It has long been claimed that poverty and disadvantage are important barriers to educational progress and that one of government’s roles is to redress the resulting inequalities in the name of social justice. In education, disadvantage has for many years been identified through eligibility for free school meals.

In the past it has been shown repeatedly that schools with higher levels of free school meals generally do less well than those with lower levels. The early versions of the Autumn Package of pupil performance information - published since 2000 to help headteachers set improvement targets - dwelt extensively on this, categorising schools into groups with similar percentages of pupils eligible for free school meals.

The differences in performance between schools in different groups were so large they led many to claim that league tables were an inappropriate way to hold schools to account for their pupils’ outcomes. Lower levels of performance were only to be expected from schools with higher levels of free school meals.

Two recent Department for Education and Skills publications have challenged these assumptions and shown that this measure of disadvantage may not operate in exactly the expected manner.

The first is an analysis comparing school performance in England from 1998-2003. In 1998, 529 secondary schools were classified as severely disadvantaged, with more than 35 per cent of pupils eligible for free meals. By 2003 this number had reduced to 359, more than 30 per cent.

However, though this reduction suggests that the 359 are the most disadvantaged, it is these schools which have shown the largest absolute improvements in GCSE performance over the five years.

In 1998 around 18 per cent of pupils in these schools achieved five or more A* to C GCSE passes. In 2003 this had risen to more than 30 per cent. By contrast there were hardly any significant changes in performance among the most advantaged schools and only small improvements of a few percentage points in the others.

The gap in performance levels between the most and least advantaged groups has thus narrowed over the five-year period. In 1998 it was around 48 percentage points whereas by 2003 it was 38 percentage points. While this is still substantial, the fact that it has occurred through significant additional improvement in schools with the highest levels of disadvantage suggests a need to revise our views about the role of free school meals in influencing performance.

The second DfES publication, the Statistical First Review, considers a variety of factors affecting pupils’ performance. Of particular interest is the role of ethnicity combined with free school meals.

Pupils eligible for free meals are shown, as a group, to achieve lower levels of performance at each key stage, and the difference between those eligible and not eligible widens across the key stages. But when ethnicity is also taken into account, a different picture begins to emerge.

For example, pupils from Asian backgrounds form the largest of the non-white ethnic groupings in schools. The incidence of free meals among this group is more than twice the level than for their white counterparts.

However, although at KS1, 2 and 3 the overall performance of Asian pupils is lower than their white peers, the impact of free meals on performance is significantly much less.

At KS3, the gap between white pupils’ performances in English (comparing those without and those with eligibility for free meals) is 32 per cent, while the corresponding figure for Asian pupils is under 20 per cent. At every key stage and at GCSE, Asian pupils eligible for free meals outperform their white counterparts to a substantial extent.

Further analysis carried out with school-level data shows that those with relatively high proportions of Asian pupils have performances which are much less affected by the percentage of their pupils on free meals. Similar considerations also apply to other ethnic groups.

A further disturbing factor is that, within the most disadvantaged group of schools - the 359 with more than 35 per cent eligible for free meals - the gap between those doing relatively well at GCSE (the upper quartile of performance) and those not doing well (the lower quartile) has widened substantially, from around 12 percentage points in 1998 to more than 18 percentage points in 2003.

This suggests that either the group of schools falling further behind either faced even more intractable problems than most, or they failed to take full advantage of improvement strategies that had been successful in similar schools.

It might be that higher disadvantage really does equate to lower levels of performance. But there is substantial evidence that disadvantage does not always equate to lower levels of performance. More careful analysis may reveal unexpected possibilities for improvement.

The increased levels of achievement in many disadvantaged schools suggest they may have undertaken a radical reappraisal of pupil potential. Maybe other schools should do the same.

Professor David Jesson is at the centre for performance evaluation and resource management in the department of economics at the University of York