top of page

CSPAR Background

















There has been a vigorous debate about class sizes in schools. On one side of the debate are the enthusiasts who feel very strongly that smaller classes lead to better teaching and more effective learning. On the other side of the debate are the sceptics who argue that the evidence for the efficacy of class size reductions is in doubt and that there are likely to be other more cost effective strategies for improving educational standards.









It is clear that the overwhelming professional judgement of teachers is that, other things being equal, smaller classes allow more effective and flexible teaching and the potential for more effective learning. As we have described in our writings (see publications) there are a number of methodological limitations with much research on class size. There have been a number of influential secondary and econometric analyses of class size effects on pupil outcomes, but these are limited, e.g. because they often involve measures of pupil teacher ratios not class size and often do not have data on what goes on in schools and classrooms, which would help account for any effects found (or not found). Surprisingly, given its public importance for education, there has been little dedicated research conducted on this topic in the UK and elsewhere.


We believe that the CSPAR project therefore offers a distinctive contribution to the literature on class size effects. It started in 1996 and was initially funded by a consortium of Local Authorities, in conjunction with the Institute of Education. Later on, the research was funded by a grant from the English Government and the Economic and Social Research Council (ESRC).














The Key Stage 1 (KS1) study involved tracking over 10,000 pupils in over 500 classes in over 300 schools from school entry (at 4/5 years) to the end of KS1 (at 6/7 years), that is, over their first three years of school. It featured an impressive multi-method approach that involved the collection of information on a range of factors including: class size (as experienced by pupils on a moment by moment basis and as on the class register), the number of staff and other adults in the class, information on teacher and pupil behaviour in class from systematic classroom observations and teacher ratings of individual pupils, teacher estimates of time in different curriculum areas and activities, end-of-year reports by teachers of effects of class size differences, and detailed case studies of individual classrooms. Deliberately, the study set out to integrate quantitative and qualitative research.












The aim was to provide an integrated and comprehensive account of how class size affects classroom learning and behaviour.


Subsequently the study was extended into Key Stage 2 (7-11 years), following the progress of many of the same pupils from Key Stage 1, together with the addition of new pupils to the sample.


In a later study – the Deployment and Impact of Support Staff (DISS) project - we were able to examine the effects of class size on teacher pupil interaction and pupil classroom engagement across both primary and secondary sectors. This site has a separate section to this part of the study (link)

For many years there has been a widely perceived problem of large classes at primary level in the UK. At the time of the redesign of this website in 2015 there was still a heated debate about class size and a concern with the number of primary classes with over 30 pupils, despite the cap of 30 on class sizes at reception and Key Stage 1 introduced in the 1990s by the Labour Government.


  • to establish whether class size differences and pupil adult ratios affect pupils' academic achievement.
  • to study connections between class size and classroom processes which might explain any differences in attainment that we found. We looked at several main processes: within class grouping practices, teaching, individual support for reading, pupil concentration and peer relations.  

In contrast to the famous USA STAR project, it was conducted using a non-experimental design; in other words, rather than assign children and teachers to different class sizes, we measured the effects of natural variations in class size. We did this with a longitudinal follow-up study of children from school entry, and employed sophisticated multilevel regression statistical analyses and carefully designed methods of data collection, in order to determine effects of class size controlling for other potentially confounding factors, such as pupil prior attainment.

The project had two main aims:

bottom of page