Abstract
We designed a statistical support system of which the knowledge is based on the experts' conceptual model for the choice of a statistical analysis method. To protect this system from the fate most statistical support systems suffer—somewhere on a lab shelf—we set out to find possible differences in conceptual models of experts and nonexperts by studying accounts by researchers on the method-choosing task. Results showed that there are important differences between the two models, the most important being the importance attributed to the research problem: Nonexperts tend to link the research problem as a whole directly to a method, whereas experts first of all categorize characteristics of the analysis problem. We conclude that there is a need to incorporate nonexperts' concepts into the statistical support system in order to support them in translating their view into the experts' view. After this translation the experts' concepts needed for solving the problem can be dealt with. Keywords : statistics, statistical expert systems, statistical knowledge, nonexperts' statistical knowledge, conceptual models.
Get full access to this article
View all access options for this article.