Statistical computing : individual differences in the acquisition of a cognitive skill
The rate at which individuals acquire new cognitive skills may vary quite substantially, some acquiring a new skill more rapidly and efficiently than others. It has been shown through the analysis of think aloud protocols that learning performance on a map learning task, for instance, is associated with the use of certain learning procedures. In the domain of mathematical problem solving, it has also been shown that performance is associated with strategic as opposed to tactical decision making. Previous research on learning and problem solving has tended to focus on tactical processes, ignoring the role of strategic processes in learning and problem solving. There is clearly a need to examine the role of strategic processes in learning and to determine whether they might be an important source of individual differences in learning performance. A related question concerns teaching thinking skills. If it is possible to determine those learning procedures that differentiate good from poor learners, is it then possible to teach the effective procedures to a group of novice students in order to enhance the rate of skill acquisition? Results from the experiments reported here show that novices differ, and that learning performance is related to the use of certain learning procedures, as revealed by subjects' think aloud protocols. A follow-up study showed that novices taught to use the procedures differentiating good from poor learners performed at a higher level than two control groups of novices. A coding scheme was developed to explicitly examine learning at macroscopic and microscopic levels, and to contrast tactical with strategic processes. Discriminant function analysis was used to examine differences between good and poor learners. It was shown that good learners more frequently use executive processes in learning episodes. A study of the same subjects learning to use statistical packages on a microcomputer corroborate these findings. Thus, results extend those obtained from the first study. A study of the knowledge structures possessed by novices was complicated by differences in levels of statistical knowledge. Multidimensional scaling techniques revealed differences between novices with three statistical courses behind them, but not among those with only two statistical courses behind them. Among those novices with three statistical courses behind them, faster learners' knowledge structures more closely resembled those of experienced users of statistical packages than did those of slower learners.