When is it best to learn with all worked examples?

Bruce McLaren, S. Isotani

In: Gautam Biswas , Susan Bull , Judy Kay , Antonija Mitrovic (Hrsg.). Artificial Intelligence in Education; Proceedings of the 15th International Conference on Artificial Intelligence in Education. International Conference on Artificial Intelligence in Education (AIED) June 28-July 1 Auckland New Zealand Seiten 222-229 6738 ISBN 978-3-642-21868-2 Springer Berlin 2011.


Worked examples have repeatedly demonstrated learning benefits in a range of studies, particularly with low prior knowledge students and when the examples are presented in alternating fashion with problems to solve. Recently, worked examples alternating with intelligently-tutored problems have been shown to provide at least as much learning benefit to students as all tutored problems, with the advantage of taking significantly less learning time (i.e., more efficiency) than all tutored problems. Given prior findings, together with the prevailing belief that students should be prompted to actively solve problems after studying examples, rarely have all worked examples been tried as a learning intervention. To test the conventional wisdom, as well as to explore an understudied approach, a study was conducted with 145 high school students in the domain of chemistry to compare alternating worked examples / tutored problems, all tutored problems, and all worked examples. It was hypothesized that the alternating condition would lead to better results (i.e., better learning and/or learning efficiency) than either all examples or all tutored problems. However, the hypothesis was not confirmed: While all three conditions learned roughly the same amount, the all worked examples condition took significantly less time and was a more efficient learning treatment than either alternating examples/tutored problems or all tutored problems. This paper posits an explanation for why this (seemingly) surprising result was found.

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence