The effect of interface design on item selection in an online catalog

The effect that content and layout of bibliographic displays had on the abi lity of end-users to process catalog information was tested using a 2 x 2 f actorial experimental design. Participants were asked to perform two relate d tasks during the course of the experiment. In the first task, they were a sked to select a set of items that they would examine further for a hypothe tical paper they must write, using a simulated online catalog to make their assessments of relevance. In the second task, they were asked to examine 2 0 bibliographic records, decide whether they would choose to examine these items further on the shelf, and identify the data elements that they used t o formulate their relevance decision. One group viewed bibliographic records on art interface similar to current online catalogs, one that used data labels and contained data elements comm only found. A second group viewed these records on an interface in which th e labels had been removed, but the data elements were the same as those in the first. The third group viewed these records on a labeled display that i ncluded enhanced data elements on the brief record display. The final group viewed these records with the same brief record data elements as the third group, but with the labels removed, using ISBD and AACR2 punctuation stand ards. For the first task, participants using enhanced brief screen interfaces vie wed more brief screens and fewer full screens than their counterparts. Scre en durations for the second 10 screens were found to have dropped from thos e of the first 10 screens. Statistical analyses comparing demographic varia bles to the screen frequencies uncovered many significant differences. Part icipants using the enhanced-content interfaces made fewer selections from i ndex and full screens, and more selections from brief screens. For the seco nd task, participants who used enhanced-content interfaces were able to mak e sort of relevance judgment more frequently than those who used standard-c ontent interfaces.