College and Research Libraries Patterns of Searching and Success Rates in an Online Public Access Catalog Sammy R. Alzofon and Noelle Van Pulis A survey of 430 users of the online catalog (LCS) at Ohio State University Libraries found that most users are undergraduate students (68 percent) and that the majority of users (95 percent) choose the online catalog as their first source of information. Of the users surveyed, fewer per- formed known-item searches than previous catalog use studies have shown, and their success rates for these searches generally were higher than reported in earlier studies. Success rates also suggest that there is a group of online catalog users who are highly motivated to seek out instruction and learn to use the catalog well. n July 1, 1982, Ohio State Uni- versity Libraries announced the closing of the card catalogs, ter- minating the dual catalog sys- tem which had existed since 1975, when the first online public access terminal be- came available. In anticipation of the change to a single, online catalog the li- braries conducted an intensive informa- tional and instructional program, and had in place more than 115 public-use termi- nals. Because the few existing studies of OSU' s online catalog are either quite lim- ited in scope or predate the changeover by several years, the closing of the card cata- logs raised a number of questions regard- ing the use of the Library Control System (LCS) as an online catalog. 1 The questions which this study addressed are: • What are the characteristics of the on- line catalog users? • Are LCS users aware that the card cata- logs are closed? • Do patrons use LCS first, in preferel'lce to the card catalogs? Do they check the card catalogs if a desired item is not found in LCS? • What types of searches are performed in the online catalog and how successful are they? How do search patterns and success rates compare to previous cata- log use studies? • What effect, if any, do length of experi- ence and group instruction have on search success? At this point, it is appropriate to de- scribe briefly the online catalog . at Ohio State University. THE ONLINE CATALOG The online catalog at OSU evolved over a period of more than twelve years. 2 LCS first became operational in 1970 as a circu- lation system for a large, decentralized li- brary. As such, it was a command-driven system designed for staff use only, which allowed remote searching and charging using brief records converted from the shelflist. In 1975 the first public-access ter- minal was made available and in June 1978 the capability for full MARC storage was added. Eventually, full bibliographic rec- ords became available in LCS for OSU ti- tles added since 1974. In addition, in 1979 Sammy R. Alzofon and Noelle Van Pulis presented an earlier version of this article, titled "Patron Success in Searching an Online Catalog," at the 12th ASIS Mid- Year Meeting, May 1983. Both authors are affiliated with Ohio State University, Columbus. 110 full records were added for all State Li- brary of Ohio books. The system also lists OSU titles on order or in process, as well as some special microform collections. Access to all records is available by au- thor, title, and call number. Access by added entries and subject headings is available for OSU titles added since 197 4 and for all State Library of Ohio titles. As indicated in the list of search commands in figure 1, all searches require a three-letter command, followed by full words or a search key. This list includes only those commands for which patrons are given in- structions. OSU maintained a dual catalog system, card and online, until July 1982. In prepa- ration for the changeover to the online cat- alog, a campus-wide informational and in- structional program was undertaken. The program was designed to perform several functions: inform faculty and students of the changeover (i.e., the freezing of the card catalogs), promote the use of LCS, al- lay fears about rumored removal of the card catalogs, and provide an avenue of voluntary instruction for anyone who needed it. The opportunity for instruction was considered especially important with the changeover. Patrons now had no choice in looking for the latest material. Since 1979, all freshmen have been receiving basic LCS training through a required library in- struction program. A workshop, intended primarily for upper classmen and gradu- ate students, has been offered since Janu- ary 1980. Brief printed instructions have LCS Initial Search Commands Patterns of Searching 111 been available at the terminals since 1975, and a longer manual since 1980.3 Online help displays, still under development, were first added to LCS in 1982. Given this variety of options for learning to use LCS, the study was designed to identify how many of the users surveyed had had group instruction and what effect, if any, this instruction had on search success. METHODOLOGY The Main Library of the Ohio State Uni- versity Libraries system was chosen as the survey site, as there are twenty-four pub- lic access terminals in a large open area near the Circulation Department as well as three terminals in the union card catalog area. The survey questionnaire was dis- tributed from 9:00 a.m. until 5:00 p.m. over a period of four days in November 1982 to five hundred patrons who were asked to supply the following information while they worked at the terminals: col- lege major or subject discipline, sex, uni- versity status, length of use of LCS (weeks, months, year or more), and whether, if the item searched is not found in LCS, the patron will check the card cata- log. The next section asked what the pa- tron was looking for (known item and/ or subject), if it was found and, if so, what the call number was, and what commands were used for the online search. Subsequent questions were: Had the patron tried to locate any of the desired items in the card catalog before using LCS? Had the patron received LCS in- struction in a user education class or li- Author AUT/ name, all or part Title TLS/ 4 + 5 search key Author and title ATS/4 + 5 search key Subject SUB/ subject heading, all or part Call number DSC/ call number, specific item Call number SPS/ call number, shelf position (browsing) LCS Secondary Commands and Options Short record DSL/line number Full record FBL/line number List of authors TBL/line number or subjects Serial modifier TLS/ 4 + 5 /SER ATS/4 + 5/SER FIGURE 1 (follows AUT/ or SUB/) (limits search to journals, magazines, and other serials) LCS Public Search Commands 112 College & ·Research Libraries brary workshop? Was the patron aware that only LCS lists newly acquired books? Comments also were solicited. At the end of the survey period, all the questionnaires were checked, replicating the search patterns the patron specified, and 430 questionnaires were judged to be complete enough for analysis. Although the questionnaires included responses about subject searching, this report fo- cuses on known-item searches. RESULTS AND DISCUSSION User Characteristics In response to the first group of ques- tions, it was found that characteristics of the participants are similar to those re- ported in previous studies (see table 1). Most catalog users are undergraduates and, in this survey, 68 percent identified themselves as such. Another 26 percent were graduate students ." Only 3 percent were faculty, while another 3 percent did not give their status or were not OSU stu- dents or faculty. The results of the Council on Library Resources online catalog use study indicate that most online catalog us- ers are male. 5 In this survey, 47 percent marked this category. Another 32 percent identified themselves as female and 21 percent chose not to indicate their sex. Nearly half (49 percent) of the respon- dents indicated they have used LCS for a year or longer. Another 23 percent have used LCS for several months, and 28 per- cent for less than a month. With respect to their awareness of the changeover to an online catalog, 75 percent indicated they were aware that only LCS lists newly March 1984 added titles. That 95 percent did not check the card catalog before using LCS indi- cates a definite preference for or willing- ness to use the online catalog as a first source of information. However, 58 per- cent said that if their Lcs· search failed, they would try the card catalog. This sug- gests distrust of LCS or lack of confidence in the searcher's own ability to use it suc- cessfully. The results of the question regarding group instruction were of special interest. Of the responses analyzed, 35 percent had had LCS instruction in the undergraduate user education program or other class, and another 13 percent had attended a library-sponsored workshop. This is nearly half of the online catalog users who were surveyed. Search Patterns Previous card catalog studies have .shown that users prefer to search by au- thor rather than title, even when both pieces of information are available. LCS search commands permit the use of either one or a combination of both. As shown in table 2, of those indicating the command used, only 18 percent searched by title. Another 14 percent used author only and 16 percent used a combination of author and title. Thus, at least 30 percent used some form of author access. Altogether, known-item searching represents 48 per- cent of the total numoer of searches per- formed. Subject searching represents 35 percent of the total. This result is similar to previous card catalog use studies but is lower than the results of the CLR study. 6 If TABLE 1 CHARACTERISTICS OF ONLINE CATALOG USERS Sex (N = 430) Status (N = 427) Male 47% Undergraduate student 68% Female 32% Graduate student 26% Unspecified 21% Faculth 3% Other Unspecified 3% Len~th of use (N = 422) GrouE instruction (N=418) Month or less 28% Class 35% Several months 23% Workshop 13% Year or more 49% Neither 52% New titles only in LCS (N = 410) Use card catalo~ Yes 75% Before LCS (N = 421) 5% No 25% After LCS (N=410) 58% N = Nurnber of responding online users . Patterns of Searching 113 TABLE 2 SEARCH PAITERNS AND SUCCESS Patterns Success Type of Search N % %-U %K N % Author 96 14 17 30 74 77 Author and title 112 16 19 34 95 85 Title 124 18 22 37 100 81 Known-item 332 48 58 101 269 81 Subject 244 35 42 165 68 Unspecified 122 18 81 66 Total 698 100 515 74 N=Number of searches performed; %=Percent of number of searches. %- U =Percent of number of searches performed, excluding unspecified. %K=Percentage of the number of known-item searches. Percentages have been rounded to the nearest whole figure . unspecified searches (representing 18 per- cent of the total} are discounted, known- item searches go up to 58 percent, and subject searches increase to 42 percent, a pattern similar to the CLR results and to that reported by Moore in her 1979 study ofLCS. When only known-item searches are considered, patrons show a slight prefer- ence for searching by title, which ac- counted for 37 percent of all known-item· searches. Author-title searches followed at 34 percent, and author searches were the lowest number at 30 percent. 7 Pease and Gouke, in a recent study, found that 70 percent of online catalog searches in the OSU Main Library were for known items. 8 They also found that a much higher per- centage (68 percent) of search approaches were by title. The reasons for the higher percentages of known-item, and specifi- . cally title searches, in the Pease and Gouke study are not known, but might be attributable to their small sample size, to differences in research design and data analysis, or to some other unkn9wn fac- tor. Success Rates The success rates for known-item searches in this study are similar to, or slightly higher than, those reported in previous card and online catalog use stud- ies. Overall, known-item searches had a success rate of 81 percent. The author-title search had the highest success rate (85 percent), and the author search had the lowest rate of 77 percent. This difference might be due to the more ''forgiving'' na- ture of a search key; that is, the user only needs to know, and enter correctly, four letters from a last name and five from the first title word to perform an author-title search, but the author search requires more complete information. The title search success rate of 81 percent is higher than that found by Pease and Gouke. This could be due to some of the factors men- tioned above or to others, such as a larger number of experienced users in this study or to improved LCS system features. It also should be noted that, in this study, most failures to find known items were user failures rather than collection failures (i.e., item not owned). The low success rate of 68 percent for subject searches was not surprising. The LCS subject search requires the use of a correct Library of Congress subject head- ing and, at this point, the online catalog has limited cross-references for the subje~t file. The results of the CLR study indicate that subject searching, while more popu- lar than previous studies have shown, also is problematic for many users. In- depth analysis of the subject searching data from this survey is being completed for a separate study. The analysis of the effect of length of use and group instruction on search success produced some interesting results. Viewed separately, neither of these fac- tors has much effect on search success, but there does appear to be a wider variation in success rates when group instruction is linked to length of experience. As shown in table 3, users who have not had group instruction seem to do as well as, or 114 College & Research Libraries TABLE 3 TYPE OF INSTRUCTION AND SUCCESS RATE FOR KNOWN-ITEM SEARCHES Oass (N=91) Workshop (N=59) Neither (N = 145) 79% 83% 86% N=Number of searches performed . %=Percentage of successful known-item searches . slightly better than, users who had LCS training in a class or workshop. All still hover around the overall success rate of 81 percent for known-item searches. A simi- lar pattern appears with respect to length of experience with the online catalog (see table 4). Users with only weeks of experi- ence appear to be as successful as users with more than a year of experience, with success rates respectively of 81 percent and 82 percent. However, when length of experience was analyzed in conjunction with instruc- tion, it was found that users with a year or more of experience who also had attended 100 "' Gl 95 ..r: ~ 90 0 Jl ~ 85 ~ 80 ~ 75 0 70 Gl 01 65 ~ c: Gl 60 ~ Gl 55 a.. 50