College and Research Libraries Research Notes User Characteristics of Keyword Searching in an OPAC Pat Ensor Keyword and Boolean searching modes are now becoming more commonly available on online public access catalogs ( OPACs), and questions have arisen regarding their use by library patrons. Which patrons use keyword searching, and which do not? This study attempts to begin providing answers to this question in the context of an academic library that uses the Northwestern Online Total Integrated System (NOTIS) online catalog. Ill any Northwestern Online Total Integrated System (NOTIS) li- braries that had their online catalog available prior to the advent of keyword searching tended to look upon the capability as a frill, an advanced function to be taught after other forms of searching. This neglect was en- couraged initially by slow response time for keyword searches, by questions about how many people could use key- word simultaneously, and by the complex- ities of keyword. This unenthusiastic response to keyword searching was not, indeed, limited to NOTIS libraries; librar- ians at other institutions approached it in a similarly cautious manner. 1 After the keyword mode became available at Indiana State University (ISU) Libraries, questions began to arise. The percentage of searches performed in keyword mode rose steadily, from 15.6% in November 1988, to 21.4% in Novem- ber 1989, when it leveled off. Did certain users prefer keyword searching all the time and not use the other modes? Who tended to use keyword searching more? The author studie.d patron percep- tions and demographics related to key- word searching on NOTIS to try to answer some of the questions posed above. The guiding thesis of the study is: The use or nonuse of keyword searching on LUIS is related to variables such as age, computer experience, subject area, status, and frequency of searching the OPAC. (The full project report, submit- ted to ERIC, details other aspects of pa- tron keyword searching.)2 The findings of this study represent users' early reception of keyword search- ing. Future studies could compare these findings to similar data collected about keyword searching and user reactions to proposed OPAC features. The University. ISU has approxi- mately 9,000 undergraduate and 2,000 graduate students. A few doctorates are Pat Ensor is Coordinator of the Electronic Information Services at Indiana State University Libraries, Terre Haute, Indiana 47809. offered in the fields of education and psychology. Master's degrees are awarded in all schools, including the college of arts and sciences, and the schools of business, education, nursing, technol- ogy, and health, physical education, and recreation. The university has approxi- mately 700 faculty members. The Library. ISU Libraries include a main library, Cunningham Memorial Li- brary, as well as a science library that covers chemistry, biology, and geology. Since March 1985, the ISU Libraries have made the NOTIS online catalog, LUIS, available to the public. It lists more than 99% of the library's holdings, with 1,751,000 bibliographic records. It also includes the holdings of two nearby smaller institutions-Rose-Hulman In- stitute of Technology, an engineering school, and St. Mary-of-the-Woods Col- lege, a liberal arts institution. Keyword Searching. ISU Libraries made the keyword mode of searching available on LUIS in the late spring of 1988; thus, it had been available for al- most two years when this study was con- ducted. Prior to the introduction of keyword/Boolean searching, NOTIS had three modes of searching available: author, title, and subject. Keyword searching on LUIS is exe- cuted in its most basic form by entering "k=[word or phrase]." More elaborate searching may be done using the syntax of BRS search language. LUIS has a se- ries of eight easily accessed keyword help screens. The keyword searching mode is listed on the LUIS welcoming screen, along with author, title, and sub- ject searching options. LITERATURE REVIEW The author did not discover similar attempts to survey patrons extensively about their use of keyword searching on any online catalog. A relevant Council on Library Resources study report ap- peared in November of 1982; Joseph Matthews surveyed users of six com- puter systems in seven libraries, includ- ing the Mankato State University's (MSU's) OPAC, which provided early keyword searching. About 45% of the Keyword Searching 73 searching on MSU' s catalog was key- word searching, as compared with about 19% subject heading searching.3 The overall Matthews report about the mas- sive 29-institution CLR OPAC study re- ports that keyword searching was used frequently when available but was not usually a requested future enhancement where it was not. Systems with the key- word/Boolean feature logged more sub- ject searching and were more successful in known-item searching than systems without keyword.4 The percentage of respondents who learned of the availability of keyword searching from library instruction classes or workshops drops as indi- viduals progress from underclassmen to upperclassmen .,.nd from graduate students to faculty. In 1983, researchers at Bell Labora- tories built two online catalog systems for their library: one was a menu-based system, using a hierarchy based on Dewey Decimal categories, and one al- lowed keyword searching of auth~r, title, and subject heading terms. The key- word system was overwhelmingly pre- ferred (80% of all searches). Keyword searchers tended to do simple one- to two-word searches. 5 The users of this system would, of course, have been quite sophisticated technologically. In 1984, Nancy C. Kranich and others from New York University reported the results of interviews carried out with pa- tronswhoused thelibrary'sGeacOPAC, the card catalog, or both to find informa- tion. Of 34 OPAC users, only one per- formed a keyword search. The researchers believed that the users were not generally aware of this option. The authors hy- pothesized that the users confused key- word and subject heading searches.6 In a study that surveys faculty use of subject searching in card and online catalogs at the University of Houston- University Park, Carolyn Frost found that 27.5% of the faculty used keyword searching "always" or "frequently." She 74 College & Research Libraries noted that "the percentage of frequent users of the keyword search was twice as high among humanities and social sci- ence faculty as it was among science and engineering faculty." 7 An interesting United Kingdom study about retrieval modes for pictorial infor- mation on videodisc shows that 51% of the users prefer keyword modes to browsing and using a joystick. Of the four groups of users, librarianship stu- dents and librarians were twice as likely to do a keyword search as where school children and postgraduate students. 8 METHODOLOGY The Questionnaire The author concluded that a question- naire would be the most useful and fea- sible way to obtain the information desired. Questionnaires have some shortcomings but are more affordable than individual interviews. Because the author wished to have a large sample size to work with, individual interviews were not feasible. Seventy-three percent of all respon- dents had done a keyword search, and more than 20% of those who had not done ~ keyword search planned to do one. The final form of the survey had 27 questions, with one question having ten parts. Questions 1 through 16 were de- signed for both users and nonusers of keyword/Boolean, and the first ten of them were designed to elicit demo- graphic and other user characteristics. Question 17 was aimed at nonusers of keyword/Boolean searching, and the rest were written for users of keyword/Bool- ean searching. (Copies of the question- naire are available from the author.) Questionnaire Administration The author personally administered the questionnaire from the end of Janu- ary 1990 to April1990. Users of the main LUIS terminal cluster on the first floor of the university library were approached January 1992 and asked to fill out the survey, primar- ily during evenings and weekends. Li- brary workers were not approached to answer the questionnaire. An attempt was made to approach anyone using a LUIS terminal during the sampling period, although with only one questionnaire administrator, some users _ were inevitably missed. Even though the campus has a science library, question- naires were not given out there because the collection covers a narrow range of subject areas. Nevertheless, approxi- mately 25% of the respondents were in a science or technology area anyway. The only exception to the above proce- dure was made in an attempt to elicit adequate faculty response. Because few faculty members were doing searches at the main cluster, the author chose ap- proximately 60 faculty in all disciplines that she knew to be LUIS searchers and mailed the questionnaire to them. This ef- fort produced a nearly 50% response rate. . Ultimately, 400 usable questionnaire responses were obtained. The author esti- mates the response rate to have been about 35%. The data retrieved were processed using the Kwikstat statistical program. RESULTS FOR DEMOGRAPIDC AND OTHER USER CHARACTERISTICS Keyword Search Performance Seventy-three percent of all respon- dents had done a keyword search, and more than 20% of those who had not done a keyword search planned to do one. These figures indicate widespread know ledge of the concept of keyword searching and its availability. The user characteristics varied significantly with performance of keyword searching, fu- ture keyword searching plans, or lack of keyword searching. Status Status (type of student, faculty) proved to be a useful factor for classify- ing respondents. The raw numbers and percentages are shown in table 1. Aggre- gate totals are 312 students (78%) and 60 faculty (15%). Analysis using only ISU re- spondents showed meaningful variation with use of keyword searching. In some TABLEt STATUS OF RESPONDENTS Status No. (%) ISU freshman/ sophomore 126 (31.5) ISU junior I senior 114 (28.5) ISU graduate student 49 (12.3) ISU faculty 57 (14.3) ISU staff I administration 5 (1.3) Blank (.3) Rose-Hulman student 21 (5.3) Rose-Hulman faculty 2 (.5) St. Mary's student 2 (.5) St. Mary's faculty (.3) Other 22 (5.5) Percentages do not exactly equal100% because of rounding. TABLE2 KEYWORD SEARCHING USE BY ISU STATUS Haven't Have Used Used Keyword Keyword ISUStatus No.(%) No.(%) Freshmen/ sophomores 112 (88.9) 14 (11.1) Juniors/seniors 96 (84.2) 18 (15.8) Graduate students 36 (73.5) 13 (26.5) Faculty 37 (64.9) 20 (35.1) TABLE3 AGE RANGES Age No. (%) Under 18 3 (.8) 18-22 216 (54) 23-30 60 (15) 31-40 69 (17.3) 41-50 25 (6.3) 51-60 20 (5) 61 or above 5 (1.3) Blank 2 (.5) Percentages do not always equal100% because of rounding. Keyword Searching 75 cases, small numbers in each status make results only indicators. A steady regression in keyword use from ISU underclassmen to faculty exists (see table 2). More than one-third of the ISU faculty responding had not used key- word searching, versus no more than about one-fourth of any group of students. Because other survey results showed that the different groups have not used LUIS for significantly different lengths of time and because freshmen and sophomores used LUIS significantly less frequently, LUIS experience would not seem to account for this difference. One possibility is that faculty are doing more known-item searching. One question in this study asked what re- spondents entered on their last LUIS search. The answers did not vary signif- icantly based on status, but faculty tended to do more author searching and less subject searching. Thirty-six percent of ISU faculty reported doing an author search the last time they used LUIS, ver- sus 13% to 16% of the different student groups. Only 44% of the ISU faculty re- ported searching for a Library of Con- gress subject heading or topic words, versus 54% to 62% of the students. Another possibility is that faculty have not received as much instruction in this area. The percentage of respondents who learned of the availability of key- word searching from library instruction classes or workshops drops as individu- als progress from underclassmen to up- perclassmen and from graduate students to faculty. Almost half ( 48%) ofiSU fresh- men and sophomores learned of keyword instruction classes and workshops, com- pared with 26% of juniors and seniors, 27% of graduate students, but only 14% of the faculty. ISU freshmen and sophomores had been introduced to key- word searching in their English classes in the last two years, but the other groups have been taught in the same way. Age Range Age differences are similar to the sta- tus-based differences. In some cases, older respondents from a returning stu- dent population performed similarly to 76 College & Research Libraries TABLE4 FREQUENCY OF LUIS USE Frequency Several times a week No more than several times a month No more than several times a year Only a few times Never Unusable answer (%) (22.8) (44) (14.8) (16) (2) (.3) faculty. The small under-18 category was dropped, and the highest three catego- ries, 41 to 50, 51 to 60, and 61 or above, were collapsed (see table 3). Whether or not the respondent has performed keyword searching varies significantly with age, and the results are somewhat similar to those for status. At the highest level, 87% of the 18- to 20- year-olds had done keyword searching. The range that searched at the lowest rate was 23- to 30-year-olds, at 39 of 60 (65%). Fifty-one of 69 (74%) 31- to 40- year-olds had done keyword, and 35 of 50 (70%) of the oldest range had done it. Around 70% of all the older groups had done keyword searching, versus almost 90% of the youngest group. Frequency of LUIS Use Question six elicited the reported fre- quency of LUIS use. Basic numbers for different replies are given in table 4. The tendency to have done a keyword search drops steadily with less frequent LUIS January 1992 use. Ninety percent of those who search LUIS a few times a week have done key- word searching. Of those who searched LUIS no more than a few times a month, 83% had done a keyword search, as had only 70% of those who searched it no more than a few times a year. Even 69% of those who had used LUIS only a few times had done keyword searching. Forty-four percent of those who had never searched LUIS before that day had, nonetheless, still done a keyword search. The most frequent users of LUIS were much more likely than other groups to say they planned to use keyword again-more than 90% (see tableS). Gen- erally, the lower the frequency of LUIS use, the less likely the respondent was to plan to use keyword searching again. The steady drop was interrupted by those who had searched LUIS only a few times; a higher percentage of them planned to do a keyword search again than those who searched LUIS a few times a year. The less frequent the use of LUIS, the more ambivalent the respon- dent was about whether keyword searching would be used again. Years of LUIS Experience Thirty-one percent of the respondents had zero to one years of LUIS experi- ence, 30% had one to two years, and 39% had three to five years. Years of LUIS experience related to a difference in rea- sons for not having used keyword searching. Because very small numbers were being dealt with, however, there- sults should be looked at with caution (see table 6). TABLES FUTURE KEYWORD SEARCHING PLANS BY FREQUENCY OF LUIS USE Frequency Yes(%) No(%) Don't know(%) Several times/week 82 (90.1) 3 (3.3) 6 (6.6) Several times/ month 139 (79.4) 3 (1.7) 33 (18.9) Several times/year 37 (62.7) 1 (1.7) 21 (35.6) Only a few times 43 (67.2) 3 (4.7) 18 (28.1) Never 4 (44.4) 0 5 (55.6) Percentages do not always equallOO% due to rounding. Keyword Searching 77 TABLE6 REASONS FOR NOT USING KEYWORD SEARCHING BY LENGTH OF LUIS EXPERIENCE Reason 0-1 Yrs. (%) 1-2 Yrs. (%) 3-5 Yrs. (%) It is too difficult to learn 4 (10) 0 3 (6.7) It takes too long to learn 0 0 3 (6.7) I don't need to; other forms of searching are adequate 10 (25) 13 (40.6) 10 (22.2) forme I don't search very often, so I wouldn't remember 2 (5) 3 (9.4) 6 (13.3) how to do it I haven't been able to go to a library instruction 8 (20) (3.1) 3 (6.7) session on it There hasn't been staff available to assist me 2 (5) 2 (6.3) 0 I wasn't aware of this form of searching at all 14 (35) 10 (31.3) 20 (44.4) Its response time is too slow 0 3 (9.4) 0 Percentages do not always have to equallOO% due to rounding. No one with one to two years' experi- ence said keyword searching is too diffi- cult to learn, as opposed to 7% for the more experienced and 10% for the less experienced. The only people who said keyword searching takes too long to learn were those who had been using LUIS for three to five years. A compara- tively high percentage of those with the longest LUIS experience said they did not search often and tended to forget how to do keyword searching. Under- standably, a comparatively high percent- age of the least experienced group said that they had not had a chance to go to a library instruction session on keyword searching. A surprisingly high percent- age of the most experienced LUIS search- ers said they were not aware of keyword searching at ttll. Previous Computer Experience Respondents' previous computer ex- perience is reported in table 7. Those who had no other computer experience were less likely to plan to do keyword searching in the future. Sixty percent of those with no other computer experience said they planned to do keyword search- ing in the future, as opposed to 78% of those who had other computer experi- ence. Eight percent did not know, as op- posed to 2% of the latter group. Thirty-two percent of those who had not searched other computer systems did not plan to do keyword searching in the future, versus only 20% of those who had searched other systems. Users of OCLC (who were probably in a beginning library science course, where they have to do OCLC and key- word exercises) were much more likely to have used keyword searching on TABLE 7 PREVIOUS COMPUTER EXPERIENCE Experience No. (%) No other computer systems 37 (9.3) Another library's computerized 88 (22) catalog CD-ROM databases (like 140 (35) ERIC, ABI/Inform) Online dial-up databases 32 (8.0) (like Dialog, BRS) . OCLC terminal 41 (10.3) Computer at home 211 (52.8) Computer at work 185 (46.3) Computer at school for a 178 (44.5) noncomputer course Computer at school for a 191 (47.8) computer course Other 15 (3.8) Each percentage given is from the total 400 respondents, since any number of options could be checked. 78 College & Research Libraries TABLES ENTRY 1YPE OF LAST LUIS SEARCH Entry Type No. (%) Author's full or last name 67 (16.8) Author's first name 3 (.8) Complete title or the first 86 (21.5) part of it Part of the title other than the 5 (1.3) first part Library of Congress Subject 69 (17.3) Heading Topic words 150 (37.5) Unusable response 13 (3.3) Blank 7 (1.8) Percentages do not always equal 100% due to rounding. LUIS. Ninety-five percent of them had done a keyword search. Those who did not say they had used an OCLC terminal had done keyword searching at a rate of 77%. . Users of a school computer for a non- computer course were somewhat more likely to have done a keyword search than the rest of the respondents-85%, as opposed to 75%. Users of a school computer for a computer course were more likely to have done a computer search-86% had, versus 73% of those without that experience. Those using a computer for a com- puter course varied in their future key- word-searching plans. Eighty-three percent planned to do keyword searching in the January 1992 future, compared with 71% of the rest of the respondents. Only one person with this kind of ~omputer experience re- ported plans to do another keyword search, whereas nine of the other respon- dents did plan another search. Twenty- five percent of those without this type of computer experience said they did not know if they would do another keyword search, versus 17% of those with school computer experience in a computer course. LWS Search Type Table 8 shows the entry type of the last LUIS search executed. Performance of keyword searching varied significantly (to .05) with this factor (see table 9). Non- keyword users tended to search by title or the first part of the title more than the keyword users. N onkeyword searchers also did substantially more searching with LC subject headings than did key- word searchers, and they did less topic word searching (searching by words that are not necessarily LC subject headings), but they still did some; more than one- fourth of them checked this option. Keyword Searching Plans Seventy-six percent planned to do a keyword search in the future. Only 3% did not plan to; 21% did not know. Ninety-one percent of those who had done keyword searching before planned to do it again in the future. Only 1% of them did not plan to do it again; 8.2% did TABLE9 PERFORMANCE OF KEYWORD SEARCHING BY ENTRY TYPE OF LAST LUIS SEARCH Entry Type Author's full or last name Author's first name Complete title or first part of it Part of the title other than first Library of Congress Subject Heading ToEic words Percentages do not always equallOO% due to rounding. No. 56 3 63 5 48 130 Keyword Srch. Done (%) (18.4) (1) (20.7) (16.4) (15.7) (42.6) Keyword Srch. Never Done No. (%) 11 (14.7) 0 23 (30.7) 0 21 (28) 20 (26.7) Keyword Searching 79 TABLE 10 REASONS FOR NEVER HAVING USED KEYWORD SEARCHING .. Reason No. (%) It is too difficult to learn 7 (5.9) It takes too long to learn 3 (2.5) I don't need to; other forms of searching are adequate for me 34 (28.6) I don't search very often, so I wouldn't remember how to do it 12 (10.1) I haven't been able to go to a library instruction session on it 12 (10.1) There hasn't been staff available to assist me 4 (3.4) I wasn't aware of this form of searching at all 44 (37) Its resEonse time is too slow 3 (2.5) ,. Respondents could check up to two items; percentage is of total number of reasons checked, 119. Percentages do not always equallOO% due to rounding. not know. More interestingly, 21% of those who had not done a keyword search before planned to do one in the future. Only 9% said they would not ever be doing one. Seventy-one percent were not sure. This means that 84% of the respondents had either done a keyword search or were planning to do one. Only 2% of the respondents had never done a keyword search and never planned to. Reasons for Never Having Done Keyword Searching Respondents were asked to check up to two reasons for never having done a keyword search. The most frequently cited response was that people were unaware of the existence of keyword searching. This implies that these people might do key- word searching in the future and not that they are opposed to it (see table 10). CONCLUSION The reported use of keyword search- ing in this study varies significantly with status, age range, frequency of LUIS searching, use of an OCLC terminal, school computer use for computer and noncomputer courses, and type of last LUIS search entry. Faculty, older respon- dents, and less-frequent LUIS searchers were less likely to have done keyword searching. Those who had used an OCLC terminal or a school computer used keyword more, as had those who said their last LUIS search entry was for topic words. Those who searched LUIS frequently, those who had used a school computer, and those who had already done a key- word search were more likely to plan to do one in the future. Those who had never used a computer system before were less likely to plan to do a keyword search in the future. Those who had not done keyword searching before were quite likely not to have been aware of its existence, but some thought that they did not need to do it. Whether or not they have done keyword searching seems to relate most to age factors and computer experience, with gender and subject area not relating to significant variations in keyword searching usage. Future studies could profitably examine the use of keyword search capabilities in other OPACs, among other audiences, and with peri- odical article databases to see if these conclusions are supported. REFERENCES AND NOTES 1. Randolph L. Whitson, "Keyword Access to a Library's Records in an Online Catalog," Tennessee Librarian 38:28-29 (Winter 1986). 2. Keyword/Boolean Searching on an Online Public Access Catalog: Patrons and Their Percep- tions has been submitted as an ERIC document. 80 College & Research Libraries January 1992 3. Joseph R. Matthews, A Study of Six Online Public Access Catalogs: A Review of Findings- Final Report (Washington, D.C.: Council on Library Resources, 1982). ERIC, ED 231 389. 4. Gary S. Lawrence and Joseph R. Matthews, Detailed Data Analysis of the CLR Online Catalog Project: A Final Report (Washington, D.C.: Council on Library Resources, 1984). ERIC, ED 242 332. 5. Valerie Geller and Michael Lesk, "An On-Line Library Catalog Offering Menu and Keyword User Interfaces," in Fourth National Online Meeting Proceedings (Medford, N.J.: Learned Information, 1984), p.159-65. 6. Nancy C. Kranich and others, "Evaluating the Online Catalog from a Public Services Perspective: A Case Study at the New York University Libraries," in The Impact of Online Catalogs, ed. Joseph R. Matthews (New York: Neal-Schuman, 1986), p.89-117. 7. Carolyn 0. Frost, "Faculty Use of Subject Searching in Card and Online Catalogs," Journal of Academic Librarianship 13:86-92 (May 1987). 8. S. Batley, "Visual Information Retrieval: Browsing Strategies in Pictorial Databases," in Twelfth International Online Information Meeting Proceedings (Oxford, England: Learned Information), p .373-80. 11 Extraordinary materials on U.S. policymaking Dr. William LeoGrande, The American University A magnificent achievement Dr. James Blight, Harvard University Unparalled in extent Published by Chadwyck-Healey and the National Security Archive, this series has been lauded in nearly every media-from Library Journal to Nightline. These extensively indexed microfiche collections reproduce previously classified and unclassified government documents. Most of this primary source documentation has not been published anywhere else. Dr. Nikki Keddie, UCLA The series contains nine individual subject No research library should sets, the latest of which is South Africa: be withOUt it 11 The Making of U.S. Policy, 1962-1989. Dr. Barnett Rubin, Columbia University The Making of U.S. Policy series has earned praise since the publication of the first in the series, El Salvador: The Making of U.S. Policy, 1977-1984. To receive a brochure or sample fiche, call Melissa Henderson at Chadwyck-Healey Inc. at (800)752-0515. Or write to Chadwyck-Healey Inc., 1101 King Street, Suite 380, Alexandria, Virginia 22314.