| The creation and distribution of library research guides would appear to be beneficial, but this has not been documented. In this study, a group of existing library research guides were loaded in a Campus Wide Information System (CWIS) and linked to the Internet by a Gopher system. The use of the files was monitored for three years (1993-1995) and a Transaction Log Analysis (TLA) undertaken. Results were analyzed in order to determine whether (1) searchers would independently discover the guides; (2) whether a pattern of discovery based on geographic proximity or Internet domain could be established; and (3) whether retrieval of specific topics might be might differ geographic proximity or Internet domain. The transaction log used in this study contains basic elements about individual search states including date, time, hostname or numeric address (IP), and the directory/file sequence.; General interest profiles based upon geographic proximity and domain were identified without violating the privacy of individuals. By requiring searchers to enter the files either through an alphabetical or a subject menu, searcher preferences for either method by file and by topic were identified. Tracking the use of library research guides over time proved to be a reliable indication that the development of the Internet has penetrated the geographic barrier to discovery and use of library research guides. Patterns of discovery of library research guides differ based on both geographic proximity and domain. However, statistical and qualitative analysis of the logs for searchers outside the United States yielded different results, and indicated that system time-outs hindered access to information.; It was also found that the primary user group accounted for over 50 percent of the total use. Therefore, library research guides were used by the primary user population as well as by users with similar needs and interests. An analysis of the frequency of retrieval by topic indicated that rate of retrieval changed depending upon whether the files were accessed using the alphabetical or subject menus. Copies of the FORTRAN, PERL, TURBO PASCAL and JAVA scripts used, as well as the SPSS coding, are included in an appendix. |
