extent of Wireless Seating and Postproject Changes in Facility Use (n = 90) Percent Wireless Exit Count Change Seating (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total 0% 4 7 7 6 3 15 42 1-10% 0 2 0 2 0 0 4 11-24% 0 0 2 0 2 0 4 25--9% 0 0 1 2 0 0 3 50-44% 0 1 1 1 0 1 4 75-99% 0 0 3 0 0 2 5 1009 0 4 5 6 8 5 28 Total 4 14 19 17 13 23 90 Does the Building Still Matter? 137 TABLE 1S Number of Public Access Workstations and Changes in Postproject Facility Use (n = 90) Number of Public Access Exit Count Change Workstations (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total > 10 2 1 0 1 1 2 7 10-19 1 4 3 2 0 3 13 20-39 0 1 2 6 3 4 16 40-59 0 2 2 1 4 5 14 60-99 0 4 4 5 2 1 16 100 or more 1 2 8 2 3 8 24 Total 4 14 19 17 13 23 90 declines were concentrated more heavily in nonwireless libraries, with 27.5 percent (n = 11) experiencing postproject reduc­ tions in usage, whereas only 14.0 percent (n = 7) of the wireless libraries suffered declines. Interestingly, usage changes in wireless-enhanced libraries cluster in the middle, with nonwireless libraries report­ ing both the greatest increases in use and the greatest declines. The investigators also asked respon­ dents to indicate the percentage of seats in their facilities offering wireless net­ work access. The responses to that ques­ tion are cross-tabulated with postproject usage data in table 14. Once again, the findings are concen­ trated at the extremes. Only 17.9 percent (n = 5) of the libraries with wireless con­ nectivity at all user seats reported 100 percent usage increases, whereas 35.7 percent (n = 15) of the libraries lacking any public wireless access reported in­ creases at this level. However, those li­ braries without wireless network connec­ tions also confirmed the highest percent­ age of usage reductions, with eleven of forty-two libraries in this category (26.2%) confirming declines in use. There is no statistically significant relationship be­ tween the percentage of user seats with wireless access and postcompletion usage levels. Most library planners consider the in­ clusion of wireless network access to be desirable, as either the principal mode for end-user access or a complement to a wired network installation. However, as the ambiguous findings in this section indicate, there was no clear association be­ tween wireless connectivity and postproject usage among the libraries answering this question. Public Access Workstations Most enhanced libraries have expanded their number of public access worksta­ tions, whether in a conventional OPAC configuration or as an Information Com­ mons. Prior to project completion, 84 of the 171 responding libraries (49.1%) had fewer than ten public access workstations. After project completion, only 15.8 per­ cent (n = 27) of these libraries still had fewer than ten public computers, whereas 67.8 percent (n = 116) of the postproject libraries had more than twenty public access computers, 41.0 percent (n = 70) had more than sixty public workstations, and 24.6 percent (n = 42) had more than one hundred such devices. The presence of public access worksta­ tions would appear to be essential for encouraging student use because they are normally the starting point for catalog, database, and Internet research. In many libraries, students also are able to check e-mail and use applications software at public access workstations. Table 15 sug­ gests the existence of a mild, positive re­ 138 College & Research Libraries March 2004 lationship between the number of public workstations in a facility and increases in postproject usage, but it is not statistically significant. Interestingly, one-third (n = 8) of the mostly larger libraries offering 100 or more public access computers experi­ enced a 100 percent increase in usage fol­ lowing project completion. In compari­ son, 40 percent (n = 8) of the libraries pro­ viding nineteen or fewer public access computers suffered usage declines in the most recent measurement year. Even though digital divide concerns are dimin­ ishing in many academic libraries as com­ puter prices come down to affordable lev­ els for most students, an investment in pub­ lic access computers (including loaner laptops) appears to have a positive influence on postproject usage. Library Instruction Facilities Information literacy instruction has be­ come a major focus of academic library services today. Most respondents identi­ fied inadequate (or nonexistent) instruc­ tion facilities as a major shortcoming of the preproject library. The number of worksta­ tions and overall quality of a dedicated li­ brary instruction lab or electronic class­ room can have a significant effect on instruction. Where more workstations are available, students are able to participate more actively in library instruction. This active learning experience should, in turn, result in more sophisticated use of library resources and, possibly, in more frequent use of the postproject library facility. Prior to project completion, 67.9 percent (n = 114) of the 171 responding libraries either lacked a dedicated instruction lab altogether or had a facility with only one instructor workstation. These libraries had a severely limited capacity to provide an active learning experience for students in the library instruction context. After completion, 75.4 percent (n = 129) of the responding libraries had at least eleven workstations in their instruction labs, and 60.7 percent (n = 102) had more than twenty computers. In comparison, only 21.4 percent (n = 42) lacked an in­ struction lab or offered an instructional facility with only one workstation. Table 16 tests the hypothesis that the number of workstations in the instruction lab would be related to usage levels in the postproject library. Unexpectedly, there was no statistically significant association between the num­ ber of computers in the instruction lab and postproject usage, with a p = .415 finding resulting from a chi-square test. Only sev­ enteen of the fifty-eight libraries (29.3%) providing usable before-and-after exit count data and offering twenty or more computers had usage increases of 100 per­ cent or greater, whereas 25.6 percent (n = 23) of the libraries with exit count data re­ ported usage improvements at that level. TABLE 16 Number of Instruction Lab Workstations and Postproject Changes in Facility Use (n = 90) Number of Workstations Exit Count Change (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total 0 1 2-5 6-10 11-19 20-39 40 or more 1 0 0 0 1 1 1 1 4 0 0 2 2 5 1 1 1 0 2 4 10 2 0 0 1 3 7 4 0 1 0 1 4 2 5 1 1 0 0 4 10 7 6 7 1 2 16 26 32 Total 4 14 19 17 13 23 90 Does the Building Still Matter? 139 Usage increases were most apparent in li­ braries where instruction labs contained twenty to thirty-nine workstations, with 38.5 percent (n = 10) of the libraries expe­ rienced 100 percent usage gains. Six of the thirteen libraries (46.2%) without an in­ struction lab or providing only an instruc­ tor workstation suffered postproject usage decline. In short, although the actual number of workstations in the instruction lab may con­ tribute to library use, it is not statistically associated with postcompletion usage. It is possible that hands-on experience in searching electronic resources may em­ power students to feel comfortable in con­ ducting their research without coming to the library. However, table 16 does not address another possible relationship, that is, the linkage between instruction lab quality and usage levels. That relation­ ship is examined in the quality improve­ ment section of this article. Findings: User Space Typically, prior to initiating a building project, academic libraries are forced to cannibalize seating space to accommodate collection growth. Although this measure has been justified to maintain and pre­ serve print and nonprint collections in the absence of other alternatives, it has af­ fected library use by reducing the num­ ber and quality of spaces available for re­ search and study. In many libraries, the extension of shelving to building perim­ eters has increased collection exposure to ultraviolet light while depriving users of seating space enhanced by natural light. Overall User Seating It was anticipated that most projects would expand the amount of general use (table, carrel, lounge) seats, and that ex­ pectation was confirmed by survey re­ spondents. Logically, it also might be ex­ pected that the amount of seating in an enhanced facility would be related to in­ creases in building usage. However, a chi-square test (chi-square = 33.304, df = 30, p = .310) confirmed that there is no relationship between the num­ ber of seats in an improved library and postproject usage patterns. Given the broad variety of projects and institutions included in this study, this finding is not surprising. In smaller institutions, the re­ placement of a very undersized facility may result in significant usage increases, whereas postproject gains may be slighter in larger institutions already possessing significant library space. It is possible that smaller, more intimate facilities may ap­ peal more to students than large facilities do. However, the data provided by this study do not allow further inferences. Tabular data about seating and usage lev­ els will be provided upon request by the investigators. It appears that a larger number of seats does not, in itself, produce significant in­ creases in facility usage. Rather, it is more likely that usage increases in relation to the library’s ability to seat a larger per­ centage of the student body at one time. However, respondents were not asked to provide exact seating capacities or enroll­ ment figures, so further inferences are not possible from the data available. Group Study Seating Group study rooms are considered an essential component of contemporary li­ brary design, given the current emphasis on active and collaborative learning in higher education. Before project initiation, 45.6 percent (n = 78) of the 171 respond­ ing libraries did not have a single group study room and 80.7 percent (n = 138) had five or fewer studies. The project investment in group study capacity was significant, but not as dra­ matic as might have been expected. Fol­ lowing completion, 44.5 percent (n = 76) of the responding libraries had eleven or more group study rooms, whereas 8.2 percent (n = 14) still did not provide any group studies and another 19.9 percent (n = 34) had just one to five studies. Only 7.6 percent (n = 13) of the responding li­ braries provided more than thirty group study rooms. Still, the researchers antici­ pated that there would be a relationship between the number of group study 140 College & Research Libraries March 2004 TABLE 17 Number of Group Study Rooms and Changes in Postproject Facility Use (n = 90) Group Exit Count Change Studies (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total 0 0 2 0 1 0 1 4 1-5 2 3 3 3 1 2 14 6-10 0 4 5 4 3 11 27 11-19 1 4 4 8 5 7 29 20-29 1 1 2 1 3 0 8 30-39 0 0 2 0 1 0 3 40 or more 0 0 3 0 0 2 5 Total 4 14 19 17 13 23 90 rooms and postproject usage levels. The cross-tabulations for group study room numbers and postproject usage levels are presented in table 17. Unexpectedly, table 17 indicates that there was no demonstrable relationship between the number of group study rooms and facility use. Only nine (20%) of the libraries pro­ viding eleven or more group studies ex­ perienced a 100 percent usage increase. One library lacking any group study rooms whatsoever reported a 100 percent increase in usage, and two libraries pro­ viding just one to five group studies con­ firmed similar increases. The largest clus­ ter of libraries with 100 percent increases was found for those offering six to ten group study rooms, rather than a larger number of collaborative study spaces. These findings were unexpected be­ cause many libraries reported—in the survey, via e-mail, and in follow-up site visits—that they had not provided an adequate number of group study rooms to meet student demand. Further exami­ nation of these findings is required be­ cause they run counter to a major higher education trend. It is possible that the use of enrollment as a control variable might better indicate the impact of group study room numbers on physical library usage. Findings: Collection Variables The depletion of shelf space for print col­ lections is a visible, readily understood ra­ tionale for library facility expansion. When books cannot be re-shelved, students and faculty members complain, and higher edu­ cation administrators understand that a space crisis has developed. Overcrowded shelves also may discourage facility use because they make both browsing and re­ trieval of known items more difficult. Survey participants were asked to in­ dicate the provisions their libraries had made for print collection growth during project planning. The investigators fo­ cused on three aspects of print collection planning—long-range stack capacity, use of compact storage, and reliance on off- site storage facilities—that could poten­ tially affect facility usage, assuming that ease of access to print resources is a fac­ tor potentially affecting students’ deci­ sions to use library facilities. Print Growth Capacity To determine how effectively facility plan­ ners had addressed long-range collection needs, the investigators asked respondents to indicate a projected date of shelf space exhaustion for print collections at current acquisition rates. These projections incor­ porate shelf space in both conventional and compact shelving. Postproject usage levels are cross-tabulated with project shelf space exhaustion dates in table 18. As indicated by a chi-square test (chi- square = 26.244, df = 30, p = .663), there is no apparent relationship between shelving capac­ Does the Building Still Matter? 141 TABLE 18 Projected Date of Shelf Space Exhaustion and Postproject Changes in Facility Use (n = 87) Projected Exhaustion Exit Count Change Date (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total 2003-05 0 3 3 2 3 2 13 2006-10 0 4 7 2 2 1 16 2011-15 1 0 3 4 2 5 15 2016-20 2 4 1 4 1 6 18 2021-25 0 0 2 2 1 4 9 2026-30 0 0 1 1 1 1 4 2031- 1 2 2 1 3 3 12 Total 4 13 19 16 13 22 87 ity and postproject facility usage. Although 50 percent (6 of 12) of those facilities provid­ ing collection growth capacity until 2031 or later experienced 50 percent or greater us­ age increases, the investigators also found that one-quarter (n = 3) of the libraries pro­ viding this capacity experienced reductions in postproject usage. This somewhat am­ biguous finding may reflect decreased us­ age of print materials. Although investing in long-term stack capacity appears wise for collection management, it does not appear to significantly affect facility usage. Compact Shelving Use Although it has traditionally been viewed as a high-density storage option for low- use materials, compact shelving has been employed in the University of Kentucky and Nova Southeastern University librar­ ies, among others, to stretch shelving ca­ pacity and capture space for user seating, service points, and other needs. Slightly more than one-half of the responding li­ braries made some use of compact shelv­ ing, but only 12 percent (n = 20) of the 165 libraries answering this question used it for one-quarter or more of their collec­ tions. Cross-tabulation data on compact storage use and exit count are shown in table 19. As table 19 confirms, most libraries providing exit count data did not use compact storage for a large part of their TABLE 19 Percent of Collection in Compact Storage and Changes in Postproject Facility Use (n = 89) Percent in Compact Exit Count Change Storage (25-49%) (0-24%) 0-24% 25-49% 50-99% 100+% Total None 3 6 8 8 6 8 39 1-10% 1 3 4 6 3 4 21 10-24% 0 3 3 2 2 7 17 25-49% 0 1 1 1 1 2 6 50-74% 0 1 1 0 1 1 4 75-99% 0 0 1 0 0 1 2 100% 0 0 0 0 0 0 0 Total 4 14 18 17 13 23 89 142 College & Research Libraries March 2004 collections. Also, as the chi-square results (chi-square = 10.153, df = 25, p = .996) for this table indicate, there is no relationship between use of compact shelving and postproject facility use. Interestingly, 41.2 percent (n = 7) of the libraries housing 10 to 24.9 percent of their current collections in compact storage experienced 100 per­ cent increases in postproject usage. This limited finding does suggest that a well- organized, easily operated compact stor­ age system does not discourage collection use. In fact, such a system may facilitate long-term collection access by permitting libraries to place their entire collections into one physical facility, rather than over­ crowding conventional shelving or rely­ ing on off-site storage. Off-site Storage Use From the standpoint of user access, off-site storage is the least desirable of the major storage options. Most (72.7%, n = 120) of the 165 libraries completing this question were able to avoid dependence on off-site storage. However, 15.2 percent (n = 25) of the responding libraries did rely on a re­ mote site to store one-quarter or more of their print collection materials, indicating that their project did not permit (whether by design or accident) the placement of all collections in a facility permitting direct- user access. The use of off-site collection storage facilities does not necessarily dis­ courage facility use because it may enable the library to provide more on-site seating or otherwise deliver a more pleasing learn­ ing environment. The findings indicate that off-site stor­ age is not associated statistically with postoccupancy usage patterns. Only eigh­ teen libraries using off-site storage also provided pre- and postproject usage data, and 38.9 percent (n = 7) of these libraries reported usage declines. Three libraries using off-site storage did experience 100 percent postproject increases in usage, but none of those libraries had more than 10 percent of their collections shelved in a remote location. Off-site storage itself probably does not contribute directly to frequency of library facility use, particularly if low-use mate­ rials are selected carefully for that site. However, in many cases, libraries relying heavily on off-site storage have not pro­ vided sufficient shelf space to accommo­ date collection growth. This may mean that higher-use materials are stored in cramped spaces, thereby reducing ease of print collection use, which is still a major reason for visiting a physical library. TABLE 20 Impact of Non-Library Facilities on Postproject Changes in Library Use 50% Use 50% Use % of 50% % of 50% Type of Nonlibrary Increases Increases Increases Increases Facility Included with without with without in Same Building Facility Cn) Facility Cn) Facility Facility General Computer Lab en = 68) 26 10 38.2% 45.5% Snack Bar/Cafe en = 31) 12 24 38.7 40.7 General Classrooms en = 35) 16 20 45.7 36.4 Seminar Rooms en = 47) 21 15 44.7 34.9 Multimedia Prod. Center en = 41) 20 16 48.8 32.7 Auditorium en = 20) 9 27 45.0 38.6 Conference Room en = 78) 32 4 41.0 33.3 Art Gallery en = 23) 12 24 52.2 35.8 Writing Lab en = 20) 11 25 55.0 35.7 Bookstore en = 5) 1 35 20.0 41.2 Research Institute en = 6) 2 34 33.3 40.5 Educational Tech.Center en = 26) 11 25 42.3 39.1 Findings: Nonlibrary Facilities As noted in table 4, new, stand-alone li­ braries comprise only 33.3 percent (n = 30) of the ninety libraries providing usable exit count data. New, multipurpose facilities (12 of the 90 facilities with meaningful exit count data) also are becoming increasingly popular as a means for institutional cost containment, and they may contribute to the creation of strategic partnerships po­ tentially benefiting both the library and its coinhabitants.8 In other instances, inclu­ sion of the library in a multipurpose facil­ ity may be the only politically viable strat­ egy for securing institutional commitment toward a major facility enhancement. The trend toward inclusion of nonlibrary facilities within the building has changed the character of recently improved librar­ ies in significant ways. Rather than having a few nonlibrary units occasionally occu­ pying space in the building, the physical library is increasingly becoming the home base for a wide variety of operations. As such, it is becoming a more complex facil­ ity that attracts students for multiple pur­ poses and must accommodate the needs of nonlibrary units. Table 20 lists twelve types of nonlibrary facilities found frequently in postproject buildings and shows possible associations between the presence of particular facili­ ties and usage patterns. The first column lists the nonlibrary facility types and their frequency of occurrence (maximum n = 90) in projects for which usable exit count data were available. The investigators focused on libraries experiencing 50 percent usage gains to provide a better indication of fa­ cility impact than would be possible by examining only the 100 percent increasers. The second column indicates the num­ ber of libraries (maximum n = 36) includ­ ing a specific nonlibrary facility that ex­ perienced 50 percent or greater increases in postoccupancy usage. The third col­ umn lists the number of projects without that facility that reported 50 percent us­ age increases. The fourth and fifth col­ umns indicate the percentage of libraries with and without a particular facility that reported 50 percent increases. Does the Building Still Matter? 143 Conference rooms (n = 78) and general computer labs (n = 68) were found most frequently in postproject buildings pro­ viding before-and-after exit count data. As a result, they would be expected to show the greatest frequency among the thirty-six libraries experiencing 50 per­ cent usage increases. That expectation is confirmed in the second column. How­ ever, because these types of facility were included most frequently in library im­ provement projects, their relative impact on usage is shown more clearly by the percentage data in the two right-hand columns than by the raw data in columns two and three. Interestingly, writing labs and art gal­ leries appear to be slightly associated with increases in postoccupancy usage, with more than one-half of the libraries includ­ ing those nonlibrary facilities reporting 50 percent usage increases. Somewhat sur­ prisingly, fewer facilities with general com­ puter labs (38.2%) experienced 50 percent postoccupancy usage growth than facili­ ties without general computer labs (45.5%). General-use classrooms and seminar rooms appear to be slightly associated with higher increase levels, but there is no major usage difference between librar­ ies with and without them. Somewhat unexpectedly, the presence of cafes and snack bars was not associated with increases in us­ age, despite their current popularity in li­ brary design. The associations noted above are mod­ erate tendencies based on a limited num­ ber of cases. None of these associations is statistically significant. On balance, there is no evidence that the presence of particular nonlibrary facilities has a significant impact on library exit counts. There may be good reasons for including various nonlibrary facilities in a project, but there is no indi­ cation from this study that their presence has a significant impact, either positive or negative, on facility usage. Findings: Facility Quality A final set of eleven questions addressed facility quality and librarian satisfaction with their libraries before and after project 144 College & Research Libraries March 2004 TABLE 21 Facility Features Rated "Excellent" by Respondents and Changes in Postproject Library Use 25-49% 0-24% 0-24% 25-49% 50-99% 100%+ Facility Feature and Decrease Decrease Increase Increase Increase Increase Number of Responses (n = 4) (n = 14) (n = 19) (n = 17) (n = 13) (n = 23) Facility Layout (n = 42) 2 4 4 10 9 13 Service Point Location (n = 47) 2 6 7 11 8 13 Instruction Lab (n = 53) 2 6 8 9 9 19 User Seating (n = 66) 2 9 11 13 11 20 Collection Storage (n = 49) 1 6 11 9 7 15 Public Computers (n = 66) 2 7 13 13 11 20 Telecommunications (n = 58) 2 8 10 9 9 20 Artificial Lighting (n = 42) 2 5 7 9 7 12 Natural Lighting (n = 62) 1 8 10 11 10 22 HVAC (n = 26) 1 2 5 3 6 9 Overall Ambience (n = 69) 2 12 10 13 10 22 completion. These respondent percep­ tions were expected to indicate the degree of facility improvement resulting from the project. They also permitted cross-tabulation of quality ratings for particular facility fea­ tures (seating, lighting, HVAC, etc.) with postproject usage to determine the rela­ tive impact of specific building elements on facility use. Respondent assessments of the postproject library ambience were consis­ tently positive, with 70.4 percent (n = 119) of 169 participants assigning an “excellent” rating and 94.3 percent (n = 160) consider­ ing the facility ambience either “excellent” or “very good.” Significantly, no respon­ dents rated their postproject facility ambi­ ence as either “fair” or “poor.” To assess the usage impact of specific facility improvements, regardless of project type, the investigators cross-tabu­ lated respondent ratings of “excellent” with the usage change levels. The find­ ings from this cross-tabulation are re­ ported in table 21. The excellence–usage relationships depicted in table 21 are significant for aca­ demic library planning. The left-hand column indicates the number of “excel­ lent” ratings assigned to a particular fa­ cility feature, with 90 being the highest possible number. The remaining columns show varying degrees of postproject us­ age change for libraries where a specific feature was rated “excellent.” If a particu­ lar facility feature were associated with high levels of usage, the investigators hypothesized that most libraries giving an “excellent” rating for that facility would experience 50 to 99 percent or 100+ percent increases in postproject usage. In the 100+ percent column, the maximum score would be 23, indicating that all li­ braries experiencing usage increases at that level also would have assigned that facility variable an “excellent” rating. It is apparent from the number of re­ sponses for specific features in the 100 per­ cent column that most facilities with in­ creases of that magnitude offer both an excellent ambience and excellent natural lighting for users. The quality of user seat­ ing, public access computers, and telecom­ munication infrastructure also is very im­ portant, with the instruction lab coming just slightly behind (n = 19) in association with 100 percent usage increases. Al­ though all eleven facility features probably influence postproject use levels, the follow­ ing features were considered “excellent” by 80 percent or more of the libraries ex­ periencing 100 percent increases, with n = 23 being the highest possible value: Does the Building Still Matter? 145 • overall facility ambience (n = 22); • natural lighting (n = 22); • user seating quality (n = 20); • public access computers (n = 20); • telecommunications infrastructure (n = 20); • instruction lab (n = 19). These findings suggest that librarians and facility planners should place particu­ lar emphasis on the quality of these ele­ ments when planning library improve­ ments. The quality ratings are subjective assess­ ments that can be treated as interval data, thereby enabling the investigators to exam­ ine relationships between the degree of postproject improvement and subsequent usage levels. In order to permit this type of analysis, the eleven quality variables were reverse coded with values ranging from 1 to 5, with 1 being the lowest rating (poor) and 5 the highest rating (excellent) assigned. Mean changes in quality assessment for each facility feature then were calcu­ lated by subtracting the mean of the preproject responses from the mean of the postoccupancy responses. A one-way Analysis of Variance (ANOVA) test was conducted to compare respondent per­ ceptions of postproject feature quality with their perceptions of quality prior to project initiation. For all eleven subjective quality mea­ sures, significant changes were found be­ tween respondents’ pre- and postproject ratings of the measure. Table 22 shows the relative improvement in quality for these facility features resulting from the build­ ing project, as perceived by respondents. The greatest change in quality is shown for overall facility ambience, with a mean postproject improvement of 2.827 on a five- item scale. In other words, the mean im­ provement in ambience rating was just less than 3 points on a 5-point scale, represent­ ing a change from “poor” (1) to “very good” (4) or from “fair” (2) to “excellent” (5). Mean quality improvements of 2.5 or greater also were reported for user work space (2.748), the instruction lab (2.687), the telecommunications infrastructure (2.667), and public access workstations (2.599). The location of service points (1.670) and heat­ ing, ventilation, and air conditioning (HVAC) systems (2.068) received the low­ est mean improvement ratings, but posi­ tive changes of this magnitude are still quite significant. On balance, the percep­ tion ratings indicate that facility quality was improved greatly as a result of the building project. Presumably, these across­ the-board quality improvements should be associated with greater facility usage. TABLE 22 Degree of Quality Improvement and Postproject Changes in Facility Use, Based on Scale Rankings from 1 to 5 (Postproject rating minus preproject rating) Facility Feature Rated Mean Rating Improvement Standard Deviation Quality of Layout 2.122 1.400 Quality of Service Point Location 1.670 1.312 Quality of nstruction Lab 2.687 1.260 Quality of User Work Space 2.748 1.283 Quality of Collection Storage 2.399 1.468 Quality of Public Workstations 2.599 1.161 Quality of Telecommunications 2.667 1.200 Quality of Artificial Lighting 2.117 1.358 Quality of Natural Light 2.273 1.512 Quality of HVAC 2.068 1.352 Quality of Ambience 2.827 1.212 146 College & Research Libraries Strength of Relationships Earlier sections of this paper have ex­ plored the existence of possible relation­ ships between specific variables and postproject usage levels. However, they have not specified either the strength or the direction of those relationships. This section examines those relationships more closely to determine which ones have the greatest impact on postproject usage. Table 23 presents the bivariate correla­ tion coefficients (Pearson’s r) between key independent variables for the study and changes in postproject exit counts. The table includes both numerical indicators (number of data ports in the postproject facility, etc.) and subjective assessments of feature quality by respondents. Thir­ teen of these twenty-five independent March 2004 variables were significantly related (p < .05) to increases in the postproject exit count, and some of the relationships are unusually strong. Three variables attaining the p < .05 level of significance—institution type, number of data ports, and percentage of wired seats—were identified earlier as factors positively influencing postproject usage levels. These variables, and the ten additional quality variables identified as significant at the p < .05 level in table 23, are statistically correlated with increases in postcompletion facility use. Two of these variables—quality of the instruction lab and quality of the telecom­ munication infrastructure—are related to factors examined earlier in this analysis. In those assessments, the investigators TABLE 23 Strength of Relationships between Specific Facility Features and Increases in Postproject Usage Correlation Significance Facility Feature (Pearson's r) Level (p)=)� Quality of Instruction Lab .399 .000 Institution Type (public or private) .384 .000 Quality of Layout .342 .001 Number of Data Ports .293 .005 Quality of Public Access Workstations .292 .006 Quality of Natural Lighting .282 .007 Quality of User Work Space .280 .008 Quality of Telecommunication Infrastructure .259 .014 Quality of Overall Ambience .244 .020 Quality of Collection Storage .236 .026 Quality of HVAC System .236 .026 Percent of Wired Seats .223 .034 Quality of Service Point Locations .221 .038 Number of Workstations in Instruction Lab .149 .160 Quality of Artificial Lighting .149 .162 Percent of Facility Designated for Library Use .110 .302 Percent of Seats with Wireless Access .064 .551 Number of Group Study Rooms .058 .589 Location on Campus .035 .741 Availability of Wireless Network Access -.029 .787 Size of Library (gross square footage) -.059 .581 Collection Shelving Exhaustion Date -.084 .439 Use of Compact Shelving -.098 .360 Carnegie Classification (modified) -.099 .354 Number of General-use Seats -.131 .220 Does the Building Still Matter? 147 did not find any relationship between the number of workstations in the instruction lab or the type of premises wiring and postproject usage levels. The correlational analysis confirms that such relationships do exist, when post- and preproject qual­ ity of these features is compared. For the number of data ports, this relationship is significant at the p = .005 level. In fact, improvement in instruction lab qual­ ity emerges from table 23 as the single variable most closely related to increases in postproject usage, with a Pearson’s r value of .399 and a significance level of p = .000, even though the raw number of workstations in the lab did not have a significant impact on postcompletion usage. The correlational analysis also confirms the importance of private institutional affiliation as a power­ ful factor influencing usage levels, with a similar significance level of p = .000. Several other quality variables—lay­ out, public access workstations, natural lighting, and user work space—are sig­ nificant at the p<.010 level. Effective lay­ out and the use of abundant natural light­ ing are key characteristics of functional design. Their presence as factors related strongly to facility use is strong evidence that effective architectural design does influence postproject usage. The finding of a significant relationship (r = .221; p = .038) between service point locations and facility use also indicates that the place­ ment of service points is another key fac­ tor that should influence library design. It was reported earlier that the num­ ber of public access workstations is asso­ ciated with increased building usage, al­ though the nature of that relationship was not specified. Table 23 confirms that both quantity and quality of a library’s public access computers are related to facility usage. In short, library planners are well advised to place abundant, high-quality com­ puters in a facility to encourage student use. The number of general-use seats in an improved library facility was not associ­ ated with increased usage. However, table 23 indicates that the quality of user seating and work surfaces is a factor significantly af­ fecting postoccupancy usage. It would ap­ pear that comfortable seats, plus spacious, well-lit table and carrel surfaces, are fac­ tors that have a positive impact on usage of an enhanced facility. Those work spaces also should be wired, as the finding of a significant relationship (r = .223; p = .034) between wired seating levels and postproject usage indicates. Not surprisingly, improvement in the quality of a library’s HVAC system also emerged from this analysis as a factor af­ fecting facility use. Although the mean improvement in HVAC quality was 2.068 on a 5-point scale, study respondents were least satisfied with this factor among the eleven quality variables studied. However, where the improvement was significant, better HVAC quality was as­ sociated with greater increases in postoccupancy usage. It was found earlier that the projected date of shelf space exhaustion was not as­ sociated with usage levels. In fact, table 23 confirms that there is actually a slight nega­ tive relationship between the availability of collection growth space and usage lev­ els, although it is not statistically signifi­ cant. However, the quality of collection stor­ age space is a factor influencing usage that is significant at the p = .026 level. It appears that elements other than shelf capacity (lo­ cation, convenience to user seating, attrac­ tiveness, etc.) influence usage. This find­ ing indicates that the quality of access to print collections is still a consideration af­ fecting student usage of library buildings. Interestingly, the availability of wireless network access was not correlated with postproject usage, contrary to expectations from the chi-square analysis. In fact, the Pearson’s r for this variable was slightly negative. The percentage of wireless seats also was not significantly associated with usage after project completion. The quality of artificial lighting and the number of group study rooms in the fa­ cility were likewise not related to postoccupancy usage. However, despite the absence of a significant correlation, the investigators encourage attention to these building components in academic library design. 148 College & Research Libraries March 2004 Conclusions and Significance This study has compared preproject and postcompletion facility usage patterns for major library projects completed between 1995 and 2002. The investigators received Web survey responses to a 68-item survey from 182 of 357 libraries finishing facility projects covering at least 20,000 square feet. Using changes in before-and-after-project exit count data as their dependent variables, the researchers were able to distinguish key project characteristics associated with sig­ nificant increases in facility usage from fa­ cility features less associated with postoccupancy usage growth. Major Conclusions The investigators found that 80 percent of the libraries completing a major space im­ provement project between 1995 and 2002 experienced greater facility usage in 2001– 2002 than they did in a preproject baseline year, whereas 20 percent of the responding libraries reported lesser usage. The median change in postoccupancy usage was a 37.4 percent increase. In some instances, usage grew dramatically, with eight libraries re­ porting postoccupancy usage gains exceed­ ing 200 percent and 25.6 percent experienc­ ing increases of 100 percent or greater. Nine of the eighteen “decliners” reported usage reductions of less than 10 percent in the most recent fiscal year. One institution recorded a confirmed usage increase of 1012 percent. In addition to finding a general pattern of greater library usage following project completion, the investigators identified a number of specific facility attributes as­ sociated with postproject usage gains, in­ cluding: • number of data ports; • percentage of seats with wired net­ work access; • number and quality of public access computers; • quality of library instruction lab; • quality of telecommunication infra­ structure; • quality of natural lighting; • quality of user work spaces; • quality of layout (including loca­ tion of service points); • quality of collection storage space; • quality of HVAC system; • quality of overall facility ambience. The study further confirmed that nearly all (93.1%) of the private institu­ tions completing library projects experi­ enced postoccupancy usage increases, whereas 73.8 percent of the public insti­ tutions reported greater usage. In addi­ tion, there was a tendency for projects completed during the latter part of the study period to experience more consis­ tent usage gains than projects completed between 1995 and 1998. Several key variables did not have a demonstrable impact on postcompletion usage increases. Expected associations between specific facility/institutional fea­ tures and growth in facility use were not found for the following variables: • type of project (new versus ex­ panded or renovated); • campus location; • presence of nonlibrary units in gen­ eral; • presence of high-end wiring system; • presence or coverage of wireless communication systems; • number of group study rooms; • presence of general computer labs; • cybercafes or snack bars; • long-term shelving capacity for print collection growth. Although group study rooms and wireless systems were not associated with increased facility use in this study, the investigators are not recommending that they be given a diminished priority in facility planning. Significance Discussions about the “library as place” often occur in a context of speculation uninformed by validated empirical evi­ dence. This study addresses that research gap by providing a series of findings about the impact of new and improved library spaces on use of the physical fa­ cility, as measured by exit count data. Although statistical summaries from major library associations indicate gener­ ally decreased levels of collection, refer­ Does the Building Still Matter? 149 ence, and building usage, this study con­ firms that the great majority of new and im­ proved libraries have experienced sustained increases in usage of the physical facility fol­ lowing project completion. In addition, some libraries have experienced profound increases in usage, with 25.6 percent of survey partici­ pants reporting postproject usage gains ex­ ceeding 100 percent. In short, a high-quality building does make a difference, and students continue to use an improved facility even after the novelty of a new library has worn off. This study also has identified specific facility and institutional variables associ­ ated with higher levels of postoccupancy usage. By knowing what facility features are most likely to produce usage increases, facility planners can better design librar­ ies likely to attract students and advance the academic mission of the institution. Moreover, they can better determine which features are most important to retain in “value engineering,” when budget limita­ tions require that facility features be elimi­ nated or scaled back. Although the inves­ tigators urge that the results not be used rigidly to include or exclude certain facili­ ties (group study rooms or cybercafes, for example), they do provide general guide­ lines for identifying facility features most likely to yield expanded library usage. It also is clear from this study that stu­ dents are not uniformly “deserting” aca­ demic libraries. Rather, they may be mak­ ing less use of older facilities lacking good computers, an extensive network access infrastructure for laptop computer users, and a comfortable environment condu­ cive to a variety of uses (individual and study research, instruction, social). With a median 37.4 percent increase in postproject usage, new and improved libraries are clearly not experiencing declines in usage of the physical facility. Instead, they are encoun­ tering a pattern of usage growth that runs counter to the declines reported in some articles and data compilations. Although the study did not address the specific types of use occurring in im­ proved facilities, it provides clear, empiri­ cal evidence that students can and will use a comfortable, well-equipped library, even with remote access to many elec­ tronic databases and the Internet avail­ able. This is an important conclusion be­ cause it suggests that a discerning invest­ ment in library facility improvements— whether a new library or improvements to existing space—will attract students to a specialized physical place designed to provide research and study space, teach information literacy skills, expose stu­ dents to recorded knowledge in both print and electronic formats, and make “information experts” readily accessible. Academic libraries will continue to compete with other campus units for scarce capital project, technology, staffing, and other funds. However, this article provides valid, empirical evidence that well-planned, contemporary libraries are still used heavily in an era of rapid tech­ nological change. It also provides findings important for planning facilities likely to attract student users well into the future. Ultimately, decisions on the future of academic library space will reflect changes in technology, institutional pri­ orities, teaching strategies, and librarians’ own assessments of the significance of the physical library. As Donald E. Riggs has noted, “the belief that ‘everything is go­ ing to be online’ influences decision mak­ ers not to fund requests for new librar­ ies.”9 In the meantime, the findings in this article should move the “library as place” debate beyond the anecdotal level, pro­ vide some guideposts for strategic facil­ ity planning, and demonstrate that well- designed libraries remain essential as flexible, evolving, and relevant learning centers in an increasingly decentralized information environment. Implications for Further Research Although this study examines many vari­ ables affecting facility usage following project completion, it does not address a variety of other issues in the building–us­ age relationship that merit further study. Several such issues are identified below: 1. As the library provides greater ac­ cess to e-journals and e-books, how will 150 College & Research Libraries March 2004 users secure access to and use those re­ sources? In short, to what extent will they use the physical library as a portal to those materials? 2. What specific uses (research, study, social, other) are students making of li­ brary facilities today? How will those usage patterns change over time? 3. How are student attitudes toward the physical library as a social institution changing? How will those attitudes affect subsequent student behaviors? 4. Is there any difference in physical facility use between students receiving library instruction and those not receiv­ ing it? Do usage patterns differ by disci­ pline or academic level (freshman versus master’s, for example)? 5. How do commuting and residen­ tial students use the physical library? 6. Do curriculum changes (new pro­ grams, writing requirements, etc.) affect usage of the physical library? 7. Do non-site-specific library services (document delivery, virtual reference, etc.) affect use of the library building? 8. Are older libraries (pre-1980, for example) used differently from those built in recent years? 9. To what extent does the inclusion of specific nonlibrary facilities (computer labs versus writing centers, for example) affect various uses of the library build­ ing? 10. Do students making use of nonlibrary facilities under the same roof use the library differently from students not utilizing those services? 11. Does the creation of an Information Commons or the provision of productiv­ ity software on library computers affect building use? These issues are merely suggestive of a broader range of research questions in­ volving use of the physical library in a changing environment. The investigators anticipate that the current study will stimulate further research into the use of library buildings, thereby giving librar­ ians and planners a sound base of empiri­ cal evidence for understanding emerging usage patterns and creating physical “places” that will effectively address fu­ ture library space requirements. Notes 1. For a detailed discussion of current, recent, and planned projects in the California State University system, see Gordon Smith, “Libraries for the 21st Century,” Newsletter of the California State University Libraries 2 (fall 2003): 1–8. Available online from http://www.calstate.edu/LS/ Fall2003.pdf. [Cited 4 December 2003]. 2. Harold B. Shill and Shawn Tonner, “Creating a Better Place: Physical Improvements in Academic Libraries, 1995–2002,” College & Research Libraries 64 (Nov. 2003): 431–66. 3. Scott Carlson, “The Deserted Library,” Chronicle of Higher Education 48 (Nov. 16, 2001): A35–38. 4. Scott Bennett, Libraries Designed for Learning (Washington, D.C.: Council on Library and Information Resources, 2003). 5. Virginia Kelsh, “USF Law Library,” e-mail to Harold B. Shill, Feb. 10, 2003. 6. Comprehensive Carnegie classification listings are available online at the Carnegie Foun­ dation for the Advancement of Teaching Web site: http://www.carnegiefoundation.org. 7. The Pennsylvania State University, Faculty Advisory Committee on Academic Computing, “FACAC Student Survey 2002” (University Park, Pa.: The Pennsylvania State University, 2002). 8. For a discussion of the political advantages of strategically selected partnerships, see Harold B. Shill, “Strategic Positioning and the Building Project: Penn State Harrisburg’s Library of the Future,” in Racing toward Tomorrow: Proceedings of the Ninth National Conference of the Association of College & Research Libraries, ed. Hugh A. Thompson (Chicago: Association of College & Research Libraries, 1999), 370–79. 9. Donald E. Riggs, “New Libraries Remain an Excellent Investment,” College & Research Libraries 63 (Mar. 2002): 109. http:http://www.carnegiefoundation.org http://www.calstate.edu/LS