Top menu

Teaching graduate statistics online? Some thoughts on pitfalls and opportunities

The following blog post was written by Dr. Jay Verkuilen. This post is part of a blog series about transitioning to distance learning during the COVID-19 pandemic. 

I’ve taught graduate statistics since 2007 and a good bit of undergraduate statistics before that. I’ve only taught online a few times due to necessity. COVID-19 has pushed a lot of instructors to have to adapt very suddenly. I’ve been fortunate enough to be on sabbatical this year and thus not teaching right now but have good reason to suppose that instruction in the fall and likely the spring next year will be online. More broadly, more and more of my instructional work going forward will be online (see below). As such, I’ve been thinking of how to adapt, as well as benefitting from the experiences of others grappling in real time with this problem.

There are, of course, many class videos available out there and I have watched many, mostly ones from the pre-COVID-19 era. I’m not actually sure they’ve done much beyond record what they normally do to a room full of students. This isn’t unusual more broadly: New technology tends to be used more or less in the same fashion as the old one was until people start learning how to do new things. This takes a while and we really are only about a decade into this process. Think of how little most researchers take advantage of the possibilities of online data gathering, for instance to generate planned missingness designs to reduce respondent burden or to make use of survey experiments. Instead they tend to simply dust off the same pencil-and-paper instruments using answer formats that wouldn’t surprise Rensis Likert in 1930 and put them online.

We know from a lot of research that a course which is all direct instruction is not likely to be that useful, even when students express satisfaction with it, such as listening to a very engaging lecturer. Students need to engage with the material and ask questions, do things, write notes, and so forth, to gain a deeper understanding of the material. In a lot of ways, a good statistics class is structured and interactive. The instructor’s job is to structure it and both the instructor and students need to interact, be it synchronously (i.e., during class or lab) or asynchronously via the students doing assignments.

Pure lecturing for two hours is unlikely to work even in person—I mean, it’s boring—but it would be much worse online, where it’s even easier to tune out due to the loss of nonverbal cues, various internet distractions, etc. Asynchronous lectures may help, though, because students could hit rewind or listen at their own pace. Asynchronous lectures also help reflect the fact that students’ schedules and lives may be very different in the COVID-19 world and may have issues such as childcare, computer use, and work schedules.

It doesn’t really fit what I do anyway. I don’t run through prepared slides. Part of that is I tend to have a lead foot when I’m going through slides, which means students can’t actually keep up. Board writing keeps me on the same pace as a student who’s also writing notes. I actually don’t much enjoy writing or talking to slides, though I do often prepare and hand out demos that we work together. Finally, I go off-script a lot based on what students bring into the class. I’ve even thrown out the script of a class completely when students were clearly confused about some big picture issues. By contrast, having a collection of slides creates an implied obligation that we will get through them. The ability to go off-script is something I’d like to preserve but, ironically, it’s going to take some work and pre-planning to make that happen due to how much more reliance on technology forces things into channels.

GC’s class periods are two hours. As such, I am thinking of the following:

  • Prepare a lecture that lasts about 45 minutes. Students should watch it in advance, much as they should do the reading. The goal in this case isn’t for them to understand everything, but to get an orientation, the highlights of the reading or my thoughts on the general topic. In addition, prepare some short voice-over demonstrations of software. Post these three or four days before class with the understanding that students should have watched these before class begins.  
  • Solicit questions from the students two days in advance of the class, possibly with some questions or discussion points posted as seeds. 
  • Use the class time—maybe a shortened version of class—to answer these questions, have a class discussion, or run a software demo that’s more interactive. This would be recorded so that students who may not be able to attend would have the ability to see and hear what was discussed. Students would have access to the code in advance. Of course, attendance in real time has the benefit of allowing students to follow up or elaborate.

One issue involves recording. Quite honestly, I’m rather microphone- and camera-shy so I am not actually particularly happy to be recorded, although I’ve come to an uneasy truce with it. Furthermore, I know a lot of students don’t want to be recorded for posterity either. As such, I think I’d be much more comfortable having any class discussion periods that are recorded exist only for the length of the class and not in perpetuity. The lectures can be reused, though of course one shouldn’t let them get too stale.

Many professors are already well ahead of me, especially folks who’ve been teaching “flipped classrooms.” While some of this seems like fad chasing or ways for administrators to turn classes into cash cows, all of a sudden, it’s become quite necessary and there’s going to be a lot more attention being paid to this problem. Taking more courses online solves a lot of issues for them: It helps deal with the issue of space and facilities. It may well even help bend the cost curve of higher education induced by Baumol’s cost disease. It’s also likely that more of modern life will have to move online to adapt to climate change and to deal with issues like transit congestion that feed it. This is a particular issue in a metropolitan area like New York City, where commutes are often breathtakingly long, and transit is already congested, but is a huge issue in many other areas where gridlock traffic is rife. In sum, we should expect online instruction, and online life more broadly, to grow in importance in the future, even after the current acute phase of COVID-19 recedes. It behooves us to learn how to do it well.


Jay Verkuilen is Associate Professor of Educational Psychology in the Quantitative Methods track. He is also affiliated with the Quantitative Masters in the Social Sciences program. His Ph.D. is from the University of Illinois at Urbana-Champaign and he has been a faculty member at Graduate Center since 2007. His research is on methodology as well as substantive applications, primarily in the psychometrics of anxiety disorders and job burnout. He teaches courses on Linear Models (EPSY 833, Fall annually), Generalized Linear Models/Categorical Data Analysis (EPSY 835, Spring annually), Item Response Theory (EPSY 832, Fall biannually), and Mathematics for Social Scientists (EPSY 740, Spring annually), and is planning a course on Multivariate Statistics with Machine Learning (Fall biannually).

No comments yet.

Leave a Reply

Powered by WordPress. Designed by WooThemes

Skip to toolbar