This is an opinionated post based on how I teach my undergraduate econometrics course. It will not be for everybody. The title applies mostly to anyone who wants to do data science or econometrics with R. This is the second time I have taught this course with R, and I have changed it around in many ways that I think optimize the process for students. In this post, I’ll cover just two major changes:
This summer, I am overhauling my econometrics class in many ways, in part because I was pleased to recieve a teaching grant from my college to make more R resources for my econometrics class. Last Fall was the first time I had taught it using R, and I’ve learned a ton since then. Expect a flurry of posts in the coming weeks more on those topics.
This post, however, explores some of the trends that I have been thinking about in teaching econometrics, and something monotonous that I have been struggling with that encapsulates the tension in these trends: what to name my course.
One of the advantages of using Stata for linear regression is that it can automatically use heteroskedasticity-robust standard errors simply by adding , r to the end of any regression command. Anyone can more or less use robust standard errors and make more accurate inferences without even thinking about what they represent or how they are determined since it’s so easy just to add the letter r to any regression.
As I have mentioned in other posts, this semester I trying to maximize the number of intuitive visualizations for my students to master economic and mathematical concepts more in my teaching, as well as develop my own R skills. For my intermediate microeconomics course, I recently put together an interactive Shiny App with R that demonstrates the effect of a tax on consumers and producers.
Simply input your own (inverse) demand and supply functions and the size of the tax, and the graph and summary will update with the market equilibrium and the surpluses lost by the tax.
For my econometrics course this semester, I have been using R to help students visualize linear regression models. Running a regression in R is quite simple, as is intepretting the results, with a little bit of training. However, I emphasize that I want students to understand what is happening “inside the black box” of regression. I discourage blindly trusting R's opaquely simple input and output, and get students to learn what R is doing under the hood, even if they will never have to manually estimate the model themselves.
This semester, I am posting my lecture slides for my Econometrics class on GitHub. While I usually post my lecture slides for my courses on my website or link to my Dropbox, those are the final PDF documents. On GitHub, I am posting both the final PDFs as well as the source .rmd files. I am primarily doing this for my econometrics students, who will be learning R and R Markdown for their assignments (which is how I write my slides), but this is also open to anyone.
With the new school year in full swing, I have redesigned my personal website, and plan to make occasional posts on the tools I use in my research and teaching. Over the summer, I made the full conversion to using R, R Markdown, and Github for nearly everything I do in my professional life (including managing my website with Hugo/Academic).
This semester, I am teaching econometrics to my students using R for the first time, and optionally nudging them to use R Markdown for their homeworks and paper assignment.