Drawing gears and chains of reasoning

This dissertation discusses the design and evaluation of GearSketch, a learning environment for the gears domain aimed at students in the final years of primary school. GearSketch is built around a computational model of the gears domain that represents how gears and chains can be connected to each other and how these connections transmit motion between objects. When working with GearSketch students use a pen-based or touchscreen computer to draw their own gear and chain systems. The learning environment ensures that only valid systems (i.e. systems in which the gears can turn) can be created. Students can explore the behavior of configurations they have drawn by starting a simulation and watching an animation of the gears and chains in motion. GearSketch offers tutorials and practice problems to facilitate learning.
The effects of different features of the GearSketch learning environment on learning outcomes were examined in three experimental studies. The first study examined the value of GearSketch’s ability to interpret and animate students’ drawings. This study showed that students who worked with a version of GearSketch with these features learned significantly more than students who practiced with a learning environment that lacked these features. In the second study learner modeling capabilities were added to GearSketch. Modeling learners’ knowledge as they progressed through practice problems, allowed the learning environment to adaptively select practice problems based on learners’ current understanding of the gears domain. However, the results of this second study showed that learners for whom practice items were adaptively selected did not learn more than learners who practiced with a fixed set of items. A possible explanation for this finding was that learners relied more on trial and error during problem solving than on deliberate reasoning. The third study examined the effect of encouraging students to use more deliberate reasoning while working on the practice items. The results of this study showed that the guidance that was offered during the practice items improved students’ performance during practice, but did not affect learning outcomes as measured by a posttest. Additionally, this third study showed that the knowledge about the gears domain of both the guided students and those in a control group was significantly improved by practicing with GearSketch. Together these studies show that a learning environment based on a drawing-based simulation can be effectively used by primary school students to learn about the gears domain, but that modeling and guiding students’ reasoning in this domain is still an open challenge.