The head of the OECD PISA surveys, Andreas Schleicher, has been called the most powerful man in education. On the BBC website this month he described 7 big myths about top-performing school systems. Myth number 4 in Schleicher’s list is that small classes raise standards. He argues that “everywhere, teachers, parents and policy makers favour small classes as the key to better and more personalised education.” In contrast, he argues, high performing education systems invest in better teachers and high performing countries (many in East Asia) have large classes – so the size of a school class can’t be important.
Far from being a myth, however, my sense is that the view that class size is unimportant is in fact becoming more and more accepted by many involved in educational policy and planning, including think tanks, and politicians. It‘s been the view of OECD for a number of years, and is also accepted by the influential Sutton Trust toolkit and other leading reports including those from McKinsey (2007), Gratton Institute (2012), and the Brookings Institution (2011).
Given the huge importance of class size for educational resourcing it is vital that we are clear about the reliability of the evidence on the effect of class size. The logic behind the conclusions drawn from the PISA surveys seems reasonable. Countries and regions performing at the higher end of the attainment chart, particularly East Asian countries and regions like Hong Kong and Shanghai, have relatively large classes in comparison to the OECD average. It is therefore argued that class size cannot be important. What is more, these countries score higher in some other characteristics, such as teacher salaries, and it is therefore argued that these characteristics are more important.
But the argument here is limited in ways not often made clear. Essentially the problem is the familiar one of mistaking correlation for causality. There are a host of reasons why high performing countries do well: parental support, cultural factors, private tutoring, homogeneous demographics and so on. And there can also be cultural and other factors behind the status and pay of teachers, which affect academic performance. So to conclude that class size is unimportant on the basis of this kind of cross country comparison is not warranted.
It is important to recognise that the PISA studies are not dedicated studies of the impact of class size but secondary analyses of data collected by others at just one point in time. These kinds of one-off surveys (‘cross sectional’ designs) are much less powerful than studies over time (‘longitudinal’ designs) and are highly questionable as evidence of the causal relationship (or not) between class size and pupil attainment.
What is needed to address the causal effect of class size are dedicated studies with strong research designs, which deal with the problem of potentially confounding factors, like pupil and teacher characteristics. Two main studies do this, each using a different approach: the experimental STAR project from the USA (pdf), and a large scale longitudinal study in the UK (CSPAR [pdf]). These studies show that smaller classes do have a positive effect on pupil progress especially in the first 2 or 3 years of schooling. Recent French research has found similar effects.
The preoccupation of some in the West with the effect of class size on pupil academic attainment scores is limited in another way. Teachers’ reasons for liking smaller classes are denigrated by some as self serving. But in my experience teachers see the benefits of smaller classes in less obvious though still important ways: in terms of allowing the creation of better classroom relationships, classroom discipline, catering for pupils with special educational needs and so on. Interestingly in high performing East Asian countries like China, Hong Kong and Taiwan there has been a move recently to introduce smaller classes. This is not to improve academic attainment but to help with a move from teacher directed to student centred approaches, meeting individual pupils’ diverse educational needs, and the development of critical thinking and collaborative approaches.
The exclusive focus on class size and pupil attainment is also limiting because it ignores ways in which class size can affect important aspects of what goes on in classrooms on a moment by moment basis. In dedicated systematic observational research at the IOE we found that smaller classes at primary school level led to more individual attention for pupils and a more active engagement in classroom activities. In later research we also found that at secondary level it was low attaining pupils who benefited most from smaller classes in terms of reduced classroom discipline and their classroom engagement. So the view here is that performance on tests is not the only worthwhile educational outcome.
In one sense of course Andreas Schleicher is right – the quality of teaching is more important than class size. Class size reduction is simply what it says on the tin – it reduces the number of pupils in a class. In itself it is not an educational initiative and this is one reason why comparisons with interventions like one to one tutoring and metacognitive training (as in John Hattie’s well known ‘Visible Learning’ meta analysis) are not a fair test. The problem with Andreas Schleicher’s analysis, I feel, is that it considers the class size issue as a binary choice: either class size reduction or quality teaching. But teaching does not occur in a vacuum and teachers and pupils will be influenced by the contexts within which they work, one feature of which is the number of pupils.
I think the most important educational questions are about how to adapt teaching to make the most of having fewer (or more) pupils in a class. It amazes me that there is next to no research which evaluates the benefits of class size changes along with specified changes to teaching, for example, the introduction of collaborative group work, which might well benefit from smaller classes.
This is the background to an international network on class size and effective teaching, funded by the Leverhulme Trust, which I am directing. The project involves established researchers on class size effects from the USA, UK, France, China and Hong Kong each developing a distinct empirical and conceptual angle on class sizes and effective teaching. The aim is to move things on by developing an integrative framework with important practical benefits for education world wide. It is exploring the benefit of small classes for pupils with SEN in mainstream schools, pupils’ well being, and high quality classroom teaching and dialogue.
It strikes me that this type of focus is far more useful for education than perennial arguments over whether or not class size is related to pupil attainment.
If class size does not matter at all why not classes of 200 and pay the teacher 100 grand a year? One snag with large classes is that it increases the workload of the teacher. It might not affect results if teachers generally just absorb that but there must come a breaking point. What is it? 30, 40, 50…..200? Of course in the UK weaker children tend to be put in smaller classes so I should think that raw attainment would intuitively be higher in bigger classes simply because there are brighter children in them. Where differences in class size are marginal eg on a national basis in similar countries, sure probably other factors will have more effect but that is not the same as saying its a good idea in an individual institution to have class sizes of 40. Damn lies and statistics as usual.
A thoughtful article, thank you.
As I recall the arguments at the time, when the new Labour government legislated to limit Key Stage 1 classes in or around 1997, billions were spent in capping class sizes at 30. It was considered by some that the investment would have been better targeted elsewhere. There was little evidence to support the view that small class sizes raised achievement generally, unless the sizes were reduced by much more than could possibly be afforded – say by about half. Other initiatives to raise standards and reduce burdens on teachers were more cost-effective.
Class sizes of 15 are affordable if you manage to get rid of all the rest of the associated bureaucracy. The tendency is to increase rather than decrease support systems so it seems unlikely that much will change except at the margins and change at the margins is most likely to produce no real change in the outcomes.
A very interesting article. I picked up the same BBC report and commented on it in Local Schools Network.
I completely agree with you Peter, “to conclude that class size is unimportant on the basis of this kind of cross country comparison is not warranted.” My argument is that Schleicher does often confuse correlation with causality. I don’t know if he believes he is the most powerful man in education, but there are far too many, especially those in politics who have to apply quick fixes to ‘cure’ education, that do.
On the issue of impact, the evidence from America in particular identifies that with adjustments to teaching, to make good use of class-size reduction, it has to be in the region of halving the present general figure to around 15-17 before benefits become identifiable. Under this regime, with the size stable over about the first four years of schooling (YrR – 3) girls and pupils from disadvantaged backgrounds in particular show benefits in cognitive development up to high school at least.
You might also be interested in another posting on LSN, which discusses the impact of International Large-scale Assessments.
Good luck with the Leverhulme Trust project.
In Britain , as 16-18 school funding has dropped by 10% in the last few years and as other post 16 funding has fallen in real terms such as in FE, class sizes have inexorably increased at an exponential rate particularly for A levels and in many cases for level 2 provision in many FE colleges. as a consequence, any previous support in the classroom is the first to be cut which has an impact on science in particular ( setting up experiments etc.). Also, senior leaders tend to look at class sizes and instead of having two 15 pupils in each class there is now one 30 class. my experience and common sense says this cannot be beneficial for the learners be they above average, average or requiring more specific attention; feedback is harder to provide, homework takes longer to mark, the impact of independent learning is harder to recognise etc. many leaders quote Hatty but there is far too little current research on the impact of class sizes on overall attainment and achievement (including readiness for employment, social skills as well as subject grades)
Reblogged this on Chaos Theory and Human Pharmacology and commented:
When I was at school, I got on with my work and liked the relative anonymity that a class of 30 gives. I disliked having someone constantly watch over my shoulder as it puts you on edge and made me more likely to make mistakes. So I think that each pupil is different and a good teacher can teach accordingly. So no, I don’t think smaller and larger classes always mean better outcomes.