An Interactive Shiny Application to Build Intuition for Statistical Modeling with Interaction Terms

Presented by David McGowan
Masters Candidate in Biostatistics

Plan B Adviser: Ashley Petersen

Active learning is an essential component of a productive statistics classroom, even at the undergraduate and graduate levels. In recent years, Shiny apps have become a popular method of developing interactive statistical tools and lessons. Instructors can use R and Shiny to create webpages where students can engage with material on any statistical topic. This talk presents a new, interactive learning activity, housed within a Shiny app, for students to practice using interaction terms in linear regression. Using two publicly available datasets, students are guided through two activities and build their intuition for interaction models, predictor centering, and the effects of both on model interpretation. This activity is designed for a graduate course in biostatistics taken by incoming students of that program, some of whom have no background in statistics. After completing the activity, students are also able to download example solutions and the answers they themselves wrote. Extensions of this app, in terms of both content and technology, will also be discussed.