Using Retelling Data to Study Elementary School Children's Representations and Solutions of Compare Problems

In an attempt to explain subjects' difficulties with particular types of compare problems, Lewis and Mayer (1987) have presented a model that simulates the comprehension processes when solving these problems. The crucial component of their model is the “consistency hypothesis,” according to which students are more likely to experience representational difficulties—and consequently to make errors—when the presentation of the terms in the relational statement of the compare problem is not consistent with their preferred format. To test this model, an experiment was carried out in which 40 fifth graders were asked to solve and retell a set of one-step compare problems. The accuracy data, solution times, and, especially, the retelling data yielded strong evidence in favor of the consistency hypothesis.