You have /5 articles left.
Sign up for a free account or log in.
One of the limitations of political science as a science is that it’s impossible to rerun history to see what would have happened. Where would the U.S. be now if Palm Beach hadn’t used butterfly ballots in 2000? (This is our version of the “butterfly effect.”) No matter how good your budget, or how indulgent your IRB, it’s just not realistic to build another America and run the simulation. So we have to take information where we can find it.
This semester, as an outgrowth of [waves hands wildly], we’re running what social scientists call a natural experiment. It’s natural in the sense that we didn’t set out to run an experiment; circumstances dictated some changes that amounted to one. But it offers an uncommonly direct way to test some long-standing intuitions.
In normal years, we set certain enrollment figures that sections would have to hit in order to run. When enrollment was higher, that number was higher, on the theory that a classroom that held a small class could have held a larger one. As enrollment has dropped, we’ve lowered the threshold steadily, on the theory that the opportunity cost of a small section has shifted from a larger section to an empty room. But there was still a reasonable minimum in most cases. (The minimum always had exceptions -- the only section of a class required for graduation, or the only evening section that fulfills a given requirement, say. But those exceptions were understood as exactly that.)
This year it was much harder than usual to predict enrollments. We’d had online classes for a long time, and that area has been growing for years. In 2020 we added “remote live,” or synchronous online sections, for obvious reasons. Going into this fall, we had no way of knowing in April what things would be like in September, so we ran sections in multiple formats.
A few things happened. We got some extremely welcome federal money that helps offset the enrollment decline that COVID accelerated. We saw students split their enrollments in new ways, including a frustrating number switching from full-time to part-time. And the overall enrollment decline combined with the shift of some classes online to greatly reduce the classroom crunch. Suddenly the opportunity cost of a small in-person section was at an all-time low.
So this semester we’re running more sections with single-digit enrollments than we’ve done in recent memory. Under the circumstances, the economic argument against it is unusually weak.
Of course, there are still academic arguments in some cases. A public speaking class with two students in it really doesn’t make sense. And if a class only has a student or two registered, it’s entirely possible that they could drop the class in the add/drop period, at which point the professor is left high and dry. So it’s not carte blanche. But the larger point stands.
This is an accidental chance to test the theory that smaller sections will result in greater student completion and retention. We didn’t design the semester to do that, but here we are.
It’s not a perfect experiment, obviously. COVID is a beast of an intervening variable, and small sections aren’t distributed entirely at random; some programs are more highly enrolled than others. But natural experiments are almost never perfect; within social science, that’s accepted as the price of admission.
I can envision several possible outcomes. Maybe success increases in the smaller sections but not the larger ones; that would tend to support the theory. Maybe it increases in both, which is a somewhat murkier finding. Maybe it declines because the effects of the pandemic overpower any effects from smaller class size. Maybe class size winds up making little or no discernible difference. Maybe it helps certain groups but not others.
That matters for a host of reasons, not the least of which is that federal money isn’t permanent. When it goes away, it would be helpful to know whether smaller sections pay for themselves through increased retention. Here’s an accidental chance to find out.
The trick with a natural experiment is to recognize its existence quickly enough to capture its teachings before they’re washed away. This time, at least, we see it. We can’t rerun 2020, but we can give the institutional research office a pretty good workout …
