Just Simulate It! Introducing Computing Through Simulation Studies in a Second Statistics Course

Abstract

Introductory data science courses provide students with a computational foundation beyond traditional introductory courses. Even in the intro stats course, many students are using R to support their coursework rather than applet or “point-and-click” software systems. Introducing computation earlier in the statistics and data science curriculum enables students to work deeper and sooner with real data sets. Changes in early statistics and data science education have a ripple effect across the curriculum. As the introductory courses are modernized, the later courses must change too. The class described in this paper is a second-semester statistical modeling course with a modern, post-data science flair. Regression models are introduced separately (multiple regression, Poisson regression, logistic regression) before being generalized as the generalized linear model (GLM). In this class, learners studied the patterns and behaviors of these models through targeted labs leaning heavily on simulated data. This course emphasizes the development of statistical intuition through hands-on learning experiences, rather than a set of rules for each situation.