• Confessions of a Community College Dean

    In which a veteran of cultural studies seminars in the 1990s moves into academic administration and finds himself a married suburban father of two. Foucault, plus lawn care.


The Self-Selection Variable, (Mostly) Eliminated

What happens when everyone takes online classes?

October 29, 2020

It’s been a while since I posted a piece suggesting a possible research topic for scholars of higher education. It’s time for the drought to end!

We have years’ worth of data comparing student performance in online classes to student performance in on-site classes. But the data have always been plagued by questions of self-selection. Students generally chose their own classes, so it could be difficult to disentangle the effects of the modality from the effects of the students who chose the modality. Yes, we could control statistically for factors like race, age, sex and GPA, and those shed light on the question. But students are more than their demographics. Anecdotally, I’ve heard enough variations on “I’m too busy for a real class, so I’ll just take it online” over the years to wonder.

This semester, online education (and its new cousin, remote live education) has become ubiquitous. We’ve run every section of every English, psychology, history, business, math and biology class virtually this semester. Students didn’t have the option of taking the class in person.

Put differently, as the result of an external shock, we’ve (mostly) eliminated the self-selection variable this semester.

That doesn’t necessarily mean the data could generalize, of course. Any reasonable observer would conclude that a shift in modality isn’t the only thing stressing students out these days. But in an awful way, at least in this instance, COVID has been a great equalizer. High achievers, low achievers, the tech-savvy and the tech-phobic are all in the same sections.

Some thoughtful faculty have mentioned that their online classes are going unusually well this semester. I was pleasantly surprised. Then one particularly thoughtful one -- I’ll withhold the name to protect the innocent -- suggested that the comparatively low reputation of online classes had previously tended to deter the high achievers from taking them. As a result, online sections had been disproportionately populated with students who chose to go online because they thought it would be easier. But this semester, the high achievers have been pushed into online classes by default, raising the overall level of the class.

I believe this professor’s account of their own classes; I’d like to know if something like that is happening more broadly, or if they just got a good batch this year. With everyone shunted into online classes by default, is their average achievement level rising? This strikes me as an empirical question, of the sort that calls for some ambitious sociologists of education (hint, hint) to address.

If the self-selection hypothesis is substantially correct -- that differences in success rates by modality have less to do with the modalities themselves than with the students who choose them -- then the conversation about online education may have to change.

“Remote live” sections are another matter. This is the first full semester that we’ve done them at any meaningful scale. Admittedly, they add a degree of self-selection to the mix, so it’s not a perfect natural experiment. But natural experiments are rarely perfect. I’d refrain from drawing any significant conclusions about remote live classes at this point, given that this is the first time they’ve been done for an entire semester.

Any given class, or discipline, or even campus, can be an outlier. I’d like to find out whether the self-selection hypothesis is proving itself across states and disciplines. And if it is, is it doing so reasonably equitably, or is it reinforcing the existing stratifications we know so well?

As always, any insights from my wise and worldly readers would be welcome …


Be the first to know.
Get our free daily newsletter.


We are retiring comments and introducing Letters to the Editor. Letters may be sent to [email protected].

Read the Letters to the Editor  »



Back to Top