FSM – past their eat-by date?

Everyone uses free school meals as a gauge of school pupil disadvantage. Is it time to rethink this generally accepted metric? asks Stephen Gorard.

Meals being served up on traysEligibility for free school meals (FSM) has long been used as a measure of potential disadvantage for education research, policy and practice, in England and the other home countries. This is because figures for FSM have been collected by local and central government since at least 1989, and FSM has a simple binary legal definition, of a pupil coming from a family living below the poverty line.

FSM has been used by the Department for Education and Ofsted as context for school performance figures and inspection results, and is the main basis for the distribution of extra funding via the pupil premium. Schools with higher levels of pupils eligible for free school meals tend to have lower overall results in terms of statutory tests and pupil qualifications. So FSM eligibility is currently an important measure. However, it has some problems as a measure, and there are some criticisms in terms of how it is used in practice. My recent paper, Who is eligible for free school meals? Characterising free school meals as a measure of disadvantage in England, looks at the issues raised by using free school meals as a measure of disadvantage in schools.

Records show that under 20 per cent of the school population has received (‘taken’) free lunches at their school in the past 26 years. In order to receive the lunch (and for the school to be reimbursed for it effectively), these pupils must be registered as eligible for FSM, for example, via the school having certification that the pupil’s family is in receipt of Income Support.

The first problem is that 20 per cent is a minority of the school population. So using FSM as a tool for educational analysis means that the system has no information on the relative family incomes of the other 80 per cent. An alternative measure such as family income itself, or a proxy such as occupational or social class or even parental education, is preferable in this respect because it can provide evidence on all pupils. It would also be more flexible as an input for analyses. Instead of computing gaps in thresholds of attainment between FSM and non-FSM pupils, for example, it should be possible to correlate attainment scores with family income.

In addition, FSM inherently works by imposing a threshold, meaning that it may not be fair to treat all FSM-eligible pupils as equal. Some will be just below the threshold and may move in and out of eligibility during their school career. Once the full welfare package is taken into account, they may also be seen to come from families with slightly higher incomes than those just above the threshold. And with the introduction of universal free meals in early primary years, FSM as a measure of disadvantage is in danger of disappearing altogether.

Perhaps the biggest problem with using FSM for analysis, policy or practice is that the figures are incomplete. In any year, around 11 per cent of pupils do not have a yes or no value for the relevant variable in the National Pupil Database (NPD). Around 7 per cent of these are in fee-paying schools, while the other 4 per cent are just missing. The remaining NPD variables show that these missing FSM pupils are more likely to be living in care, to be recent arrivals at their current school, to be entered for very few public examinations, and to have very low levels of recorded attainment at school. They represent a kind of super-deprived group, and the schools that they cluster in are therefore trebly disadvantaged. These schools are taking pupils with generally low attainment, but receive no funding for them via the pupil premium, and no allowance from Ofsted during inspection or calculation of school performance figures.

One argument for FSM is that no other indicator of disadvantage is available nationally over 26 years. Because of this long record, it can be used to track school intake patterns to see whether poorer children have access to the better schools (Gorard & See, 2013). It is universal for pupils at state-funded schools, except for the missing group mentioned above, and therefore involves no sampling error. And if analytical or policy concern is for the poorest in society, its threshold nature does not matter as much. Other measures such as social class are sparse, not routinely collected, and involve considerable judgement to code. They have no clear definition, and their meaning tends to change over time and place, and to vary between age groups and sexes. In surveys, and even official statistics such as those collected by the Higher Education Statistics Agency, they have at least as high a proportion of missing cases as FSM, and that proportion has been growing over recent years. The situation for postcode data as an indicator of multiple deprivation is even worse. More pupils are missing postcodes on NPD than are missing FSM data. And in any case, attributing a modal local population characteristic to every person in an area is a fundamental mistake.

So I predict that FSM will survive for research, policy and practice; not because it is ideal, but because there is nothing better. It is an indicator that gives us an indication of something valuable – but no more than an indication. For example, it might be used to allocate the pupil premium to schools, but then perhaps schools should use that funding on the basis of educational need rather than being restricted to applying it only to activities for FSM-eligible pupils.

Stephen Gorard is Professor of Education and Well-being in the School of Education, Durham University.

References: 
  1. Gorard, S., & See, B. H. (2013). Overcoming disadvantage in education. London: Routledge.

Share this page