Research on abstract categorical reasoning has shown that students with "A" grade point averages (GPA) have limited, better and often less, reasoning ability than students with lower GPA's. A similar effect was also found when testing "bright" and "dull" subjects. An analysis of this effect was conducted using the component model of categorical reasoning. Findings indicated that most subjects employed either basic or advanced reasoning, but not complete reasoning. In the standard instruction condition, high GPA students were less likely than low GPA students to employ a complete reasoning strategy. In the 50% instruction condition, high GPA students were more likely than low GPA students to employ a complete reasoning strategy. Results suggest that the failure or success of research in this area may be dependent on exactly what problems are used by the investigator and the exact form of the instructions given to the subjects. (Author/KMF)