On the Subjective Representation of Statistical Problems

Montgomery, H. & Allwood, C. M. On the Subjective Representation of Statistical Problems. Scandinavian Journal of Educational Research 22, 107‐127. Think aloud data from 19 subjects solving three statistical problems are reported. The data indicated that the problem solving process could often be regarded as a successive development of the problem formulation where each formulation tended to be on the same or on a higher qualitative level than the preceding formulation. Good problem solvers were found to differ from the poor solvers inasmuch as they (a) more often made an exhaustive definition of an essential concept mentioned in the problem, or (b) appeared to attend more to what was asked for in the problem. It was tentatively concluded that some of the erroneous problem solutions were due to a pure lack of reflection on how the solution of the problem should be related to the information asked for in the problem. For a problem, which was formulated in a seemingly ambiguous way, subjects’ knowledge of the underlying statistical model was positively related to the correctness of their problem interpretation.