Presentation at AERA 2000
John Fitz and Stephen Gorard
Cardiff University School of Social Sciences
King Edward VII Avenue
In 1988 the Education Reform Act enabled all parents in England and
Wales to express a preference for any school for their child. This, and
other measures such as age-weighted per capita funding created a market-like
situation for schools within which their survival supposedly depended on
a regular supply of students. Theoretical arguments and previous
small-scale empirical studies in the UK have suggested that such a market
would lead to increasing socio-economic segregation between schools. In
contrast, the investigation reported here has found that socio-economic
segregation between schools has actually declined in several respects since
1988. The study uses school-level data relating to free school meals, ethnicity,
first language use and special educational needs, for every school in England
and Wales. All of these indicators at each level of aggregation are in
agreement. Student segregation between schools has decreased over time
in these terms. There is, however, little evidence as yet that this powerful
social movement is related to market forces, and some indications are presented
here that while markets clearly do not cause socio-economic segregation,
they do not seem to be causing the desegregation either.
Background to the study
Reinventing the principles and organisation of the allocation of public services has been a recent feature of public policy in the UK, Australia, New Zealand, the USA and some nations in continental Europe and Scandinavia. Characteristic of these reforms has been the creation of competition between, and within, public sector institutions, the advocacy of choice for newly constituted consumers of public services and the consequent manufacturing of client-provider relationships in the pursuit of efficiency gains in public service provision (Osborne and Gaebler 1993). For the UK, these themes are manifest in the 1988 Education Reform Act. The creation of markets in education, increasing parental choice, advancing the autonomy of educational institutions and the implementation of a per capita funding regime exemplify one aspect of this so-called 'new public management'. For scholars interested in the restructuring effects of market forms of provision of public services generally, and education particularly, the UK has become something of a ‘social laboratory’ by virtue of the extent to which policies promoting ‘competition’ and ‘choice’ have been developed.
The construction of markets in education in the UK involved the creation of interlocking policy initiatives aimed at forcing competition between schools and increasing parental choice. The institutional policy instruments employed in market creation have been discussed fully elsewhere (e.g. in Fitz, Halpin and Power 1993, Whitty et al. 1998). Briefly though, in England and Wales a ‘limited market’1 at the school level was created and advanced by legislation such as the Assisted Places Scheme 1981, the Education Reform Act 1988, and subsequent legislation in 1992 and 1996. The prior nation-wide system of allocation of students to places by Local Education Authorities (LEAs) based on locally defined neighbourhood or catchment areas was intended to be replaced by parents exercising ‘choice’ in a ‘market’ featuring a ‘diversity’ of schools. A decade on, it is interesting to consider what marketisation has yielded in education? It is from that perspective that this paper offers commentary on, and some answers to, a recurrent theme in the American and other literature on ‘choice’ policies - namely the extent to which they are associated with, or cause the socio-economic segregation of school intakes.
The academic response to this and related questions about markets, choice and competition in education has been a substantial corpus of research, drawing on a number of perspectives (see for example Gorard 1997a, Gorard 1999a). Some writers have been primarily concerned to theorise the nature of a system of parental choice of schools (e.g. Le Grand and Bartlett 1993), others have wished to describe and analyse the micro-political process of choosing a new school (e.g. Gewirtz et al. 1995). Some have been concerned to find out which members of the family are involved in the process of choosing (e.g. David et al. 1994), while others have considered the implications for schools (e.g. James and Phillips 1995). Some writers have been avowedly in favour of the strengthening of market forces in the system of educational planning (e.g. Tooley 1994), but perhaps the majority of British research has emphasised the negative consequences of the policy (e.g. Glatter et al. 1997), with some observers apparently taking the results for granted (e.g. Hatcher 1998). Some of the outcomes of this work are that the nature of limited markets is better understood, that the difficulties of choosing for some sections of the community have been emphasised, and that the criteria used to make choices have been well-rehearsed.
In Britain, advocates of increased choice and diversity in education have tended to concentrate on the purported beneficial effects of programmes of choice on school standards. Their argument can be summarised as suggesting that most families are able to make wise educational choices. Therefore 'good' schools will prosper and 'bad' schools will have to either improve or close through lack of students. Although advocates of choice in the USA and elsewhere have suggested that there may be social justice benefits as well (i.e. fairer access to good schools), in Britain not much has been made of this argument. In fact, the reverse has happened. It is the opponents of market forces in education who have promulgated the social justice agenda, and they have emphasised the extent to which the market operates as a more refined mechanism of social selection and exclusion. Such writers generally suggest that allowing parents to choose schools will additionally privilege those who are already privileged, since making a good choice, perhaps making any choice at all, requires families to have financial and/or symbolic resources which are unequally distributed in society. Such resources might include literacy, confidence, taste, knowledge of legislation, leisure time, and private transport. This argument can be very simply summarised as suggesting that poor families will not be able to 'play' the market successfully. Choice will therefore further advantage those families who are already financially or culturally privileged. Reay (1998), for example, claims that the 'market system of education provides the middle-classes with a competitive edge, of which they will increasingly take advantage' (p.1). Gipps (1993) states that 'the concept of market choice allows the articulate middle and educated classes to exert their privilege whilst not appearing to’ (p.35). Theoretical models have, therefore, predicted a growth in social stratification between schools as a result of increased market forces in school placements (Bowe et al. 1994, Bourdieu 1997).
The introduction of programmes of parental choice of school, linked to funding arrangements where finance follows pupils, has been shown by several studies to have implications for school intakes and their social class composition. In general, it has been suggested that the process of choosing requires abilities and resources that are unequally distributed in society, and that, therefore, schools will become more uneven in socio-economic terms. In effect, it is claimed that social segregation between schools is increasing, leading some disadvantaged schools into a 'spiral of decline', and creating a system of winners and losers. The findings of small-scale empirical studies of school choice in urban areas of England have found evidence that supports these predictions (Blair 1994, Gewirtz et al. 1995), and the results from studies of school choice in England, Scotland, Israel and New Zealand have provided confirmation (Ambler 1997, Woods et al. 1997, Willms and Echols 1992, Goldring 1995, Waslander and Thrupp 1997, Lauder et al. 1999, also see Gorard 1999a for fuller discussion of this issue). British research by Levacic and Hardman (1998) suggests that within a system of choice schools those with high levels of students from poor families tend to lose numbers, and therefore budget share, over time. Levacic and Hardman (1999) also suggest that the policy of allowing grant-maintained schools to opt out of LEA control has increased polarisation between institutions. Bagley and Woods (1998) report that the families in their study were avoiding schools on the basis of current student background characteristics such as race, religion and ability. Much academic writing has argued, therefore, that markets in education have a stratifying impact (Conway 1997, Hatcher 1998).
In the USA, where large scale studies and research which is ‘quantitative’ in character are more frequently reported, and where programmes of choice have been more limited in application, this consensus is not so clear (Coons and Sugarman 1978). In another light, markets can be seen as extending to wider participation a privilege that some members of society already have (Witte 1990). Here though ‘participation’ means something slightly different in this different national context. ‘Choice’ programmes in US usually involve, in one form or another, measures intended to enable socially disadvantaged and low income parents to choose private fee-paying schools through some form of fees waiver. Increased ‘choice’ and ‘participation’ was therefore justified by some as an antidote to social stratification, which it was argued, routinely takes place in schools systems where allocation of students to places is based on a catchment or neighbourhood system (Maynard 1975, Spring 1982). Compulsory state education schemes in the USA (as in the UK and elsewhere) have up to now appeared powerless to do anything about (Cookson 1994). In the UK 'participation' has moved beyond this and now focuses on expanding choice for all parents within the state sector, enabling families to select schools outside their neighbourhood and to choose between local authority, grant-maintained or religiously affiliated ‘voluntary schools. In these circumstances choice advocates claimed that the right to choose ‘good’ schools has been extended to those families previously denied them by the zoning effects of admission policies based on ‘neighbourhoods’ or ‘catchment areas’. In both the USA and the UK however, the main emphasis of the reforms and the research that followed it has been on the same group of socially disadvantaged and low income families.
US choice programmes which give greater emphasis to access to fee-paying schools are claimed by some observers to be especially popular with these disadvantaged sections in many communities, such as immigrant, minority, and one-parent families, who have been deserting some large inner-city schools (Levin 1992, Wells 1995). Witte (1998) reports that a voucher scheme in Milwaukee attracted primarily very low-income families, with considerably below-average incomes for local public schools (even below the average of those eligible for free lunches). These families were mainly Black or Hispanic in origin, and often had one parent, suggesting that choice might therefore lead to successful desegregation by income and ethnicity over time.
Nevertheless, even in the USA, the cumulated conclusions about choice and equity are confusing (see Lee et al. 1994). In the USA especially, there have been many small policy experiments, but the resulting information is very limited, and the outcomes, in terms of equity, (as well as standards) are still in dispute (Archbald 1996). Witte (1998) found that the poorer families attracted to the voucher scheme had mothers with higher than average educational backgrounds (cf. the findings of the 'artificially poor' in England: Edwards et al. 1989). This 'controversy exists for a reason' (Witte 1998, p. 248) because different studies have produced different answers to what is apparently the same research question. Sometimes the reasons for these differences could be the nature of the choice programme being studied, and sometimes the nature of the methods used, the sample selected, or the timing. A voucher scheme is not the same as a policy of open enrolment, while a few hundred interviews cannot encapsulate socio-economic movements within a national school system, and a change of policy can produce markedly different effects in the early and more established stages of implementation, for example. It is also becoming clear that national policies properly applied can lead to very different schemes for allocating school places in different regions (White et al. 1999a, 1999b).
The methods used, and the timing of research, into the impact of markets are therefore crucial.2 Mainstream educational research in the UK has been primarily 'qualitative' and small-scale in nature, focusing on the process of choice, and, as in the USA, methodological debates have developed over the validity of some of the most prominent results (e.g. Gewirtz et al. 1995, Ball and Gewirtz 1997, Gorard 1997b, Tooley 1997, Waslander and Thrupp 1997, Gorard and Fitz 1998a, Gorard 1999b, Gorard 1999c, Gibson and Asthana 2000, Gorard 2000a). The problem for smaller-scale studies is that changes of socio-economic segregation between schools may have arisen simply from changes in the number of schools in many authorities since 1988, coupled with changes in the socio-economic characteristics of the families in each area.
For example, a school principal may have witnessed a significant increase in the number of children eligible for free school meals, or student formally identified with ‘special educational needs’ (SEN), in one school. Even if this growth takes into account changes in the number on roll, bearing in mind that schools may have grown in size for any number of reasons, including the closure of adjacent schools, the local rise in indicators of disadvantage is still not evidence of increasing segregation. Even if the school is now taking a larger share of disadvantaged children compared to its neighbouring schools than it used to, this is still not evidence of greater segregation. For example, the school in question may have started the period with less than its 'fair share' of poor students and be simply catching up with its more disadvantaged neighbours, so leading to less segregation in fact. Indeed, our figures suggest that it is likely that the great majority of principals have witnessed a rise in indicators of disadvantage over the last ten years, so increasing segregation can only be said to take place only when all of the changes in number of students and indicators are calculated relative to other schools in the region of analysis. In short, we need to separate the effect of changes arising from demographics shifts from the results of market-based policies (and this is precisely what our method allows, see Gorard 2000b).
What has been missing until now has been a robust comparator, of the kind which would enable researchers to track, over time, the stratifying effects of markets in comparison with the situation pre-1988. By implication, post-1988 markets in education have been compared with the status ante, which has been variously referred to as ‘state monopoly schooling’ (Chubb and Moe 1988) or ‘selection by mortgage’, whereby the housing market effectively determines who is entitled to go where (cf. Hirsch 1997). As we noted earlier but it is worth repeating, these systems were capable of socially stratifying schools. For sure, this has been recognised by some commentators. Waslander and Thrupp (1995), for example, state that 'those endowed with material and cultural capital will simply add to their existing advantages through choice policies' (p. 21). However, there has been no direct comparison of the extent to which social stratification, which undoubtedly occurred under the catchment or neighbourhood area system, has been transformed by the post 1988 impact of market led principles of educational provision. In practice, it would be possible for markets to have a clearly stratifying effect but for them to still lead to less stratification than a pure catchment area system. A limit case of this claim is the relatively rapid desegregation by ethnicity following the replacement of apartheid with more liberal market policies in the universities of South Africa, (MacGregor 1999).
This paper considers evidence about this equality argument in relation to market reforms, and it does so by examining the entire school system in England and Wales, from the year following the passing of legislation introducing market principles nation-wide via the Education Reform Act 1988 to the present day. The subject of the enquiry was therefore not a small-scale or short-term experiment on the US model but an established national system, and this study faces no issues of sampling as it considers all state-funded schools over ten years. In terms of the official indicators used, this is what happened in England and Wales after all parents were given the right to select any school for their child. The data on which they are based are now public, and the method of analysis is transparent and well-publicised. The results are surprising.
The paper sets out to consider two related points. The first is: has
there been increased segregation between schools since 1988, to which the
answer would appear to be 'no' (at least not until very recently). Overall
segregation in terms of poverty, ethnicity and special educational needs
has actually declined in England and Wales since 1988, as it has done in
most regions and most schools within England and Wales. This phenomenon
could be due to the breaking of links between housing and local schools
caused by the introduction of increased parental choice since the 1980s
(cf. Chubb and Moe 1988), or an outcome of the ongoing decline in the proportion
of academically selective schools since the 1970s, or perhaps it is more
likely to be due to a variety of local, regional and national factors.
The second point the paper sets out to consider is: are these changes related
to market forces? The answers to this question would also appear to be
'no'. The paper therefore begins to outline some more complex potential
explanations, and ways in which these are now being researched.
The data used in this study comes from the following sources:
• UK Department for Education and Employment - School-level intake data on all secondary schools in England 1989-1999, and all primary schools in England 1995-1998
• UK Population census 1991 - Local education authority (LEA) data on population density and characteristics
• 41 LEAs - School-level data 1991-1998 including some figures from primary schools, via personal observation of official STATS1/Form7 archives
• The Education Authorities Directory and Annual (1998) - school names and types, details of local government reorganisation
• DES (1990, 1991, 1992), DfE (1993, 1994, 1995), and DfEE (1996, 1997, 1998a, 1998b) - LEA-level data on number and types of schools, figures for independent schools and City Technology Colleges, student ethnicity , exclusions and statements of special educational need (SEN).
School-level data, whatever its limitations and whenever it is available, is clearly preferable to the use of summary or population census data about the population characteristics of each school's theoretical 'catchment' area, as used for example by Gibson and Asthana (1998) and advocated for this purpose by Parsons (1998). The dangers of using 'local area' data instead of school-based data emerges in socio-economic research based on post-code (zip-code) location, such as comparative research into the relative effectiveness of those schools in Wales where the language of instruction is Welsh alongside the majority of schools where the language of instruction is English (see Gorard 1998a, pp. 462-463). There, it was demonstrated that while Welsh language schools exist in the same post-code locations as their more common English language counterparts it does not mean that their intakes are similar in terms of family background. Local population figures cannot, therefore, be used to show school-level segregation. The school level data used in this paper on the other hand consists of student eligibility for, or take-up of, free school meals (see below), students who have statements of special educational needs and student ethnicity, and first language use.
In brief, free school meals are available to school students in very low income families (defined as eligible for state-funded Family Income Support). They are a widely used indicator of poverty in the UK. Overall, about 18% of the student population fall into this group, although they are unevenly distributed geographically and by institution (around 20% are non-white, 8% speak English as a second language, and 2.5% have a statement of special educational needs). There are a few minor problems in the recording and use of these indicators, the solutions to which have been discussed elsewhere (Gorard and Fitz 2000).3 In general, the method of analysis, the number of triangulating indicators, and the sheer scale of the evidence, we argue, overcomes the problems encountered.
In each case, the raw figures per school are converted into a segregation index and a segregation ratio (see Gorard and Fitz 1998b for further details). The index is defined as the proportion of students who would have to change schools for there to be an even spread of disadvantage between schools within the area of analysis (in this paper nationally, regionally, by LEA, and then by school).
More precisely the segregation index is the sum of :
(|dps - (dpa * tps/tpa)|) / 2*dpa
Where dps is the number of disadvantaged pupils in a school
dpa is the number of disadvantaged pupils in the area
tps is the total number of pupils in a school
tpa is the total number of pupils in the area
The sum is multiplied by 100 for convenience, and to make it easy to distinguish from the ratio figures. The segregation ratio, on the other hand, is defined as the proportion of disadvantaged students within a school over or below its 'fair share'.
More precisely it equals: (dps / dps) / (tps / tpa)
The first gives an area summary of the between-school segregation on
a measure of social or educational disadvantage, and would be zero if there
was no segregation. The second is used to trace the 'trajectory' of segregation
for individual schools, and would be one for all schools if there was no
segregation in a particular year. The related indices used here are very
similar to other measures of segregation such as the concentration index,
the disparity ratio, or the Hoover coefficient, as used to assess inequality
of income in a society. In a comparison of several indices of inequality,
Kluge (1998) suggests its use 'if you need to know, either which ratio
of the total wealth or which ratio of the population has to be redistributed
in order to reach complete equality. The Hoover coefficient is the normalized
mean deviation from proportionality. It strictly indicates plain disproportionality
only. Thus, it is the least disputable coefficient among the other coefficients
described here' (p. 4). Both of these indices, and several other measures
of segregation and inequality including the more common dissimilarity index,
have been used with the data from the current project, and all give similar
results (e.g. Gorard 1999d).
Summary of findings
In summarising the findings of our research so far, and presenting the most up-to-date picture of the relationship in England and Wales between markets and segregation, it is only possible for this paper to give illustrative examples of our evidence at national, regional, local or school level, supported by references of more detailed tables and figures published elsewhere.
The national picture
When the analysis is applied to all secondary schools in England (those catering for children aged 11 and above) using the most reliable and complete indicator of disadvantage (eligibility for free school meals), segregation between schools has declined from a high of 36% in 1989 to around 30% by 1997, and now appears to be climbing again (Figure 1). Therefore around one third of students would have to change schools in order for their to be no stratification. The values in Figure 1 are calculated as the total for each school in relation to the relevant national figures for families in poverty. The change in figures for all primary schools are almost identical, as are the more limited figures for families from ethnic minority backgrounds, first language other than English, and students with special educational needs (see Tables I, II). The relevant legislation, enacted in 1988, can only have begun to take effect in schools in 1989. In the following year, there was a slight increase in between-school segregation (see below), followed by a marked decline which appears to have flattened out and settled at a significantly lower level by 1995. This pattern is very different from that predicted by many previous writers, and from that expected by the current authors.
Figure 1 - Change in FSM segregation over time in England (secondary)
Table I - Alternative indicators of segregation in all English secondary
Indicator 1996 Index 1997 Index 1998 Index 1999 Index
SEN statement 29.1% 28.1% 27.6% 27.4%
English ad. lang. 67.8% 67.3% 66.1% 65.7%
Non-white - 14.9% 13.0% 12.5%
Table II - Alternative indicators of segregation in all English
Indicator 1996 Index 1997 Index 1998 Index
SEN statement 39% 38% 36%
English second lang. 70% 68% 68%
Ethnicity non-white - 16% 15%
On the basis of these new findings, one interpretation is that whatever the stratifying effects of market forces and competition may be, the effects of catchment areas or zoning and 'selection by mortgage' may have been a good deal worse (we are not yet in a position to comment on the very recent increase in segregation by poverty).4 What some commentators have appeared to assume, is that the situation was somehow less stratified before 1988 in England and Wales. One might only expect the introduction of schemes of choice to lead to segregation if they started from a relatively well-integrated system. They did not (Hirsch 1997). The picture presented in Figure 1 indicates a considerable degree of social segregation before the impact of policies of parental choice. In other words, prior to 1988 in England and Wales, local patterns of use and preference already led to clear segregation by income and presumably social class. When policies of choice and competition were superimposed on this pattern of use, there was often a noticeable, but temporary, increase in the segregation index for a year or two (nationally in 1990). Segregation then declines and settles at a lower level than before, as the market becomes 'established'. If some sections of society are more aware of changes in policy and more attuned to their new rights as 'consumers' ('alert clients'), one might expect that they would produce a shift towards stratification in the immediate aftermath of choice reforms whatever the long-term outcomes. Put simply, after a change in legislation, some sections of society will be quicker off the mark in utilising any new-found rights, and it is likely that these sections will comprise those who are already more privileged in some sense.
The explanation of this temporary increase may be that the more privileged families at the outset may be more ‘alert’ to their new rights under choice legislation, and more capable of realising their choice in the short term. However, in the longer term choice is also likely to be exercised by the less advantaged sections of any community (cf. Echols et al. 1990, Cookson 1994). In the UK, this participation could have involved, in the decade under scrutiny, selection of a school in a catchment area where the individual in question would have been unable to afford to live, or sending a child to a fee-paying school via the now defunct Assisted Places Scheme. However well, or poorly, these schemes worked out in practice, they do at least extend rights, in principle, to other groups that have always been available to the socio-economically privileged. It is also the case that choice policies may undermine the prevailing source of social stratification in education, namely the catchment-area system of allocating school places.
This interpretation is confirmed to some degree by Department of Education and Employment (DfEE) data on the growth of appeals. These figures represent the number of families who exercised their right to appeal in cases where they failed to achieve their child’s admittance to their first preferred school. Given the growth over four years from 3.5% to 6% of families prepared to appeal, and the indication that this growth is set to continue, and taking into consideration the fact that appeals are only relevant to that minority not obtaining their first choice of school, it can be deduced that the proportion of ‘alert’ or market-aware clients has grown significantly since 1988 (to what extent, even so, these families intersect with those defined as in poverty is an important question to be addressed by the second phase of this project). Those studies made immediately after the introduction of policies to increase school choice, such as the Education Reform Act 1988, therefore, may quite validly produce different findings from those of studies made some time later.
Of course, it could be argued that the national picture of overall socio-economic segregation between schools is too highly aggregated, and that we may be chiefly representing here little more than socio-economic convergence of the residents in different parts of the country whereas markets in schools are local in nature. For this reason the paper presents the results of similar calculations at various levels from national to school district. A regional/local analysis may also be of assistance in identifying the determinants of these powerful changes since, despite national legislation, local variations in patterns could be clues to the existence of differences in the ways in which contested places at school are actually allocated (see White et al. 1999).
The regional picture
The overall pattern of reduced segregation between schools also appears in every region in England and for the increasingly separate principality of Wales (Table III). The values in Table III are calculated as the total for each secondary school in relation to the relevant regional figures for families in poverty (again the figures for the primary sector, and for other indicators show similar declines). The greatest proportionate decreases are in Inner London, Outer London, the South East, and the South West of England. Of these the first three are the most densely populated regions, and therefore perhaps the most likely to show change in a market-like situation. It would be expected that choice of school, or any other change in the policy of allocating school places, would have less impact on patterns of enrolment in rural areas with fewer candidate schools within reasonable travelling distance for most families.
Table III - Change in FSM segregation over time in English Regions
Region 1989 1990 1991 1992 1993 1994 1995 1996 1997
North East 24 24 25 23 24 23 23 22 22
North West 31 32 33 32 31 30 29 29 29
Yorkshire/Humberside 32 33 34 32 30 30 29 29 -
East Midlands 31 32 32 30 29 28 28 28 29
West Midlands 34 35 35 34 32 31 31 32 31
Eastern 30 31 32 29 27 26 27 27 -
Inner London - 18 19 19 16 16 16 15 15
Outer London - 30 30 28 27 25 25 25 25
South East 34 34 34 33 32 30 30 30 30
South West 27 28 27 25 23 23 22 24 23
Wales - - 24 24 23 22 22 22 22
Another observation from Table III, also confirmed at a local level, is that segregation is generally lower in regions like Inner London where poverty is greatest, and in the North East and Wales where the population itself is less variable in terms of class structure, income and other socio-economic indicators (Gorard 1998b). Perhaps this is a clue that segregation depends on the local variability of potential school users as much as their allocation to schools. If so, it may be important that although the total school population of the secondary schools featured in this paper grew from 2,958,268 students in 1989 to 3,216,135 in 1997, the number of students eligible for free school meals grew from 506,066 in 1993 (the first year in which eligibility was returned by all schools) to 590,379 in 1997. At least part of the desegregation could therefore be due to this increase from 17% to 18% of students eligible for free schools meals, which represents a considerable increase in the official assessment of children from families in poverty (and this despite the fact that the method of analysis takes such an increase into account).
This overall pattern of reduced segregation between schools at national and regional levels also appears in a clear majority of the Local Education Authorities in England and Wales (see Gorard and Fitz 2000 for the full local tables). The differences between areas are interesting, reinforcing the importance of considering 'regional' differences in the sociology of education (cf. Rees et al. 1997), and highlighting the danger of attempting to generalise from a small-scale study using only one or two LEAs - a method frequently used in UK research. The current phase of our project involves fieldwork in a diverse sub-sample of these 130 areas to begin to explain the variety observed in local patterns, and thus identify further potential determinants of segregation and desegregation.
In the original six LEAs used for the study in Wales (Gorard and Fitz 1998b), it was possible to use our detailed local knowledge to group the schools into districts within LEAs. Analysed at the district level the same picture of overall desegregation with minor variations was obtained. It is therefore worth noting here that this is further evidence that the level of aggregation used in analysis cannot obscure variations at any other level. It would be impossible for desegregation at LEA level to be due chiefly to cross-border movements between adjacent LEAs in London, for example, since this would show up at the regional level. It would, for the same reason, be impossible for segregation to increase at a sub-LEA level and not show up at a higher level of aggregation unless balanced by a superior and opposite process of desegregation elsewhere. Therefore suggestions, such as those of Gibson and Asthana (1999), that segregation is actually increasing at one level of analysis while decreasing at another are necessarily incomplete until they can explain how such a paradox can be resolved.5
Special education need and segregation
Although there is no intention here to dispute that schools operating in a performance-led market may indeed feel compelled to 'privilege the academic' and thereby marginalise the needs of SEN students and their parents (Bagley and Woods 1998), the fact remains that where figures are available they do also show that the spread of SEN is becoming more even across schools. Indeed, unlike other measures of educational disadvantage such as family poverty (see above), this is not apparently related to any overall increase in the number of statemented children in secondary schools. For example, although in Bolton LEA statements of special educational need increased from a total of 213 in 1994 to 473 in 1997, in Cornwall they declined from 2,515 to 1,497, while in Wrexham they remained steady at 365 to 368. Despite these differences all three exemplar LEAs experienced desegregation over time (see Table IV). A very similar pattern emerges from the sparse data from primary schools that has so far been made available to the researchers. Segregation between schools in terms of families in poverty is decreasing over time (see for example Table V).
Table IV - SEN Segregation index (secondary) 1994-1997
LEA 1994 1995 1996 1997
Bolton 25 25 20 20
Cornwall 22 22 11 12
Wrexham - 17 16 14
Table V - FSM Segregation index (primary schools) 1994-1997
LEA 1994 1995 1996 1997
Hounslow 19 16 17 15
The school-level picture
In trying to decide what is producing the overall desegregation and its local variations, perhaps the first step should be to look at the history of individual schools. Do some LEAs display 'polarisation' whereby most schools are moving towards a more equal share of FSM pupils, but one or two schools are becoming 'sink' institutions taking the surplus students from a disadvantaged background? The first thing to note that this process is not new. Polarisation happened already, even under conditions where LEAs had a greater say in the allocation of students to places. The question then is, does polarisation happen more frequently under market conditions? In fact, this possibility is catered for by the method, and the answer is clearly ‘no’. The figures presented here are calculated as the total for each secondary school in relation to the relevant local figures for families in poverty. The number of schools moving away from the 'ideal' of a proportionate share of children from poor families, and the size of that movement, are directly related to the segregation index for their LEA. For example, the London borough of Haringey has nine schools which collectively show an increase in segregation from 13% to 19%. It should therefore be no surprise to discover that seven of these schools are further from an equal share of FSM, while only two are closer (Table VI). It is also clear, as it was in the detailed analysis of six LEAs from Wales (Gorard and Fitz 1998b), that none of these schools has become anything like a 'sink' school in a spiral of decline over the ten years (at least in terms of segregation, rather than overall numbers). In fact, the most disadvantaged school in 1989 (school b), which was the closest to a sink school in 1989, is the school showing the largest shift towards an even share of FSM pupils.
Table VI - Changes in segregation ratio for each school in Haringey
Haringey (Schools) a b c d e f g h i
1989 1.42 2.18 0.41 0.93 1.01 1.01 0.90 0.85 0.88
1997 1.37 1.47 0.27 0.71 0.73 1.46 1.25 0.84 1.32
In Islington LEA, on the other hand, a near neighbour with the same number of schools as Haringey but an overall drop in segregation from 15% to 5%, only one school (i) moves away from an equal share of FSM (and this is a voluntary-aided one moving towards a decreasing share of disadvantage). The other eight all move closer to parity in terms of segregation (Table VII). As in Haringey, the most disadvantaged school in 1990 (f) shows the largest shift towards an even share of FSM pupils, matched by an equal and opposite shift by the most advantaged school (c) which starts with nearly half of its fair share of FSM students and now takes slightly more than its fair share. In every LEA that has been examined in detail so far a similar picture emerges. As with England and Wales overall, and the regions and LEAs within them, the schools themselves are generally moving towards an even spread of FSM pupils, and there are few 'losers' in this process. As already stated, if polarisation between schools had occurred it would be picked up by this form of analysis at the school level.
Table VII - Changes in segregation ratio for each school in Islington
Islington a b c d e f g h i
1990 0.80 0.69 0.56 0.93 0.81 1.72 1.49 1.17 1.13
1997 0.94 1.12 1.08 0.96 1.10 1.10 0.96 0.98 0.76
The impact of market forces?
One key potential explanation for the socio-economic desegregation of schools since 1988 is the concurrent increase in parental choice. It may be that programmes of choice, by ending 'selection by mortgage', have allowed poor families greater access to desirable schools. It has already been suggested above that the enormous increase in appeals against allocation of a non-choice secondary school could be the result of slowly increasing awareness of consumer rights among poorer families. If this were so, one might expect the pattern of changes in segregation in any area to be related to the proportion of parents prepared to appeal. Appeals are being used here as an indicator of competitiveness in the local market in which they occur, and of the proportion of 'alert' families willing and able to appeal (Willms and Echols 1992). If the changes noted above are related to market forces, then the changes should perhaps be greater in areas where a higher proportion of parents go to appeal. In fact, this is not so. There is no significant relationship (linear or otherwise) between appeals per LEA and levels of desegregation.
It is also noteworthy, that despite the overall lack of relationship between appeals and desegregation, appeals are slightly more common in areas with higher proportions of FSM (correlation coefficient of +0.25). This is contrary to pre-conceptions which would assume that areas with higher levels of poverty would have fewer families prepared to appeal. Another way of reading the same data would be that areas with highly polarised family incomes are generating more appeals in a system of choice based on league table indicators. All of this suggests that there is no positive relationship between the observed changes in segregation and the actual level of competition for places in schools.
Although the linear correlation between the type and diversity of schools in each LEA and the change in segregation is very weak, an interesting approximately 'normal' distribution has been observed in relation to the number of selective schools, and changes in segregation. In general, areas with a large number of grammar schools show little change in segregation, and areas with large changes (in either direction) have no grammar schools. This is strong evidence perhaps, of the stratifying effects of academically selective systems, both before and after 1988. The same pattern is observed in relation to grant-maintained and fee-paying schools as well, with areas having more diversity of provision showing smaller changes in segregation over time. It may be that the elements of selection inherent in a system with more diversity are inhibiting changes in the distribution of intakes that is occurring in other areas (or that the indicators of disadvantage used here generally operate at a level 'below' the users of selective schools). If so, as with the data for appeals, this is another finding suggesting a lack of relationship between markets and desegregation. Market advocates may have expected areas with a greater diversity of schools to show the greatest changes in enrolment. This is clearly not so. In fact, it is the areas primarily populated by notionally equivalent LEA-controlled comprehensives that have shown the greatest changes (in either direction).
Some preliminary suggestions for other determinants have been identified
from the local statistical analyses, and these are summarised here. Areas
showing large changes in segregation over time generally have high population
densities, presumably allowing the possibility of travel to a non-adjacent
school, and they have experienced school closures in the early 1990s, perhaps
encouraging greater homogenisation. Areas showing little or no change in
segregation over time share many of the following characteristics (in addition
to being dissimilar to those above). They use a system of guaranteeing
places in each secondary school to all leavers from specified matched feeder
primary schools (reducing the number of places available at age 11), they
give preference to those with older siblings in the same school (reinforcing
the existing social make-up), and they have greater diversity of school
types including, in some cases, elements of selection by ability.
Having considered changes over 10 years in all of the state-funded schools in England and Wales, in terms of their student intake as assessed by four different measures of socio-economic disadvantage, at different levels of aggregation, it is clear that segregation between schools has not increased as a result of marketisation. On the evidence presented here, it can be argued that schools are now significantly more socially mixed than in 1988 in the sense that the intake to each school is now generally a better reflection of the wider society from which it recruits (at least in terms of the most disadvantaged sections of each community). The changes in school enrolment observed in previous small-scale studies would seem to be partly explicable in terms of socio-economic and demographic changes in society rather than changes in school admissions. There is little evidence here that this ongoing desegregation by poverty, language, need and ethnicity is related to the introduction of the limited market for schools. Perhaps, in time, markets may be found to have much less impact on education systems than is popularly believed (cf. Levin and Riffel 1997). The success or failure of a market policy may depend not so much on the nature of the policy itself as on the nature and scale of socio-economic segregation in the system that it replaces.
If the decrease in segregation is not due to increased competition for school places, as indicated by the growth in appeals, nor to the increasing diversity of provision in developed local markets, then what are its determinants? Some partial explanations have been suggested but all of the results so far are rather unsatisfying. In essence, if an assumption is made that the changes in segregation can be explained further, then further analysis needs to be done, particularly at a lower level of aggregation involving local reorganisations, diversity of schools, variations in admission policies, funding of free travel, handling of appeals, and availability of public transport facilities. It is, of course, also possible to consider the possibility of there being systematic differences within the groups considered here. For example, although indicators of poverty are more evenly spread, it could be that more motivated, or better-educated, or artificially poor parents among these are still being segregated. Similar considerations apply to the majority of the population not encompassed within the measures of socio-economic disadvantage used here. It would be unparsimonious to assume such a position, but unwise not to consider it as a possibility (along with many others). Collection and analysis of the necessary in-depth data for a stratified sub-sample of LEAs, schools and their users is therefore the next key task for this study. It will also be interesting to begin to examine how far, if at all, the changes noted above have impacted on school performance.
This study, so far the largest of a national system of school choice
involving all families, has demonstrated that the introduction of market
forces in education have not led to increased social segregation over time
for the most disadvantaged sections of society. In doing so the study both
challenges the findings of, and to some extent questions the methods of,
several previous small-scale studies. The findings are also more widely
significant and surprising (although not necessarily welcome), for example
in the USA and other countries experimenting with limited choice policies,
and in other areas of public life such as health provision where market
forces have been advocated. Since this is so it is important to stress
two other conclusions. First: there is growing evidence that any amelioration
in segregation is additional to, or perhaps in spite of, the outcomes of
the programme of choice. Second: by questioning the simple case of 'markets
lead to segregation' this paper leads to a more complex picture of
the determinants of social change. This should make the in-depth fieldwork
in the second phase of the project fascinating to do.
1. We use ‘limited market’ to denote the fact that schools cannot expand infinitely nor do parents have untrammelled access for their children to any school of their choice.
2. Consider three examples of research that have been incorrectly suggested to the authors as 'proving' that markets in education lead to increased segregation between schools:
Ambler (1997) ostensibly presents evidence of this in England, but on examination is seen to use data from Halsey, Heath and Ridge (1980) relating to the 1960s, and from Edwards, Fitz and Whitty (1989) which was actually concerned with choice of schools in a small part of the small private sector in the UK. All of the data cited as evidence by Ambler was collected before the Education Reform Act 1988.
Willms and Echols (1992) used a sample size smaller even than the initial Welsh study of six complete LEAs, and used data only from the first two years after relevant legislation became 'operational' (see above comments concerning the starting gun effect). They showed that parents not using designated local schools in Scotland were, in general, better educated and of more elevated social class. What they did not show, nor even attempt to measure, was increasing segregation between schools. Their work only predicts that this would happen, and only as a side issue to their main and undisputed findings about the process of choice, in the same way as Bowe et al. (1994) for instance.
Waslander and Thrupp (1995) claim to have shown that segregation increased after dezoning in New Zealand, but closer examination of their tables reveals they have actually shown that segregation was highest during zoning, and lowest after dezoning in the only year when contested places were allocated by ballot. Interestingly, these figures for segregation no longer appear in the book emanating from the same project, which nevertheless continues to claim evidence that segregation, now termed 'polarization', has increased (Lauder et al. 1999). Their previous assumption, criticised by Gorard and Fitz (1998a), that the school population in their study remained constant over time is reinforced in the book by using ethnicity and SES scores from 11 schools rather than four (with a correspondingly flatter distribution). However, it is still the same four schools for whom the actual changes over time are calculated, and on which their results in the book are predicated. The main impact of this change has therefore simply meant that the reader can no longer check their calculations, since there are no base figures for the four schools alone.
3. In Bristol LEA, for example, the number of cases of SEN recorded on the Form7 in the first year that the question was introduced was of a different order of magnitude to those in any subsequent years, possibly due to a confusion over the distinction between Columns A (statemented) and Column B SEN (assistance requested).
4. The irony, as one commentator notes, is that 'in Britain, the dominant view ... is still that selection of students by ability... is an insidious route back to elitism...., yet selection by residence is acceptable even if it is leading to the concentration of privilege among better-off families living close to more-desired schools' (Hirsch 1997, p. 163)
5. One fairly obvious explanation for the confusion in the case of Gibson
and Asthana (1999) is that their data actually shows the opposite of what
they claim in their report. For example, in their Table I (p.14) they confuse
changes over time in terms of percentage points, with percentages. This
common error has already been discussed as providing confusing and apparently
contradictory accounts in this field (Gorard 1999b). Calculated proportionately,
Gibson and Asthana have actually shown that their schools are becoming
less polarised over time in terms of both FSM and GCSE results.
Ambler, J. (1997) Who benefits from educational choice? Some evidence
from Europe, pp. 353-380 in Cohn, E. (Ed.) Market approaches to education,
Oxford: Elsevier Science
Archbald, D. (1996) Measuring school choice using indicators, Education Policy, 10, 1, 88-101
Bagley, C. and Woods, P. (1998) School choice, markets and special educational needs, Disability and Society, 13, 5, 763-783
Ball, S. and Gewirtz, S. (1997) Is research possible? A rejoinder to Tooley's 'On school choice and social class', British Journal of Sociology of Education, 18, 4
Blair, M. (1994) Black Teachers, Black Students and Education Markets, Cambridge Journal of Education, 24, 2, pp. 277-291
Bowe, R., Ball, S. and Gerwirtz, S. (1994) 'Parental Choice', Consumption and Social Theory: The operation of micro-markets in education, British Journal of Educational Studies, 42, 1, pp. 38-52
Bourdieu, P. (1997) The forms of Capital, in Halsey, A., Lauder, H., Brown, P. and Wells, A. (Eds.) Education Culture, Economy, and Society, Oxford: Oxford University Press
Chubb, J. and Moe, T. (1990) Politics, Markets and America's Schools, Washington: Brookings Institute
Conway, S. (1997) The reproduction of exclusion and disadvantage: symbolic violence and social class inequalities in 'parental choice' of secondary education, Sociological Research On-line, 2, 4
Cookson, P. (1994) School Choice , London: Yale University Press
Coons, J. and Sugarman, S. (1978) Education by choice: The case for family control, Berkeley,: University of California Press
David, M., West, A. and Ribbens, J. (1994) Mother's Intuition? Choosing Secondary Schools, East Sussex: Falmer Press
DES (1990) Statistics of education: schools 1989, London: Department of Education and Science
DES (1991) Statistics of education: schools 1990, London: Department of Education and Science
DES (1992) Statistics of education: schools 1991, London: Department of Education and Science
DfE (1993) Statistics of education: schools 1992, London: Department for Education
DfE (1994) Statistics of education: schools 1993, London: Department for Education
DfE (1995) Statistics of education: schools in England 1994, London: HMSO
DfEE (1996) Statistics of education: schools in England 1995, London: HMSO
DfEE (1997) Statistics of education: schools in England 1996, London: HMSO
DfEE (1998a) Statistics of education: schools in England 1997, London: HMSO
DfEE (1998b) Statistics of education: schools in England 1998, London: HMSO
Echols, F., McPherson, A. & Willms, J. (1990) Parental Choice in Scotland, Journal of Educational Policy, 5, 3, 207-222
Edwards T., Fitz J. and Whitty G. (1989) The State and Private Education: An evaluation of the Assisted Places Scheme, London: Falmer Press
The Education Authorities Directory and Annual (1998) The Education Authorities Directory and Annual , Redhill: The School Government Publishing Company
Fitz, J., Halpin, D. and Power, S. (1993) Education in the market place: Grant maintained schools, London: Kogan Page
Gewirtz, S., Ball, S. and Bowe, R. (1995) Markets, Choice and Equity in Education, Buckingham: Open University Press
Gibson, A. and Asthana, S. (1998) Schools, students and examination results: contextualising school 'performance', British Educational Research Journal, 24, 3, pp. 255-268
Gibson, A. and Asthana, S. (1999) Schools, markets and equity: access to secondary education in England and Wales, presentation at AERA Annual Conference, Montreal
Gibson, A. and Asthana, S. (1999) What's in a number?, Research Papers in Education , 15, 2
Gipps, C. (1993) Policy-making and the use and misuse of evidence, in C. Chitty and B. Simon (Eds.) Education answers back, London: Lawrence and Wishart
Glatter, R., Woods, P. and Bagley, C. (1997) Diversity, differentiation and hierarchy. School choice and parental preferences, in: Glatter, R., Woods, P. and Bagley, C. (Eds.) Choice and Diversity in Schooling. Perspectives and Prospects, London: Routledge
Goldring, E. (1995) Communal Opportunities to Learn, Parental Involvement and School Choice: Israel and the United States, presentation at ESRC/CEPAM Invitation Seminar, Milton Keynes 1995
Gorard, S. (1997a) School Choice in an Established Market, Aldershot: Ashgate
Gorard, S. (1997b) A choice of methods: the methodology of choice, Research in Education, 57, 45-56
Gorard, S. (1998a) Four errors.... and a conspiracy? The effectiveness of schools in Wales, Oxford Review of Education, 24, 4, 459-472
Gorard, S. (1998b) Social movement in undeveloped markets: An apparent contradiction in education policy studies, Educational Review, 50, 3, 249-258
Gorard, S. (1999a) 'Well. That about wraps it up for school choice research': a state of the art review, School Leadership and Management, 19, 1, 25-47
Gorard, S. (1999b) Keeping a sense of proportion: the "politician's error" in analysing school outcomes, British Journal of Educational Studies, 47, 3, 235-246
Gorard, S. (1999c) Questioning the crisis account: polarisation by stealth or by outcomes?, Working Paper 34, Cardiff: School of Social Sciences, ISBN 1872330312
Gorard, S. (1999d) Examining the paradox of achievement gaps, Social Research Update, 26, 1-4
Gorard, S. (2000a) Here we go again: a reply to 'What's in a number?' by Gibson and Asthana, Research Papers in Education , 15, 2
Gorard, S. (2000b) Education and Social Justice, Cardiff: University of Wales Press
Gorard, S. and Fitz, J. (1998a) Under starters orders: The established market, the Cardiff study and the Smithfield project, International Studies in Sociology of Education, 8, 3, 299-314
Gorard, S. and Fitz, J. (1998b) The more things change.... the missing impact of marketisation, British Journal of Sociology of Education, 19, 3, 365-376
Gorard, S. and Fitz, J. (2000) Investigating the determinants of segregation between schools, Research Papers in Education , 15, 2
Halsey A., Heath A. and Ridge J. (1980) Origins and Destinations: Family, class and education in modern Britain, Oxford: Clarendon
Hatcher, R. (1998) Class differentiation in education: rational choices?, British Journal of Sociology of Education, 19, 1
Hirsch, D. (1997) What can Britain learn from abroad?, in: Glatter, R., Woods, P. and Bagley, C. (Eds.) Choice and Diversity in Schooling. Perspectives and Prospects, London: Routledge
James, C. and Phillips, P. (1995), The Practice of Educational Marketing in Schools, Education Management and Administration, 23, 2, 75-88
Kacapyr, E. (1996) Are you middle-class?, American Demographics, www.demographics.com
Kluge, G. (1998) Wealth and people: inequality measures, Entropy and Inequality Measures, ourworld.compuserve.com
Lauder, H., Hughes, D., Watson, S., Waslander, S., Thrupp, M., Strathdee, R., Simiyu, I., Dupuis, A., McGlinn, J. and Hamlin, J. (1999) Trading in futures: Why markets in education don't work, Buckingham: Open University Press
Lee V., Croninger R. and Smith J. (1994) Parental Choice of Schools and Social Stratification in Education: The paradox of Detroit, Educational Evaluation and Policy Analysis, 16, 4, 434-457
Le Grand, J. and Bartlett, W. (1993) Quasi-Markets and Social Policy, Basingstoke: Macmillan
Levacic, R. and Hardman, J. (1998) Competing for resources: the impact of social disadvantage and other factors on English secondary schools' financial performance, Oxford Review of Education, 24, 3, 303-328
Levacic, R. and Hardman, J. (1999) The performance of grant-maintained schools in England: an experiment in school autonomy, Journal of Education Policy (forthcoming)
Levin, B. and Riffel, J. (1997), School System Responses to External Change: implications for school choice, in Glatter, R., Woods, P. and Bagley, C. (Eds.) Choice and Diversity in Schooling. Perspectives and Prospects, London: Routledge
Levin, H. (1992) Market Approaches to Education: Vouchers and school choice, Economics of Education Review, 11, 4, 279-285
MacGregor, K. (1999) Market forces dictating, Times Higher Educational Supplement, 16/4/99, p. 9
Maynard A. (1975) Experiments with Choice in Education, London: Institute of Economic Affairs
Osborne, D. and Gaebler, T., (1993) Reinvention of Government: How the entrepreneurial spirit is transforming the public sector, New York: Plume
Parsons, E. (1998) Parental choice and the social mix of secondary schools, presentation at BERA Annual Conference, Belfast
Reay, D. (1998) Engendering social reproduction: mothers in the educational marketplace, British Journal of Sociology of Education, 19, 2, 195-209
Rees, G., Fevre, R., Furlong, J. and Gorard, S. (1997) History, place and the learning society: Towards a sociology of lifetime learning, Journal of Education Policy, 12, 6, 485-497
Spring J. (1982) Dare Educators Build a New School System?, p. 33, in Manley-Casimir M. (Ed.) Family Choice in Schooling, Toronto: Lexington
Tooley, J. (1994) In defence of markets in education, in Bridges, D. and McLaughlin, T. (Eds.) Education and the Market Place, London: Falmer
Tooley, J. (1997) On school choice and social class, British Journal of Sociology of Education, 18, 2
Waslander, S. and Thrupp, M. (1995) Choice, competition, and segregation: An empirical analysis of a New Zealand secondary school market, 1990-93, Journal of Education Policy, 10, 1., 1-26
Waslander, S. and Thrupp, M. (1997) Choice, competition, and segregation: An empirical analysis of a New Zealand secondary school market, 1990-93, in Halsey, A., Lauder, H., Brown, P. and Wells, A. (Eds.) Education Culture, Economy, and Society, Oxford: Oxford University Press
Wells, A. (1995) African-American students' view of school choice, in: B. Fuller, R. Elmore and G. Orfield (Eds.) School Choice: The cultural logic of families, the political rationality of schools , New York: Teachers College Press
White, P., Fitz, J. and Gorard, S. (1999) The new legislation on school admissions: Working Paper 32, Cardiff: School of Social Sciences, ISBN 1872330290
White, P., Gorard, S. and Fitz, J. (1999) An analysis of local school admission arrangements: Working Paper 35, Cardiff: School of Social Sciences, ISBN 1872330266
Whitty, G., Power, S. and Halpin, D. (1998) Devolution and choice in education, Buckingham: Open University Press
Willms, J. and Echols, F. (1992) Alert and Inert Clients: The Scottish experience of parental choice of schools, Economics of Education Review, 11, 4, pp. 339-350
Witte J. (1990) Introduction, p. 1 in Clune W. and Witte J. (Eds.) Choice and Control in American Education, Volume 1: The theory of choice and control in education, London: Falmer Press
Witte, J. (1998) The Milwaukee voucher experiment, Educational Evaluation and Policy Analysis, 20, 4, 229-251
Woods, P., Bagley, C. and Glatter, R. (1997) The public market in England: School responsiveness in a competitive climate. Occasional Research Paper 1, School of Education: Open University