Here we go again. The arguments over class size are with us once more. Now there are worries about ‘supersized classes’ for young children in school, the result of pressure on school places and the current fragmented state of local educational planning.
Large classes are a recurring worry, especially when experienced by the youngest children in school. Worries about this problem led the last Labour Government to introduce a legal cap of 30 on class sizes in England. Last year there was a debate in Parliament about perceived breaches of this rule by the then coalition Government, and now the problem is receiving coverage again. The concern is that the 30 maximum protection is being relaxed and this will have a negative impact on children’s education.
The educational issue here is whether there is a threshold beyond which class sizes have a negative impact, and in particular does it matter if class sizes increase beyond 30? An allied issue is whether there is an optimal number of pupils in a class? What does the research evidence tell us about these two things?
Perhaps the most widely reported claim said to be based on research is that class size needs to be decreased to below 20 for any effect to be found. This is perhaps one reason why some are not so concerned about class sizes over 30 – because it is already well beyond a class size that is felt to be significant and so one more child will not make a big difference.
As far as I can tell the evidence for the 20 threshold view is not strong. Often cited in support is the early ‘meta analysis’ by Glass and colleagues in the USA in the 1970s. This was one of the first of the now common meta analyses which seek to combine many different quantitative studies using one common metric but it is widely known that this study included some very odd and now very dated studies. More credible is the STAR research project which was conducted in the 1980s in Tennessee, USA. This is perhaps the single most significant educational research project on the class size topic in that it used an experimental design in which teachers and students were randomly assigned to smaller and larger classes from kindergarten to grade 3. It’s an impressive study but one of the drawbacks of experimental designs is that in order to limit the variation in the variable of interest, in this case class size, it compared the progress of students in small classes of 23 and very small classes of 17. Apart from the fact it is difficult to generalise from these to other class sizes, it is also noticeable that the midpoint of the two groups compared is 20! So the conclusion about 20 being important looks to be to some extent an artefact of the research design.
In order to get a good understanding of whether or not there is a threshold or optimal effect for class size we need research which measures the effect of different classes size across the full distribution of class sizes found in the country. This is what we did in our large scale longitudinal CSPAR study based at the UCL Institute of Education. In essence this multi method study examined the effects of class sizes as they occurred naturally in schools across England (rather than controlling for class size as in experimental studies) and controlled statistically for potentially confounding factors, like student attainment levels. The study analysed the progress of the same pupils from school entry (the reception year, 4 yrs) to Y6 (11 years). There was a large sample – over 20,000 pupils – and their academic attainment and many other measures were collected at the end of each school year. There was also a careful study of classroom processes connected to class size differences, using questionnaires, interviews and systematic observation approaches.
We found an effect of class size for the first two years in school and generally speaking the effect was linear, in other words there was a relationship between class size and educational progress in literacy and maths across the distribution of class sizes, with little suggestion the effect was more or less obvious for any particular class size. So decreases in class size had a beneficial and measureable effect on the academic progress of young children in KS1.
But one main limitation of the debate about class size is that the effects of class size are likely to be seen in a range of ways, not only in terms of academic performance, and it is important to try and capture these other educational processes. Interestingly, the few systematic studies we have on class size do show clear effects on classroom processes like teacher pupil interactions and pupil behaviour in class. Perhaps the best way of getting at these classroom processes is through moment by moment systematic classroom observations – not easy or cheap to collect. In the CSPAR we conducted such a method of data collection and found that in smaller classes there was more teacher individual attention for pupils, more pupil engagement and a more interactive role for pupils. In later work we have also found that smaller classes seem to benefit lower attaining pupils in secondary schools in terms of decreased off task behaviour and teacher criticism.
But this kind of research barely scratches the surface of the effects of class size. I find it amazing that we still know so little about other classroom processes which teachers feel are connected to class size. What about the pupils’ attitude to learning, their enthusiasm and confidence, their ability to learn independently, their ability to think critically and to develop personal creativity and practical skills. And what about outcomes for teachers including the provision of individual help for pupils and the assessment of progress. What about the ethos of the class, levels of well being, the level of cooperation and, the personal relationships of all concerned? It’s perhaps significant that recent speculation about the social and personal advantages afforded by independent schools might be connected to how such qualities can be more easily developed in smaller classes.