Scaffolding Problem Solving with Embedded Examples to Promote Deep Learning
Michael Ringenberg and Kurt VanLehn
This in vivo experiment which occurred in the Physics LearnLab compared the relative utility of an intelligent tutoring system that used hint sequences to a version that used completely justified examples for learning college level physics. In order to test which strategy produced better gains in competence, two version of Andes were used: one offered participants hint sequences and the other completely justified examples in response to their help requests. We found that providing examples was at least as effective as the hint sequences and was more efficient in terms of the number of problems it took to obtain the same level of mastery.
Background and Significance
When students use a tutoring system with hint sequences, they sometimes engage in help abuse on virtually every step (citation needed). This means that the tutoring system is telling them each step, so essentially, they are generating a worked-out example. There may be nothing wrong with this for some students, as examples can be effective instructional material (citation needed).
Will robust learning be fostered if students are presented with relevant, annotated, worked-out examples instead of a targeted hint message when a learning event is encountered?
The manipulation in this study was based on the feedback the Andes system provided while participants worked on assigned homework problems covering Inductors:
- Examples Condition
- Participants were presented a relevant, annotated, worked-out example when help was requested.
- Hints Condition
- Participants were presented a targeted hint when help was requested.
Providing annotated, worked-out examples instead of hints during problem solving will promote the learning of knowledge components and help appropriately generalize the knowledge components.
Dependent variables & Results
- Near Transfer, retention
- Performance on problems involving inductors on the normal mid-term exam that were similar to the training problems. There was not significant difference in performance between the two conditions. Both conditions did better than a baseline of participants who solved no homework problems.
- Transfer task, deep structure assessment
- Problem matching task: No significant difference in performance between the two conditions; however, participants in the examples condition solved fewer training problems. Both conditions did better than a baseline of participants who solved no homework problems.
- Number of problems completed: Participants in the examples condition solved significantly fewer problems than participants in the hints condition.
- Time on task: Participants in the examples condition spent less time solving problems than those in the hints condition. Participants in both conditions spent about the same amount of time per problem.
- Ringenberg, Michael A. & VanLehn, Kurt (2006). Scaffolding Problem Solving with Annotated, Worked-Out Examples to Promote Deep Learning. Paper presented at the ITS 2006, Taiwan. Winner of Best Paper First Authored by a Student Award. 231Kb PDF