Modeling human-computer decision making with covariance structure analysis

Abstract

The development of a complete theory of human-computer interaction has been hindered by the lack of full scale testable models. To date, those models which have been tested are primarily low-level (e.g., key-stroke; Card, Moran, & Newell, 1983). Although providing clear predictions, they rarely provide information on more than just a few low level variables. Other models, such as those provided by mental representations (e.g., Gentner & Stevens, 1983), are often not testable or do not provide quantified information concerning the importance of variables and their interrelations.