Abstract
The worked example effect indicates that showing students worked examples (high guidance) is superior to problem solving (low guidance) which provides no guidance, whereas the generation effect suggests that self-generating items (low guidance) is superior to studying the externally presented answers (high guidance). This obvious contradiction between the two effects was hypothesized to be resolved by suggesting that the materials used had different levels of element interactivity. For the worked example effect, materials may be high in element interactivity, while, for the generation effect, simpler materials are used. With an increase of learner expertise, the worked example effect may be eliminated or reversed because expertise reduces element interactivity, but the generation effect should be still robust. Five 2 (levels of guidance: low and high) x 2 (levels of element interactivity: low and high) mixed factorial experiments were conducted to investigate the hypotheses. In Experiments 1 to 3, the level of learner expertise gradually increased in the domain of geometry with the results supporting hypotheses. The interaction of guidance and element interactivity was obtained with novices but the worked example effect that contributed to the interaction was eliminated or reversed with more knowledgeable students. Experiments 4 to 5 were designed to replicate the results of the first three experiments by testing students in the domain of trigonometry on both immediate and delayed tests. The results were replicated not only on an immediate test, but also on a delayed test. When combined, the results of all five experiments indicated that levels of element interactivity might be a key factor when deciding on levels of instructional guidance.