IN RECENT weeks The TES has published several articles suggesting that schools in England and Wales are more socially polarised than ever. But a study I am carrying out with John Fitz has found that the opposite is true. Segregation between schools in terms of pupil characteristics has in fact declined significantly since 1989.

We have created a database containing pupil-intake information for all schools in England. We analysed the data at a national level, and by regions, local education authorities, school districts, and schools. In each case the overall picture is the same. The socio-economic composition of state-funded schools is now more mixed.

This is true for primary and secondary sectors, and for all five indicators used in the study (eligibility for and take-up of free school meals, first language, ethnicity, and special educational need). These findings, for the years 1989-98, confirm the results already published from our study in Wales (first published as The Missing Impact of Marketisation in the British Journal of Sociology of Education 1998 ). In a sense then, this is no longer a subject for debate.

Segregation between schools still exists, of course, and is high in some regions, including the West Midlands, but lower in areas of relative uniformity or poverty, such as inner London, the North-east and Wales.

Our measure gives an estimate of the number of disadvantaged children who would need to change schools within an area for there to be an even spread of disadvantage.

In 1989, 36 per cent of pupils receiving free school meals would have had to change schools to achieve this even spread. But by 1998, the proportion had fallen to 30 per cent.

This surprising finding leads to three interesting questions: what is causing these powerful social changes; why have other commentators not found the same phenomenon; and what are the implications for the comparative performance of schools?

The increasing power of parental choice is one potential explanation. While a market-like situation in the allocation of school places may encourage segregation, the previous system of catchment areas may have been even worse in this respect.

However, as far as we can tell, the polarisation of intakes in any area is unrelated to the local level of school admission appeals by parents. The search for an explanation continues.

It is similarly difficult to explain why these findings are so different from theoretical predictions, and from small-scale studies in England, and even some larger-scale studies from further afield.

Some general points may be made:

* Commentators do not have an agreed definition of what segregation (or stratification or polarisation) actually is, making it hard to recognise in reality.

* Sometimes it is incorrectly assumed that the socio-economic characteristics of a population remain constant, and that therefore the only change is in their distribution between schools.

* Segregation has increased, at least temporarily, in a minority of authorities and schools. Therefore it is possible that some early small-scale studies were unfortunate in their choice of sample, and unable to make valid generalisations from their findings.

* The growth of postcode-based information systems has encouraged some researchers to ignore the pupils’ characteristics, and rely instead on proxy indicators from the 1991 Census arranged into artificial school “catchment” areas. Such data may be insufficient for detailed calculations on segregation. In addition, such studies assume that the 1991 indicators have not changed much.

* Finally, many commentators appear to want segregation to be increasing. It makes a better story for the media, adds credible urgency to research, and increases the chances of government funding for projects to tackle the problem.

Since we already know that educational outcomes are closely linked to pupil characteristics, if schools are becoming more similar over time, one would expect their results to become more similar also.

However, the precise relationship between changes in socio-economic segregation and school outcomes will, like the search for an explanation for the changes themselves, require local and detailed case-studies. This will be the next phase of our project.

Stephen Gorard and John Fitz are based at the School of Social Sciences, Cardiff University. Their research has been funded by the Economic and Social Research Council. “Investigating the determinants of segregation between schools”, their latest academic paper, will be published in a forthcoming issue of Research Papers in Education.