OPTIMIZING HUMAN-COMPUTER INTERACTION FOR THE ELECTRONIC COMMERCE ENVIRONMENT

Abstract

The paper investigates the interaction effects between the search strategy of software agents and the subject’s product class knowledge. The experimental study consists of a 2 (product class knowledge: high, low) x 4 (agent search strategy: elimination by aspects (EBA), weighted average method (WAD), profile building (PROFILE) , and simple hypertext (HYPERTEXT)) design with product class knowledge as the between groups variable. Significant differences were found for affective reactions of the subjects toward the agent/application depending on the level of product class knowledge possessed by the subjects. Subjects with high product class knowledge had more positive affective reactions towards agents/applications which used the WAD and EBA strategies as compared to the profile building strategy. Subjects with low product class knowledge had more positive affective reactions to agents/applications which used the profile building strategy as compared to the EBA and WAD strategies. When the systems were modified to increase the amount of information provided and to increase the degree of control provided to the subjects, their affective reactions to the agents/applications were found to be different from the original study. Subjects responded more positively to the previously less preferred strategy, thus weakening the interaction effect. This research is done in the context of consumers searching for information on the World Wide Web prior to the purchase of cars.