College and Research Libraries Research Notes Time Patterns in Remote OPAC Use Thomas A. Lucas Expanded hours of access and the reduction of peak system loads are often cited as advantages of remote access to online public access catalogs (OPACs). This argument is based on the assumption that remote users search OPACs when libraries are closed or when there is low internal use. The author tested this assumption by performing a transaction log analysis of the remote and internal use of the OPAC of the Research Libraries of the New York Public Library. The analysis showed that patterns of remote and internal use differed greatly. A large part of remote searching occurred when the Research Libraries were closed. Compared to internal searching, remote searching was distributed more evenly over the course of the day and the week. The study shows that remote access expands the hours of use of the catalog and has the potential to reduce peak system loads at the Research Libraries. he literature on online public access catalogs (OPACs) cites many potential benefits of re- mote access to these systems.1 Advocates maintain that remote access expands the hours of access to the cata- log by allowing users to search it when the library is closed. They also argue that remote access distributes use over a greater period of time, thus reducing peak system loads and improving re- sponse time.2 These arguments are, however, based on an assumption about user behavior. The assumption is that remote users search OPACs when librar- ies are closed or when there is low inter- nal use. If, however, remote users search at the same times as internal users, the two benefits described above obviously are not being realized. Although OPAC use studies now con- stitute a large literature, only two authors have examined the time patterns of remote use in any detail. One is Sally W. Kalin, who studied the remote use of Penn State's OPAC for one week in 1984. Kalin found that remote searching was heaviest between 2:30 and 5:00p.m. She also tallied remote searches by day of the week, and found that they peaked on Wed- nesday, and were lowest on Saturday and Sunday. However, she did not attempt to compare patterns of remote and internal use. Moreover, remote access was not available at Penn State when the library was closed.3 Thus Kalin's study does not reveal whether remote users at Penn State searched during hours of low internal use. During an eleven-month period in 1988-89, Thomas A. Peters studied re- Thomas A. Lucas is Supervising Librarian in the General Research Division of the New York Public Library, Fifth Avenue and 42nd Street, New York, New York, 10018. 439 440 College & Research Libraries mote use of the OPAC at the University of Missouri at Kansas City. Peters found that remote use was heaviest on week- day afternoons, and that there was com- paratively little use of remote access in the evenings and on weekends. He com- pared remote use to total system use by day of the week and found that in both cases use was highest on weekdays. Both remote use and total use were much lower on weekends. Peters did not com- pare remote and internal use by hour of day. As at Penn State, remote access to the OPAC at Kansas City was not avail- able when the library was closed.4 Two other studies treat the time pat- terns of remote use only in passing. Ber- nard G. Sloan, commenting on seven years of experience with remote access to the Statewide Library Computer Sys- tem (LCS) in Illinois, notes that "remote users access a system more frequently on evenings and weekends than do users of on-site public access terminals." Lynn L. Magrath reported in 1989 that most re- mote users in Colorado's Pikes Peak Li- brary District searched the Maggie III OPAC between 4:00 p.m. and midnight. Neither Sloan nor Magrath attempts a fuller comparison of remote and internal use, nor do they indicate whether remote access was available when their libraries were closed.5 RESEARCH QUESTIONS With the partial exception of Peters's and Sloan's work, these pioneering stu- dies of remote use do not address the following questions: How do patterns of remote and internal use compare? Do remote users search during periods of low internal use? Do they search when the library is closed? The present study attempts to answer these questions for one institution: the Research Libraries of the New York Pub- lic Library (NYPL). NYPL is a private corporation founded in 1895 by the merger of the Astor and Lenox Libraries with the Tilden Trust. NYPL operates four Research Libraries and eighty-two Branch Libraries in Manhattan, the Bronx, and Staten Island. (Brooklyn and Queens have separate library systems.) September 1993 The Research Libraries hold thirty-eight million items in four locations: the Cen- tral Research Library in Midtown Man- hattan, the New York Public Library for the Performing Arts at Lincoln Center, the Schomburg Center for Research in Black Culture in Harlem, and the Library at West 43rd Street. The Research Librar- ies serve 1.4 million people a year in person, 337,000 by telephone, and many others by mail. This study exclusively concerns the Research Libraries. Advocates maintain that remote access expands the hours of access to the catalog by allowing users to search it when the library is closed. In January 1992, the Research Librar- ies replaced an existing OPAC with an INNOPAC system. The name of the cat- alog, CATNYP, was retained. CATNYP contains 1.6 million catalog records dat- ing from 1972 to the present. Two peri- odical indexes have also been mounted on the system. CATNYP has one hundred ports which, at the time of the study, served all four locations of the Research Libraries, remote users, and the Graduate Center of the City University of New York (CUNY). During the eight- week period of the study, the system recorded 293,549 user-keyed searches, an average of 7,527 each day that the Central Research Library was open. Because the Research Libraries are lo- cated at the heart of a metropolitan area of eighteen million people, the process of making remote access to CATNYP avail- able has progressed gradually. During the study period, remote access was pro- vided through a ~ingle dial-access port. This port was available twenty-four hours a day, seven days a week. The dial-access number was released on a selective basis to librarians, students, and faculty mem- bers at three universities in New York. The dial-access number was not released to the general public, nor was its use promoted in anyway. These limitations on remote access ob- viously affected the outcome of this study. If more ports had been available to remote users, and if the dial-access number had been widely disseminated and promoted, remote use undoubtedly would have been much higher. Neverthe- less, the remote use that did occur showed a clear time pattern. The total number of remote searches during the eight weeks of the study was 3,891, an average of 69 per day. These remote searches consti- tuted 1.5 percent of all public searches, and 1.3 percent of total system use. The present study undertakes to an- swer the following four questions. First, did remote users search CATNYP during days when the Research Librar- ies were closed? Second, how did the daily patterns of remote and internal use compare? Third, did remote users search CATNYP during hours of the day when the Research Libraries were closed? Fi- nally, how did the hourly patterns of remote and internal use compare? METHODOLOGY The research method used in the study was a transaction log analysis. Release 7 of INNOPAC compiles a detailed trans- action log which is retained on the sys- tem for ten days. During the present study, the transaction log data were printed each Monday for eight weeks. The data were then transferred to Micro- soft Excel version 4.0 for further compi- lation and analysis. The eight-week period of the study began at 11:00 a.m. on May 4, 1992, and ended at 11:00 a.m. on June 29, 1992. May and June are months of somewhat lower than average use in the Research Libraries. Still, use was substantial during the study period, and showed a clear pattern. The INNOPAC transaction log identi- fies the port at which each search is entered. During the study period, each port was assigned to one of four classes of users: internal public, remote public, staff, and CUNY. Thus it was possible to separate the searches made by each of these user groups. The present study is concerned only with searches done by the internal public and the remote pub- lic. Staff and CUNY searches are not con- sidered. During the study period there Time Patterns in Remote OPAC Use 441 were 259,088 internal public searches and, as mentioned above, 3,891 remote public searches. FINDINGS Remote Use by Day of Week The first question addressed was whether remote users searched CAT- NYP during days when the Research Li- braries were closed. During the study period, the Central Research Library, where 89 percent of all internal use of CATNYP occurred, was open from Tuesday through Saturday. Central was closed on Sunday and Monday. The New York Public Library for the Performing Arts, where 6 percent of internal use oc- curred, was open on Monday and from Wednesday through Saturday. Perform- ing Arts was closed on Sunday and Tuesday. The Schomburg Center for Re- search in Black Culture, where 5 percent of internal use occurred, was open on Tuesday, Wednesday, Friday, and Satur- day. Schomburg was closed on Sunday, Monday, and Thursday. The Library at West 43rd Street, where less than 1 percent of internal use occurred, was open from Tuesday through Saturday and closed on Sunday and Monday. Thus 94 percent of all internal searches were done in Central, Schomburg, and the Library at West 43rd Street, buildings which were closed on Sunday and Monday. Figure 1 shows that there was substan- tial remote use on Sunday and Monday, when most divisions of the Research Li- braries were closed. More than 12 per- cent of remote searches were done on Sunday, and over 14 percent on Monday. Remote searching on Sunday and Mon- day was only slightly below the level that would have resulted if remote searches had been evenly divided among the seven days of the week. Sun- day and Monday together constitute two-sevenths, or 29 percent of the week, and 27 percent of remote searches oc- curred on these days. Thus it is clear that remote patrons made good use of CAT- NYP on days when the Research Librar- ies were closed. When the daily patterns of remote and internal searching are compared, the 442 College & Research Libraries S~ptember 1993 s 600 t------ - --r=========;----i e 500 +------ a r400 c3oo h e200 s 100 0 Sunday Monday Tuesday Wednesday Thursday I• Library closed 0 Library open FIGUREl Friday Remote Use of CATNYP by Day of Week (n = 3,891) p e S 20% +---------- ---1 r e c a 15% +--------- e r n c t h 10% e 0 s 5% f 0% Sunday Monday Tuesday Wednesday Thursday • Remote (n = 3,891) 0 Internal (n = 259,088) FIGURE2 Friday Remote and Internal Use of CATNYP by Day of Week Saturday Saturday differences are readily apparent (see New York Public Library for the Per- figure 2). Internal use was concentrated, forming Arts, the only Research Librar- necessarily, in five days of the week, ies location open on that day. In addition, with particularly heavy use on Tuesday, some staff members in the Central Re- Wednesday, and Saturday. Twenty-three search Library used the public terminals percent of all internal searching oc- on Mondays.) curred on Tuesday, and 22 percent on Remote use, on the other hand, was Wednesday. On Sunday, by contrast, distributed more evenly throughout the there was no internal searching, and week. During the eight-week study pe- only 3 percent of internal searching oc- riod, the heaviest remote searching oc- curred on Monday. (These Monday curred on Thursday, with 18 percent of searches are mainly attributable to the the total, and the lightest on Friday, with Tune Patterns in Remote OPAC Use · 443 350 .------------------------------------------------------------- S300 r-----------------------------------~~ e 250 t-------------- -----,--.--l a r 200 t---------------------------4 c 150 h e 100 s 50 0 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 A.M. P.M. I• Library closed 0 Library open FIGURE3 Remote Use of CATNYP by Hour of Day (n = 3,891) 12 percent of the total. A Kruskal-Wallis multiple comparison procedure was performed on the data to determine whether the differences in the number of remote searches per day were the result of chance. The procedure showed that there was no significant difference (p = 0.25) in the number of remote searches per day.6 Remote Use by Hour of Day The hours of the Research Libraries vary by division, but most divisions in the Central Research Library, where 89 percent of all internal use of CATNYP occurred, have similar hours. During the study period, most divisions in Central were open from 11:00 a.m. to 7:30p.m. on Tuesday, from 11:00 a.m. to 6:00 or 7:30 p.m. on Wednesday (depending on the division), and from 10:00 a.m. to 6:00 p.m. on Thursday through Saturday. The New York Public Library for the Per- forming Arts, where 6 percent of internal use occurred, was open from noon to 8:00 p.m. on Monday and Thursday, and from noon to 6:00 p .m. on Wednesday, Friday, and Saturday. The Schomburg Center for Research in Black Culture, where 5 percent of internal use occurred, was open from noon to 8:00 p.m. on Tuesday and Wednesday, and from 10:00 a.m. to 6:00p.m. on Friday and Saturday. The Library at West 43rd Street, where less than 1 percent of internal use oc- curred, was open from 9:00 a.m. to 5:00 p .m. on Tuesday through Saturday. Thus, in the Central Research Library, where most internal use occurred, CAT- NYP was available about eight hours a day, five days a week. Remote access, by contrast, was available twenty-four hours a day, seven days a week. Figure 3 reveals that remote users did a large part of their searching during hours when the Central Research Li- brary was closed. Fully 48 percent of remote searches occurred either before 10:00 a.m. or after 6:00 p.m.-hours when the Central Research Library was usually closed. (In figure 3, the usual schedule for the Central Research Li- brary was used to show the hours when the Research Libraries were open.) Figure 3 shows that there were three major peaks in remote searching during the day: the first in late morning, the second in late afternoon, and the third in late evening. The intervening valleys correspond to mealtimes. (Lunch and dinner are generally eaten later in New York than in other parts of the country.) Remote searching declined to a low level after 1:00 a.m., but there was some 444 College & Research Libraries September 1993 16% p s 14% e e 12% - f- r- - r- r a 10% - r- r- - 1- o r 8% c f c - r- r- - e h 6% n 4% t e s 2% 0% - r- - I• h I II • ill I II .I .I .• .• .I .- .• .I .1.. 11--,.1 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 A.M. P.M. I• Remote (n = 3,891) 0 Internal (n = 259,088) I FIGURE4 Remote and Internal Use of CATNYP by Hour of Day searching throughout the night. The re- mote users confirmed New York's repu- tation as the city that never sleeps. Figure 4 shows that the hourly pat- terns of remote and internal use differed greatly. Internal searches were bunched together, with 82 percent occurring during the six-hour period from 11:00 a.m. to 5:00 p.m. The heaviest hour for internal search- ing was from 2:00 to 3:00 p.m., when nearly 16 percent of all internal searches occurred. Remote use, by contrast, was spread more evenly throughout the day. The busiest hour for remote searching was from 4:00 to 5:00p.m., when 8 percent of remote searches occurred. DISCUSSION It is clear that at the Research Libraries of the New York Public Library, patterns of remote and internal use of the OPAC differ greatly. Remote users do a large part of their searching during days and hours when the Research Libraries are closed. Thus the study shows that re- mote access expands the hours of use of the catalog. This benefit is of particular importance to a library such as NYPL, which has comparatively limited hours. The study also shows that remote searching of CATNYP is distributed more evenly over the course of the day and the week than internal searching. This finding suggests that if internal users could be persuaded to do some of their searching from remote locations, a reduction in peak system loads might result. On the other hand, it is possible that remote access would simply in- crease the total use of the system, without reducing internal use. Further studies of this issue are needed. One ap- proach to the problem would be to ex- amine total system use before and after remote access is introduced. Another ap- proach would be to survey internal users, asking them whether they would do less searching in the library if remote access were available. Those users who indicated a willingness to switch to re- mote access might then be asked when during the day and the week they would be likely to do remote searching. The results of such studies would be valuable to librarians who must con- vince their administrators and funding agencies of the benefits of remote access in these times of financial exigency. These studies would be particularly wel- come at a time when community infor- mation systems and freenets are receiving increasing scrutiny. The present study should also be rep- licated and extended in other types of libraries, particularly academic libraries and libraries offering Internet access. Such studies would involve different clienteles, and would help to determine whether the results of the present study are indicative of remote use in libraries generally. Further studies may confirm that re- mote access is a way of expanding access to OPACs without the expense of in- Time Patterns in Remote OPAC Use 445 creasing building hours. Future studies may also show that remote access is a means of spreading the load on OPACs over a greater number of hours, thus improving response time. As libraries in- crease the demand on their OPACs by mounting additional databases on them, these benefits of remote access can only grow in importance. REFERENCES 1. See, for example, Sally W. Kalin, "Support Services for Remote Users of Online Public Access Catalogs," RQ 31 (Winter 1991): 198-99. 2. Thomas A. Peters, The Online Catalog: A Critical Examination of Public Use (Jefferson, N.C.: McFarland, 1991), 197. 3. Sally W. Kalin, "Remote Access to Online Catalogs: A Public Services Perspective," in Second National Conference on Integrated Online Library Systems, Proceedings, Sept. 13 and 14, 1984, Atlanta, Georgia, ed. David C. Genaway (Canfield, Ohio: Genaway, 1984), 208-9; Sally W. Kalin, telephone conversation with the author, July 27, 1992. 4. Peters, The Online Catalog, 170, 199, 201-5. 5. Bernard G. Sloan, "High Tech/Low Profile: Automation and the 'Invisible' Patron," Library fournal111 (Nov. 1, 1986): LC 6; Lynn L. Magrath, "The Public and the Computer: Reactions to a Second Generation Online Catalog," Library Trends 37(Spring 1989): 535. 6. W. J. Conover, Practical Non parametric Statistics, 2d ed. (New York: Wiley, 1980), 21~31.