Cause-blind solutions and the puritan sense of fairness

In social affairs, it very difficult to untangle multiple causes of a problem. Here is a recent example. Our DWF rates have risen during the pandemic. The causes are many and their relative weight is unclear. It could be that online modality is generally bad, or it could be that many of our faculty have not mastered it yet. It could be also what was happening in student lives: illness, family care, job losses, stress. Most likely, the decline in course completion rates is probably a confluence of these and other unknown factors. Disentangling the causes is a lengthy, and sometimes impossible process. True experimental studies are often impractical. The thing is – we will probably never know for sure.

Yet one can already hear opinions that, it was FOR SURE the modality, and we should all go back to what it was before. They assume that once you get students back to f2f classes, success rates will automatically improve. But we do not know if they will. We know anecdotally that some students have terrible time in online classes, while others actually prefer to stay online. We do not know how many are in each group, why they prefer one modality over the other, and whether their own explanations of their success or failure are correct. It may be the case that a student was depressed but blames Zoom for it. Or, a student aced all the courses online, but only because she or he stayed with parent, did not go out, and did not have to work at their part-time job as much. So, subjectively, they feel like the online instruction fits their style, while in reality it was something else. This is why surveys are not always helpful. We have no idea whether actual learning online is more or less robust than in f2f, and for which subjects, for which age groups, etc. The depth of our ignorance is so great that one actually has to study social sciences to appreciate it.

What happens if we jump into conclusion that is too closely tied to a presumed cause? We risk creating interventions that do not work. For example, we knew for decades that student engagement in campus life correlates with their academic success. Increasing extra-curricular activities seemed like a sure path to success. However, it is more likely that students who are more likely to engage are also more likely to be successful in classes. It is the king of all errors - confusing correlation and causation.

We should have robust hypotheses about causality, even thought we cannot test them properly. Having the menu of possibilities will help us design solutions that are cause-blind, which is to say, they are likely to work regardless of what primarily caused the problem. For example, designing more extensive incomplete policies would be an example of such a solution. It helps all students regardless of why they could not finish – because they were stressed, lost their job, could not cur it academically, were evicted from their apartment, or got high and lazy. These kinds of solutions will remove us from a moralistic assumption that only people who suffer at no fault of their own deserve help. Paradoxically, the intense interest in causality is fueled by the puritanic sense of fairness. The other approach is to help everyone, whether they deserve it or not. While in prevention, causality is important, in mitigation – not so much. Stop obsessing about the “why.”