Education-line Home Page

Ability grouping in the secondary school: the effects on academic achievement and pupils’ self-esteem

Judith Ireson, Susan Hallam, Peter Mortimore, Sarah Hack, Helen Clark and Ian Plewis

Institute of Education, University of London

Paper presented at the British Educational Research Association Annual Conference, University of Sussex at Brighton, September 2-5 1999



In England and Wales there is renewed interest in the issue of ability grouping in schools. There is a concern to raise standards, whilst retaining a system of comprehensive schooling. This paper presents results from a project exploring relationships between different types of ability grouping and both academic and non-academic outcomes for pupils. The sample comprises all Year 9 pupils (aged 13-14 years) in 45 secondary comprehensive schools in England, representing three levels of ability grouping in the lower school (Years 7 to 9). Measures of pupil attainment in national tests at the end of primary school (Year 6) and at the end of Year 9 are used to indicate academic achievement. All the pupils took tests in English, mathematics and science as part of the national assessment of this year group during the summer term. Test scores for these pupils at the end of Year 6 were collected retrospectively. Measures of self-concept and attitudes towards school were obtained through self-completion questionnaires. Data on attendance were collected from school records. The analysis of the effects of ability grouping on pupil attainment and self-concept reveal differences according to curriculum subjects. Overall, there are small effects of ability grouping on pupil attainment. Effect sizes differ from one curriculum subject to another and pupils attaining higher levels on entry to secondary school make greater progress in schools adopting setting in mathematics, but not in English or science. There are also small effects of school type on measures of self-concept, with pupils in the schools that adopt more mixed ability groupings having higher academic subject self concepts.


In England and Wales there is renewed interest in the issue of ability grouping in the context of a concern to raise standards, whilst retaining a system of comprehensive schooling. Many schools are currently reconsidering their grouping practices. The school system in the UK presents a particularly interesting context for research into ability grouping. Since research in the 1960s and 1970s indicated the disadvantages of rigid streaming and setting, there has been a reduction in the use of ability grouping and an increase in mixed ability organisation. In a recent survey, Benn & Chitty (1996) found that extensive streaming (tracking) for all academic subjects was rare, although some schools grouped pupils by attainment in most academic subjects. About half the schools in the survey had adopted mixed ability grouping for all subjects in Year 7, but by Year 9 most had introduced setting in at least some subjects. This means that there is a variety of grouping practices in the lower school.

Although there is an extensive international literature on the effects of grouping pupils by ability, there has been very little recent research in the UK. Radical changes to the education system in this country mean that much of the earlier and overseas literature may not be directly relevant in the current situation. In particular, we now have a national curriculum together with arrangements for a national system of tiered assessments of pupils at ages 7, 11 and 14 years. Pupils are assessed in relation to a framework of attainment targets within each subject, organised in levels corresponding to each year of schooling. The system enables pupils to perform at levels well above or below the normal level for their age group, while remaining in teaching groups based on age.

These changes in the education system have been accompanied by recommendations for schools to adopt setting by ability. Setting (referred to as regrouping in the American literature) is a system of grouping pupils according to their attainment in a particular subject. A pupil may be in a high set for one subject and a lower set for another. Setting may be carried out across a whole year group or within ability bands or timetable halves. Setting differs from streaming, in which pupils are allocated to classes on the basis of a test of general ability and remain with their class for most lessons. A report from the National Curriculum Council (1993) argued for greater use of setting to cover the requirements of the National Curriculum and ensure that the objectives of the Education Reform Act (1988) were achieved. More recently, the White Paper ‘Excellence in Schools’ (1997) stated that ‘Unless a school can demonstrate that it is getting better than expected results through a different approach, we do make the presumption that setting should be the norm in secondary schools. In some cases it is worth considering in primary schools’. The assessment of pupils in terms of levels also facilitates grouping by attainment. The test results at the end of primary school, Year 6 (Key Stage 2), are being used by some schools to place pupils in sets during Year 7 and to assess ‘value added’ at Year 9 (Key Stage 3).

The evidence from previous research indicates that the effect of ability grouping on pupil attainment is limited. British studies (Acland, 1973; Barker Lunn, 1970; Fogelman, 1983; Kerckoff, 1986; Newbold, 1977) and international reviews (Kulik & Kulik, 1990; Slavin, 1987; 1990) indicate mixed findings for the effects on academic achievement. Two important British studies were based on data from the National Child Development Study. The first found little difference in performance on standardised tests of achievement in mathematics and reading when ability level was controlled, but there were differences in the patterns of entry in national examinations and in access to the curriculum (Fogelman, Essen & Tibenham, 1978; Fogelman, 1983). The second study compared pupils attending four types of secondary school, secondary modern, grammar, comprehensive and private (Kerkhoff, 1986). Data were collected at ages 7, 11 and 16 years. Standardised tests of reading and mathematics, and verbal and non-verbal scores from a general ability test administered at age 11 were used. Children attending grammar schools showed relatively greater improvement in mathematics over time, compared with those in the secondary modern schools. In schools that grouped pupils by ability, there was a marked divergence of attainment, with students in remedial classes falling further behind, while those in the high ability groups increased their average performance beyond that exhibited by comparable pupils in ungrouped classes. The pattern was sufficiently clear that it was possible to differentiate between the effects of a two and a three-group system, the latter producing a greater divergence of attainment scores.

Kerkhoff concluded that there were positive effects of ability grouping on academic achievement, but his analysis has been criticised by Slavin (1990). Grammar schools and secondary modern schools exist in a selective educational system, with the more able pupils attending the grammar schools. There were significant pre-test differences between the pupils in the different types of school. Slavin argued that the significant pre-test differences could not be sufficiently controlled by the use of covariance techniques. He also claimed that the gains made by the lower ability students were actually greater than those of the high ability group. Kerkhoff himself acknowledged some methodological difficulties in the tests, indicating that there were floor and ceiling effects which would serve to reduce and distort the size of the effects.

There have been a number of smaller studies, based in single schools. Two of these investigated the impact of different forms of grouping within a single school and found that streaming had little impact on the performance of high ability pupils whereas pupils of lower ability performed better in mixed ability grouping. Rudd (1956) found no difference in the attainment of two groups, streamed and mixed ability, in the same secondary school. In a study based in one comprehensive school with mixed ability and streamed ‘halls’, Newbold (1977) found similar mean scores for a variety of measures of achievement including standardised tests of ability, examination performance and teacher devised tests. There were, however, larger standard deviations in the streamed sample, except in free writing, indicating that streaming tended to widen the range of attainment. More differences were reported within the two systems than between them. There was no evidence that the high ability pupils were performing differently in the two systems, although the low ability pupils made significant gains in the mixed ability classes. A follow up study (Postlethwaite & Denton, 1978) demonstrated better overall performance by the less able pupils in the mixed ability situation, without any reduction in the levels of attainment achieved by the more able.

In a selective grammar school, Lacey (1970) found that pupils in the top group received more attention and resources, leading to higher levels of achievement, but that this had negative effects on children in the lower groups. In a follow up study, after the school had introduced mixed ability grouping, Lacey (1974) showed that the attainment of the most able pupils was unaffected by the change, whereas the attainment of the less able pupils improved. The ability range of the pupils in this study, however, was clearly restricted by the selective nature of the intake.

Slavin (1990) reviewed 29 studies, using a method of best synthesis, and concluded that the effect of ability grouping on academic attainment was limited. Each study compared ability grouped classes with heterogeneous, mixed ability, grouping. He concluded that comprehensive between-class grouping plans have little or no effect and that the effect of ability grouping was essentially zero. The strongest evidence is provided by studies using randomised or matched groups. Taken together, these show median effect sizes for high, average and low achieving pupils are +0.05, -0.10, and -0.06 respectively, which are still very low.

Such effects as are in evidence appear to occur through affecting opportunity to learn, through differences in the curriculum or the pacing of lessons. Where groups proceed at the same pace and cover the same curriculum there is little difference in learning outcomes (Hallam & Toutounji, 1996; Ireson & Hallam, in press). Using meta-analysis, Kulik & Kulik (1982, 1992) examined research on tracking in secondary education. They found the largest effects in programs designed for gifted and talented pupils. The average effect size was 0.41 and in 22 of the 25 studies, students in the enrichment programmes achieved more. Some programmes allowed students of high academic aptitude to enter accelerated programmes which enabled them to proceed more rapidly through their schooling When compared with pupils of the same age, the accelerated children did better, with an average effect size of 0.87. When compared with older, non-accelerated groups, the effect size was negative. Kulik & Kulik concluded that the clearest effects on achievement are obtained in enrichment programmes and accelerated classes, which involve the greatest curriculum adjustment.

One of the concerns of those who argue against grouping by ability is that placement in the bottom groups has an adverse impact on pupils’ self esteem, self concept and on their attitudes towards school and schoolwork (Gamoran & Berends, 1987; Lacey, 1974; Oakes, 1985). Gamoran & Berends (1987), reviewing the international literature, suggested that there was a negative impact of ability grouping on the motivation and self-esteem of students in the lower groups. Oakes (1985) suggests that for low track students the self-concept becomes more negative as years go by and these students tend to be critical of their ability. On the other hand, Kulik & Kulik(1992) found no overall effect of ability grouping in 13 studies, but ability grouping tended to raise the self esteem scores of lower aptitude students and reduce the self esteem of higher aptitude students. The differences in these studies may be partly because many different measures have been used in the research. Marsh and his associates have demonstrated that general measures of self concept are only weakly correlated with academic attainment, whereas sub-scales of mathematics and verbal self concept correlate more highly with attainment in these content areas (Marsh, Parker & Barnes, 1985). In addition, the effects of ability grouping on self esteem may be mediated by the behaviour of teachers and peers and by the ethos of the school (Ireson & Hallam, 1999).

There is some evidence that streaming can lead to anti-school attitudes and alienation among those in the low streams (Gamoran & Berends, 1987) but it may be that negative school attitudes do not result from streaming, banding or setting but that grouping procedures merely reflect social alienation. The relationship is likely to be complex. Pupils in low streams tend to be labelled as slow or difficult and these descriptions can become self-fulfilling prophecies. Research on the social climate within the classroom also indicates that peer relationships are more supportive in high ability groups, although these classes also tend to be more competitive. Low ability classes tend to be characterised by more angry, hostile interactions (Gamoran & Berends, 1987).

On a wider level, structured ability grouping can be perceived as denying educational opportunity to particular groups of pupils. There is evidence that low ability groups tend to include disproportionate numbers of pupils of low socio-economic status, ethnic minorities, boys and those born in the summer (see Hallam & Toutounji, 1996; Ireson & Hallam, 1999). There are also difficulties associated with the allocation of pupils to streams or sets. Selection error is a particularly serious problem in a selective school system, where small differences in test performance may lead to substantial differences in opportunity to learn and in future employment. In a non-selective but streamed, or tracked, system the effects of selection error may be less marked but nevertheless significant. Allocation to groups is based not only on prior academic achievement or ability but also on school organisational constraints (Jackson, 1964). In theory movement between groups is possible, but in practice it is restricted, because of the increasing gap in curriculum covered.

Schools in the UK have for many years been responsible for the form of grouping they adopt, albeit with advice from the local education authority. Since research during the 1960s and 1970s indicated the disadvantages of streaming and setting, there has been a move towards mixed ability grouping and a reduction in streaming. In a survey carried out in 1970, 70% of schools reported streaming in the first year (now Year 7) and only 22% used mixed ability grouping (Benn & Simon, 1970). In a recent national survey the proportion adopting streaming had dropped considerably (Benn & Chitty, 1996). The survey showed that about half the schools adopted mixed ability grouping for all subjects in Year 7, but there was increased setting and banding as pupils moved through school and in Year 9 most schools grouped by ability for some academic subjects. It is clear that a considerable variety of grouping arrangements exists in the lower secondary phase and that the most common form of ability grouping is setting, in which pupils are regrouped on the basis of their attainment in different subjects.

The changes in the UK education system, outlined above, have lead to a rekindling of interest in the ways that pupils might be grouped within schools. There is a perceived need to raise standards nationally whilst retaining a comprehensive system. In addition, some schools have been experiencing difficulties in relation to the behaviour and attendance of some pupils. Grouping arrangements are needed that enable pupils of all abilities to make maximum progress without increasing alienation and disaffection. As the brief review of the literature demonstrates, researchers have tended to consider single outcomes, either related to attainment or to social outcomes, such as self-esteem and attitudes towards school. There is a need for research considering the relationships between different types of outcome in schools adopting different levels of ability grouping, to inform policy and enable schools to make the maximum use of resources for the benefit of all pupils.

There are considerable methodological difficulties accompanying research comparing grouping in different schools. The categorisation of schools as adopting one form of grouping or another is difficult when different types of grouping operate in the same school simultaneously. The use of different types of measurement also creates difficulties for comparing results across studies. The evidence also suggests that outcomes within one school are not consistent over time, across subject domains or between teachers. There appear to be complex interactions between grouping, teaching methods, teacher attitudes and the ethos of the school.

The current research project, which is funded by the Economic and Social Research Council, aims to take account of the problems in previous studies and develop an explanatory model which will further our understanding of the complex relationships between different kinds of pupil grouping, educational outcomes for pupils and the mediating processes operating at school, class and teacher level. It aims to provide information about school practices in relation to ability grouping, including the procedures for allocating pupils to groups and moving pupils between groups, the allocation of resources to groups, teachers attitudes towards grouping and teacher self reports of classroom practice with different types of group. This paper will present an analysis of the data on the impact of ability grouping on academic attainment and report briefly on the impact on pupils’ self-concept.



A stratified sample of 45 schools was selected for the study, representing a range of grouping practices, intake and location. The sample comprises three levels of ability grouping in the lower secondary school (Years 7 to 9), with 15 schools in each level.

'Mixed Ability Schools' (MA): with predominantly mixed ability classes for all subjects with setting in no more than two subjects in Year 9.

'Mixed Ability/Set Schools' (MASET): with a gradual increase in setting, with setting in no more than two subjects in Year 7.

'Set Schools' (SET) with streaming, banding or setting in most academic courses from Year 7.

All schools had received satisfactory inspection reports during the three years before the start of the project. The three groups of schools were balanced with respect to intake, using free school meals as an indicator. The mean percentages and standard deviations of eligibility for free school meals are for the Mixed Ability Schools, Mean = 16.3, s.d. = 14.95; for the Mixed ability/set schools, Mean = 13.2, s.d. = 13.99; for the Set Schools, Mean = 14.1, s.d. = 12.73. These figures indicate that the intake of pupils to the Mixed Ability schools is somewhat more socially deprived than the intake to the Set Schools, but that the distributions overlap to a large extent.

In addition, very large and very small schools were excluded from the sample and the three groups of schools were balanced in terms of size. Mean numbers of pupils on roll were 969.1 (s.d. = 212.24), 993.7 (s.d. = 212.56) and 867.7 (s.d. = 181.16), with the set schools on average slightly smaller than the other two groups, but again with good overlap of the three distributions.

Data Collection

Within the schools, the cohort of Year 9 pupils was included in the sample. Data was collected from all teachers of lower school classes of English, mathematics and science and all heads of department. Information has also been collected through interviews with the head teacher and with those responsible for the allocation of pupils to groups. The data collected at each level is as follows.

Pupil data: For all Year 9 pupils in each school, attainment data has been collected from performance in national tests at age 7 years (Key Stage 2 tests) and at age 14 years (Key Stage 3 tests). Pupils have also completed a questionnaire including scales measuring self concept, attitudes towards school in general, their views about ability grouping, their attitudes towards English, maths and science and their perceptions of teaching in these subjects, their plans for future education and an indicator of parental interest in education. Data on ethnic origin attendance and exclusions from school has also been collected from school records.

School data: Schools have been selected on the basis of their grouping practices. Further information has been gathered concerning the processes and criteria used in the allocation of pupils to groups, the extent of and procedures for movement between groups and the allocation of resources to groups. These data have been collected by questionnaires and by face-to-face interviews with members of staff with responsibility for the allocation to groups in each school.

Teacher data: Information on teachers’ attitudes towards ability grouping, their perceptions of the main problems of teaching the different groups and their classroom practice has been gathered through questionnaires.


Data from over 6000 pupils in 45 schools, 15 in each level of grouping, have been entered in to the analysis. The measures of attainment are the Key Stage 2 and Key Stage 3 test scores in national assessments of English, mathematics and science. The Key Stage 2 tests were taken by the cohort of Year 9 pupils when in Year 6, the final year of primary school. Year 9 pupils took Key Stage 3 tests at the end of the year, when the majority of pupils were aged 14 years. The test marks were collected for the research. The Key Stage 2 test results were collected from the secondary schools where possible. Some schools reported they did not have the data and the researchers then contacted the pupils’ primary schools to ascertain whether they had records. Many of the primary schools were able to supply this information. In the 45 schools, the sample of pupils for whom we have both Key Stage 2 and 3 results is 4480 in English, 4337 in mathematics and 4499 in science. The distributions of Key Stage 3 marks of the pupils for whom Key Stage 2 data were and were not available were compared, for each curriculum subject in turn. This analysis demonstrated a tendency for pupils for whom Key Stage 2 data were available to attain slightly higher marks at Key Stage 3 than pupils for whom the Key Stage 2 data were not available.

The Key Stage 3 tests in science and mathematics were tiered. This meant that pupils entered for a low tier could not obtain the higher levels, however well they performed on the test. There was some overlap between the tiers and for each subject, a single scale was constructed. The details are reported below.

A multilevel analysis was carried out for each of the three curriculum subjects in turn. As previous research indicates that gender and social deprivation both influence pupil attainment, these were entered into the model as explanatory variables. The pupils’ eligibility for free school meals was used as an indicator of social deprivation. These data sets form the basis for the first set of analyses.

English: A complete set of data on all the explanatory variables was available for 3514 pupils. The English scores were roughly symmetrically distributed and were transformed to a standard normal distribution for all multilevel analyses. The analysis suggests that pupils make more progress if they are not in mixed ability schools, and there is some evidence that set schools strengthen the relationship between KS2 and KS3 scores. The estimates of the effects of the main effects of school type on attainment at Key Stage 3 were very similar for the partially set and the set schools (0.23 and 0.21), suggesting that a simpler contrast, of mixed ability versus the rest, would be more appropriate. The main effect was then 0.22 (0.10), which was statistically significant, and the interaction was 0.08 (0.06), which was not. The analysis provides some evidence that pupils make more progress in English if they are not in mixed ability schools, and some evidence that set schools strengthen the relationship between attainment at Key Stages 2 and 3.

Mathematics: The mathematics tests were set in four separate, overlapping tiers. The marks in each tier corresponded to a level, each test allowing pupils access to several levels. A single scale was constructed from the four separate scales, using the levels allocated to each pupil as the linking information. The distribution obtained was roughly symmetrical but not normal and was transformed to a standard normal distribution for the multilevel analyses.

A complete set of data containing all the explanatory variables was available for 3481 pupils. The analysis suggested that pupils made more progress if they were not in mixed ability schools, and there was some evidence that set schools strengthened the relationship between Key Stage 2 and Key Stage 3 scores. The estimates of the main effects of school type on attainment at Key Stage 3 were very similar for the partially set and the set schools (0.03 and 0.11), suggesting that a simpler contrast of mixed ability versus the rest, would be more appropriate. The main effect was then 0.25 (0.10), which was not statistically significant, and the interaction was 0.12 (0.03), which was significant. The analysis provides evidence that set schools strengthen the relation between attainment at Key Stages 2 and 3. High attaining pupils in mathematics at Key Stage 2 do better at Key Stage 3 if they are in a set school, whereas low attaining pupils do better if they are in a mixed ability school.

Science: There were two main tier tests, giving pupils the possibility of obtaining levels 3 to 6 or levels 5 to 7. To construct a score from these two tests, the levels allocated to each pupils were used as the linking information. An underlying standard normal distribution was assumed for science attainment. The empirical distribution function defined by the levels was then used to determine cut points on this underlying distribution. A transformed score was then calculated for each pupil by linearly interpolating between the cut points defined by the levels, using their observed score from the tier of assessment taken. The observed distribution, which was roughly symmetrical but not normal, was again transformed to a standard normal distribution for all multilevel analyses.

Complete data for this analysis, including eligibility for free school meals, gender and attendance was available for 3542 pupils in 38 schools. The multilevel analysis showed that the relation between the KS2 levels and KS3 scores varied from school to school, essentially in the range 0.50 to 1.00. After controlling for free school meals, attendance and gender, the estimates for school type and their interactions with KS2 levels were as follows. The main effects of school type were 0.11 (0.09) for the MASET schools and 0.05 (0.09) for the SET schools, while the interactions of school type with KS2 levels were 0.10 (0.07) and 0.10 (0.07) respectively. These interactions were not statistically significant, neither were the main effects. Using the simpler contrast, MA against the rest, the main effect was then 0.08 (0.08), the interaction was 0.10 (0.06). Neither of these was statistically significant. There is little evidence that school organisation has an effect on progress in science, although the estimates for the interactions are in the same direction as previously.

Consideration of these findings raised the question as to whether it might be more appropriate to analyse the data in relation to the extent of ability grouping in each subject, rather than use the categorisation of schools into three types. Because the setting arrangements were decided largely on a departmental basis, pupils might have experienced mixed ability grouping in English in a school classified as set or might have been set for mathematics since year 7 in a partially set school. In only two schools in the sample were pupils in mixed ability classes for mathematics in year 9. A closer examination of the information available on each of the schools’ grouping practices in years 7 to 9 was therefore carried out. Information was available as to whether pupils were in sets or mixed ability classes in each year for each school. More detailed information had been collected about the setting arrangements in place in year 9. This revealed a variety of setting practices, from rigorous setting across the year group to the formation of one top set and several parallel sets. A scale was constructed, with 0 if the teaching had been entirely in mixed ability classes, 1 for setting in year 7, 2 for setting in years 7 and 8, 3 for setting in years 7, 8 and 9. A point was added if the setting in year 9 was rigorous, adopted across the year group with no parallel sets. The data was reanalysed for each curriculum subject in turn, using the strength of setting variable in place of school type, but otherwise entering the same explanatory variables as in the first set of analyses.

The majority of schools adopted rigorous setting for mathematics. All the set schools and 11 of the MASET schools scored 3 or 4 on the setting scale. Only 8 of the 15 mixed ability schools scored 0 or 1 on the scale. The analysis supported the previous finding that the lower attaining pupils at the end of Key Stage 2 made more progress in schools with less setting in mathematics, whereas the higher attaining pupils made better progress in schools with more setting. The difference for pupils at level 1 at Key Stage 2 between the least and most set schools is 0.12 SD units. The difference for pupils at level 5 at key stage 2 is 0.36 SD units, in the other direction.

There was much less rigorous setting in English. All 15 of the mixed ability schools and 9 of the 15 MASET schools scored 0 or 1 on the 5-point scale. 14 of the 15 SET schools scored 3 or 4. The re-analysis showed no evidence of a relation between the scale and the Key Stage 3 scores. Overall the evidence suggests that progress in English is not related to setting. In science, 14 of the 15 school mixed ability schools scored 0 or 1. Ten of the 15 MASET schools scored 2, and 13 of the 15 SET schools scored 3 or 4. There was no evidence of any effect of the setting scale on progress in science.

This analysis provides further support for the effect of setting on progress in mathematics but not in English or science. The effect of setting in mathematics appears to act to the advantage of the pupils attaining higher levels at the end of primary school, whereas pupils whose attainment is low at the end of primary school make more progress in mixed ability classes.

The relationship between ability grouping and self-concept was explored for each curriculum subject in turn, with pupils’ level of attainment at Key Stage 2, gender and a measure of social deprivation (eligibility for free school meals) entered in the regression model. The analysis revealed a strong relation between gender and self-concept in Mathematics and science, with boys having higher self-concepts in these subjects than girls. In English, girls had higher self-concepts than boys, but the difference was not as great. Self-concept was also related to level of attainment on entry to secondary school. The relation was evident in all subjects, but was strongest in mathematics.

The effects of school type were explored when gender and attainment were included in a linear regression model. This analysis suggests that pupil self-concept is influenced by the type of school, with overall a more positive effect on the self-concept in the mixed ability schools. In the next step in the analysis, these data will be entered into a multi-level model.


This research supports much of the work carried out earlier this century on ability grouping, which generally showed little or no impact on overall pupil attainment. Our findings add to the previous body of research in showing that the impact of ability grouping differs in relation to the curriculum subjects, with progress in mathematics being more strongly affected than progress in English or science.

In this country, English is a subject that is generally thought to be suitable for mixed ability teaching. A survey of the teachers in the schools participating in this research revealed that Humanities were considered the most suitable for mixed ability teaching, with English next. Even in the partially set and set schools, pupils in years 7 to 9 were less likely to be organised in sets in English than in mathematics.

There may be differences in the curriculum and teaching methods that mediate or interact with the grouping arrangements to produce these differences. There is evidence that the effect of setting on pupils’ attainment is greatest when curriculum materials are differentiated (Askew & Wiliam, 1995; Kulik & Kulik 1982, 1987). Kulik & Kulik concluded from their meta-analyses that gifted pupils’ achievement improved significantly when they were provided with programmes specifically designed to meet their needs. When pupils are grouped by ability but cover the same curriculum, there is little effect on attainment. It is probably the case that setting in mathematics is accompanied by a greater differentiation of the curriculum than in English or science. The national examinations (GCSE) and the Key Stage tests in mathematics, taken by pupils at the end of year 9, have been tiered for many years. This means that the curriculum followed by pupils in the lower sets is designed to enable them to enter for the lower grades, whereas pupils in the top sets are prepared for the top grades.

As mentioned above, schools were less likely to set for English than for mathematics. The reason why English is considered suitable for mixed ability teaching is probably that work in English can be differentiated by outcome. This means that all pupils in a class can undertake work on the same text or write about the same subject, but it is expected that they will produce work of differing quality. Teachers also expect pupils to draw on their own personal experience in English, whereas in mathematics it is expected that pupils’ work will draw on their previous learning in the subject.

The findings presented here remain to be explored in greater depth, and there are some aspects that we still wish to check and confirm. If the remaining analyses confirm these findings, then there will be some important messages for schools and policy makers. In England there has been some pressure on schools to move towards greater use of setting, as a means of raising educational standards. Many primary schools have reconsidered their use of setting in recent years, and some secondary schools have decided to introduce setting following the introduction of the national curriculum and tiered assessments. This research suggests that setting in mathematics, accompanied by curriculum differentiation, may be a means of raising the attainment of the more able pupils. The effect is not great, however, and there are some costs in terms of the progress of pupils whose attainment is low at the end of primary school. The impact on pupils’ self-concept may be important in the longer term, influencing later attainment in the subject and decisions about choice of subjects after the age of 16. These factors must also be taken into account when formulating policy on ability grouping in schools.


Acland, H. (1973) Streaming in English primary schools British Journal of Educational Psychology, 43 (2), 151-159.

Barker Lunn, J. C. (1970) Streaming in the Primary School Slough: NFER.

Benn, C. & Chitty, C. (1996) Thirty Year On: Is Comprehensive Education Alice and Well or Struggling to Survive? London: David Fulton.

Benn, C. & Simon, B. (1970) Half Way There: Report on the British Comprehensive School Reform London: McGraw-Hill

Creemers, B. P. M. (1994) The Effective Classroom, London: Cassell.

Evans, P., Ireson, J., Redmond, P. & Wedell, K. (1988) Curriculum research for pupils with moderate learning difficulties. Final report to the Department for Education and Science.

Finley, M. K. (1984) Teachers and tracking in a comprehensive school Sociology of Education, 57, 233-243.

Fogelman, K. (1983) Ability grouping in the secondary school, In Fogelman, K. (Ed.) Growing Up in Great Britain, Papers from the National Child Development Study. Macmillan for NCB.

Gamoran, A. & Berends, M. (1987) The effects of stratification in secondary school: synthesis of survey and ethnographic research. Review of educational research, 57, 415-435.

Gregory, R. P. (1984) Streaming, setting and mixed ability grouping in primary and secondary school: some research findings. Review of educational research, 57, 415-435.

Hallam, S., Ireson, J., Mortimore, P., Hack, S. & Clarke, H. (1999) Teachers’ attitudes towards ability grouping in the secondary school. Paper presented at the 8th European conference for Research on Learning and Instruction, Goteborg, Sweden, August 1999.

Hallam, S. & Toutounji, I. (1996) What do we know about the grouping of pupils by ability? London: Institute of Education.

Ireson, J. & Hallam, S. (1999) Raising standards: is ability grouping the answer? Oxford review of education.

Kerkoff, A. (1986) Effects of ability grouping in British secondary schools. American sociological review, 51, 842-858.

Kulik, C-L. C. & Kulik, J. A. (1982) Effects of ability grouping on secondary school students: a meta-analysis of evaluation findings American educational research journal, 19, 415-428.

Kulik, J. A. & Kulik, C-L. C. (1992) Meta-analytic findings on grouping programs, Gifted child quarterly, 36(2), 73-77.

Marsh, H. W., Parker, J. & Barnes, J. (1985) Multi-dimensional adolescent self-concepts: their relationship to age, sex and academic measures. American educational research journal, 22, 422-444.

Newbold, D. (1977) Ability Grouping: The Banbury Enquiry. Slough: NFER Publishing Company.

Oakes, J. (1985) Keeping track: How schools structure inequality. New Haven: Yale University Press.

Oakes, J., Gamoran, A. & Page, R. (1991) Curriculum differentiation: opportunities, consequences and meanings. In P. Jackson, (Ed) Handbook of Research on Curriculum. New York: Macmillan.

OFSTED (1995) Annual report of Her Majesty's Chief Inspector of Schools, 1993/94: Part 1, Standards and Quality in Education, London: OFSTED

Reid, M. E., Clunies-Ross, L. R., Goacher, B. & Vile, D. (1982) Mixed Ability Teaching: Problems and Possibilities. Windsor: NFER-Nelson.

Reynolds, D., Sullivan, M. & Murgatroyd, S. (1987) The comprehensive experiment: A comparison of the selective and non-selective system of school organisation. London: Falmer Press.

Sammons, P., Hillman, J. & Mortimore, P. (1996) Key characteristics of effective schools: a review of school effectiveness research, London: Office for Standards in Education.

Slavin, R. E. (1987) Ability grouping and student achievement in elementary schools: a best evidence synthesis Review of educational research, 57(3), 293-336.

Slavin, R. E. (1990) Achievement effects of ability grouping in secondary schools: a best evidence synthesis Review of educational research, 60, 471-490.

Stradling, R., Saunders, L. & Weston, P. (1991) Differentiation in Action Slough: NFER.

This paper is based on research funded by the ESRC through the Ability Grouping in Schools: practices and consequences project.

This document was added to the Education-line database on 15 March 2000