A componential model for understanding the effects of composite interface metaphors on web-based task performance and learning

Two experiments describe how a composite metaphor interface (CMI) for performing website tasks can be developed using an iterative and theory-based design approach. The CMI consisted of two semantically unrelated metaphors: the library metaphor and the travel metaphor. A new componential model for predicting the effectiveness of CMIs is described and used for deriving and testing specific hypotheses. Results show that an improved version of the CMI resulted in better declarative knowledge about metaphor-interface mappings than a non-metaphorical control interface (even before the interface was first used), and in less mental effort, as measured by the clicking behavior. It also resulted in faster performance of web-based tasks, but only if differences in pre-existing declarative knowledge were taken into account. The discussion addresses i.a. the question how mismatches between interface functions, and the metaphor they belong to (confusions) can be measured and used for improving the quality of the CMI.