Ides Nicaise

Two OCCAM contributors (John Jerrim and myself) participated in the Advisory Board of UNICEF’s latest Report Card on educational inequalities in rich countries. It was a great experience, not least because of the diversity of perspectives on the topic. There were many underlying conceptual dilemmas, data problems and difficult choices to be made in selecting the indicators and the ensuing ranking of countries. The UNICEF research team finally decided on all these methodological choices, which was fortunate in view of the lack of consensus within the Advisory Board.

Inequality versus unequal opportunities

A key issue in the philosophical and social literature on inequality relates to the level of ‘acceptability’ or ‘desirability’ of inequality. Whereas many may want to see inequalities decline, some would argue that inequalities are ‘natural’ and – to some extent – desirable. The meritocratic view claims that some level of inequality will always exist due to ‘innate’ differences between individuals; it is therefore futile and inefficient for policymakers to try to eradicate inequality completely. Moreover, according to meritocrats, inequalities in educational attainment that emerge through effort should be nurtured, not countered. The only source of inequality that most parties would agree should be tackled is inequality of opportunity by social background. In other words, each generation of children should have an equal opportunity to ‘succeed’ (e.g. in education, in the labour market), irrespective of their background – including the wealth of their parents.

UNICEF decided to adopt a neutral position by ranking countries according to their overall inequality in educational outcomes (measured, for example, as the performance gap between the top and bottom 10% of children in reading). One could obviously argue that there is a strong correlation (across countries) between overall equality and equality of (social) opportunity – and hence, praise countries at the top of the ranking for having ‘equitable’ educational systems. However, there are many other differences across countries (e.g. income distributions, share of migrants) that may explain some countries strong results; it may not be due to the education system per se.

The ‘Unfair Start’ report also provides information about the impact of socio-economic background on reading performance. The measure used for this purpose is the determination coefficient (R²) of a regression of performance on socio-economic background. The latter parameter is also used as the key indicator of inequality in the OECD’s PISA research: rather than the overall degree of educational inequality in a country, it reflects the proportion of that inequality that is explained by social origin. Whether a country has 10% or 30% working class children in its school population does not affect the determination coefficient.  

An Unfair Start’ shows a rather different picture of inequality in a country depending upon which measure is used. Take Hungary, for example. It is a country with ‘average’ levels of inequality is terms of the difference between the top and bottom 10% of readers, but is one of the worse performing countries in terms of the social determination coefficient.

Hidden inequalities

The international large-scale assessments in education such as PISA, TIMSS, PIRLS and ICCS have been criticised for using biased samples which exclude truants and school dropouts or where students with special educational needs are under-represented. Without ignoring these issues, I would like to stress a major difference between PISA and other datasets, namely age sampling (as opposed to grade sampling).

Grade-based samples compare the performance of students of the same grade across countries, as their main aim is to assess curricula. Figure 28 in ‘An Unfair Start’ displays the huge cross-country variation in school arrears due to grade repetition among 15-year olds: from 1.1% in Iceland to 34% in Belgium. Moreover, the risk of grade repetition, just like under-performance, is very unequally distributed across social groups within countries.

It is no surprise, therefore, that country rankings by social inequality in education differ strongly between PISA and other international large-scale assessment studies. PISA correctly reflects the part of inequalities that are shaped through grade repetition, whereas other datasets tend to dissimulate this part of the picture. The position of Belgium and France in PIRLS and PISA provides a nice illustration of the resulting bias: both countries belong to the top 15 in equality according to PIRLS (figure 8 in ‘An Unfair Start’, and to the bottom 15 in PISA (figure 18 in ‘An Unfair Start’; Admittedly, the different rankings may also reflect shifts in social inequality between primary and secondary education.).

All in all, ‘An Unfair Start’ sketches a very comprehensive picture of educational inequalities from early childhood to tertiary education. It also synthesises a lot of scientific research that explains, qualifies and enriches the picture. It is important for users to take all these qualifications on board in debates about the ranking of countries.

About the author(s)