Using Multimodal Learning Analytics to Model Student Behavior: A Systematic Analysis of Epistemological Framing

Abstract

Research has shown that students respond to social expectations of interviews by engaging with the content in distinct ways that may or may not be productive. These structures of expectations with respect to knowledge are referred to as epistemological framing.  In this study our goal is to introduce a systematic way to analyze student behaviors and describe how they cluster together to reflect different epistemological frames. In analyzing the data statistically, frames are regarded as a latent variable that accounts for the co-occurrence of behaviors.  We use a computer clustering algorithm to systematically identify behavioral clusters in videotaped data of early elementary students that construct explanations of biological systems in semi-structured interviews. We also examine the relationship between these behavioral clusters and mechanistic reasoning as a way to investigate the importance of the inferences made based on these identified frames. Results show that there is a clear association between epistemological framing and student reasoning. By providing a statistical model of student framing, our approach can support the ongoing refinement of theory around epistemological frames and their impact on learning.