Empirical Investigation of Assistance Dilemma with a Tutoring System that Can Control Levels of Support

Abstract

We experimentally investigated the assistance dilemma with a learning system that can control levels of support (LOS). Our system supports learning of natural deduction (ND) and was established based on the client-server framework. An experiment was performed. In the first, half of the participants learned ND in the high LOS condition, and the other half learned in the low LOS. In the learning phase, the solution time was shorter and the trial and error steps until solution were greater in the high LOS group than in the low LOS group. However, in the posttest, more participants successfully reached the solution in the low LOS group.