By Daisy Philtron (Colorado School of Mines)
Information
Interaction terms in a linear model allow the effect of one predictor on the response to vary based on the value of a second predictor. They are often among the least understood concepts in a linear modeling course. In this work we present results from a study designed to assess the effectiveness of two different activities exploring interactions. The class environment consists of 25 to 40 students at an engineering school. The students are almost entirely math, statistics, or data science students in their second, third, or fourth year. The class will be randomly divided into two groups. Students are assigned one activity to work through based on their group assignment. The activities both explore the same two data sets, each with an interaction present. One activity will rely on visualization, and other on mathematical manipulation of fitted models. After the activity, both groups complete the same low-stakes assessment to assess understanding of interactions. The following class period, each student will complete the other activity. We further assess interaction terms on the next exam.