Measuring Usability: Preference vs. Performance.

The article analyzes the broader issue of the relation between user preferences and performances across a wide range of computer systems and usage situations, in which the exact measures used as proxies for preference and performance varied according to whatever was most appropriate for specific circumstances of each study. Besides the theoretical interest in how users' subjective experiential feelings reflect their actual performance, the relation between these two usability variables is also of interest from the practical discount usability-engineering perspective. Simply asking users how they like an interface must be one of the cheapest usability-engineering methods available, and it would thus be beneficial if one could expect the users' answers to also reflect their performance, which is normally more expensive to measure. Because of data availability constraints, the present metaanalysis only considers the relation between the average subjective satisfaction of a group of users and the average performance of the same group. It is thus most appropriate for usability-engineering scenarios in which a single choice has to be made for all intended users of a certain group.