ABSTRACT
Finding the sources of misconceptions possessed by students is a difficult task because it is impossible to see what is happening in their heads. The only directly observable outcomes are the students’ responses to the test items. Studying their think-aloud protocols is one method for discovering how students solve or think through problems. Several computer programs that are capable of diagnosing students’ misconceptions have been developed in the past decade (Brown & Burton, 1978; Marshall, 1980; Ohlsson & Langley, 1985; Sleeman, 1984; Tatsuoka, Baillie, & Yamamoto, 1982; VanLehn, 1983). The common ground for these cognitive diagnostic systems is that they infer the unobservable cognitive processes from logical interrelationships among cognitive tasks, subtasks, and goals involved in problems representing the domain of interest. It is important that we be able to retrieve invisible things from the “black box” and put them into a useful form so that valuable information can be obtained for improving educational quality.
