In our first intensive week of Econometrics and Data Analysis, we covered the basics of probability. My students can now model uncertain processes mathematically and use their models to answer questions. One prototypical example I teach is about an ATM machine. We assume a number of individuals make withdrawals in a given day and model the probabilities associated with the possible amounts of each withdrawal. We use the model to figure out how much money the bank should put in the machine at the beginning of the day such that it is pretty unlikely that it will run out. We also talk about the weaknesses of the model and how we might extend it to allow for an uncertain number of customers.
Now that we have the basics of probability down, it’s time to start analyzing data. I find this incredibly exciting and empowering. This is when we learn to estimate the parameters of the models we talk about in the first week instead of just taking them as given. If you have questions about how the real world works (and who doesn’t?), here is where you start getting real answers.
The strange thing is that many students come into classes on probability and statistics with the idea that it’s going to be boring. They are expecting tedious math and dumb examples. There are several concrete things you can do as an instructor to change these expectations.

Share your enthusiasm for analyzing data. I have found that two kinds of people teach statistical methods. The first group are people who actually analyze data for a living. They do it because they love figuring how to answer interesting questions and getting the answers. These people just need to be open about their enthusiasm and share their experiences in the trenches. The second group are people who develop new statistical methods for a living. They get excited about the math (which is great for them) and teach the class as a math class. This usually goes badly for the majority of the students who see statistics as a means to an end. I’m not saying the math isn’t important–it’s critical! I’m just saying most students need more than the beauty of the math as motivation.

Pose interesting questions first. Traditionally, statistics instructors start with a method, and then show how to apply it with a few fairly threadbare examples. It’s much more effective to describe a rich scenario in which we need new tools to answer a question. I didn’t follow this advice on Monday and regretted it. I should have started the class with a discussion of the relationship between age and health and wealth. Then, as we walked through real data from the Health and Retirement Study, it would have been clear why we were there.

Empower students to actually do analysis. At the end of the term, I want my students to have two concrete skills. They should be able to read and critique other people’s basic empirical research, and they should be able to conduct their own. At a minimum, that means you have to show them how to compute descriptive statistics and simple linear regressions with Excel. Personally, I think it’s worth the extra effort to teach more powerful software (e.g., Stata) so they have room to grow after the class is over.

Use data they care about. Economics majors are a surprisingly diverse bunch. I try to work with data across the board so I have something for everyone. It skews a little to my tastes in labor and development, but we also look at aircraft leasing firms, innovation in healthcare, and the stock market.

Make them practice. Students should be analyzing data early and often, so they get comfortable with the methods and the software.

Be patient. Probability and statistics are layered–you start with simple concepts and quickly combine them and build on them. These connections become obvious when you have years of experience, but they take time to develop. Students will need to hear the same things more than once in different ways. You (or a teaching assistant) should allocate as much of your time as possible to listening and gently guiding students along.