By Yiyun Gong, Dennis Pearl, Matthew Beckman, Neil Hatfield, & Shunqi Zhang (The Pennsylvania State University)
Abstract
Web-based software applications are often used in statistics teaching environments. Collecting and analyzing log files as students interact with web apps allows for the evaluation of their understanding of concepts. Bayesian Networks are a valuable tool for this task. Over the past three years, students in an introductory statistical concepts course worked with a web activity associated with two interactive songs, “My Family’s Mean” and “Super Bowl Poll” at project SMILES (www.causeweb.org/smiles). This poster will demonstrate the use of Bayesian Network methodology to estimate the students’ competency in concepts associated with these songs based on their task performance and onsite behavior. By having replicated these Bayesian Network models of log file data from three different years under various learning settings (e.g., in 2020, the class was taught online synchronously due to COVID-19), we can compare and contrast the different student cohorts in their concept knowledge and ability to complete tasks related to those concepts successfully. Participants will see the utility of log files and the value of Bayesian Network models for assessing students’ understanding of materials and concepts in the low stakes environment of web apps.