Working in Small Groups: Online Econometrics, Day 13

I am a huge believer in peer-to-peer learning and I love interspersing my in-person classes with small group exercises. It’s a nice break from lecturing and allows the students to actively work with the concepts I’m teaching that day. I walk around the room during these exercises and try to unstick any groups that get stuck.

Last summer I tried this technique a few times in my online class, and it didn’t go very well. I would introduce the activity in our virtual meeting room, and then everyone would move to their assigned break-out room. Watchitoo, our video-conferencing software, called these “playgrounds,” and it took about 30 seconds and several button clicks to get to one. That friction noticeably reduced the amount of wandering around the class that I could do.

The main problem with the Watchitoo playgrounds was the difficulty students had in sharing their work. In a physical classroom, one member of the group would usually write or type while the others looked over their shoulder to contribute. The closest my Watchitoo students could come was laboriously transcribing their math into a text chat.

Because Zoom doesn’t have a break-out room feature now (it is supposedly entering beta soon), I didn’t try any small group exercises this summer until Monday. That’s when I realized that all we really needed was multiple class meeting rooms. Our peerless online learning program manager set up two new rooms, and we went for it. Here’s what I sent my students on Sunday night:

Who Survived the Titanic?

On April 15, 1912 RMS Titanic rammed an iceberg in the
North Atlantic and sank more than 12,000 feet to the bottom
of the ocean.  About two thirds of her 2,224 passengers and
crew perished.  Tomorrow you will analyze passenger data to
see what characteristics predicted survival.

Start by downloading the titanic3.csv and titanic3info.txt
files from the course web site.  These files were prepared
by Thomas Cason of UVa using (primarily) information found
in the Encyclopedia Titanica.  They are distributed by Frank
Harrell of the Department of Biostatistics at Vanderbilt
University.

On Monday morning I told my class they had 45 minutes to go through the following five steps, and divvied them up among our three meeting rooms:

  1. Import the csv data and decide what variables are reasonable candidates for predicting survival. Don’t use whether or not they got into a lifeboat—this variable is almost perfectly collinear with survival.
  2. Create versions of these variables that are appropriate for inclusion in a regression.
  3. Estimate a simple model of survival that includes only sex and age. Don’t include interactions but do choose what you think is an appropriate functional form for age.
  4. Choose and estimate an expanded specification you think makes sense. You may explore multiple specifications in your search.
  5. Be prepared describe what you learn from your regressions.

In each room one student shared their Stata window so the rest of the group could see what they were typing and contribute to the analysis. We hit a few bumps on the road as the data files didn’t download cleanly from the course website (Canvas wanted to display the file in a textbox), and the exercise itself was just too long, but on the whole, I deemed it a success. It was fun, informative, and I was able to move between the rooms to help out much more quickly than I could last summer.

I imagine that with the new break-out room feature, this will be even better. I would love to be able to switch between rooms with a single click inside the app instead of exiting meeting rooms and entering others by following a link on a web page. Hopefully I will be able to try it as a beta tester in the next few weeks, but even without it, I’m definitely doing more small group exercises next time I teach this class.