End‐users, mediated searches, and front‐end assistance programs on Dialog: A comparison of learning, performance, and satisfaction

We trained 40 doctoral students to search Inspec or ERIC on Dialog using either the Sci‐Mate Menu or native commands. Each subject then performed a free search on his or her chosen research topic. In comparison with 20 control subjects—doctoral students for whom a free search was performed by a skilled librarian—the experimental subjects were no less satisfied with their retrievals, which were smaller but higher in precision than the retrievals produced by the intermediaries. End‐users retrieved as many relevant references as the intermediaries. Use of the menu interface did not affect quality of retrieval or user satisfaction, although subjects instructed to use native commands required less training time, and interacted more with the databases than subjects trained on the Sci‐Mate Menu. Inspec subjects placed a higher monetary value on their searches than did ERIC subjects, indicated that they would make more frequent uses of databases in the future, and interacted more with the database. Among the ERIC subjects, women made more errors than men, interacted more frequently with the database than men, placed a higher monetary value on their searches and indicated that they actually intended to consult a higher proportion of the references listed than men. © 1990 John Wiley & Sons, Inc.