Setting by 'ability': a philosophical problem
Updated: Nov 8, 2019
Despite the proliferation of evidence from educational research that it is harmful to student outcomes, setting by 'ability' remains popular with teachers, senior leaders, parents, and even students.
But it's not just empirical evidence which stands against setting and streaming; there are philosophical and ethical reasons why we should avoid it as well.
In philosophy we always begin by defining terms. But the concept of 'ability' is itself problematic. People generally mean something like 'academic potential' when they use it to justify setting, and 'previous academic attainment' when they put setting into practice. Which are totally different things!
The potential problem
'Potential' is more of an trajectory, albeit an optimistic one, than an accurate predication of a student's progress. But many students fall short of this 'potential': they might not work hard enough, it's true, but they might simply hit a plateau in their learning which stalls their progress for some time. Learning is not a linear process, so any trajectory has significant levels of error built into it.
This is clearly problematic when we consider the effects of setting on individual students. If we put a child into a set, we are, to some extent, determining the course of their future: students in higher sets tend to achieve at, or even slightly above their trajectories, and students in lower sets tend to attain grades lower than their trajectories. If there is error inherent in determining trajectories, then this can have an extremely strong negative influence on a student's future.
Ability to learn can't be assessed
Sometimes people suggest that 'ability' is more about how quickly and easily students pick up new ideas, and how well they can apply these in new contexts. I have to say that this is more what I mean by ability when I use the term. The problem is that this propensity can't be assessed, so can't be used to determine which classes students should be in.
Why do I think that this can't be assessed?
Firstly, new ideas come in all shapes and sizes. At any given point, a student may be able to pick up new ideas in one domain more rapidly than in another. And this is even within a single subject; so, in Chemistry, a student may pick up practical ideas very easily, but struggle to apply ideas about chemical bonding.
So, unless we determine the proportion of each different kind of idea making up each subject, and then balance out each student's rate of learning for each kind of idea, then any kind of assessment is based on a hunch, which isn't really good enough when we are talking about determining students' futures.
Secondly, I return to my original point that learning is not a linear process. In education we often talk about threshold concepts. Once a student has picked up a threshold concept then they can solve all kinds of problems which were previously inaccessibly to them. One way of thinking about learning as non-linear is to think about threshold concepts. These don't necessarily appear at monthly junctions in the curriculum, and students don't necessarily 'get' them the first time around. And a student who struggled with a previous concept, like balancing equations, which stalled their progress for a time, may then 'get' the concept of the mole, which then illuminates balancing for them, giving them a huge jump in their learning after a long period of plateau.
So, if learning isn't linear, then any 'assessment' of ability to learn, is ephemeral: it's only true for that lesson, not the week, the term, and certainly not for the year, which is how long most students spend in a particular set.
So how do we set?
When students are put in sets it is done on the basis of their previous assessment results, sometimes tempered by a bit of the 'personal touch' (I think she has a lot more potential than she shows in tests, or I think he picked up a lot of those marks by fluke in the test).
Prior assessment bears very little relation to 'ability' - either potential trajectory, or propensity to learn new ideas. The former because of the error issues. And the latter because performance is backward-looking and learning is forward-looking: they are fundamentally distinct.
The human touch is supposed to temper the error inherent in the prior attainment approach; the students who under or over perform. But it is still a hunch, and all the evidence suggests that teachers are more likely to overestimate the potential of boys vs girls, or under estimate the potential of students from underprivileged backgrounds. So the ‘human touch’ puts societal prejudice back in to the system. And that is ethically problematic when students’ futures ride on the decisions teachers make.
But should we set?
So assuming that it is possible to determine the 'ability' of students accurately (which it isn't), the question remains as to whether or not we should set students.
Since our purpose as teachers is to ensure the best outcomes for our students, I look briefly at two consequentialist theories of distributive justice (where education is the commodity to be distributed).
A utilitarian approach
The evidence suggests that setting by so-called ability has an overall negative effect on student outcomes for middle or lower attainers, and a minor benefit to higher attainers. The utilitarian approach suggests choosing the system with the greatest benefit for the greatest number: which in this case would be to ditch the sets.
(There are more students in the middle and lower group and the negative effects for them are not compensated by the minor benefit to higher attainers.)
The main argument people give me against mixed-attainment grouping is that it's "not fair" on higher attainers who are "held back" by the lower and middle attainers. But this critique has no power if you subscribe to a utilitarian view which is a non-individualist perspective. Essentially, if your main aim is to maximise student outcomes, then you should disband the sets.
What would you choose?
But the critique is powerful: what about those students who have lower outcomes without the sets? (Of course, we could also ask about all the students who have lower outcomes because of the sets!)
So let's turn to the ethical position known as 'justice as fairness'. Under this view, the policy to be adopted should be the one that a person would subscribe to if they were ignorant of a range of pertinent facts about themselves, such as their gender, academic potential, social class, sexuality, disability, etc. The idea is that behind this 'veil of ignorance' persons are unlikely to choose policies which oppress women, as they have a 50% chance of being a woman.
Rawls argues that behind the veil of ignorance, social inequalities are only allowed if the worst off would be better off than they would have been under a more equal distribution.
Setting by 'ability' increases inequality. Of course it does. It means that different students experience completely different styles of teaching. Teachers go faster with the 'high ability' groups, and more slowly with the 'low ability' groups, so the idea that the bottom sets will catch up is ludicrous.
So setting would be allowed if, and only if, it made the worst off (lowest ability) students better off. But it doesn't. Those are exactly the students most penalised by setting.
Imagine you are behind the veil of ignorance: you have no idea whether you are an A set, or an L set student, or female (whose ability is constantly underestimated by teachers), or from an impoverished background (even more significantly underestimated): would you choose rigorous setting? Having seen the self-fulfilling prophesy of students put into bottom sets, who give up on themselves and suffer with low self-esteem, I know I wouldn't.
So even if we could accurately determine students' ability, there are strong ethical reasons why we should embrace mixed-attainment and ditch the sets. And we can't even do that. So there.