It seems inevitable that many conversations in education sink into an argument of correlation versus causation. As social science is complex and involves so many related variables, it is easy to exclaim “correlation does not equate to causation” and then stop discussing the matter.

It’s true that many correlations do appear between unrelated facts. Just take a look at the humorous and informative Spurious Correlations website by Tyler Vigen. It tabulates high correlations for seemingly completely unconnected facts, like between the per capita consumption of mozzarella cheese and the number of civil engineering doctorates awarded.

The organiser of the Programme for International Student Assessment (Pisa) test also has an amusing graph in which there seems to be a moderate correlation between the mean country score on the reading scale and a country’s ice-cream consumption.

It is unlikely that eating ice cream makes students smarter, and it also seems unlikely that doing better in tests increases their ice-cream consumption.

But take another look at the above example. Might there be a third explanatory variable? It is not unreasonable to think that richer countries perhaps do better on the Pisa test (again, for various reasons) and that richer countries consume more ice cream.

It’s an interesting discussion at the moment because of the proposed links between knife crime and exclusion. Whenever anyone proposes a connection, someone else claims correlation is not causation. And they do so in an effort to end that discussion.

This is unfortunate, and I think that we should approach things differently. When correlations are claimed, it seems most useful to first establish whether they might be related or whether there might be a “hidden” explanatory factor. And if they do seem related, we should then not jump to conclusions regarding causality.

We should, in short, be more open-minded.

Maybe the most reasonable explanation is that there is a bidirectional nature. I have previously written about this regarding motivation and achievement. Rather than nip discussions in the bud by assuming one direction or the other prematurely, we should look more deeply at the variables at play.

Can exclusions lead to increased knife crime committed by those excluded, and can knife crime lead to exclusion of those perpetrating it? The answer to both is probably “yes”.

Instead of dismissing claims, why should we not simply say “more research needed”?

Christian Bokhove is associate professor in mathematics education at the University of Southampton

This article originally appeared in the 6 March 2020 issue under the headline “Correlation vs causation: an easy shutdown to tough talk”