Exploring Online Learning Response Data Using Stochastic Processes with Dependence

• Activity Number: 684
• Type: Contributed
• Date/Time: Thursday, August 13, 2015 : 10:30 AM to 12:20 PM
• Sponsor: IMS
• Author(s): Hongwen Guo*
• Keywords: online learning and tutoring ; process data ; fractal dimension ; strong dependence ; confidence interval

Abstract

Data collected for one student are expected to show strong and positive dependence because of knowledge-based dependency, or persistence of learning behavior. When response data show little or negative correlations, it is suspected that the student may be randomly guessing or inconsistent in learning behavior, and these may be a sign of learning problem. With popularity of online learning/tutoring and the abundance of online processing data, there is a need to understand basic data structures of online learning data. In the study, we show that the dependence of online data may be characterized by fractal dimensions (FD) locally and globally. For educators who use percent of correct responses as an index to instruct teaching and learning, the dependency will affect accuracy of the index. It is necessary to adjust the confidence interval (CI) by FD. It is recommended that researchers/instructors monitor percent correct, its CI, and global and local FD of a student's responses. The combined information may show a student's learning behavior in short and long learning windows, and it may tell whether the student needs early intervention in the learning process.