P2-04: Bayesian Network Analysis of Log File Data from Students Using SMILES Interactive Statistics Songs


By Dennis Pearl, Matthew Beckman, Neil Hatfield, Yiyun Gong, & Sean Klavans (Pennsylvania State University); Larry Lesser (University of Texas at El Paso)


Abstract

In fall 2019, at Penn State University, approximately N = 170 college introductory statistics students did activities involving four interactive songs as a homework assignment to review for the final exam. The songs are part of the NSF-funded Project SMILES collection at https://causeweb.org/smiles/. The creation, design, and learning objectives of the interactive songs are detailed in the November 2019 issue of Journal of Statistics Education (https://tandfonline.com/doi/full/10.1080/10691898.2019.1677533). The four songs used in the current phase were "Height of Confidence", "Super Bowl Poll", "Everything's Unusual", and "Correlation Does Not Imply Causation." Log files tracking each student action were collected anonymously (approximately 17,600 interactions across 784 sessions) and analyzed using Bayesian Network methodology to estimate the probability that students have competency in the learning objectives given their performance on the tasks at hand.


Recording