Grammar schools and pupil achievement

The quality of debate about grammar schools is poor. Most commentators’ opinions on the matter seem to consist of arguing ‘I went to a grammar school and I did well out of it. Now I am forced to pay for my son/daughter to go through private education’.

There are two problems with this argument. First, it represents a sample size of one, from which we can infer little about the effects of selection in society as a whole. Two, journalists are unable to describe the effects of their having attended a grammar school relative to the counterfactual – in our case this would be attending a comprehensive in a non-selective system. It’s not surprising they are unable to do this – our school experience makes up such an important part of our self-image. However, it is necessary we make these comparisons to understand what difference a selective education system could make to our country.

Going to school in a selective versus non-selective system might have all sorts of effects on us throughout our life in terms of our self-image, motivation, aspirations, education path or wages. But arguably the most important thing we need to understand is whether selective schooling makes a difference to our achievement in exams at the end of compulsory schooling. This is because all other aspects are related to this: if we have low motivation as a result of failing an 11+ style exam, this should lower our achievement at 16; if we do well at 16 because we attend a grammar school, this should feed through into higher lifetime wages.

We can use the presence of 164 grammar schools in England today to tell us a little about the effect of selective schooling on GCSE performance. We have to be careful in this type of analysis because of the following methodological issues with comparing selective and comprehensive areas today. First, grammar schools are located in more affluent areas of the country, so pupils who live in these areas have family backgrounds that mean they would tend to make good progress at secondary school, quite independently of school attended. Second, one-in-five pupils who attend a grammar school today actually live in a different local authority to the grammar school. These porous local authority boundaries mean we don’t really have distinct selective schooling and non-selective schooling areas of England. Finally, the presence of state grammar schools alters who goes into a private school. Areas with grammars (like Kent) tend to have private schools for primary aged children and less academic private schools for those who failed the 11+. Areas with comprehensives tend to have academically selective ‘elite’ private schools. This is a problem because when pupils move into the private sector they don’t appear in administrative records of schools that researchers use to analyse the education system.

Despite these problems, we can say something about pupil performance in a selective schooling system. Pupils who get to attend a grammar school, rather than a comprehensive, will achieve somewhere around 4 better grades at GCSE. So, for example, they might get 6 A and 4 B grades, as opposed to 2 A and 8 B grades at a comprehensive. Pupils who end up at a secondary modern school, rather than a comprehensive, might achieve around 1 worse grade at GCSE. That is, perhaps, 4 B and 6 C grades, as opposed to 5 B and 5 C grades at a comprehensive. Since there are between two and four times more pupils at secondary moderns than at grammar schools, the aggregate effect is little different. My judgement is that these effects are smaller than most people would expect. The scale of improvement in exam scores that grammar school pupils achieve could probably be matched by attending a comprehensive and employing a private tutor for a few weeks before GCSEs to cram in four subjects.

We can say a little more about who benefits from grammar schools, though these statements are more contested in the literature. Most researchers agree that it is those who ‘just’ pass the 11+ who disproportionately benefit; those who are very able appear to do well in all education systems. Also, boys benefit more than girls, and this might be due to the greater propensity for boys to get ‘distracted’ in comprehensive schools. Alternatively, they may simply respond better to a more competitive academic environment.

So, we know that selective schooling produces a distributional effect whereby some are better off and others are (only slightly) worse off. The question is, is this a better system? There are two dimensions to this question. The first is to ask whether the system is more or less efficient at delivering the distribution of skills that we need in the workforce. The second is to ask whether the system is equitable.

Employers often complain about the low levels of numeracy and literacy skills in young employees, and point to the education system as being the cause of this. The question for us is, in what part of the workforce do we have a particular problem? If we think our economy is being held back by inferior skill levels achieved by the top 25% of the workforce (i.e. vacancies are left unfilled by skill shortages, or that the economy would shift towards higher skill industries in the future, if the supply of skills were forthcoming), then we should move to a selective system. However, when employers complain about skills in the existing workforce, it would seem that they are mainly referring to employees in the bottom 75% of the skill distribution. This would imply that a selective system could not improve the efficiency of the workforce, and might even cause it to deteriorate. This is not to say that we would not benefit from improvements in skills at the top of the distribution; rather problems further down the skill distribution appear to be likely to help labour productivity in the short term.

Selection by ability involves splitting a population of children into two groups based on certain criteria and there would seem to be four concerns about this from an equity perspective. The first issue is that it is entirely arbitrary where we draw the line between winners and losers and this violates the principle of horizontal equity – the equal treatment of equals. For example, the 25th and 26th percentile children by ability are, to all intents and purposes identical. If we allow only the top 25% into grammar schools, the system treats them very differently. This horizontal equity principle is particularly strong violated because the consequences of passing or failing are larger for this group of children than for any others, according to research.

The second equity argument contends that the test itself will rank pupils inconsistently and so is unfair. Firstly, suppose we asked a set of pupils to take five 11+ tests over the course of a week and select the top 25% of pupils on each of the five occasions. Some pupils will appear in the top 25% in every test, but others will ‘pass’ or ‘fail’ depending on the vagaries of the test. This variation in performance for the individual pupil is impossible to avoid, thus we cannot devise a consistent measure of who should, and should not, attend a grammar school. Another inconsistency results from the timing of the test. It was well-documented in the 1960s that if we selected at 11, we would chose a completely different set of pupils than if we selected at 14, 16 or indeed 7. The choice of age 11 is arbitrary and favours relatively early developers.

The 11+ plus test, as with all tests, assesses achievement and uses this as a (poor) proxy for ‘intelligence’. The third equity argument says that it is arbitrary how we assess achievement in a test, and that it matters very much how this is done. For example, the modern 11+ test places a greater weight on literacy skills than SATs taken in the same academic year that slightly favour numeracy and problem solving skills (English is only one-third of overall SAT assessment at age 11). The social class gap for literacy skills higher than for numeracy skills, so the 11+ test favours middle class children more than a SATs-style assessment would. Neither test is more ‘just’: it is not clear how we should make the decision as to how much relative weight to put on the different dimensions of ‘intelligence’. There is a related question as to what type of pupils we should make allowances for. Should new-comers to the country be allowed to sit the test in their mother tongue? Should dyslexic children get extra time or even lower pass thresholds, and what about children with other special educational needs? We do allow lower pass thresholds for August born children compared to October born children, but should we have lower pass thresholds for children who have missed out on months of primary education for other reasons such as illness? In short, achievement at 11 is only a good proxy for ‘intelligence’ where all children have had the same opportunity to learn up to this point. If we do not compensate for lost opportunities, it is no longer selection by intelligence.

The final equity argument (related to argument three) is the one that is most commonly made: that is, a test of achievement at 11 is necessarily socially selective. Obviously achievement at 11 is correlated with social background in any schooling system, but the argument made here is that an 11+ test favours a child from a higher income family over a child with the same achievement from a lower income family. There is clear evidence for this in the data: take two children who score exactly the same on a Key Stage 2 test and the one who is eligible for free school meals is much less likely to gain a place at a grammar school than the one who is not eligible for free school meals. Competency in the 11+ test can be purchased in one or more ways. A higher income family can simply send their child to a private primary school for coaching through the test – it appears that around 15% of children in state grammar schools had a private primary education. Alternatively, a high income family can purchase private coaching in the years leading up to the test. This is clearly very prevalent, though it is difficult to get statistics on it: the private tuition industry is a ‘grey’ part of the economy, with many tutors giving tuition only a few hours a week and not declaring this extra income. Even without this purchased help, a family from a higher social background is much more likely to be able to help their children through the test than a lower social background family.

So, overall, there are quite significant gains to attending a grammar school compared to a comprehensive school (though not as large as many commentators suggest). But there are potentially small losses from attending a secondary modern compared to a comprehensive school. It is difficult for us to assess whether this distributional shift in the skills base of the population would be better or worse matched to the skill needs of the economy, so the efficiency argument in favour of grammar schools is unproven. Against this there are quite serious equity arguments against selection by ability and it is simply impossible to overcome these regardless of the type of assessment test we set our children.

However, stating that academic selection is inequitable is not the same as claiming that comprehensive schooling is equitable. The quality of the peer group a child experiences in a comprehensive school mostly depends on the neighbourhood the child lives in. That said, comprehensive schools are vastly less socially stratified than selective schooling systems. So, if improving the fairness of the education system is our goal, we should look to reform priority to school by place of residence or religion in order to reduce social stratification in schools, and thus produce a more equitable school system.


Atkinson, A., Gregg, P. and McConnell, B. (2006) The result of 11+ Selection: An Investigation into Opportunities and Outcomes for Pupils in Selective LEAs, CMPO Working paper, 06/150.

Jesson, D. (2000) The Comparative Evaluation of GCSE Value-Added Performance by Type of School and LEA, Discussion Papers in Economics: University of York.

Levačić, R. and Marsh, A. J. (2007) Secondary modern schools: are their pupils disadvantaged? British Educational Research Journal, 33 (2), 155-178.

Schagen, I. and Schagen, S. (2003) Analysis of National Value-added Datasets to Assess the Impact of Selection on Pupil Performance, 5, 29 (4), 561-582.

One thought on “Grammar schools and pupil achievement

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s