Using a Triad Judgment Task to Examine the Effect of Experience on Problem Representation in Statistics

This research investigated whether the differences found between novices and experts in using surface and deep structures to categorize problems applied to the domain of statistics. Also explored was whether the methodology of a triad judgment task was reliable in discriminating how beginning and advanced students represent statistics problems. The task was designed in which source problems shared either structural features (t-test, correlation, or chi-square) or surface similarity (story narrative) with the target problem. Graduate students (N = 101) with varying levels of experience in the domain of statistics were asked to chose which source problem “goes best” with the target problem for each triad. Students with advanced experience in statistics tended to represent the problems on the basis of deep, structural features while beginning students tended to rely on surface features. Discussion on the effectiveness of the methodology employed and potential uses in other domains is presented.