A Test of the Interaction Hypothesis: Joint-explaining vs. Self-explaining

It is widely assumed that students learn more from interactive learning situations than non-interactive; however, recent evidence from human tutoring suggests this generalization may not always hold (VanLehn et al., 2007). Instead of merely contrasting interaction with non-interaction, a more appropriate test may be to hold the type of interactive activities constant, while manipulating the degree of interactivity. Explaining was chosen as the interactive activity because it is well known that explanation is an effective, individual learning strategy that is easily trained, which produces deep learning (McNamara, 2004). The effect size is estimated to be between d = .74 – 1.12. Furthermore, there is mounting evidence that explanation activities during collaboration can also result in deep learning (Hausmann, Chi, & Roy, 2004). The effect size for peer collaboration, in general, is estimated to be between d = .21 – .88. The present study tests the interaction hypothesis by explicitly contrasting two interactive activities (i.e., selfvs. joint-explanation). The interactive hypothesis predicts jointexplanation will lead to better problem-solving performance because there are more opportunities to be interactive.