The map currently displays every mainstream state funded school in England with secondary age pupils (i.e. secondary, all-through and middle schools, but not sixth forms) that is open as of November 2013.

For admissions the percentage displayed is the precise percentage allowed to be selected on religious grounds by the school's admissions criteria, if the school is sufficiently oversubscribed.

For free school meal eligibility and English as an additional language we display the ‘best’ and ‘worst’ 1%, 2%, 3%, 4%, 5%, 10%, 20%, 30% and 40% of schools. In each case this means that the school’s place in the rankings lies between the figure given and the next lower figure: i.e., ‘Best 30%’ indicates a school ranked between 20% and 30% from the most inclusive of the list by proportion of FSM or EAL pupils compared to the local area, while ‘Worst 3%’ means a school ranked between 2% and 3% from the least inclusive end of the list by proportion of FSM or EAL pupils compared to the local area. We also display ‘Similar to area’ for those in the 40-50% least and most inclusive. The reason we are more specific at the very highest and lowest ends than in the middle is because schools form a bell curve, with the extreme outliers being more spread out. We don't display more precise rankings as differences may be down to a few pupils which could just be chance.

On admissions, we colour fully inclusive (i.e. 0% selective) schools green, schools that partially (i.e. 1-99%) select yellow, and schools that fully (i.e. 100%) select red.

On FSM and EAL, roughly speaking we colour schools green if they are more inclusive than would be expected for a school in their area (i.e. have more pupils eligible for FSM or who speak EAL than would be expected), yellow if they are about or slightly less inclusive (i.e. have a few fewer pupils eligible for FSM or who speak EAL than would be expected – which might just be down to chance), and red if they are highly uninclusive (i.e. have a lot fewer pupils eligible for FSM or who speak EAL than would be expected). More details on how these scores are calculated are set out in the next section of the FAQs.

The rankings are not intended to be any kind of definitive judgment on how inclusive or uninclusive particular schools are, but merely present in an understandable form what our data shows.

On the school admissions tab, the colours show how religiously selective the oversubscription criteria are of schools in each local authority (with green being more inclusive and red being less inclusive). On the free school meal eligibility and English as an additional language tabs, the colours show how segregated the schools in the local authority are on the measure (i.e. how much the schools in the local authority vary – their standard deviation).

Take a look at the ‘Overall averages’ tab for information.

You can find some of this information on the ‘Overall averages’ tab, and for local authorities you can currently get a good idea by clicking on them on the heatmap.

A few schools are missing FSM and EAL data because they have opened since January 2013 and so such data is not publicly available yet.

Some schools that have been found to be unrepresentative of the areas in which they are located may have legitimate reasons for not being representative. However, we are not seeking to make value judgements on the legitimacy or otherwise of different schools’ rankings – simply to present the data and allow people to draw their own conclusions.

As we explain in the methodology section below, a school's local area percentage is calculated by looking at its middle super output area and other MSOAs in the same ward, first half of postcode and local authority. If two neighbouring schools are in different MSOAs then they may well end up with slightly different local area percentages.

However, we group schools into percentile bands (i.e. lowest 1%, 2%, 3, 4, 5, 10, 20, 30…) and do not display the exact score each school has been assigned precisely in order to not overstate (and stop over-interpretation of) the precision of our results: while a school’s score might be altered slightly by peculiar ‘border effects’, a school's band is unlikely to be.

This is quite probably because pupils move between different areas when applying to school, especially for the FSM figures. It might also in part be because we are only displaying mainstream state schools, while the figures for FSM pupils in an area also include those at special schools (which are very small in number so this is unlikely to have much impact), and for EAL they also include those at special, alternative provision and independent schools (of which there are more but still unlikely to have a massive impact).

The heatmap measures the standard deviation of the difference between schools and their local areas within a local authority, i.e. the extent to which schools vary in how different they are from their local areas. Schools in a local authority may gain a positive (or negative) score overall for the reason given in the previous FAQ, and yet there still might be quite a bit of variation between those schools as to just how inclusive/segregated they are: it is the latter that the heatmap is measuring.

For the simple reason that there are no secondary schools in the City of London.

Because there are two small islands in the Bristol Channel that are part of the local authority.

Every effort has been made to ensure that the data is accurate and fairly presented, and that as many schools are included as possible. Contact us and we'll see what we can do.

School profile data comes from Edubase and the annual school census (most recently January 2013). Information on admissions criteria comes from the 2013-14 admissions directories produced by local authorities. We will update this to 2014-15 in due course.

Local area free school meals data comes from the most recently published data (2010) available on Neighbourhood Statistics, whereas local area English as an additional language data comes from the 2011 Census available on nomis. Adjustments have been made to account for the differences in age of the various datasets and, in the case of EAL, the fact that pupils are more likely to be recorded as speaking EAL by their school than on the Census.

Finally, information on which Middle Super Output Area, ward and local authority each postcode is in comes from the National Statistics Postcode Lookup.

Note that for a small number of schools there are suppressed figures (due to very small numbers of pupils being involved): here we assume the true figure is the average of the suppression range. There are also 30 religious schools whose admissions criteria take every pupil in a catchment area before starting to religiously select: we assume that these criteria allow 20 percent of pupils to be selected on faith. Finally, four Church of England schools give priority to pupils from all CofE primaries locally, without naming the individual schools (a breach of the Admissions Code): we assume that these criteria allow half of places to be selected on faith.

We intend to make this data available in due course.

For each school, a local profile is constructed taking into account the age of the pupils at the school and the difference in free school meal eligibility between primary and secondary schools. We do this by first of all taking appropriately aged pupils from the Middle Super Output Area in which the school is situated (an MSOA consisting of 5,000-15,000 people). If there are fewer pupils in the MSOA than are at the school, we then gradually grow the local profile by adding to this pupils from other MSOAs in combinations of the same ward, first half of postcode (known as an outcode) and lower local authority (i.e. London borough, unitary authority, metropolitan or non-metropolitan district, as they were prior to the 2009 reforms) until our local profile has at least as many pupils as are at the school.

The below Venn diagram shows the six different areas used to build up our local profile – we start with MSOA, then other MSOAs in the same ward and outcode, and so on, gradually growing our local profile outwards from the areas closest to the school until we have enough pupils to match its intake. This process therefore attempts to recreate what the school’s intake would look like if it was simply admitting its local community.

Our approach is consistent with those in the literature; Allen and West simply take all the pupils living closest to the school until they get to the number of pupils at the school, reporting that ‘Of course, the choice of neighbourhood measurement is necessarily somewhat arbitrary, but many have been tested and the substantive results reported are not sensitive to alternative measures of neighbourhood. Alternative neighbourhood calculations include use of postcode areas (e.g. Chamberlain et al., 2006; Sutton Trust, 2006), a three-kilometre walk-zone or the local authority.’

In our previous work we have simply looked at the pupils living in the MSOA and not factored in the wider area, and didn’t weight by school size in averages. We think that our new approach is more accurate, but it does not significantly affect findings we have published up to now.

Once the school’s local profile is created, the school is then compared to its local profile by subtracting the percentage of pupils having the characteristic (FSM eligibility or EAL) in the school’s local area from the percentage in the school (known as the absolute difference). Schools are then sorted based on this difference, and our percentile ranks and colour coding are calculated from this score.

In the ‘Overall averages’ tab, as well as presenting absolute differences, we also sometimes present proportional differences, where we divide the percentage of pupils having the characteristic in the school by the percentage in its local area, and then calculate how much the resulting percentage differs from 100. Figures in this tab are weighted by school size and normalised so that the average across all schools is 0.

We will add English primary schools shortly (although won’t be able to present information on admissions criteria due to the large number of religious schools making it too time consuming to check them all).

We would like to add Welsh schools but haven’t yet established if the school or local area data is available.

We have data on English special and alternative provision schools and sixth forms but don’t have local area data; we would have to obtain this before adding them, which may not be possible.

We have basic data on each private school but not on how many pupils are eligible for free school meals or speak English as an additional language. Such data is not collected and so we cannot add these schools.

We have ethnic breakdowns of each school and local area and will consider adding ethnic profiles in due course.

We don’t have local area data on special educational needs or academic performance so this would be harder to add.

The following text is extracted from the Fair Admissions Campaign’s upcoming report.

In 2012 Shepherd and Rogers found similar patterns of low numbers of pupils eligible for FSM in English faith schools. 76% of Catholic primary schools and 65% of Catholic secondary schools were found to have a smaller proportion of pupils eligible for FSM than was representative of their postcode. 63.5% of Church of England primary schools and 40% of Church of England secondary schools were also found to have a smaller proportion eligible for FSM than was representative of their postcode. Although 40% perhaps does not sound particularly high on first reading, only 29% of secondary schools without a religious character were found to take a smaller proportion of pupils on FSM than was representative of their postcode. This means that both Catholic and Church of England secondary schools are significantly more likely than secondary schools without a religious character to have student bodies which under-represent students eligible for FSM, which corroborates the Fair Admissions Campaign research presented here.

Many studies have also found evidence that those faith schools which are their own admissions authorities (which are more likely to be religiously selective) exhibit a greater degree of socio-economic selection than other faith schools, and this is corroborated by our findings on faith schools that have not had any external restrictions on how religiously selective they can be. In 2007 Tough and Brookes found that ‘Faith schools which are their own admission authorities are ten times more likely to be highly unrepresentative of their surrounding area than faith schools where the local authority is the admission authority.’ They also found that ‘Non-religious schools which are their own admissions authorities [which, at the time, were predominantly grammar schools] are six times more likely to be highly unrepresentative.’

In 2011, Dr Richard Harris found that ‘The proportion of pupils in the London data who were eligible for FSM in 2008 was 0.266. The mean (and median) proportion in… VA CoE schools… was 0.242 (0.181), in VA RC schools, 0.201 (0.174), and in schools of the other faith group, 0.138 (0.128). Each of these school types is, on average, recruiting disproportionately few FSM-eligible pupils, with the proportion for VA CoE schools closest to the expected value. Insofar as FSM eligibility is a marker of economic disadvantage, it is hard to avoid the conclusion that… faith schools, on average, are socially selective.’

Extensive research on this issue has also been conducted by Dr Rebecca Allen and Professor Anne West. In August 2011, they reported that ‘schools with a religious character (or faith schools) have fewer FSM pupils and more top ability pupils and that, in general, they are more affluent in their intake than the neighbourhoods they are located in.’ In 2009, they concluded that ‘It is clear from our analysis that many religious secondary schools in London are not serving the most disadvantaged pupils. Overall, religious schools educate a much smaller proportion of pupils eligible for free school meals and their intakes are significantly more affluent than the neighbourhood in which they are located.’ And in 2008, when being interviewed by the House of Commons Children, Schools and Families Committee, Rebecca Allen noted that ‘In my most recent research… I was able to show that religious schools have higher ability and lower free school meal intakes compared with the neighbourhoods in which they are located. To give you an idea of the magnitude of those effects, if we take a community school and a voluntary-aided religious school, both located in a neighbourhood with exactly the same levels of deprivation, the community school is likely to have about 50% more free school meal children than the voluntary-aided school... We can show that there really is a direct correlation between the number of potentially selective admissions criteria that schools use, and the extent to which their intakes are advantaged.’

The following text is extracted from the Fair Admissions Campaign’s upcoming report.

It is worth addressing criticisms of the use of free school meals data for this kind of analysis as advanced by the Catholic Education Service for England and Wales. The CES’s arguments essentially boil down to the following three points:

1. ‘Free school meals’ is one measure among many. More Catholic pupils come from deprived areas than pupils at other schools: Relying on the Income Deprivation Affecting Children Index (IDACI), the CES said that ‘18.6% of pupils at Catholic primary schools live in the 10% most deprived areas, compared to 14.3% of pupils at primary schools across England. At secondary level, 17% of pupils at Catholic secondary schools live in the 10% most deprived areas compared to 12% of pupils nationally.’
2. The catchment areas for Catholic schools are geographically wider than other schools – even than local authorities. Therefore schools are being compared to geographic areas that are too small.
3. Free school meals require pupils to register. Many do not, skewing subsequent statistics. The Catholic commentator Christina Odone similarly claims that many in faith communities would be unlikely to claim FSM due to the social stigma of doing so.

Taking these three arguments in turn:

1. It is the case that pupils in Catholic schools are more likely to come from more deprived areas. For this reason, Catholic schools look relatively better on the IDACI, a measure that looks at how deprived an area is, than on comparisons of numbers of pupils eligible for free school meals. But this difference simply reflects the fact that Catholic schools are more likely to be in inner cities. It is not the same as saying that the pupils are actually more likely to be the most deprived people within those areas. To work this out, you would need to look at a measure of deprivation for the individual families – not simply for the areas in which they live. This is what the free school meal eligibility measure does. Catholic schools overall admit relatively fewer pupils eligible for free school meals than the average for their local areas.
2. Catchment areas for religiously selective schools are sometimes wider than local authorities, especially for secondary schools in unitary authorities, although it is debatable the extent to which this is justifiable as an excuse. However, primary Catholic school intakes are generally not larger than local authorities, and they also take significantly fewer pupils eligible for FSM than others in their LAs. In addition, as a rough approximation for secondary schools, if we focus just on Catholic schools in large counties, then we find that they have 77% as many pupils eligible for FSM as the other schools in their LAs: in line with Catholic schools in general. This is hardly surprising, as one would expect that where there are peculiar local affects due to catchment sizes (i.e. a school happening to be in a poor/rich part of a richer/poorer local authority), these are equally likely to affect a school positively as negatively – and so any such local affects are likely to cancel themselves out in the aggregate, which they do not.
3. This argument is the reason why FSM eligibility, and not FSM take up, is used in research such as ours. With that said, only those registered are counted as eligible. But even so, there is no evidence to suggest that it would affect religiously selective schools more than other schools. The Department for Education has the following to say on free school meals, when discussing its use in determining allocation of the Pupil Premium (which is intended to go to schools for each pupil they have who is from a disadvantaged background):

   But isn’t FSM an inaccurate measure of disadvantage?

   FSM is the only pupil level measure of deprivation available. The link between FSM eligibility and underachievement is very strong and data on FSM is easily collected and updated annually. The FSM indicator best fits the rationale for the premium.

Indeed, there is further evidence that religiously selective admissions policies are socio-economically selective that cannot be explained away by questioning FSM as a measure. Allen and West found in 2011 that ‘higher-income religious families are more likely to have a child at a faith school than lower-income religious families’, writing that ‘Significantly, within the groups of both Church of England and Roman Catholic families, children from top quartile households are statistically significantly more likely to attend faith schools, though the differences are not very large (9 versus 8% for Church of England families and 52 versus 47% for Roman Catholic families)’. In 2009 they also examined ten highly socio-economically selective Roman Catholic and Church of England secondary schools in London (five of each), and found that they had fewer pupils eligible for FSM than at neighbouring schools of the same denomination; they also found that ‘In all five [Catholic] schools, criteria and practices that could enable pupils to be ‘selected in’ or ‘selected out’ were used. According to the 2001 survey of school admissions, all five schools interviewed pupils and/or parents, one used banding and one used the academic record of siblings as part of the admissions process to decide which pupils of those who applied should be offered a place at the school.’ And ‘All five [CofE] schools required the demonstration of religious adherence. In addition, three interviewed pupils and/or parents; two used banding (in one case this was banding skewed towards higher ability pupils); one selected a proportion of pupils on the basis of aptitude in languages; two gave priority to children of former pupils; one used ‘compassionate’ factors; and one gave priority to pupils with pastoral reasons. Only one school used none of these admissions criteria or practices.’ Some of these practices have since been banned by the School Admissions Code.

Finally, in their 2010 report Unlocking the Gates, Barnardo’s found that their ‘services in Bradford and Luton have found themselves advising increasing numbers of newly arrived eastern European families in recent years. While these families are often devout Catholics and wish their children to attend a faith school, they can struggle to meet the priority admissions criteria for local Catholic secondary schools. In Luton for example, some have only recently arrived or have moved around the city and therefore have not had consistent enough attendance at a particular church to be able to gain the required reference from a priest; others are denied admission because they failed to gain entry (particularly if they arrived mid-year) into a Catholic primary school which operates as a “feeder” to the secondary school.’

The following text is extracted from the Fair Admissions Campaign’s upcoming report.

It is worth considering whether these data are affected by the propensity of different religious groups to be poor or wealthy. Data compiled by the National Equality Panel for the Government Equality Office, during the passage of the Equality Act 2010, can shed light on this. It shows that Jews in the UK are much wealthier than average, both in terms of income and total wealth; conversely, Muslims are poorer than average, especially in terms of total wealth. This sheds interesting light on the Jewish and Muslim figures, perhaps explaining why Jewish schools are the very worst in terms of socio-economic selection. However, a comparison of the two dominant groups in the UK, namely Christians and those of no religion, shows less difference: Christians are significantly wealthier, perhaps reflecting the fact that they are typically older, but actually earn slightly less per hour on average.

We can control somewhat for these differences by looking at how change of selection within a particular denomination alters the figures. Thanks to the great variety of types of establishment and in the degree to which their oversubscription criteria allow them to select, we can do this for Church of England secondary schools. Here we can see that socio-economic and ethnic selection increases as religious selection increases.

The figures in this table are weighted to school size and exclude grammar schools, University Technical Colleges and Studio Schools. They have been normalised so that the average across all remaining schools is 0.

Figures are weighted by school size and have been normalised so that the average across all secondary schools is 0.

Figures are weighted by school size and have been normalised so that the average across all secondary schools is 0.

Figures are weighted by school size and exclude grammar schools, University Technical Colleges and Studio Schools. The FSM segregation measure shows (in an absolute sense) the average difference between school and area on free school meal eligibility across religiously selective schools, minus the average difference between school and area on free school meal eligibility across schools that do not religiously select.