ABSTRACT
One of the aims of intelligent tutoring system (ITS) research is to develop sophisticated means for individualized assessment of student knowledge and cognition (Wenger, 1987). In this regard, we have attempted to develop a generally applicable cognitive measurement theory (CMT) as well as specific measurement models. For the most part, this has involved attempts to integrate extensions of Rasch (1960) measurement theory with production system theories of cognition and learning (Anderson, 1993; Pirolli, 1991; Pirolli & Recker, in press). Here, we present a particular CMT model designed to capture factors affecting the development of cognitive skills in programming. The model is designed to measure the acquisition and strengthening of independent elements of cognitive skill, individual differences in propensity for learning such skills, and the effects of pedagogical manipulations on the acquisition of subsets of the skill elements. We report on the application of the model to data collected from four studies (Bielaczyc, Pirolli, & Brown, 1991; Pirolli & Recker, in press; Recker & Pirolli, 1992) involving the LISP Tutor (Reiser, Anderson, & Farrell, 1985), as well as the broader framework for the model, possible extensions, and how the measurement model may be used to feed into an online diagnostic model.
