You are here: Publications and Records > Commons Publications > Select Committees > Children, Schools and Families > Children, Schools and Families

UNCORRECTED TRANSCRIPT OF ORAL EVIDENCE To be published as HC 311-i

House of COMMONS

MINUTES OF EVIDENCE

TAKEN BEFORE THE

CHILDREN, SCHOOLS AND FAMILIES COMMITTEE

 

 

Diversity of School Provision

 

 

Wednesday 30 January 2008

DR STEVE GIBBONS, DR TOM BENTON and SIMON RUTT

Evidence heard in Public Questions 1 - 78

 

 

USE OF THE TRANSCRIPT

1.

This is an uncorrected transcript of evidence taken in public and reported to the House. The transcript has been placed on the internet on the authority of the Committee, and copies have been made available by the Vote Office for the use of Members and others.

 

2.

Any public use of, or reference to, the contents should make clear that neither witnesses nor Members have had the opportunity to correct the record. The transcript is not yet an approved formal record of these proceedings.

 

3.

Members who receive this for the purpose of correcting questions addressed by them to witnesses are asked to send corrections to the Committee Assistant.

 

4.

Prospective witnesses may receive this in preparation for any written or oral evidence they may in due course give to the Committee.

 


Oral Evidence

Taken before the Children, Schools and Families Committee

on Wednesday 30 January 2008

Members present:

Mr. Barry Sheerman (Chairman)

Annette Brooke

Mr. Douglas Carswell

Paul Holmes

Fiona Mactaggart

Mr. Andy Slaughter

Lynda Waltho

 

 

Examination of Witnesses

Witnesses: Dr. Steve Gibbons, Research Associate, Centre for the Economics of Education, London School of Economics, and Dr. Tom Benton, Senior Statistician, and Simon Rutt, Deputy Head of Statistics, National Foundation for Economic Research, gave evidence.

 

Q1 <Chairman:> I welcome Dr. Steve Gibbons, Dr. Tom Benton and Simon Rutt to our session this morning. Our topic, the diversity of school provision, is one that the Committee particularly wanted to look at, because it is time to assess the effectiveness of the Government's policy in this area. We do not have any hard and fast views, but we hope to add value by pursuing this inquiry, and it is down to us to find out all the facts before making any decisions.

I am not sure whether any of you would like to say a few words to start the conversation. I will turn to Dr. Gibbons first. Steve, I will refer to you by your first name, if I may-this is not a very formal session. Tell us, in a nutshell, where you are in this work.

<Dr. Gibbons:> Over the past few years, I have been doing a range of research using the national pupil database looking at questions regarding the segregation of pupils into different schools, where people are sorted into different schools according to their achievements, school competition, choice and performance-a range of questions related to the issues that you are concerned with here. I have a range of different areas of study. I suppose the kind of thing I am being asked to speak on is the achievement of kids when they enter secondary school. The story is that you get a wide diversity in terms of the average achievement of children who have entered secondary schools.

<Chairman:> Order. We seem to be having some trouble with the sound this morning, and the acoustics in this room are bad. I ask everyone to speak up.

<Dr. Gibbons:> We looked at the ages and achievements of children from secondary schools, and compared those schools in terms of the average achievement of the children in them. We found that you get a wide spread in terms of the achievement of children when they go to secondary school. If you imagine a range of pupils, from the lowest achievers to the highest achievers, the range of spread you get across community schools is around a third of that distribution. In a range of achievers across all types of schools-including voluntary aided schools and grammar schools-it spans about 60% to two thirds of that distribution.

That appears quite wide, but the number of schools at the top and bottom are quite small, and the share of variability and achievement that is due to differences across schools is actually quite small. Around 90% of the variability in achievement across pupils is within schools, not between schools. If you are interested in differences in achievement between pupils, the place to look is within schools, because there are big differences between pupils in the same school that swamp the differences between the average pupils in different schools.

Q2 <Chairman:> What would you say to policy makers who say, "Look, why do we have two schools with the same social composition, where one is achieving all its targets-five GCSEs at A to C, or whatever they might be-and another, with a very similar social distribution, is not getting anywhere near that?" What would you say to someone who says that the whole job is to get the not-so-good school up there with the good school, and that it should be possible, because they have the same sort of intake?

<Dr. Gibbons:> Generally speaking, achievement is very closely related to the composition of the school, demographically and in terms of prior achievement.

Q3 <Chairman:> Fool's gold, is it? Basically, if you know the social composition going into the school, broadly you know what the result will be.

<Dr. Gibbons:> It is a good guide. Clearly, there are schools that do well with a given intake and schools that do badly with a given intake, but in general intake will dominate performance in the end. I have not looked in great detail at what drives the effectiveness of schools. This is the holy grail, to try to define what makes certain schools work better than others. We are not really in a position to answer that.

The message from a lot of educational research is that it is very hard to find the facts that make the difference to the achievement that you were talking about-the value added, if you like. The factors people are looking at now are the leadership skills of the head teacher and certain qualities of the teachers within the schools. It is hard to pin down. It is not to do with resources, and it is not do with certain policy issues; it is to do with unobservable factors that we cannot isolate, given the data we have at the moment. That is my assessment.

Q4 <Chairman:> We will drill down on that. That is most interesting.

Turning to Dr. Benton. What research are you doing that is relevant to this Committee and how does it square with Steve's? I have a vested interest in that I am a governor of the London School of Economics. We seem to have a lot of LSE academics at the moment, but these witnesses were all chosen without consultation.

<Dr. Benton:> I have been doing a lot of work looking at the relationship between school type-faith schools, selective schools and specialist schools-and the outcomes for pupils in terms of achievement, performance and, to a certain extent, their attitudes. I have been investigating whether those things are related to school type. That was in the papers that were sent ahead.

I agree with all that Steve said, and we have had a lot of the same results in our work. We find that there is a lot more variation between schools in terms of the way pupils achieve and the extent to which they do better or worse than you would expect. That happens a lot more than schools as a whole doing much better or worse than expected. This is very much in line with the work that Steve has been doing.

A lot of our work has been focused on the different ways of assessing how good schools are by looking at their achievement data, and how we can take account of differences in their intakes. We find that simply looking at a school's raw results, such as the five A to C grade percentage in simple league table form, can be very misleading as to how well a school is doing. Some 90% of the differences between schools in terms of raw results can be attributed to intake-the nature of the students turning up. It is very important whenever we talk about school type or assessing the quality of the school in terms of achievement that we take into account the types of pupil within the school to start with and the differences in intakes.

On school type, generally speaking it is very hard to find any large effects, and we did not find major differences between one type of school and another. The only exception to that rule is a small number of selective grammar schools-selective state schools within the system-where pupils who are just clever enough to pass the entrance exam to get into a grammar school appear to do a lot better at key stage 3 than pupils who just failed the exam but who are very similar. When we look at the difference in their achievement over time, it can be very different. That is the only thing that has come out as a major effect. Other than that, school type has been found to have only a very small effect on achievement.

Q5 <Fiona Mactaggart:> Is that the work of Chardon and Chardon? What I am not clear about is whether the comparison in that work was between grammar schools and genuine comprehensive schools or between grammar schools and secondary modern schools, because such schools often describe themselves as comprehensive.

<Dr. Benton:> The work was actually done both ways. You can compare either secondary modern schools and comprehensive schools in the same area or two schools in different areas that do or do not have a selective system. Both times you will find that gap.

Q6 <Paul Holmes:> When I was teacher training, the received wisdom was that quite a large chunk of the pupils do not do well in grammar schools because, although they are in the top 20 to 30% of the ability range, they are regarded as being at the bottom of the tree rather than at the top of it.

<Dr. Benton:> Yes, that is correct. It seems that the lowest achievers who get into grammar schools get the biggest effect and overachieve compared with what they might have done elsewhere. There are alternative possible explanations for that. It could be that the grammar schools' selective tests are better than the national curriculum tests at picking out the cleverest pupils and the not-so-clever pupils. In other words, their method of selection could be more effective than the Key Stage 2 results, in which case comparing people who are similar in terms of Key Stage 2 intake would not adequately take account of differences in how clever they are. That is a possible explanation.

The other possible explanation is that, when it comes to Key Stage 3, grammar school pupils are far more likely to be entered for higher-tier examinations. That is another possible explanation in that area.

Q7 <Annette Brooke:> What about parental impact?

<Dr. Benton:> Parental impact is certainly very important. It is difficult to measure parental impact, but we try to take into account factors such as socio-economic status. Parental impact may affect prior intake as well, so by adjusting to those factors you would hope that you can take into account differences in parental support. You are certainly right that there are more factors.

Q8 <Chairman:> Simon, how does your research differ from Dr. Benton's?

<Simon Rutt:> My research looks at the step before: what happens in schools once the kids get there, and what schools do with kids. I am particularly interested in looking at which children go to which schools. I know that there are a lot of issues surrounding the fact that children do not get into their local schools, that they have to travel so far and whether covert or overt selection is taking place. There are lots of issues and queries from parents about what is happening.

By using the national pupil database, where fortunately we now have national coverage of pupils, we were able to identify schools in a community and the pupils who live around that community. We looked at whether children go to the local school or to other schools. The most important part of that was identifying the community, which was very difficult. To do that research perfectly, we would need to know each school's catchment area, of which you can get maps from the local authority-the area covers certain roads and goes out to this river and that road. We did not have that information; we only had postcode information. So for primary schools, we had the postcode of where the school sat, and we took the first four digits-for example, SW6 9-as the description of the community around the school. That covered about 2,600 households, which was described as the school's community.

We were able to look at the pupils who lived in that community and see where they went to school-how many pupils went to that local school and how many went to other schools. We then looked at the background characteristics of those pupils and aggregated that up to the level of the local authority and, in particular, the school type to see whether there are any differences for the national average of the make-up of those characteristics. What we found, which held for primary and secondary, is that the main difference is between community schools and voluntary aided schools, where there is a distinct difference in admission policies and what they can do.

We found that pupils at voluntary aided schools tended on average to come from more areas, so there was a wider dispersion of their pupils. The voluntary aided schools took less of the intake from their community. If 80% of pupils lived in the local area, community schools took a higher proportion of those pupils. You would expect them to take similar proportions, but actually voluntary aided schools tended to take less from the local area.

One factor that has appeared in some of the papers on the subject is that voluntary aided schools tend to be religious schools. Clearly, if you are a Roman Catholic school, you have got to take a lot of Roman Catholic pupils, who may not be centred around the school. When the school was first set up, there may have been a large Roman Catholic or religious congregation around the school, but with mobility, especially in urban areas-particularly London-the population has moved out further, so the schools have had to go out further to get their Roman Catholic pupils.

Q9 <Chairman:> You would get transport costs, in an advantageous way, if it were the only Catholic faith school. You can travel further.

<Simon Rutt:> Apparently so. Yes, you can travel further.

One of the interesting things that we wanted to look at was what proportion of pupils in the community were on free school meals, which is the main socio-economic indicator used on the national computer database. We expected the schools to take a similar proportion, but we found that voluntary aided schools were taking a lower proportion than one would have expected. At the individual school level, you can find schools that will take far more and schools that take less, but looking at the national averages, the distinct characteristic of voluntary aided schools is taking less than one would have expected.

One of the reasons is that the schools have to go further, they have to look outside-there may not be that many Roman Catholic people resident in the communities, so the schools have to go further out. With the travel costs covered, you would expect that the schools would still be taking free school meal pupils. We then looked at the areas where those pupils come from-they do not come from the community of the school, so where do they come from? We looked at what proportion of free school meal pupils lived in those communities and whether the schools took the same proportion or a higher proportion. Again, we found that voluntary aided schools took fewer free school meal pupils from those communities as well. Overall, community schools were taking slightly more free school meal pupils than one would have expected, and voluntary aided schools were taking slightly less. I have not looked at whether the prevalence of free school meals within the Roman Catholic and C of E faiths is less-it may well be that the proportion of those on free school meals in faith groups is lower. I do not know, but one would think that that is not the case, so you would expect a similar distribution of free school meals.

We also looked at special educational needs in the same way, at ethnic minorities and, for secondary schools, at key stage 2 attainment and the proportion of pupils who had achieved level 4 and above within the community. Looking at community and voluntary aided schools, we generally found that voluntary aided schools took lower proportions of pupils on free school meals and lower proportions of pupils with special educational needs.

Ethnicity was very similar between types of school. What was interesting was that it seemed that ethnic minority pupils travelled further to get to their school of choice or to the school they ended up in, rather than actually going to their local school. In secondary schools, voluntary aided schools tended to have a higher proportion of pupils reaching level 4. That seems to show nationally-I am looking at the average statistics-that there is a difference between the schools and their intakes, which feeds automatically into what they do and the characteristics of the ultimate impact on final attainment in those schools. We tend to find voluntary aided schools in particular categories where we find less SEN, and these things have a fairly major relationship with final attainment.

Academies were introduced particularly to address high-deprivation areas. From the number we had on the national computer database at the time, we found that the academies were in areas of high deprivation and that they took a higher proportion of pupils on free school meals from the communities that they served. They were set up in areas of deprivation, which I believe is their purpose, and they take a higher proportion of pupils on free school meals.

The main differences were between community schools and voluntary aided schools-private sector schools obtained similar results-but you do observe a lot of differences when you look at very urban areas, where there was a lot of mobility between sectors, and rural areas, where there was less mobility and pupils went to their local school.

<Chairman:> That has warmed us up. Thank you for those introductory remarks.

Q10 <Fiona Mactaggart (Slough) (Lab):> Tom, you said that 90% of the difference in outcomes for children is connected to the intake of pupils. I recognise the difference between comparisons within schools and comparisons between schools, but we are interested in the differences between schools here, even though we recognise that your research shows that there might be greater variation within a school. Are you saying that the most significant predictor of the outcomes in a school-let us leave aside the 11-plus at the moment, because it is distorting-is the intake of pupils?

<Dr. Benton:> Yes, absolutely, in terms of not only prior attainment, but free school meals and special educational needs. Taken as a whole, those are a very good predictor.

Q11 <Fiona Mactaggart:> Simon, your work says that those schools that have their own admissions authorities cherry-pick their pupils.

<Simon Rutt:> There appears to be a difference in the characteristics of those schools. I cannot not say whether they cherry-pick, because I would need to know who applies to go to the school and who gets in.

Q12 <Fiona Mactaggart:> Why do you not have that information?

<Simon Rutt:> As far as I have heard, information on who applies and who gets into every school is not available. The local authorities hold certain amounts of such information-I believe that London had a consortium to combine admissions policies-but to say whether you are statistically less likely to get into certain types of schools if you have free schools meals or special educational needs, or if you are a certain type of pupil, we need to know who applies and who gets in, but that information is not currently available. It is not a data set that I know is available.

Q13 <Fiona Mactaggart:> Have you looked for it?

<Simon Rutt:> Loosely, yes. I have not dug too deeply, but it is not something that I am aware of as available nationally.

Q14 <Fiona Mactaggart:> Should it be?

<Simon Rutt:> Yes.

Q15 <Fiona Mactaggart:> Is any of the difference between schools accounted for by the level of spending on the pupils within them? None of you seems to suggest that that is particularly significant.

<Dr. Gibbons:> I cannot answer in terms of spending on individual pupils within schools, but in terms of the average expenditure of different schools the evidence that we have is that it does not make a huge difference, given the levels of expenditure at the moment. The problem is, of course, that the expenditure is somewhat targeted towards disadvantaged schools, so it is hard to tease out causal linkages between expenditure and pupil achievement. There is some work on specific programmes, such as the Excellence in Cities programme, which suggests that there are some positive benefits, but in general if you look at the basic statistical analyses that are available on expenditure and outcomes, you find nothing. That is a fair assessment of not only the literature from this country, but the international literature.

Q16 <Fiona Mactaggart:> One of the things that we have been looking at is collaboration between schools. It seems to me, looking at the research, that collaboration between schools happens between secondary and primary schools but not particularly between secondary schools. Have you done any work on collaboration between schools?

<Dr. Gibbons:> No.

Q17 <Fiona Mactaggart:> Are you saying that it would be helpful, and that you would be able to tell us much more about school-level effects, if there were a data set that showed who had applied to schools and who had got in? Would that be complicated to produce?

<Simon Rutt:> It would be extraordinarily difficult to collect that data at a national level. At a local level, I know some local authorities have that information, but to have a national database, it would be extraordinarily difficult to collect. I am not saying it would be impossible, and it would be extremely powerful and very useful to dig into data on admissions, which pupils go to which schools, how pupils get in, whether schools are taking in balanced admissions and whether schools are taking pupils from particular areas.

That would be a very strong database to use, but lots of issues might not come out of it. It would have to be combined with a lot of qualitative research to look at the process of applying for a school place and what goes down as first choice, second choice and third choice. Second or third choices, or up to six choices, go out of the window for a lot of parents, because you have to put down your first choice as school X-if you do not do so, you will not get in it, which happens.

A database that allowed you to look at the choices parents make about which school their children go to and information on which pupils actually end up in a particular school and which of their choices it was would be very powerful. You could look at the flow of pupils around local authorities to see who goes where and be able to say once and for all whether schools are overtly or covertly selecting and to fix their intake. Such a database would be very powerful. The information would be difficult to collect nationally, but as a statistician, I would revel in the opportunity to analyse it.

<Fiona Mactaggart:> Under the new schools admissions code, it is actually impermissible for a school to give advantage to a child who puts it first. If that continues to happen, it will be a breach of the new code. That would at least clean your database, were we able to ensure that you get it.

Q18 <Chairman:> In its previous incarnation, this Committee presented a report that many of us believe changed the role of the Schools Commissioner. Will it be possible for the Schools Commissioner to conduct an evaluation of the social composition of schools every two years? That is one of the roles of the Schools Commissioner. Part of the job is regularly to evaluate the balance of the social intake of schools. Is that possible?

<Dr. Gibbons:> I presume that it will involve the kind of data that we have been using to answer these questions-you just look at the national pupil database, I guess. You look at the characteristics that are in there and how they are distributed across schools. That is what people will be looking for, I think.

Q19 <Chairman:> I was getting a rather negative picture of the possibility from Simon.

<Simon Rutt:> It is possible. From the national pupil database, we have the ability to look at who arrives at those schools and who is in them, but we do not have the information on who applied. From my research-I am just looking at the numbers-if a large proportion of free-school-meals pupils are at a school, I cannot say whether there has been any selection of those pupils, because I do not know who applied. If there are 200 free school-meal pupils in a school, it might be that only 200 of them applied, in which case the school would have been very fair in taking all the pupils who applied to it. Similarly, a school might take all the SEN pupils who apply to it.

I would like to have a data set of who applied to the school where you can look at how many pupils applied to the school who were on free school meals. If the expectation or the assumption is that they should be taking similar proportions in the community or nationally, then the question is why are they not doing so. Another thing that is not on the national pupil database at the moment, or was not when we carried out our research, is the religious affiliation of a pupil, which would be good additional information.

<Chairman:> We will be drilling down on that; it is fascinating.

Q20 <Paul Holmes:> Politicians in search of the holy grail have said that the answer to problems with pupil attainment and school improvement is diversity through the provision of CTCs, faith schools, trust schools or whatever they are called. However, all the evidence that you have given seems to indicate that that does not matter and that it is the intake of pupils that makes the difference. Is that a fair summary of what you said?

<Dr. Benton:> Sure, it is a fair summary of all the research that we have done. Furthermore, if you look at the outcomes for schools in terms of different subjects-for example, English results or maths results-you can ask whether the schools that are overachieving in maths are the same ones that are overachieving in English. When you do that, you find out that there are very different schools. There is a relationship between the two-there is a correlation between overachieving in one and overachieving in the other-but if you look at different subjects, you get different results, which indicates that whole school changes may not be the most important thing in driving results. It may be that subjects work more individually than that. You have to think, how do we improve English, how do we improve maths? A lot of it could be achieved at the subject level rather than the whole-school-approach level. Certainly in terms of school type, that is a major driver.

Q21 <Paul Holmes:> Okay. You have said that where you have diverse schools, especially if they are in control of their admissions, they start to select by academic selection, social selection and so forth. Is there any evidence that diversity and selection of various kinds have an adverse effect on other schools in the area?

<Simon Rutt:> On the admissions side, we looked at communities that have more than one school. So if a voluntary aided school is taking an unfair proportion of pupils on free school meals, we found in a number of areas the knock-on effect appeared to be that the community school in the same area had a higher proportion of free school meals and SEN pupils than the selective schools. It appears that if one school takes fewer pupils on free school meals and SEN than you would expect, the other schools in the area take more. That, in turn, has a knock-on effect on attainment.

<Dr. Benton:> The programme for international student assessment in OECD countries has consistently said that the two best performing countries in the world are South Korea and Finland. The one thing that they have in common is that they have local schools, and not much else besides. Is the lesson that the comprehensive system of local community schools is better than diversity? When it comes to the PISA countries comparing countries, there is a vast number of differences in the education systems in different countries. Immediately saying "These countries are the best, and it must be because of the comprehensive system they have both got," is probably too much of a leap to be certain about.

Q22 <Paul Holmes:> If you look through the PISA studies, you generally find that the countries that have selection, such as Germany, England and the USA, do very well with academic pupils but have a huge tail of underachievement compared with the countries that have more non-selective systems. It is not just about the top two countries.

<Dr. Benton:> Sure, I understand what you are saying. However, you have not got an enormous number of countries in those studies, so statistically it is difficult to see how you can draw robust conclusions. Although it is interesting to speculate along those lines, you could not see that in any way as being a proof that a comprehensive system is the better one.

Can I return to your previous point about the negative influence of selection?

<Paul Holmes:> Indeed.

<Dr. Benton:> There has been some further research on the positive effect that selective schools seem to have on pupils who get in them. When we compared them with secondary moderns, we found that there seemed to be a converse negative effect of a much smaller size that affects a greater number of pupils. If you consider a local authority as a whole and consider the relationship between the percentage of pupils who are selected and overall achievement within the local authority, I think that the effect would more or less balance out. It appears that the positive effect on those pupils who get into selective schools is perhaps balanced out by the effect on surrounding schools.

Q23 <Paul Holmes:> You mention inquiries into the evidence on academies in the written evidence. Part of the problem with looking at academies is that they have not been running that long, so it is hard to tell the long-term impact on intake and certain areas. However, there is evidence that in those 24 academies the admission of pupils from deprived backgrounds fell from 42% in 2002 to 36% in 2006. Is that just a readjustment, because the academies were replacing failing sink schools that had too high a proportion of pupils from such backgrounds, or is it that the academies have started to become socially selective? Is it too early to say?

<Simon Rutt:> I suggest that it is too early to say. From those statistics, it is difficult to say whether they are balancing themselves out to take account of pupils applying to the school or whether they are now being selective. By having that information we would be able to determine whether they are starting to operate selection policies on pupils getting into schools.

Q24 <Paul Holmes:> Academies are relatively new, so we will have to see how the situation pans out, but CTCs have been around for a lot longer. Are there any studies of the CTCs, some of which have been in existence for 15 years or more, examining those effects?

<Dr. Gibbons:> I have looked at the intake of CTCs compared with other schools. Between 1996 and 2002, CTCs had a much more compressed intake in terms of the distribution and level of achievement of the kids coming in. They were selective and they had higher achieving pupils as well. The CTCs were in our estimation de facto selective, but the mechanisms through which that is working are not completely clear. They claim to have a comprehensive intake but we found evidence that they do not. They are extremely selective, and, although that is not to the same extent as a grammar school, it is still significant.

Q25 <Paul Holmes:> So, there is clear, uncontroversial evidence that CTCs have become selective in some ways?

<Dr. Gibbons:> Yes.

Q26 <Chairman:> The original framework that the CTCs were given included the ability to band. You three have all told the Committee that in order to give a school a fair chance of achievement, you need a balance of abilities that reflects the community rather than distorts the community. Is that the case?

<Dr. Gibbons:> The process by which CTCs admit pupils is admittedly mysterious to me. I am not quite sure. When I trawled through the admissions policy of the CTCs, it was said that they were trying to pick a balanced intake from the London community, yet they were allowed to slack on aptitude and specific skills. How those two matters square, I do not know. In the end, the policy winds up being slightly selective.

Q27 <Chairman:> Earlier, you told Paul that, to obtain achievement, a balanced intake is needed.

<Dr. Gibbons:> I did not mean to say that.

Q28 <Chairman:> I thought that you said 90% of the results from a school depend on its intake. You said that if we represent our community and get a fair balance of the community, we can do wonderful things to raise levels of achievement. However, what about a preponderance of children who are on free school meals, have SEN or are looked-after children? We visited schools with 100% free school meals, let alone 65% SEN. They find it difficult to raise levels. Is that the truth or is it not?

<Dr. Benton:> That is not quite what we are saying. We are looking at individual pupils, so we can see their characteristics and know what we expect them to achieve. If we look at them at an individual level, that is where 90% of the difference is. The make-up of the school is not so important, but each individual's characteristics affect each individual's chances of achieving later on. That is where most of the differences between schools lie. Do you see what I am saying? From a pupil's point of view, the people around the pupil are not as important as the pupil's characteristics.

There is some evidence of schools with a higher average intake doing better than other schools. There is an effect of having high ability kids around other pupils in terms of the other pupils' achievements, but that is smaller than the 90% figure, which is based on an individual's characteristics affecting an individual's chances.

Q29 <Paul Holmes:> I have talked about politicians looking for the holy grail, as in diversity. Another holy grail that is often trotted out is super-heads who, through their dynamism and personality, can transform a school regardless of its intake. Is there qualitative, statistical evidence to back up that statement or contradict it?

<Dr. Benton:> We did a survey of all schools in London. We asked teachers how good they considered leadership in the school. We could then relate that to the attitudes of pupils in terms of whether they liked the school and were committed. We found a significant relationship between the two things. Although what I said earlier about schools achieving differently in different subjects might be expected to have some effect on the quality of the school overall, it cannot be the holy grail. It cannot be the only thing that drives performance forward. Schools do differently in different subjects, so something must be going on within individual subjects. There is certainly evidence that the quality of leadership has a significant impact.

<Simon Rutt:> A lot of the statistical analysis that we carry out on the national pupil database allows us to explain a certain amount of the variation between schools and within schools and what is actually happening by using pupil effects and school characteristics. One thing that individual research has attempted to get at, but what we rarely have at national level, is parental involvement in education. Being able to put that into some of the models that we undertake would be powerful and would allow us to look with more variation at pupils to find our their parental involvement and home environment. At the moment, we have very few socio-economic indicators, but free school meals is not an indicator of how committed a parent is to their child's education. That would be another powerful piece of information to put in, because it would help to explain variations between pupils in schools with regard to parental factors and things outside a school's control.

Q30 <Chairman:> Surely, there must be a body of research that has looked at parental influence.

<Simon Rutt:> I am sure that research projects have looked at it, although I do not know of them as such, but it would be a very powerful piece of information to have.

Q31 <Chairman:> Is that something that you would like to have in order to further your research?

<Simon Rutt:> Absolutely.

Q32 <Lynda Waltho:> I want to drill down to what you think is missing with regard to statistics and information. How useful is the pupil level annual school census in the work that you are doing, what gaps are there and what else do you feel that it would be helpful to have information on?

<Simon Rutt:> That is an extremely useful dataset. I think that it has improved the analysis of educational research, and having that pupil information at a national level has allowed much more robust and sophisticated analysis. With regard to the information in it, such as information on behaviour, attendance and exclusions, that will increase the information and power of the dataset when it becomes fully incorporated into the national pupil database. Attendance has just started to be gathered at pupil level.

I worked on the Excellence in Cities evaluation, for which we collected information on pupil attitudes and attendance, and I think that there are lots of things in the research that could be used. On a number of occasions, it became evident that schools were actually having an impact on some of the other measures, despite not impacting on attainment straight away. It takes a little bit of time for a change in culture and ethos within a school to have an impact on attainment, but it might have a more immediate impact on behaviour, attendance and attitude to school. Change the attitude to school first so that children want to come to school and learn and turn up enthused by education, and then the attainment will change.

In the Excellence in Cities evaluation, we tended to find that some of those things were having an effect first and that some of the behaviour was changing, which would eventually, hopefully, lead to changes in attainment. Therefore, when the attainment things come through, along with fixed-term exclusions, which have been difficult to get on there, that will also make the information powerful, along with looking at behaviour in the school.

<Dr. Gibbons:> I support those requests entirely and think that the information on behaviour and attendance is important. I will go back to what Simon said earlier about information on admissions and on which schools pupils put down as their second and third choices. That information would be really valuable for understanding what really drives the selection processes and makes different kids go to different schools. As it stands, we only know which school a child ends up at, but not which school he or she would have preferred to go to. We cannot really work out whether the selection takes place on the parents' side or the school's side, so those two things together are the most important things that I would like to see.

<Dr. Benton:> I agree with everything that has been said. One thing that is also to be said is that a lot of data have already been collected within schools-we have talked a little about the attitudes data that we already have-for one purpose or one evaluation. That information could be used in secondary analysis of the data that already exist, and it could be reanalysed for a new purpose, such as looking at school type, selective schools and so on. There is a lot of potential, therefore, for further analysis of the data that already exist and for looking at some of the questions that we are considering.

<Simon Rutt:> One addition would be English as an additional language. We used to collect information on fluency in English at various stages of fluency up to being bilingual. A lot of very powerful analysis was done because bilingual pupils tended to achieve higher than native English speakers, and those who are new to the country and have low levels of English tend to struggle with the curriculum and underperform. The national pupil database, at the moment, only collects information on whether those pupils have English as an additional language. In running analysis, an effect tends to come out that you know is not the same for all pupils with English as an additional language. So, if we could get fluency levels back on to the national computer database, it would be a powerful resource, particularly for urban areas where there are many refugees and asylum seekers.

Q33 <Lynda Waltho:> That was what I was going to ask about. Language is quite a big issue. I am the daughter of a school secretary and I can just imagine what my mother might think now, after listening to all the extra things that are going to be required. I know that school secretaries take on much of the burden, so, sorry mum.

That is great, thank you very much.

<Dr. Gibbons:> Another study, the "Longitudinal Study of Young People in England" is not about the population. It is not the pupil level annual school census, or PLASC, but it is very useful. I would put in a word asking for that to be continued and extended because it contains a lot of the more detailed parental background information. It has detail on parental involvement and attitudes to school. At the moment it only follows one cohort year by year. I do not know what the plans are with regard to extending it, but it would be useful to see it followed up for different cohorts, and perhaps also to see its scope extended, as it is a valuable data source for answering these questions.

Q34 <Mr. Slaughter:> The Government's contention is that academies are either replacing schools or being placed in low-achieving, often socially deprived areas, with the idea of making a significant change in the format. They also contend that, at least at national level, the percentage of children having free school meals in faith schools is not much different from in community schools; it is only slightly higher. I know that because I heard Lord Adonis say it on the "Today" programme this morning.

Therefore, I had a quick look at one of my local education authorities to see if that fact was borne out. Looking at your CV, Mr. Rutt, it is an authority with which you will be familiar. In brief, the percentage of free school meals in four community schools was 56, 50, 42, and 41; it was 21 in one C of E academy; and it was 6, 6, and 2 at three voluntary aided schools. That is not, I would submit, a minor difference. It is an extraordinary difference. It does not necessarily equate to a system of comprehensive education, as I would understand it.

My question to you all is: how do you get to such an extreme system of stratification? That may be more extreme than other LEAs; I do not know. If it is in any way representative, it is clearly more extreme in relation to faith schools, which are an established part of the school family, rather than academies, although, significantly, academies seem to be in there as well. First, is it a problem? Is it something that we should not have ended up with? If it is, is it the local education authorities, politicians, parents, or the schools themselves that lead to that degree of difference?

<Simon Rutt:> Given that I know the local authority to which you are referring, I think that the difference between the schools within that area has come about because of the parental ethos and the culture in some of the schools. Many people have not applied to go to those schools because of pre-conceived perceptions of what the school is about, what it is like, and how it will be for their children. I know that applications to some of those schools are extremely high and that they do take a balanced intake of those who apply; they split the performance of those pupils into groups and their lowest performing groups perform higher than the highest performing groups in other schools. Is it the school's fault that such pupils apply, or is it the local authority's fault for not ensuring that a broader range of people apply? Hopefully, with the changes in admission policies, a broader range of people will apply for those schools. That situation has evolved over the years to become how it is. Parental perceptions and the cultural ethos have allowed that to develop.

<Chairman:> Dr. Benton is looking unhappy.

<Dr. Benton:> No, I am not.

Q35 <Mr. Slaughter:> I find the last point difficult to accept. Are you saying that there is self-selection in terms of parental applications?

<Simon Rutt:> I believe that there is a degree self-selection for the schools that we are talking about.

Q36 <Mr. Slaughter:> Another aspect is that even though voluntary aided schools make up half of the schools in the LEA area, only 5% of children from the LEA area go to those schools. Clearly, their catchment area must be wider because a much higher percentage of pupils are going to community schools. Is that a general feature of academies or faith schools?

<Simon Rutt:> Looking at voluntary aided schools nationally and at a local level, they tend to have a much wider dispersal of pupils. On average, they will come from more communities than those in community schools. In London, they come from even wider areas. There is a difference between inner London and outer London and other urban areas. In inner London, pupils will come from many more communities than that which the school is in. That area tends to be wider for voluntary aided schools than for community schools.

Q37 <Mr. Slaughter:> What about academies? Academies can be selective for 10% of their intake. That may or may not be significant. Does the ethos of an academy, by having an element of selection, a relationship with a sponsor or other factors, have the same effect of discouraging applications from a wider cross-section of parents?

<Simon Rutt:> I was reminding myself about academies. Given that there were only a few academies in the database that I was looking at, pupils came from a similar sort of proportion of areas as in community schools. Many pupils are from other communities, but not as many as in voluntary aided schools. The intake was from smaller areas around academies; a little bigger than community schools, but not as big as voluntary aided schools.

Q38 <Mr. Slaughter:> That does not answer my question. Your answers slightly surprise me. I do not know what the answer to my question is, but I would be surprised if you were correct. What you effectively seem to be saying is that we have a comprehensive system of education and parents choose to turn that into a selective system for their own reasons. If that is the case, I am asking whether it is likely to apply to academies as well as to voluntary aided schools.

<Simon Rutt:> I have not done any research to identify that, but I would think that as schools get better, they will have more applications from parents wishing to send their children to them, so the level of applications will be higher. Does that lead to schools selecting pupils from that application list? I have no evidence of that.

<Dr. Gibbons:> I do not have any evidence that would tell you anything specifically about academies, but it is self-evident from the relationship between community schools and house prices that an element of self-selection goes on. Parents create a selective system out of the comprehensive system by moving nearer to schools that are seen as good. That drives up house prices and keeps out lower-income families. There is a lot of evidence that school policy has a causal relationship with house prices. That in itself is evidence that this selection process is going on. Indeed, the distribution of achievement on intake into community schools, where there is no element of selection, is evidence that something like that is happening. Part of that is driven by the geographical location of schools and the kind of communities in which they are located, but there is a bit on top of that which will be generated by people selecting themselves into schools according to the kind of kids that are in there.

Q39 <Mr. Slaughter:> Two points come out of that. I think we all know that what you have said is a truism. Is it more true in densely populated areas such as in London where there is a smaller geographical catchment area, and there is not a local comprehensive serving a smaller community? More significantly, should not the types of schools we are talking about be less prone to that? In other words, if voluntary aided schools are selecting on the basis of religion and taking from a wider catchment area, should not they be less prone to the house price lottery? If academies are being targeted on deprived communities, should not they be less prone to social selection in that way? Neither seems to be the case.

<Dr. Gibbons:> Thinking about the voluntary aided sector, and the Church schools in particular, you are right. You would expect the impact on house prices to be less for those. The simple reason why they pick from larger areas is because distance is not usually a criterion that is used when rationing places. Usually there is a list of oversubscription criteria so that when the school has too many applicants for its number of places all those over-subscription criteria come in. For community schools, living near is a key one but it is not the dominant criterion for faith schools. Clearly the house price effect there will not kick in for the faith schools. But there are still differences in the preferences for those types of schools among different types of families, even if there is no house price linkage.

I was using the house price linkage as evidence that that takes place in the community school sector. If you step aside from that and just think of the voluntary aided schools, clearly different people have different preferences. This might be what is driving the parent side selection into those kind of schools. Some people just do not want to go to those schools and some people do. There are differences between those types of people in terms of their background and achievement.

Q40 <Mr. Slaughter:> Is the answer on the academies that it is too early to tell whether there is a trend towards taking a more exclusive social intake or not? If there is a trend, how would you explain it?

<Dr. Gibbons:> I have not looked at academies at all so I could not comment on them.

<Dr. Benton:> With the numbers of academies it would be hard to summarise that finding. In the last report there were 27 academies. Within those there are some where the percentage taking free school meals is going up and others where it is going down. We cannot generalise from that to say that academies mean more selection. There are simply not enough of them at this stage to be able to make that statement.

<Chairman:> We will drill down on that in a different way.

Douglas, on school diversity and collaboration. Oh, Annette, do you want to come in here?

Q41 <Annette Brooke:> I am sure that Douglas will take these questions further, but I would like to start by looking at the choice model and competition. My first questions will be directed towards you, Steve. If we have choice and the competitive model, is it just a matter of sorting out all the imperfections in the market to address the problems that we are talking about this morning?

<Dr. Gibbons:> The problems in terms of the differences between schools, or are you thinking of overall levels of achievement?

Q42 <Annette Brooke:> If we had a perfectly competitive model, would we not end up with a set of schools that were all of equal performance?

<Dr. Gibbons:> There is a diversity of opinion on that, and there are two views. First, if you have a school system that admits purely on the basis of where people live and takes only people from their local community, the make-up of the school and the achievement of pupils in that school are dependent on the kind of kids who live in that community. There are differences between communities for reasons other than schooling, such as the quality of housing and the environment. In turn, if people start paying for a good school-through house prices-it will drive sorting of a different kind in a neighbourhood, and you will wind up with a very unequal system in that scenario. If you opened up the competition and allowed people to choose any school, it would break down that linkage and you could wind up with a more even distribution of achievement across schools.

The other view is that if you open schools up to competition and allow parents to choose more widely, the most motivated parents-those with the willingness and ability to pay to travel across the borough by car to drop their kids off-will make the effective choices, which could exacerbate the inequalities. It could go either way, so the jury is still out on this one.

Q43 <Annette Brooke:> Do we not just need to identify all those imperfections and tackle them one by one? Is that possible? If we were to address transport costs, it could truly facilitate choice. To a certain extent, that is in the new legislation. Choice advisers might fill the gaps in terms of parents not perceiving the best choice for their child. Can we just keep drilling into all the imperfections and remove them? Would we end up with the perfect competitive model, which I do not actually follow, under which a poor school that is not performing will just wither away and something will come in its place?

<Dr. Gibbons:> There are two objectives: one is to raise the level of achievement; and the other is to equalise achievement across different schools. To start with, let us think about equalisation. You are right that if you designed a system of choice very carefully, and subsidised transport and provided information, you could come up with a system that would make everyone equally likely to make the right choices and wind up with a very even distribution of people across schools. There would be a lot of unintended consequences-there would be a lot more travelling, so you would create a whole set of new problems-but if the objective was to level the playing field, it would probably work with a lottery system coupled with transport facilities.

Whether that would do anything to push up achievement levels generally, and whether competition is an incentive on schools and actually raises achievement, are slightly different questions. It could work in two ways: through people finding schools that better suit their needs; and because-as in the example that you just gave-the schools that do not succeed will just wither away and die.

However, an inevitable feature of that model is that there must be inequality of achievement because otherwise those schools will not die out. I presume there is a transition that involves a lot of inequality of achievement in that kind of model. Perhaps, in the end, you wind up with better performance that is equally spread out, but it is very hard to say. The transitional consequences could be quite extreme.

Q44 <Annette Brooke:> So, the period of transition might be too painful. You said in a paper to which you contributed that although the competitive model might even out ability, it would have downsides. Am I right?

<Dr. Gibbons:> We were looking at primary schools in that series of papers and considered two matters: first, whether the performance effect of competition between schools, and parents having a lot of schools to choose from, raises achievement; and, secondly, the inequality aspects. Our conclusion was that there was generally no evidence that competition and choice made any real difference to performance. There was some evidence that that worked in the voluntary aided sector, where the incentives might be more correctly aligned for that model to work, but the impacts were quite small. Where the costs came in, the downside that we referred to was that we had some evidence that that tended to increase inequality. In areas in which there is a lot of choice among schools and a lot of closely-located schools so that people can choose among them, there is actually more stratification and more sorting-and more segregation, if you like-across schools. The downside is the inequality.

Q45 <Annette Brooke:> Do you mean in terms of socio-economic background? I was not quite sure which inequality you were talking about.

<Dr. Gibbons:> Yes. It is inequality in terms of achievement, but as we have been discussing, achievement is closely linked to the prior achievement and background of the children, so the two are virtually synonymous.

Q46 <Annette Brooke:> So, is your conclusion that we do not raise standards for the very children for whom we want to?

<Dr. Gibbons:> The evidence suggests that the effects are marginal. The international evidence is not exactly convincing on the idea that more competition increases the performance of schools.

Q47 <Annette Brooke:> May I address some questions to Tom and Simon? Is there any available evidence on collaboration in any areas, or is it that any collaboration that might exist is rather cosmetic? If we are going to have choice and not go the whole way with the model, in order to support other objectives such as equality, collaboration must be an important part of the model.

<Dr. Benton:> You are asking about general measures of collaboration.

<Annette Brooke:> Yes.

<Dr. Benton:> I do not know any general ways of doing that. Within particular evaluations or programmes there could be a purpose to collaborate. For example, I have done an evaluation looking at delivering vocational qualifications at key stage 4, and we can look at evidence there of schools helping each other. If one school cannot deliver an NVQ in a particular subject, they could get together with another school and send pupils backwards and forwards. We have some data from particular programmes, but nothing global about how much collaboration schools are involved in as a whole.

Q48 <Annette Brooke:> And whether it makes a difference, I suppose.

When the Schools Commissioner visited our Committee he gave examples from Kent, where the implication was that through the Building Schools for the Future programme, there was encouragement for grammar schools and secondary moderns to work together. Do you see that that might have a positive outcome, or will it just be cosmetic-sending a few pupils here and there?

<Dr. Benton:> It is certainly possible that it would have a positive outcome, but I do not have any evidence on that.

<Simon Rutt:> On added evidence for that, part of the evaluation of Excellence in Cities looked at partnership-level information where local authorities worked to develop these sorts of collaborations and partnerships between schools. We had lots of information about different levels-leadership, management and so on. There were a number of different indicators to inform partnership-level collaboration. When they were introduced into the model, they had no effect over and above the pupil-level effects that we have discussed. There was no added benefit of having a good partnership score or a low partnership score. Over and the above the pupil-level and other school effects that were already there, we did not see anything else. It was not the greatest possible measure in the world, but that is the only thing that I have seen and been involved with that used this sort of collaboration. There was no effect over and above lots of the other pupil-level and school information that we have.

<Annette Brooke:> Thank you. That is rather gloomy really.

Q49 <Mr. Carswell:> I am interested in the idea that competition does not necessarily raise standards. If that is the case, this must be about the only sector in socio-economic activity where more competition does not enhance outcome. I was at a recent lecture given by someone from the Milton Friedman Foundation who produced a lot of evidence to the contrary.

I want to explore the idea of spreading good practice. To spread best practice, one basically does the good things that other people do. To do that, you create an incentive to do what others do. Surely competition, rather than collaboration, is the best way of spreading best practice? For example, in the business world, the practice of putting airbags in cars was spread by companies competing with one another, rather than collaborating. Is it not the case that if you really want to spread practice, competition is a better way of doing that than collaboration? If you disagree with that, I would be interested in why you think that education is different from virtually anything else.

<Dr. Gibbons:> I do not have any evidence on the effectiveness of collaboration. The only evidence that I have is on the effectiveness of having a range of schools to choose from in the London area, so I could not really say. Without knowing whether collaboration works or not, I cannot comment on that. I think that schools are different from commercial activity and the market sector-there is a difference here. There are a lot of reasons why you might expect competition not to work especially well. It might be better if kids are brought up in environments in which teachers are not put under those kinds of pressures. I am not arguing for that, I am just saying that there are a lot of-

Q50 <Mr. Carswell:> Why would it be better for teachers not to have competition?

<Dr. Gibbons:> That argument is put forward. I do not know the way that teachers operate, but I would imagine that if you have a classroom of kids, there are a lot of things that you have to deal with that are not just to do with thinking about raising their standards. There is a lot of classroom management and other educational activities that go on, so if you have this tunnel vision on raising standards to try to beat the nearest school up the road, that is perhaps not very productive.

<Mr. Carswell:> Could not MPs say the same? If, as a politician, I did not have competition in terms of having to stand for election, I could spend my time doing other things. Surely you need competition to get the best out of teachers?

<Dr. Gibbons:> I am not really arguing against that; I am just saying what the evidence is, generally speaking. There is evidence that you could find, particularly from Caroline Hoxby in the United States, that would support the idea that competition works, but the bulk of the other international evidence suggests that it does not. I am just stating the evidence.

Q51 <Mr. Carswell:> Do any of the other witnesses wish to comment?

<Dr. Benton:> On the issue of sharing best practice between schools, one of the things about education that is different from making cars, for example, is that identifying best practice within schools involves a lot more debate. It is a lot harder to say clearly, "This is the way such and such should be taught and all other ways are wrong." It is difficult to identify those things, so you might not expect that to work in the same way as in other sectors.

Q52 <Mr. Carswell:> The word "collaboration" itself is interesting. If I learned that Ryanair and British Airways were collaborating, I would assume that they were ripping off the customer. Is there not a case for saying that collaboration is another way of describing a "non-compete" agreement between schools, and that that is a convergence of the producer interest and that, by definition, the consumer interest will suffer?

<Dr. Gibbons:> It is not collusion on price or anything, is it? It is just about sharing practice and techniques. I do not have any evidence on this, apart from the fact that I have worked as a school governor and I know that the school's teachers and head teacher seemed to value their visits to other schools and the contact that they had with other schools, and that they learned things from those experiences. However, I could not give you any evidence on whether that is effective or not, and I do not see that it is equivalent to collusion in the way that you imply.

<Dr. Benton:> Also, the collaboration in some of the evaluations that I was talking about, again with vocational degrees, is not collusion. It is particular expertise in one area, or, indeed, facilities for the teaching of a particular subject that are not present at another school.

Q53 <Mr. Carswell:> Does it mean less diversity?

<Dr. Benton:> In terms of the school types?

<Mr. Carswell:> Yes.

<Dr. Benton:> Not necessarily. It could mean more diversity because you do not need every school to be able to teach every subject. You can work together. There might be two people who want to do an NVQ in engineering or motor care or something, but you would not need every school to have the facilities for that. There could be collaboration with schools and further education colleges to ensure that the necessary expertise was shared.

Q54 <Mr. Carswell:> The question I have written down is: "How can collaboration between schools be encouraged?" I want to change that slightly. If collaboration is such a good idea, why does it need any encouragement at all from the state?

<Dr. Gibbons:> I have not said that collaboration is a very good idea. I do not know whether it is. I cannot see that it would be harmful, but I have presented no research that indicates that collaboration has any positive outcomes.

<Simon Rutt:> I have no evidence, apart from a little from Excellence in Cities partnership working, which showed no major effects over individual pupil-level factors. I have no great evidence that it works, but one would think that it ought to be encouraged, with practice shared between schools. What they are working with-the pupils-is different, and the environments are different. One would have thought that shared practice ought to be encouraged, but I have no real evidence of it having a major impact.

Q55 <Chairman:> This is not a reflection on your evidence, but I am feeling a bit depressed after this session. If diversity and competition do not make any difference, what does your research lead us to say about policy? In a sense, you are saying that this love affair we have had with competition and choice, going across all parties, of course, is not getting across.

<Paul Holmes:> Leaders, not "we".

<Chairman:> Not we, no.

<Mr. Carswell:> I do not accept that.

<Chairman:> No, we are hypothesising here. If this is a dead end in terms of policy, what does your research say can make a difference? I read you as saying that nothing makes a difference and that there is nothing we can do about this: poor kids from poor homes will not attain very well, so what can we do about it? Am I misinterpreting you?

<Dr. Benton:> I think our research is about overall school management and school structures, and whether that has an effect on achievement. It is very difficult to find things in that area. Certainly, there is research on classroom practice and things on the ground, with enormous amounts of evidence showing that there are things that make a difference to pupils' achievement at that level. I am not very involved in that research, but at conferences, a lot of people present teaching methods that are effective and good for pupils along those lines. All we are talking about is the big structural things and whether there are any big structural things you can do that affect pupils. It is harder to find things at that level.

Q56 <Chairman:> Okay. Let me bounce something through that. I have an idealistic view: when the chief master of King Edward's School in Birmingham or the high master of St Paul's School in London tells me that comprehensive education is not very good, and I look at their schools-high competition, all sifted kids from middle-class backgrounds-I would be really upset if I were a parent and my kids at one of those schools did not achieve very high standards indeed.

On the other hand, my view has always been that, if a school reflects the community in which it sits and it is a balanced community, you have a much fairer chance of getting good results for all the children in that community. Is there any evidence that this view of mine is correct? Would that lead on to, say, a banding system, where there is a duty on schools to take a fair proportion of children with special educational needs, looked-after children and children on free school meals? Would that improve educational outcomes overall?

<Simon Rutt:> There is no evidence to suggest that would happen. It would be interesting to do some work on that. I believe that a local authority in East Sussex-perhaps it is somewhere else-has started a lottery for admissions to schools, with a random selection of pupils. It will be interesting to look at the pupils in those schools and see what the effect is: whether low-ability pupils have been dragged up, because of mixed ability and mixed characteristics, with medium, high and low-ability pupils, and whether low-ability pupils are moving up the scale, rather than just having a few of them in a high-ability school or a school with an awful lot of them. It will be interesting to look at local authorities that do that to see whether it has had an effect on all pupils, particularly pupils at the lower end of the ability scale.

Q57 <Chairman:> But researchers have loads of examples of schools with a balanced intake-we were given some by Andy Slaughter-as opposed to some schools that only have certain kids. Some of us visited a school in Maidstone, where 100% of the pupils received free school meals and 65% were SEN. Are you telling me that the opportunities for a decent education for kids who go to a school where 100% of children have free school meals are no different from what they are for children who go to a school with, say, 35% free school meals?

<Dr. Benton:> No. Certainly, all the research shows that going to a school with a low percentage of free school meals is beneficial. Having pupils around who are clever or from middle-class families has an impact on the whole school.

Q58 <Chairman:> I was not saying that; I was talking about a balanced intake.

<Dr. Benton:> Sure, that is right. You were asking whether being in a school with 100% free school meals is just as good for you as being in a school with 0% free school meals. It is not. I was just answering that question first. In terms of having a balanced intake, that would be somewhere in the middle. If you did that to all schools, and gave every one a balanced intake, you would find that some pupils who were previously in a school with 100% free school meals would be better off. But, equally, the evidence appears to show that pupils who were initially in a school with 0% free school meals would be a bit worse off. So it would not appear to make an overall improvement to the system. There is no evidence to show that that would immediately improve things.

Q59 <Chairman:> What about other evidence? Research has been carried out in Kent showing that the children who go to the grammar schools get a much better education, but if you take all the children in Kent together they get a worse education in terms of the totality of children in that area.

<Dr. Benton:> That is right.

Q60 <Chairman:> Does that contradict your view that the ones who gain are balanced by the ones who lose?

<Dr. Benton:> That is only one local authority. When you look at it as a whole, it comes back to the point I made earlier about relating the percentage of pupils who are selected to their achievements. The relationship is slight. When you look at that nationally across the country, and ask, "Is there a relationship between the percentage of pupils in a local authority who are selected and the results?" you see that there is not much of a relationship. Although there might be one situation in Kent, as a whole, looking at the situation across the country, that does not seem to make too much difference overall.

<Dr. Gibbons:> To come back to your point about the balanced intake issue, one point of confusion is that when you are talking about a balanced intake relative to a school at which all pupils get free school meals, that is an improvement in intake. But if you compared your balanced intake with a school with no free school meals, that would be a worsening of the intake. So we are saying that perhaps the shifting-up of the average characteristics of the pupils when they come in has an impact on achievement, although that is a bit unclear. There is some evidence that it does, and some evidence that it does not. In terms of balancing or having a mix, I do not think that there is any evidence that that matters in itself. Having a mix is better than having all free school meals, but it is worse than having no free school meals.

Q61 <Chairman:> So parents are absolutely logical in seeking a school with the fewest poor and SEN children.

<Dr. Gibbons:> I can talk a bit about that. There are probably different views to a certain extent, and there are different views in the literature. But it is a question of the impact of peer groups; it is a peer group effect story. Whether being among high-achieving classmates impacts on someone's own achievement is a very hard thing to measure, because of the problem that high-ability kids are sorted into schools with other high-ability kids. Separating out whether there is any causal linkage is difficult.

I have written a paper that investigates the issue and looks at secondary schools. We concluded that the link is in fact very small. Given a child's age 11 achievements, if they go to a secondary school with other kids who are high age 11 achievers, they do only marginally better by the time they reach key stage 3 at age 14. There is a tiny difference. In fact, we have extended that research and have tried to measure that difference by considering a primary school and a secondary school. In any year, there is a flow of kids from one school to the other, and year after year we can follow what that flow looks like and explore how kids who make the same primary to secondary school transition differ in relation to the composition of the secondary school to which they go. Changes over time can be used to see whether the kids who make the same primary to secondary school transition do better in years when the secondary school has a high average intake from the local primary schools. You get nothing from such research. People come in and are sorted into schools with people of a certain age 11 ability, and they come out at age 14 at the same point in the distribution. According to our research, it does not seem to have any impact whatsoever.

That prompts the question: why do parents want to choose schools that have low free school meal intakes and high-achieving kids? That question is not easy to answer. It is clear that a lot of things that go along with education and being in school are not to do with achievement. Parents value the safety of their kids and the child's well-being, and there are many other considerations that come into play. The pure search for value added is not a big issue for most parents, most of whom probably accept that their kids have certain skills and abilities and they will either do well or will not do well, whatever school they go to. Many other issues that inform school choice are probably not about pure value added in test skills.

Q62 <Chairman:> Tom's research shows that the kids who just get into a selective school did better. Doesn't that contradict what Steve just said about it not making much difference?

<Dr. Benton:> That was to do with selective schools. I think what Steve was saying was not particularly focused on selective schools; it was about general school composition within any school. Selective schools are only a very small number of schools. They are separate pieces of research. In terms of what Steve has said about whether the composition of the class has an effect, I would agree with him. There is a lot of debate in the literature about the effect that that has and there certainly are differing views.

<Chairman:> As policy makers, that does not give us much of a steer.

Q63 <Mr. Slaughter:> The point was made about value added, and the fact that it might not be much of an issue for parents. Should it be an issue for us? You are saying that instead of having some schools moving towards 100% free school meals and some towards 0% if all schools moved towards 50%, or at least a mixed intake, that would make no difference to overall performance and would not result in some children doing better. You are admitting the correlation between social stratification and results, but not admitting that readjusting it would produce any overall increase in performance. You are the experts-my evidence is all anecdotal-but I am quite surprised, because the trend is that the schools with a low percentage of free school meals often have good exam results and tend to coast along. A high percentage of free school meals can often reflect a great deal of mobility in the school population, with a lot of quite challenged children and many children coming in for whom English is a second language. That makes it very difficult for the school to sustain improvements. You often get schools that have false take-offs and then go down.

If there was a greater social mix, it would be easier to hold together the grist in those schools. My observation is that that does happen and you get better results by doing that. There is a trade-off to be made, and individual families will not necessarily like it, but I am slightly surprised to hear you giving that view.

<Dr. Benton:> I understand what you are saying about the perception of teachers within a school. However, our research looks at the thousands of schools across the country. We can see whether a pupil at a certain level in one school does a lot worse than a pupil of that level in a different school. We can see what the difference is for children who go to schools with high free school meal intakes. You can compare any two children you like out of the half a million or so from the national data, and you will see that, on the whole, there is not much difference. As that is based on a lot of data, that is where our conclusions come from.

On the changes in results that you are talking about, one of the problems of looking at raw results is that, in schools that are doing very badly in terms of the percentage of pupils achieving five GCSEs at grades A to C, there tend to be more fluctuations in results. Such results may be more down to a statistical phenomenon than to the issues that you have raised.

Q64 <Mr. Slaughter:> Have I understood this correctly? You are saying that, if you take two similar children from similar ability levels who go to very different schools in terms of ethos, performance and social intake, they will do similarly.

<Dr. Benton:> That is the way that the evidence seems to point.

Q65 <Paul Holmes:> The Chairman said that he is very depressed at the evidence that we have received, but I am quite pleased with it, if we believe in evidence-based policy making, which supposedly we do. You are saying that your evidence and most of the British and international research says that what matters in school attainment is not who the head is, whether it is a faith school or an academy, but the intake of kids, their background and so forth. If we have identified that the problem is not diversity, but the family background and prior attainment of the kids, does that not mean that we should be focusing all the extra effort, initiative, input and money into the problem areas, not into rewarding successful schools, which are already successful because they have good intakes of kids?

<Dr. Gibbons:> The conclusion that I would come to from the evidence is that we need to tackle the disadvantages of the kids at the point that they enter the school. Schools provide some kind of vehicle for delivering whatever policies you want to put in place to reach those children. It is not the school-level differences that are important, but using schools as a way to get to the disadvantaged kids within them.

I come back to the point that I made at the beginning: the variation within schools is enormous compared with the variation between schools. Therefore, if you are worried about low achievement, you need to tackle low achievers within every school. There are some schools with very few low achievers, but 95% of schools have someone from the bottom 5% of the distribution of achievement. You need to tackle these problems in every school. Extended schools ideas seem to be sensible; they are vehicles for delivering services to families via the school.

Q66 <Chairman:> What you are saying points to our investment in pre-school, to early years and to Sure Start, because you are saying that it is too late once the child is in school.

<Dr. Gibbons:> Yes, it comes back to basic differences in family background. Obviously, differences in innate ability must play a role here as well. The initial conditions that kids come into schools with are driving the extremes in terms of achievement. It is not a question of the failures of schools to do things.

<Chairman:> Everyone now wants a question.

<Paul Holmes:> No, I have made my point.

Q67 <Fiona Mactaggart:> I have three quick-fire questions. I will ask them all at once, but I do not expect them all to be answered, because they are factual, I hope.

First, is the pattern of achievement being so directly connected to family income an international pattern or is it worse in Britain? Secondly, are there any long-term figures? We have talked about results within a school and within an age range, but I am interested in what happens to those children when they are 21 and 25-do you know? Does anybody know? Thirdly, when we were discussing differences between schools and school admissions, we focused on the difference between community schools and voluntary aided schools. Is the pattern the same for foundation schools as for voluntary aided ones? I do not feel that we teased that out.

<Dr. Gibbons:> I think that the fact that background is linked to achievement is generally an international phenomenon; it is universal.

Q68 <Fiona Mactaggart:> Is it worse in Britain than elsewhere?

<Dr. Gibbons:> I do not know the magnitude off the top of my head, but it is of a similar order. Literature produced by some of my colleagues at the LSE on inter-generational ability suggests that perhaps there is a stronger link in Britain than elsewhere. It is very difficult to get evidence on this, because the number of surveys that cover the issue is rather limited.

Q69 <Chairman:> Your colleagues told us quite strong things about social mobility in the UK when they were here last week.

<Dr. Gibbons:> Yes, that it is worse.

<Chairman:> And do you think that the two are related?

<Dr. Gibbons:> Yes. Clearly, the links between background and achievement are directly linked to the social mobility question.

Q70 <Chairman:> So a greater percentage of our population is poor, low-achieving and non-aspirational regarding its children's education than other countries, to put it crudely?

<Dr. Gibbons:> No, the evidence is that, in the long run, kids seem to progress up the income distribution scale from lower levels less well in this country than in other countries. Off the top of my head, I do not know how to compare background and achievement internationally, but there is a strong link everywhere. That is well known. You can see that in PISA-the OECD Programme for International Student Assessment.

Q71 <Fiona Mactaggart:> The second question was about long-term results. Do we know about these children when they are older? Do they end up going to prison; do they end up going to university? I am interested in whether we know that these things have results in adult life and in terms of success in the world, or whether some of the results that we are talking about are short-term.

<Simon Rutt:> I have no direct evidence of research that has been carried out, but we are just undertaking some research where we are tracking pupils through their secondary education and then looking at what has happened in further and higher education, to see whether there is any relationship between the schools pupils are in, academic attainment at 16 and what happens in further and higher education. This is through the Aimhigher initiative, which was recently introduced.

We have a lot of pupil attitudinal data as well, looking at aspirations in terms of higher education and aspirations and attitudes regarding school and education in general. It will interesting to be able to plot that through and look at pupils to see who ends up in further and higher education, and how what happens in statutory education affects what happens in post-16 education.

Q72 <Fiona Mactaggart:> So you are saying, "Watch this space."?

<Simon Rutt:> It is due to happen very shortly.

<Dr. Benton:> There are certain bits of research about post-16 and onwards. For example, we are looking at some stuff on the youth cohort study at the moment, which shows the links between achievement at school and the chances of being in education and training later in life. So there are sources of data, but as people get older, it gets harder and harder to track them.

Q73 <Fiona Mactaggart:> The people who drop out are the people who succeed least in my experience.

<Dr. Benton:> That is right.

Q74 <Fiona Mactaggart:> Samples are so selective that they are inaccurate. And the point I made about foundation schools?

<Simon Rutt:> Foundation schools-looking at the tables again-seem to be very similar to voluntary aided in their admissions and the type of pupils they take, as in the communities they serve, the communities where they sit, the proportions of characteristics within those communities and who ends up going to foundation schools. They seem to be more similar with voluntary aided than with community.

Q75 <Chairman:> What is interesting for us is the joined-upness of this research. We have had a session on social mobility, and we are trying to link that to the stuff you are telling us and to relate that to policy direction and policy decisions.

We are also trying to research back down the chain. When you go to schools now they will tell you-and local education experts will tell you-that they are now able to predict as a child comes into the school whether they are going to end up as a failure, a NEET or whatever. They know extremely early. Have you done research on how early you can tell a child's level of achievement?

<Dr. Benton:> We can predict it fairly early, but it is not that accurate. We were talking about the very large variation between pupils within schools. So you would not be able to predict all that accurately when a pupil arrives at secondary school what is going to happen to them by the time that they leave.

Although, as we have talked about, we can predict 90% of the differences between schools, within schools knowing which children are going to succeed and which are not is much more tricky. To predict on the individual level, who you are makes quite a big difference.

Q76 <Annette Brooke:> Two things. Can I come back to Steve on whether this competition makes any difference? In your article, you cover secondary schools that are close together in an urban area. You suggest that strong competition in an urban setting can deliver better results. Can you comment on that?

<Dr. Gibbons:> There are two strands to the research. One paper looked at primary schools, where we looked very carefully at the potential effects of choice and competition. From that, we found little evidence that it made any difference. In general, we found some evidence that it worked for voluntary aided schools.

A separate paper, looking at secondary schools, does something a bit more general, looking at whether schools in dense settings in urban areas perform better or worse than schools elsewhere. We were trying to get at the question of whether city schools are failing schools, or whether they are not doing well because they have a low-quality intake, if you like.

Q77 <Annette Brooke:> Could I just say that that is not true competition, because we have not got a rural situation where there are not very many choices?

<Dr. Gibbons:> It is not true competition. No, that is right. That paper was really about the effects of density on performance. We found that schools in high-density settings perform better, marginally so, in terms of the value added between the age of 11 and GCSEs. We cannot pin down what that is to do with. It is closely linked to the number of neighbouring schools, rather than more general things, such as population density or proportion of built-up environment. It seems to be something to do with the schools, but we can only conject what it is. It could be collaboration; it could be competition. One candidate explanation is that it is a competition-generated effect. That is a bit of positive evidence, but we could not pin it down to be specifically due to competition.

Q78 <Chairman:> This has been a very interesting session. We have about a minute remaining. We have really appreciated your expertise. You seem to be giving a strong message today. If you were in a fantasy land where Fiona Mactaggart or Douglas Carswell was Secretary of State and you were the Permanent Secretary, and you did not think that the diversity and choice policy direction would raise the standards for most students in our country, which policy areas would you look at? About what would you say, "Minister, this is where I would be looking, based on my research."?

<Dr. Gibbons:> I could not really say much more than I have already about focusing on the differences in kids within schools, rather than trying to focus on between-school differences. The diversity, the schools targeted to try to raise performance in one school relative to another, is a red herring. We should focus within schools.

<Dr. Benton:> I would suggest looking at things at classroom level that can improve learning and borrowing from ideas in medical research, such as proper randomised control trials, to work out the most appropriate methods of teaching, and sharing that.

<Simon Rutt:> I reiterate what both my colleagues have said, but there are also some questions about the information available on admissions policies to look at the choice that parents have and who goes where and the impact that that has on schools.

<Chairman:> It has been a good and very thought-provoking session. Sometimes, it felt like a seminar, and it was all the better for that. Thank you very much.