What types of feedback enhance the effectiveness of self-explanation in a simulation-based learning environment?

In this research, self-explanation was prompted and feedback was supplied to help learners activate prior knowledge, detect misconceptions, and replace unscientific mental models with correct scientific models. The research investigated the effects of two types of tutor feedback on learning and conceptual change in a simulation inquiry environment: Elaborative feedback incorporated tutor explanation and knowledge of results feedback provided only confirmation or disconfirmation of learners' statements. Sixty-eight undergraduate students, with low prior knowledge in the physics of waves, were randomly assigned to receive either (a) self-explanation prompts with no feedback (NF), (b) self-explanation prompts with knowledge of results feedback (KRF), and (c) self-explanation prompts with elaborative feedback (EF). A pretest-posttest design was used to investigate participants' knowledge gain and conceptual change resulting from learning tasks they performed by interacting with a physics simulation and explaining what they observed. The simulation, learning tasks, and knowledge tests focused on five fundamental principles of wave physics, four of which are often subject to misconceptions. Chi-square tests of association followed by pairwise Fisher's exact test comparisons revealed elaborative feedback was advantageous, but only for two of the four concepts prone to persistent misconception – the mechanism of sound propagation and the medium-speed relationship. The findings suggest that prompting learners to self-explain can be sufficient for learning, but only for concepts whose acquisition is not hindered by persistent misconceptions. For concepts prone to such misconceptions, elaborative feedback may be necessary for understanding phenomena at deep structural levels. It is proposed that self-explanation combined with elaborative feedback may be a highly effective instructional strategy across many scientific domains, especially in the context of simulation-based inquiry learning.