Infant classes of 20 or under needed to close the achievement gap

OECD figures out today show how poorly the UK continues to compare internationally on class size. Primary class sizes rank 4th largest at 25.8 (compared to the OECD average of 21.5). Additional government figures reveal that in England's primary schools in 2007/08 the average class size was even higher, at 26.2 pupils per class. According to the evidence, this matters most in infant classes (for 4-7 year-olds, Reception to Year 2), which rose from 25.6 in 2006/07 to an average of 25.7 pupils per class.

England's 'small' infant classes oversized

• In 1997 the government pledged 'small' infant class sizes, setting a legal limit of 30 pupils per class: this definition of 'small' is invalid according to research
• Government figures show that the number of infants in classes with over 31 pupils rose from 21,060 in 2006/07 to 22, 810 in 2007/08

Defying the evidence

• The evidence shows that classes of 20 pupils or under for the first three years of school produce long-term benefits for literacy and numeracy, especially for lower-achievers
• Low-income pupils make particularly significant achievement gains in small infant classes

Millions and potential wasted

• Millions which should be being spent on teachers are being misspent on teaching assistants, whom evidence show have little impact on learning
• Despite ten years of increased investment under Labour the achievement gap persists and many pupils leave primary without the basics

'The government must finally commit to a proper class size reduction programme for infant classes if it wants to see real results,' commented Anastasia de Waal, Head of Family and Education. 'When the first years of primary school lay the foundations for pupils' future achievement, cutting corners rather than class size is a huge mistake.'

Whilst there is continuing debate over the value of cutting class sizes for all age groups, there is broadening consensus within the academic research that small classes are a vital investment (see Notes to Editors for evidence). This consensus is manifest in the fact that even the New Labour government, which has said that classes of 70 are viable, has made a statutory commitment to small infant classes. However, the details of this commitment reveal serious shortcomings.

The government's definition of 'small':

'We know small class sizes in the early years are important, which is why the legal class size limit for [infants] is no more than 30.'(DCSF spokesperson, Telegraph, 16th May 2008)

Academic research on class size, however, defines 'small' as being between 15 and 20 pupils in a class, with a minority of studies including up to 25 pupils. A finding supported by teachers:

'Having a class of 30 last year [2007/08] meant each child had much less quality teaching focus. High ability children rarely got challenged enough as there just wasn't enough time to get around to every child. With 80 per cent of the class EAL [English as an Additional Language] and coming into Year 1 at a very low level, one teacher and one teaching assistant wasn't enough to cater to all the needs in class and so progress was definitely hindered.' Year 1 teacher Rosie Collis

Furthermore, the government has failed to honour even this flawed commitment. Despite it now being illegal for infant classes to have over 30 pupils, loopholes over so-called 'exceptions' have meant that nearly 23, 000 infants last year were taught in classes of over 31, according to government figures. In the vast majority of cases the 'exceptions' were allowed on the grounds that pupils had been admitted outside the 'normal admission round'. Whilst this may be legally acceptable it is not educationally acceptable: the reality for thousands of pupils last year was oversized classes.

Why small classes matter

'In larger classes the CSPAR research [a longitudinal study on the impact of class size in England], for example, found that pupils were more likely to passively listen to the teacher as 'one in the crowd', whereas in smaller classes they were more likely to be asked questions and interact with the teacher.'

The Primary Review, Interim Reports: Research Survey 9/2, 'Classes, Groups and Transitions: Structures for Teaching and Learning,' 2008, p26

In small classes research (see Notes) shows that:

• There is more individual interaction between teachers and pupils
• There is more teacher support for learning per pupil
• There is more attentiveness to the teacher and therefore less disruptive behaviour from pupils
• Teachers spend more time teaching rather than 'managing' pupils (both behaviour-wise and in terms of assessment duties)
Teachers are better able to identify pupils' problems and specific needs

Increased opportunity for interaction between teacher and pupil leading to more 'on task' behaviour helps to explain the observed relationship between smaller classes and higher achievement.

Teaching assistants do not compensate

Between 1997 and 2007 there has been an increase of fewer than 35,000 teachers but an increase of 104,120 teaching assistants. Whilst this investment in teaching assistants is a clear admission that the government does indeed consider one adult with a large class to be unmanageable, the evidence shows that teaching assistants do not counteract large classes. Research from the largest class reduction programme to date, the Tennessee STAR Project, found that teaching assistants made no difference to either pupil performance or behaviour in large classes, a finding also replicated in research in the UK. (Gerber, S.B. et al. 'Teacher Aides and Student Achievement,' Educational Evaluation and Policy Analysis, Vol. 23, pp 123-143, 2000; CSPAR (see Notes))

Andrew Adonis recently admitted that New Labour's policies had failed to narrow the achievement gap at primary level:

'We still have what I term the stubborn 20 per cent who aren't reaching the standards we expect in literacy and numeracy by the age of 11.' 'Primary pupils without skills highlight Labour's biggest failure, says schools' minister,' Guardian, 21st August 2008

This underperformance in primary school is in turn leading to further underperformance in secondary schools.

Research evidence from both the US and the UK suggests that the achievement gap could be shrunk by cutting infant classes to a maximum of 20. Underachievement in England is strongly connected to low-income background; the majority of evidence showing a positive impact of class size reduction has shown particularly significant gains for low-income pupils (see e.g. the Tennessee STAR Project and CSPAR in Notes). Even the most influential critic of class reduction strategies, Stanford University's Eric Hanushek, has conceded that smaller classes positively affect low-income pupil achievement.

Yet the government is instead wasting resources on remedial measures. For example, last week Ed Balls announced 'catch-up' classes for struggling primary pupils:

'By intervening early and using the kind of personalised tuition and support through trained teachers that parents want, we're on the verge of something truly exciting happening in our classrooms...'

Truly effective 'early intervention', which would avoid the need for remedial strategies, would be to shrink infant classes. Furthermore, parents do not want 'catch up' classes but the consistent personalised tuition achievable in small classes. This explains why smaller classes are cited as the main reason parents move their children to the independent sector, where the average primary class size is around half that of the state sector.

'Academic research on class size backs parents' intuition - smaller classes are better because pupils get more attention from teachers,' commented Anastasia de Waal. 'For struggling pupils, this attention is especially vital to both their academic and wider development.'

Small classes are not a 'cure-all' however they are an essential prerequisite for maximising learning opportunities amongst infants.

Cutting infant classes down to a maximum of 20 pupils is within reach:

• There are currently a significant number of teachers working in supporting rather than class teacher roles. These supporting, or so-called 'pull-out' teachers would be better deployed as additional class teachers (American Educational Research Association, 'Reducing Class Size, What do we Know,' March 1999)
• 'Redemptive strategy' budgets, such as those for 'personalised learning', could be reallocated to expenditure on additional class teachers (Odden et al. 'Reallocating Resources: How to Boost Student Achievement without Asking for More,' Thousand Oaks: Corwin Press)
• Expenditure on teaching assistants, at least in the infants, could be spent instead on additional teachers
• As well as saving on remedial strategies, shrinking class size can save on teacher costs by lowering attrition rates (see Notes)

Research suggests that 'overnight' class size reduction programmes can be problematic, for example in relation to acquiring a supplementary pool of satisfactory teachers. Therefore a more effective way would be to phase small classes in. This has the additional benefit of enabling a reorganisation of resources, rather than requiring a sudden cash injection.

Phasing in can be achieved by:

• Initially targeting resources to greatest needs: focusing on schools with high proportions of low-income pupils, as measured by free school meal eligibility (O'Connell et al. 'Capitalizing on Small Class Size,' Eric Digest, Vol. 136, April 2000)
• Using 'floating' literacy and numeracy teachers, who go from class to class: for example utilising existing teachers in supporting roles.

Notes to editors:

Civitas is an independent social policy think-tank. It receives no state funding either directly or indirectly and has no links to any political party. Civitas's education research seeks to take an objective view of educational standards in Britain. It aims to offer an improved perspective on how best to deliver equitable and high standards of education for all.

For more information ring: Anastasia de Waal, Head of Family and Education 020 7799 6677 / 07930354234

Higher achievement shown in the below examples of the largest experimental US studies:

• STAR Project (The Student/Teacher Achievement Ratio), Tennessee (1985): in this 'experimental' research, the impact of classes of 13-17 ('small') were compared with classes of 22-26 ('regular'), using 7000 pupils in 79 schools. Pupils were randomly assigned to a small, regular, or regular class with a teaching assistant and each school had at least one of each of the three. The researchers found that pupils in smaller classes significantly outperformed their counterparts in the regular classes, including those with teaching assistants; minority and low-income pupils' performance found to be initially impacted on the most significantly; special educational needs were found to be identified by teachers earlier on in the small classes (American Educational Research Association 2003: Word, E. et al. 'STAR: Tennessee's K-3 Class-Size Study', Nashville, Tennessee State Department of Education, 1990). The Lasting Benefits Study, a follow-up study looking at pupils who had been involved in the project, found the achievement and behaviour benefits for those pupils who had been taught in small classes to still be evident after a further three years in regular sized classes. Project Challenge then implemented smaller classes in Tennessee schools with a high proportion of low-income pupils and found that results in both reading and mathematics went up (McRobbie, J. et al. 'Class Size Reduction: Lessons Learned from Experience,' Policy Brief No. 23, Wested, 1998)
• SAGE (Student Achievement Guarantee in Education), Wisconsin (1996): class size reductions (to 15 pupils or fewer) for the first three years in school were phased in to school districts with high numbers of low-income pupils. Results found that in maths and reading pupils outperformed their counterparts in larger classes; the gap between low-income and more affluent pupils narrowed (Molnar, A. et al. 'Evaluating the SAGE Program: A Pilot Program in Target Program in Pupil-Teacher Reduction in Wisconsin,' Education Evaluation and Policy Analysis, Vol. 21, No. 2, pp165-177, 1999)
• North Carolina, Burke County (1990): classes in the first three years of school were cut to 15, and professional development activities introduced for teachers. Compared to their counterparts in the county, pupils in smaller classes did better in both reading and maths tests; teaching time was found to increase and time spent controlling behaviour to decrease (American Educational Research Association, 'Reducing Class Size, What do we Know,' March 1999)
• Project Prime Time, Indiana (1981): early grade class sizes were reduced from 25 to 18 pupils in 24 elementary schools. The achievement of pupils in the smaller classes was higher in both reading and maths compared to pupils in larger classes in previous years (Hansen, A., Northern Arizona University, 'Research Brief: Class Size and School Size: The Major Studies in Class Size Reduction' for The Principals' Partnership, 2005)

Evidence from meta-analysis of empirical research on class size reduction:

• Smith and Glass, 1978: undertook a meta-analysis of 77 empirical studies looking at the relationship between class size and outcomes. They found that, overall, small classes, with small constituting under 20, were associated with higher achievement in all age groups if small classes lasted for a significant period (US Departments of Education: 'Reducing Class Size, What Do We Know?,' March 1999)
• Robinson and Wittebols, 1986: reviewed over 100 research studies using a so-called 'related cluster analysis' approach. They found that the most significant effects of small class size were evident in kindergarten and the first three years of school; they also found that reduced class size had a specifically significant impact on disadvantaged and minority pupils (US Departments of Education: 'Reducing Class Size, What Do We Know?', March 1999)

Some of the US research has been criticised methodologically; however reanalysis of the data finds that the main findings are supported - and apply also to this country.

For example:

• Krueger, Princeton University: re-analysed the Tennessee STAR data, correcting for methodological weaknesses. He found that smaller classes were systematically related to high pupil performance (Krueger, A., 'Experimental Estimates of Education Production Functions,' Quarterly Journal of Economics, November 2003)
• Goldstein and Blatchford, Institute of Education, University of London: also re-analysed the STAR data, correcting for methodological weaknesses. Their findings concurred with the original, with smaller classes having a positive effect on pupil performance (Goldstein H. and Blatchfod, P., 'Class Size and Achievement: A Methodological Review,' IoE, University of London, 1998)

Significantly less research on the effect of class size has been carried out in England, however the largest study to date, the Class Size and Pupil Adult Ratio (CSPAR) project undertaken by researchers at the University of London's Institute of Education, has shown a strong relationship between small classes and greater achievement. CSPAR analysed a sample of over 10,000 pupils from school entry until the end of Key Stage 1. The researchers identified a 'clear effect' in literacy and numeracy attainment, even after adjusting for other 'possible confounding factors'. Pupils entering school with low literacy levels progressed the most. The researchers concluded that the effect was comparable to that reported by the STAR project, meaning that the impact of class size reduction is supported by both 'experimental' (STAR design) and 'non-experimental' research (CSPAR design) (Blatchford, P. et al., 'Are class size differences related to pupils' educational progress and classroom processes? Findings from the Institute of Education class size study of children aged 5-7,' British Educational Research Journal, Vol. 29, 2003, pp 709-730; The Primary Review, Interim Reports: Research Survey 9/2, 'Classes, Groups and Transitions: Structures for Teaching and Learning,' 2008)

Further England-based evidence comes from Maria Iacovou, Institute for Social and Economic Research: using the National Child Development Study Iacovou found a 'significant and sizeable' association between smaller classes and higher attainment in reading in the early years of school (Institute for Social and Economic Research Paper 2001-10, 2001)

In addition and contributing to higher pupil achievement, the following benefits have been found as a result of class size reduction. (Benefits not necessarily picked up in studies focusing solely on test scores):

• Reduced teacher attrition (especially in low-income serving schools). Evidence from New York State showed that cutting classes to 20 lowered teacher leaving rates in the district by 4.2 percentage points (Pas, E., 'The Effect of Class Size Reduction on Teacher Attrition and Recruitment: Evidence from Class Size Reduction Policies in New York State,' PhD Dissertation, Syracuse University, June 2007)
• Improved pupil behaviour (Blatchford, P., et al., 'Do Low Attaining and Younger Students Benefit Most from Small Classes? Results from a Systematic Observation of Class Size Effect on Pupil Classroom Engagement and Teacher Pupil Interaction, Paper to Symposium: 'Class Size Effects: New Insights into Classroom, School and Policy Processes,' American Educational Research Association Annual Meeting 2008)
• Lower drop-out rate at the end of school: US research found that low-income students who were in small classes in the first four years of their education, were 18 per cent more likely to graduate from high school than their low-income counterparts who attended average-size classes (Finn, J., D., et al. 'Small Classes in the Early Grades, Academic Achievement and Graduating from High School,' Journal of Educational Psychology, Vol. 97, No. 2, 2005)
• Teacher workload reduced, therefore impacting on recruitment and retention and associated costs (e.g. retraining, loss of government investment in training)
• Better pupil social development (Belsky, J. et al. 'Does Class Size in First Grade Relate to Children's Academic and Social Performance or Observed Classroom Processes,' Developmental Psychology, 2004, 40, 651-664)

Additional sources: Krueger, A., et al., 2002 in Chubb and Loveless, 'Bridging the Achievement Gap,' Brookings Institute; Word, E. et al. 'STAR: Tennessee's K-3 Class-Size Study', Nashville, Tennessee State Department of Education, 1990; Hansen, A., Research Brief: 'Class Size and School Size,' Northern Arizona University for The Principals' Partnership, 2005