tenopir.p65 234 College & Research Libraries May 2000 Patterns of Database Use in Academic Libraries Carol Tenopir and Eleanor Read Database usage data from a random sample of academic libraries in the United States and Canada reveal patterns of use in selected types of libraries. Library users tend to use commercial online databases most frequently early in the week, at midday, and at times that correspond to the academic calendar (November in this six-month sample). On aver­ age, relatively low numbers of users are simultaneously logged on to research databases at any size of library. A questionnaire sent to these same libraries identified many other factors that might influence data­ base use, including level of instruction, availability of remote log-in, and placement of a database on the library’s home page, although none of these factors was found to be statistically significant. cademic librarians today must consider many factors when making the difficult decisions about what electronic re­ sources to purchase for their users. Their decisions are based not merely on deter­ mining what content is best or most unique or even what system offers the most user-friendly or most powerful ac­ cess software. To best allocate budgets and select the options that will serve the greatest number of users, librarians also must weigh pricing options and licens­ ing restrictions and predict likely levels of use for each database selected. Predict­ ing usage patterns and levels of use is difficult, but important, because prices or licensing restrictions are often based on amount of use, total number of potential users, or number of simultaneous search­ ers allowed. Predicting likely numbers of simulta­ neous users is especially difficult without a history of prior usage. The library that is leasing a new product or offering online access for the first time often has to guess how much the database will be used. How­ ever, usage data from other academic li­ braries may help similar libraries to pre­ dict levels and patterns of use. This paper reports on a two-phase study of academic libraries to identify patterns of database use and the factors that might influence this use. Online data captured from ninety-three academic li­ braries reveal how many users are logged Carol Tenopir is a Professor in the School of Information Sciences at the University of Tennessee; e-mail: ctenopir@utk.edu. Eleanor Read is a Graduate Research Assistant in the School of Information Sciences at the University of Tennessee; e-mail: eread@utk.edu The authors would like to thank Danielle M. Green for her able assistance and Cari Springer for her statistical consulting. George Banks and Leslie Preston were involved in the project at its early stages, especially in the overwhelming task of figuring out how to deal with tens of millions of online data points and to present the data in a meaningful way. Shawn Collins, technology coordinator at the School of Information Sciences, has offered expert support throughout. 234 mailto:eread@utk.edu mailto:ctenopir@utk.edu Patterns of Database Use in Academic Libraries 235 on simultaneously to selected online re­ search databases and what time of day, week, and month academic users are searching most often. Examination of these patterns will help other academic libraries negotiate simultaneous usage li­ censes and estimate the number of work­ stations and ports required. The researchers asked each library about its specific environment for online access and gathered informa­ tion about factors that may influence online use. However, usage data do not show what each individual library is doing, if anything, to encourage use of these data­ bases, so the usage data were supple­ mented with a survey. The researchers asked each library about its specific envi­ ronment for online access and gathered information about factors that may influ­ ence online use. Review of the Literature The ALA’s Office for Research and Statis­ tics participates in studies regularly to gauge amounts of library use. In 1997, in cooperation with ACRL, ALA published a survey that showed how all types of academic libraries have embraced elec­ tronic services, although the survey did not consider when the resources were being used or how libraries measured use.1 Earlier usage studies for automated resources often were done to help librar­ ies determine how many terminals were required when they first brought up an online catalog (just as even earlier stud­ ies tried to predict an appropriate num­ ber of chairs to provide in the library). A 1983 report incorporated queuing mod­ els to recommend appropriate numbers of terminals for online catalogs.2 Turnstile counts have been used to optimize refer­ ence department staffing or pickup sched­ ules for shelving. These show that peak usage periods in academic libraries cor­ respond to the academic calendar and daily class schedules.3,4 To improve user services, specific li­ braries often focus on how electronic re­ sources are used by their patrons. The Biological Sciences Library of The Ohio State University, for example, conducted a four-and-a-half-year study of the use of CD-ROM databases in its library. It found that a majority of users recorded all use within a one-month period and more than a third of all users recorded all use on only one day.5 The New York Public Library Research Libraries compared patterns for remote usage of their OPAC with patterns of usage from within the libraries. They found that patterns for remote searching of the OPAC are distributed more evenly over each day and within each week than in-library searching and that a “large part of remote searching occurred when the Research Libraries were closed.”6 In a call for better computerized monitoring of remote users, Thomas A. Peters discussed the widespread belief that remote users will likely follow “diurnal” usage pat­ terns but suggested that this may only be true in metropolitan areas “already noted for diurnally diffuse human behavior.”7 However, data such as turnstile counts or other measures of amount of use are not enough to see why a particular resource is used. Anne-Marie Belanger and Sandra D. Hoffman surveyed academic library users to see if demographic or other factors in­ fluence how often they use ERIC on CD­ ROM.8 Judith A. Adams and Sharon C. Bonk found that, for faculty, not knowing what electronic resources are available is the major barrier to use. Faculty believe that training classes for electronic re­ sources are a high-priority need.9 Charles T. Townley and Leigh Murray found in a study of six southwestern academic librar­ ies that database use does not fit a predict­ able pattern across libraries and databases. Database use is influenced by the length of time the database has been available (available more than one year promotes heaviest use), limiting the forms of access to the database, the source of the database (locally loaded and Internet-based data­ bases were preferred), and the availability of user instruction in the library.10 http:library.10 236 College & Research Libraries Librarians’ attitudes or factors such as convenience, accessibility of a database through menu screens, or the availability of training materials might be expected to influence the amount of use that any specific resource such as an online data­ base receives. Carol Tenopir and Lisa Ennis analyzed reference librarians’ atti­ tudes toward electronic reference prod­ ucts and changes in university reference services throughout the 1990s.11,12 Refer­ ence rooms in libraries grew busier throughout the decade. Many more work­ stations were added in libraries, while at the same time remote access was made widely available. Librarians reported that users prefer electronic reference products over print, both from within the library and through dial-up access; and library instruction classes now most often focus on electronic resources. End-user online searching of commercial databases is now an integral part of library reference ser­ vices and users have come to expect online access through their libraries. Ref­ erence librarians in these research librar­ ies feel busy but, for the most part, have positive attitudes toward technology. Collection development and reference librarians now spend much of their time evaluating and selecting online resources, a task more complex than, but not unlike, traditional selection and collection build­ ing. Added to the traditional task of evalu­ ating quality, content, coverage, and ease of use is the need to select the best pos­ sible pricing options from among many alternatives and to negotiate online licens­ ing agreements.13,14 The best pricing op­ tion varies from library to library and, within a library, may be different for heavily used resources than for those used less often. Many online resources are purchased with some type of user-based license. In simultaneous usage licenses, librarians must commit to how many online users will be permitted to be connected at any one time to a given database or a family of databases. Providing for too many si­ multaneous users will waste the library’s money and allowing too few will cause May 2000 frustration among users who must wait to access a desired database. The information reported here is part of a larger study that examines patterns and factors of use for both academic and public libraries.15 This study demon­ strates that although the exact amount of use varies with the size and type of li­ brary, the usage patterns in academic and public libraries are similar. Although all libraries have some unique situations, learning from other, similar libraries may be the best way to estimate how many users will likely use a database and when that usage will occur. The usage pattern data in this study were first reported in 1999.16,17 (The present article updates, cor­ rects, and expands on some of the data in these earlier reports.) Usage data show how often online databases are used in academic libraries; the survey informa­ tion begins to look at why they are used. Methodology: Phase 1—Usage Data Measuring online usage is more complex than collecting turnstile counts, sampling workstation queues, or even measuring access to a single library’s online catalog. Online use from many libraries searching comparable databases must be captured over a period of time to give a broad pic­ ture that will allow prediction of patterns for individual libraries. To get both remote and in-house online activity for commer­ cial databases from many libraries, a ma­ jor database producer and aggregator agreed to provide the researchers with usage data for all of its online databases. This database aggregator provides online access to many bibliographic, full-text, and directory databases, thirty-eight of which were used by one or more academic librar­ ies in this study. This database aggregator provides more than a hundred database titles, many of which contain overlapping in­ formation aimed at different audiences. For example, the same journals and maga­ zines may be available in an indexing- only version, a full-text version, or a com­ bination version. Versions indexing thou­ sands of titles may be sold to university http:libraries.15 Patterns of Database Use in Academic Libraries 237 libraries; small colleges may prefer ver­ sions with fewer, selected titles. Rarely does the same library purchase overlap­ ping titles, although a library may pur­ chase separate current and backfile ver­ sions or choose a combined version. Some databases are subject specific (business journals, for example); others are aimed at a general interest academic audience. Online usage for every library is captured automatically by the database provider in five-minute intervals, twenty-four hours a day. A random sample of a hundred aca­ demic libraries in the United States and Canada was taken from this company’s list of more than 1,200 libraries. From this sample, usable online usage data were available for ninety-three libraries. The ninety-three libraries represent every Carnegie Class of parent academic insti­ tution, with the largest number being from Master’s Universities and Colleges I. Table 1 shows how Carnegie Class was distrib­ uted in the study sample. For purposes of analysis, similar Carnegie Classes are re­ ported together. All data are reported in the following six groups: Baccalaureate Colleges, Master’s Universities and Col­ leges, Doctoral Universities, Research Uni­ versities, Associate of Arts Colleges, and Schools of Business and Management. Online usage for every library is cap­ tured automatically by the database pro­ vider in five-minute intervals, twenty-four hours a day. Even with a sample of only ninety-three libraries, a year ’s worth of data would yield more than ten million data points for every database. Therefore, for this study, usage data were sampled once per hour (on each half-hour), for six­ teen hours per day (8 a.m. to midnight, eastern time), for a period of six months (July to December 1997), resulting in more than 282,000 data points per database. To report all sample times in local time, the researchers have converted the hourly data into the local time for each library and taken into account daylight savings time for those areas that observe it. Because the database company’s computer was set to eastern time, data from libraries in west­ ern time zones begin and end earlier. Usage data reveal how many simulta­ neous users are logged on to any one da­ tabase at any of the sampled times. Time TABLE 1 Carnegie Categories for Academic Institutions Carnegie Class No. in Sample Surveys Received Baccalaureate Colleges 16 8 Baccalaureate (Liberal Arts) Colleges I 7 3 Baccalaureate Colleges II 9 5 Master's Universities and Colleges 35 22 Master's (Comprehensive) Universities and Colleges I 33 21 Master's (Comprehensive) Universities and Colleges II 2 1 Doctoral Universities 9 6 Doctoral Universities I 2 2 Doctoral Universities II 7 4 Research Universities 18 11 Research Universities I 14 6 Research Universities II 4 5 Associate of Arts Colleges 12 7 Schools of Business and Management 3 3 Total 93 57 Source: www.carnegiefoundation.org http:www.carnegiefoundation.org 238 College & Research Libraries FIGURE 1 Average Simultaneous Use Patterns by Time of Day, Academic Libraries by Type of Institution stamps on the data allow patterns to be drawn that show average numbers of us­ ers by time of day, day of the week, and month for each class of library and each database or database group. In this analysis, the thirty-eight data­ bases used by the libraries were examined in four different groupings: (1) all thirty- eight databases together; (2) eight data­ bases that together cover general maga­ zines and journals for academic libraries (including full-text, indexing-only, backfiles, and current files); (3) the single most-used database among these aca­ demic libraries (a current general maga­ zine and journal title); and (4) eleven di­ rectory or bibliographic databases that specialize in business information. Methodology: Phase 2— Questionnaires Usage data reveal typical patterns of use within classes of academic libraries but do not identify why or how the specific environment might influence online use. To begin to answer the questions of how and why, information beyond usage data is needed. Information about unique en­ vironmental factors that may influence online use in individual libraries was May 2000 sought by sending a ques­ tionnaire to each of the aca­ demic libraries in the sample. Survey questions were grouped in four catego­ ries: (1) Information about Your Library; (2) Informa­ tion about the Databases You Provide; (3) Information about Databases from … [the aggregator]; and (4) Other Factors. Fifty-seven libraries responded and all Carnegie Classes were represented, as shown in table 1. Analysis of Usage Patterns The number of simultaneous users for all thirty-eight da­ tabases in all ninety-three li­ braries ranges from zero us­ ers (the mode and median) to sixty-six users (the mean is .28 and the standard deviation is 1.37). Not surprisingly, online usage in academic libraries follows the regular rhythms of academic life. Although the range and exact number of users vary by class of library, the patterns of use, in­ cluding peaks and valleys, are quite simi­ lar. The busiest time for online research in or from academic libraries is between 11 a.m. and 5 p.m. (local time), on Mondays and, to a slightly lesser degree, Tuesdays, in November. (It is likely that April and May also would be high-use months as the spring semester is coming to an end.) Use in all types of libraries dips in August and then begins a steady rise through the se­ mester until after November. Although these usage patterns corre­ spond to the peak hours for library use, 75 percent of these libraries also offer re­ mote online access. It appears that not many college students choose to be work­ ing online at times other than the normal hours that the physical library is open, even if they can be working from a dor­ mitory, office, campus lab, or home. Fig­ ures 1, 2, and 3 show these patterns of use aggregated for all databases, but sepa­ rated by type of institution. Patterns of Database Use in Academic Libraries 239 FIGURE 2 Average Simultaneous Use Patterns by Day of Week, Academic Libraries by Type of Institution Patterns of use for eight general data­ bases together show similar patterns of use (figures 4, 5, and 6). The mode and median are zero, but the mean is .56 and standard deviation is 2.15. The patterns of use for the single most-used general database and the eleven business data­ bases are very similar and so are not pre­ sented here. However, amounts of use are higher and vary more for the single most-used database, with a mean of .84 simulta­ neous users and a standard deviation of 2.77 (the mode and median are zero). As might be expected, the amount of simultaneous use for business databases is higher for schools of busi­ ness than for other types of institutions. The overall mean for business databases is .14, with a standard devia­ tion of .57 (the mode and median are zero). Average numbers of us­ ers do not show the true im­ pact of multiple users on workstations, online ports, and staff. Patterns do a bet­ ter job by helping to identify times of heavy demand. An­ other way to show impact is to measure how often mul­ tiple users are logged on. Tables 2, 3, 4, and 5 show how many simultaneous us­ ers are logged on at any one time for all databases (table 2), the eight general interest databases (table 3), the single most-used database (table 4), and finally, the business databases (table 5). (Note that the tables do not include every discrete num­ ber of users.) Unlike use of a library’s catalog, simultaneous usage of a reference database is relatively uncommon. Pro­ viding access to only one user for a general research database, for example, would be satisfactory 82.8 per­ cent of the time in research libraries and 95.2 percent of the time in baccalaureate colleges. Allowing five simultaneous us­ ers would be satisfactory 94.9 percent of the time in research libraries and 99.8 per­ cent of the time in baccalaureate institu­ tions (table 3). FIGURE 3 Average Simultaneous Use Patterns by Month, Academic Libraries by Type of Institution 240 College & Research Libraries The single most-used database in these libraries also was analyzed alone to mitigate any effects on the data from seldom-used files, such as backfiles. Table 4 shows the simulta­ neous use figures for the general magazine/journal database used most often in these libraries. Clearly, databases that are expected to be used by students and faculty in a variety of aca­ demic disciplines and that have general, current- event appeal will attract more simultaneous users. Providing access to only one user for this database, for example, would be sat­ isfactory only 76.4 percent of the time in research uni­ versities and 93.8 percent of the time in baccalaureate colleges. If five simultaneous users were provided for, baccalaureate institution users would FIGURE S Average Simultaneous Use Patterns by Day of Week, Magazines and Journals for Academic Libraries, Academic Libraries by Type of Institution FIGURE 4 Average Simultaneous Use Patterns by Time of Day, Magazines and Journals for Academic Libraries, Academic Libraries by Type of Institution May 2000 be accommodated 99.7 percent of the time and research library users 91.6 percent of the time. However, when selecting a maximum number for si­ multaneous use capability, the law of diminishing re­ turns applies. A satisfaction rate of 99.5 percent could be achieved with four simulta­ neous users in baccalaureate college libraries and with twenty-two simultaneous users in research libraries, but to achieve 100 percent satisfaction would necessi­ tate eighteen simultaneous users in baccalaureate col­ lege libraries and sixty-six in research libraries (table 4). Aiming for 100 percent sat­ isfaction is neither reason­ able nor cost beneficial. In using these usage charts to determine number of users for a simultaneous use contract, librarians should consider not only the Patterns of Database Use in Academic Libraries 241 FIGURE 6 Average Simultaneous Use Patterns by Month, Magazines and Journals for Academic Libraries, Academic Libraries by Type of Institution class of their institution, but also the type of database they are purchasing. In most academic libraries, general interest titles will get more use than more specialized databases. For example, as shown in table 5, most types of libraries could achieve 98 to 99 percent satisfaction rates for business databases by accommodating only two si­ multaneous users. Not surprisingly, busi­ ness colleges are an exception, requiring the capacity for six simultaneous users to achieve 99 percent satisfaction. Multipurpose workstations that allow access to the library’s online catalog and other online databases are the rule. Carnegie Class is one way to catego­ rize academic libraries; number of stu­ dents enrolled is another. This is an im­ portant distinction because not all insti­ tutions within the same Carnegie Class have the same number of students. Table 6 accounts for this difference by showing how many simultaneous users are re­ quired to meet demand 99 percent of the time at each academic li­ brary based on student en­ rollment. As might be ex­ pected, the number of simul­ taneous users required gen­ erally increases as the stu­ dent population increases. The exception to this rule, shown in table 6 for schools with 45,000 to 89,999 stu­ dents, should be interpreted with caution because it is based on data from only two universities. Analysis of Questionnaires The fifty-seven libraries that responded to the question­ naire offer a variety of elec­ tronic media for end-user searching. More than 90 per­ cent offer CD-ROM, com­ mercial online, or World Wide Web access. Many pro­ vide access to several commercial online services (table 7). In terms of the ways libraries influence database use, 75 percent provide remote access in addition to in-house access (table 8). Approximately 95 percent of the librar­ ies that responded to the survey offer ac­ cess to the reference databases on ten or more workstations (data not shown). Eighty-eight percent of the libraries offer access to this company’s databases on three-quarters or more of their public workstations, but only five percent have workstations dedicated to these databases. Multipurpose workstations that allow ac­ cess to the library’s online catalog and other online databases are the rule. Many librarians believe that the data­ bases analyzed in this study are among some of the most popular with library us­ ers. The largest number of libraries (46%) reported that these are the most popular databases (libraries in six of the ten Carnegie Classes said they were the most popular). An additional 21 percent believe these databases rank second in popularity (data not shown). Some libraries actively 242 College & Research Libraries May 2000 TABLE 2 Simultaneous Use for All Databases Aggregated No. of Users Cumulative Percentage Type of Institution (Carnegie Class) 1 2 3 4 5 6 7 8 9 10 15 25 Bacc. 97.3 98.9 99.5 99.8 99.9 99.9 99.9 99.9 99.9 99.9 99.9 Master's Doctoral 94.5 95.4 96.9 98.1 98.1 99.1 98.8 99.5 99.2 99.7 99.4 99.8 99.5 99.9 99.6 99.9 99.7 99.9 99.7 99.9 99.9 99.9 99.9 Research 91.5 94.7 96.5 97.5 98.2 98.7 99.0 99.2 99.4 99.5 99.7 99.9 Associate 96.1 98.2 99.2 99.6 99.8 99.9 99.9 99.9 99.9 99.9 99.9 99.9 Business 90.5 94.9 97.3 98.6 99.3 99.7 99.9 99.9 99.9 99.9 99.9 Overall 94.2 96.8 98.1 98.8 99.2 99.4 99.6 99.7 99.7 99.8 99.9 99.9 Maximum Simultaneous Users (100%) 18 55 24 66 35 23 66 promote one or more of these databases or, at least, make these databases easier to find. Of the libraries that responded to the survey, 66 percent note these specific da­ tabases on the library system’s main menu, nearly 20 percent post signs that promote them, 60 percent provide handouts that de­ scribe them, and 82 percent offer training that specifically mentions this company’s databases (table 8). After a user logs in or sits down at a workstation, just one to three steps are required to reach the databases in at least 87 percent of the libraries (data not shown). TABLE 3 Simultaneous Use for Magazines and Journals for Academic Libraries No. of Users Cumulative Percentage Type of Institution (Carnegie Class) Bacc. Master's Doctoral Research Associate Business Overall 1 95.2 89.2 92.3 82.8 91.7 85.6 89.1 2 97.9 93.0 96.1 87.7 95.6 92.0 93.1 3 99.1 95.4 97.9 91.0 97.7 95.9 95.4 4 99.6 96.9 98.8 93.3 98.9 98.1 96.9 5 99.8 97.8 99.3 94.9 99.4 99.1 97.8 6 99.9 98.4 99.6 96.1 99.7 99.6 98.4 7 99.9 98.8 99.8 97.0 99.8 99.8 98.8 8 99.9 99.0 99.9 97.7 99.9 99.9 99.0 9 99.9 99.2 99.9 98.1 99.9 99.9 99.2 10 99.9 99.3 99.9 98.4 99.9 99.9 99.3 15 99.9 99.6 99.9 99.0 99.9 99.9 99.6 25 99.9 99.6 99.9 99.8 Maximum Simultaneous Users (100%) 18 55 24 66 35 22 66 Patterns of Database Use in Academic Libraries 243 TABLE 4 Simultaneous Use for the Most-Used Database (General Magazine/Journal) No. of Users Cumulative Percentage Type of Institution (Carnegie Class) Bacc. Master's Doctoral Research Associate Business Overall 1 93.8 84.3 87.6 76.4 86.7 85.3 84.7 2 97.4 89.6 93.3 82.1 92.6 92.2 90.0 3 98.9 93.1 96.2 86.2 96.0 96.1 93.2 4 99.5 95.3 97.8 89.3 98.1 98.1 95.2 5 99.7 96.6 98.7 91.6 99.1 99.1 96.5 6 99.9 97.5 99.2 93.4 99.5 99.5 97.4 7 99.9 98.0 99.6 94.8 99.7 99.7 97.9 8 99.9 98.3 99.8 95.8 99.8 99.9 98.4 9 99.9 98.6 99.9 96.5 99.9 99.9 98.6 10 99.9 98.8 99.9 97.0 99.9 99.9 98.8 15 99.9 99.3 99.9 98.2 99.9 99.9 99.3 25 99.7 99.2 99.9 99.7 Maximum Simultaneous Users (100%) 18 55 24 66 31 22 66 Close ties with academic classes also should influence how often a database is used. Three-quarters of the libraries said that all of this company’s databases they subscribe to have subject matter related to academic classes and all of the librar­ ies said that at least some are class related. Although many librarians do not know for sure if the databases are mentioned in academic classes, 66 percent reported that they are mentioned specifically and 58 percent said that specific class assign­ ments require use of the databases. Although database use undoubtedly is influenced somehow by library policies, none of these factors proved to be statis- TABLE 5 Simultaneous Use for Business Directories and Periodicals No. of Users Cumulative Percentage Type of Institution (Carnegie Class) Bacc. Master's Doctoral Research Associate Business Overall 1 99.3 97.4 95.8 95.2 98.7 84.2 96.7 2 99.9 99.1 98.7 98.1 99.6 90.9 98.7 3 99.9 99.7 99.6 99.4 99.9 94.8 99.5 4 99.9 99.9 99.8 99.8 99.9 97.1 99.8 5 100.0 99.9 99.9 99.9 99.9 98.5 99.9 6 99.9 99.9 99.9 99.9 99.4 99.9 7 99.9 99.9 99.9 99.9 99.7 99.9 8 99.9 99.9 99.9 99.9 99.8 99.9 9 99.9 99.9 99.9 99.9 99.9 99.9 10 99.9 99.9 99.9 99.9 99.9 99.9 15 99.9 99.9 99.9 99.9 99.9 99.9 25 99.9 99.9 99.9 Maximum Simultaneous Users (100%) 5 28 20 29 17 23 29 244 College & Research Libraries May 2000 TABLE 6 Number of Simultaneous Users for 99 Percent Coverage No. of Students No. of Libraries No. of Simultaneous Users Enrolled in Enrollment Group Needed for 99 Percent Coverage 1-2,499 13 2 2,500-4,999 15 3 5,000-7,499 16 4 7,500-9,999 7 4 10,000-14,999 15 5 15,000-19,999 7 5 20,000-24,999 7 7 25,000-44,999 8 7 45,000-89,999 2 4 90,000 or ,ore 2 27 tically correlated to amount of use. In or­ der to take into account the size of the stu­ dent population in statistical testing, a variable for average use per 10,000 stu­ dents was created. Because these new data normally were not distributed, the nonparametric Mann-Whitney and Kruskal-Wallis tests were employed to test for effects on usage. The independent variables tested included: availability of remote log-in, number of workstations provided, whether the aggregator ’s da­ tabases were noted on the main menu, number of steps required to reach the databases, whether signs were posted, whether handouts were available, and percentage of workstations that allowed the company’s databases to be searched. Conclusions Knowing when and how different types of databases are likely to be used in any library will help librarians determine ap­ propriate levels for simultaneous use con­ tracts, optimum number of usage ports, and how staffing can best be assigned. It will come as no surprise to academic ref­ erence librarians that peak online usage follows clearly defined patterns. The greatest number of users is online early in the week, at midday, in the month when term papers are due. A majority of TABLE 7 Electronic Media for End Users Media Types Used by Respondents Percentage Type of Institution (Carnegie Class) Bacc. Master's CD-ROM 87.5 100.0 Locally loaded 12.5 31.8 Loaded on another library's computer 75.0 31.8 Commercial online from a vendor 100.0 95.5 Commercial online from an OPAC company 25.0 27.3 World Wide Web 87.5 100.0 Doc. 100.0 33.3 50.0 100.0 0.0 83.3 Res. 100.0 63.6 36.4 100.0 27.3 90.9 Assoc. 85.7 57.1 28.6 71.4 14.3 100.0 Bus. 100.0 33.3 33.3 100.0 33.3 33.3 Overall 96.5 38.6 40.4 94.7 22.8 91.2 Patterns of Database Use in Academic Libraries 245 TABLE 8 Ways Libraries Influence Database Use Influential Factors Percentage Type of Institution (Carnegie Class) Bacc. Master's Doc. Res. Assoc. Bus. Overall Remote access 62.5 85.7 100.0 81.8 28.6 66.7 75.0 Workstations dedicated to these databases 12.5 0.0 0.0 0.0 14.3 33.3 5.4 Databases noted on library system's main menu 87.5 66.7 50.0 54.5 85.7 33.3 66.1 Signs 25.0 33.3 0.0 0.0 28.6 0.0 19.6 Handouts 37.5 71.4 50.0 50.0 71.4 66.7 60.0 Databases mentioned specifically in training 75.0 90.5 83.3 81.8 57.1 100.0 82.1 Databases mentioned specifically in academic classes 75.0 75.0 66.7 27.3 71.4 100.0 65.5 Specific class assignments require use of these databases 50.0 65.0 33.3 54.5 57.1 100.0 58.2 academic users are accessing databases at the time they typically use the library. For all types of academic libraries, there are clear valleys and peaks for online use, following the rhythms of academic life. Even for general interest magazine da­ tabases that are available both in-house and through remote access, often no one is online. Peak usage can be quite high in some libraries, but average usage for any one database or group of similar data­ bases is low. Students may be in chat rooms or surfing the Net at all hours, but few are likely to be searching research databases. Although none of the variables the re­ searchers tested were found to be statis­ tically significant predictors of amount of use, a combination of factors may influ­ ence use, many of which are difficult to measure or capture on a questionnaire. Librarians influence use in a variety of subtle and obvious ways, including men­ tioning specific products in a user instruc­ tion class, advocating use of a specific da­ tabase in specific class assignments, re­ ferring to a database on a library’s wel­ come screen, or otherwise reminding us­ ers about a specific database. Although no single reference database may be in use by a large number of simultaneous users day or night, making sure that databases are available most of the time when users need them, and instituting policies that make this access easy, is a responsibility of all academic libraries. Notes 1. American Library Association, Electronic Services in Academic Libraries (Chicago: ALA, 1997). [Executive summary is available from www.ala.org/alaorg/ors/elecsvcs.html.] 2. John E. Tolle et al., “Determining the Required Number of Online Catalog Terminals,” Information Technology & Libraries 2 (Sept. 1983): 261–65. 3. Marjorie E. Murfin, “National Reference Measurement: What Can It Tell Us about Staff­ www.ala.org/alaorg/ors/elecsvcs.html 246 College & Research Libraries May 2000 ing,” College & Research Libraries 44 (Sept. 1983): 321–33. 4. William E. McGrath, “Periodicity in Academic Library Circulation: A Spectral Analy­ sis,” Journal of the American Society for Information Science 47 (Feb. 1996): 136–45. 5. Bruce A. Leach, “Identifying CD-ROM Use Patterns as a Tool for Evaluating User In­ struction,” College & Research Libraries 55 (July 1994): 365–71. 6. Thomas A. Lucas, “Time Patterns in Remote OPAC Use,” College & Research Libraries 54 (Sept. 1993): 439–45. 7. Thomas A. Peters, “Remotely Familiar: Using Computerized Monitoring to Study Re­ mote Use,” Library Trends 47, no. 1 (summer 1998): 7–20. 8. Anne-Marie Belanger and Sandra D. Hoffman, “Factors Related to Frequency of Use of CD-ROM: A Study of ERIC in an Academic Library,” College & Research Libraries 51 (Mar. 1990): 153–63. 9. Judith A. Adams and Sharon C. Bonk, “Electronic Information Technologies and Re­ sources: Use by University Faculty and Faculty Preferences for Related Library Services,” Col­ lege & Research Libraries 56 (Mar. 1995): 119–31. 10. Charles T. Townley and Leigh Murray, “Use-based Criteria for Selecting and Retaining Electronic Information: A Case Study,” Information Technology and Libraries 18 (Mar. 1999): 32– 39. 11. Carol Tenopir and Lisa Ennis, “The Digital Reference World of Academic Librarians,” Online 22 (July/Aug. 1998): 22–28. 12. ———, “The Impact of Digital Reference on Librarians and Library Users,” Online 22 (Nov./Dec. 1998): 84–88. 13. Eileen Abels, “Pricing of Electronic Resources: Interviews with Three Vendors,” Journal of the American Society for Information Science 47 (Mar. 1996): 235–46. 14. Carol Tenopir, George Banks, and Leslie Preston, “Pricing Options for End User Prod­ ucts in Libraries,” in Proceedings of the 19th Annual National Online Meeting, New York, May 12–14, 1998, (Medford, NJ: Information Today, 1998): 419–32. 15. Carol Tenopir and Danielle Green, “Patterns of Use and Usage Factors for Online Data­ bases in Academic and Public Libraries,” in Proceedings of the 62nd Annual Meeting of the Ameri­ can Society for Information Science, Washington, D.C., October 31–November 4, 1999, (Medford, NJ: Information Today, 1999): 616–27. 16. Carol Tenopir, “Factors That Influence Database Use,” in Proceedings of the Ninth Na­ tional Conference of the Association of College and Research Libraries, Detroit, April 8–11, 1999, (Chi­ cago: American Library Association, 1999): 68–75. 17. Carol Tenopir and Danielle M. Green, “Simultaneous Usage of Online Databases in Aca­ demic and Public Libraries,” in Proceedings of the 20th Annual National Online Meeting, New York, May 18–20, 1999, (Medford, NJ: Information Today, 1999): 459–68.