There is currently a conference taking place at Queensland University of Technology (QUT) Australia called the “National Summit on Student Engagement, Learning and Behaviour.” A lot of the folks in my Twitter timeline seem to be in attendance. The conference website makes it clear that it has been set up in reaction to calls for more traditional discipline in the classroom. So the conference is on a mission.
Whether as a result of this mission or not, people have been tweeting some pretty dodgy claims. During one session about school exclusion, the claim was made that, “Schools with higher suspension rates have poorer climates, higher drop out, lower achievement, more time on discipline.” This is clearly meant to imply that high suspension rates cause these other problems but it is a classic example of attributing a cause to what is actually a correlation. For example, it is quite possible that the poor climate (however that is measured) causes the high suspension rate or, alternatively, there may be some unmeasured or unmeasurable factor that causes all of these issues. A high level of social deprivation in the local community could plausibly cause all of these issues, so too could an inadequate behaviour management policy.
In another tweet the claim was made that being excluded once increases the likelihood of further exclusions. I hardly find this surprising. What is the point of this claim? It certainly doesn’t prove that exclusions cause exclusions because, again, there is likely to be a latent factor – e.g. propensity towards violence – that causes particular students to be repeatedly excluded. Perhaps the argument is that exclusions don’t cure bad behaviour. Is that the purpose of exclusion? I had assumed that it was primarily to keep other students safe and allow them to learn.
What if we control for all other factors and find that exclusion is the greatest ‘predictor’ of poor school climate and so on. Have we now proved cause? Well the language is seductive but all that it essentially means is that of the things we’ve measured, exclusion and school climate have the greatest correlation. This still does not tell us which way around this interaction works or whether there might be a factor that we haven’t measured that causes both; either of which scenario seems far more plausible that attributing cause to exclusions.
I am not making an argument for exclusion here. I am just pointing out that you can’t really argue against it with these correlations. If you think exclusion is morally wrong then make that case.