Ordering Worked Examples to Promote Categorization

We report an investigation of novices' ability to categorize probability word problems. Individuals can effectively learn from worked examples (Ward & Sweller, 1990), especially if multiple examples with the same structural features are presented (Gick & Holyoak, 1983). We were interested in whether the order of worked examples could affect a learner's ability to categorize new problems. Participants studied 5 types of probability word problems; each type had 2 worked examples. Two orders of worked examples--blocked and random--were used. For isomorphic test items, the random ordering reduced participants' total number of categorization errors and their categorization error rates. For transfer test items, ordering did not affect categorization errors. We argue these results demonstrate that the random ordering facilitated construction of problem schemas, which subsequently improved participant's categorization ability. However, the random ordering appeared to have no effect on learners' ability to restructure problem schemas as required by transfer items.