Jump to ContentJump to Main Navigation
Unequal AttainmentsEthnic educational inequalities in ten Western countries$

Anthony Heath and Yaël Brinbaum

Print publication date: 2014

Print ISBN-13: 9780197265741

Published to University Press Scholarship Online: May 2015

DOI: 10.5871/bacad/9780197265741.001.0001

Show Summary Details
Page of
date: 26 May 2020

Ethnic Differences in Early School-leaving

Ethnic Differences in Early School-leaving

(p.95) 4 Ethnic Differences in Early School-leaving
Unequal Attainments

Jan O. Jonsson

Elina Kilpi-Jakonen

Frida Rudolphi

British Academy

Abstract and Keywords

In this chapter we study the differences between ethnic groups in early school-leaving in six of the countries: England and Wales, Finland, France, the Netherlands, Sweden, and the USA. We find sizeable gross differences in early school-leaving between the majority group and some, but far from all, ethnic minority groups, mainly to the disadvantage of minorities. Most differences disappear when we compare those with similar social origins, however, and once we also control for educational performance (grades, or test results) a substantially important disadvantage remains for only one minority group out of the 42 we study. In particular, except for those from the Middle East, Asian minority groups have very high continuation rates into upper secondary education. There is little evidence to suggest that there is any intrinsic or cultural ethnic disadvantage that discourages minority students from staying on in school, or that discrimination or unfair treatment pushes them out.

Keywords:   early school-leaving, drop-out, educational inequality, ethnic minorities, cultural theories, choice theories


FOLLOWING THE GENERAL CONCLUSION FROM CHAPTER 3—that children of immigrants on average perform at a slightly lower level than pupils of the majority population—we expect their educational careers to be shorter. However, we also learned that children of immigrants do not form a homogeneous group, so in following pupils' educational careers we are well advised to continue distinguishing several ethnic minority categories. This chapter will, like the three that follow, analyse ethnic minority students' educational careers, seen as a sequence of educational transitions. We concentrate on ethnic minority students whose parents immigrated, but who themselves were either born in the destination country or arrived there before starting school; that is, the groups that are commonly called the ‘second generation’ and ‘1.75 generation’, respectively (Rumbaut 2004).

In many countries, the transition to (upper) secondary school, at around age 15–16, is a watershed. Most of those who do not make the transition to upper secondary school are likely to have reached the end of their school career and, if so, they will in many cases find it hard to compete for jobs in a strained youth labour market; and for those who do find jobs, career opportunities are limited without higher education. While the large majority of those who do not make the transition have the formal right to leave school, they can, for practical purposes, be regarded as early school-leavers (we sometimes refer to this phenomenon as non-continuation or drop out, though these concepts have slightly different meanings across countries). We study modern Western countries, predominantly with mass education systems, and the sheer size of a student cohort that the early school-leavers represent suggests that in (p.96) several countries they constitute a conspicuous minority. We concentrate on countries for which we have suitable data, and for which early school-leaving can be defined. Because there are few datasets that are immediately comparable when it comes to studying ethnic differences in making educational transitions, the number of countries—six in total—is substantially smaller than in the analysis in Chapter 3.

It is commonly believed that early school-leaving causes great problems for individuals and for society. Previous research unanimously shows correlations between leaving school early and ending up in crime, substance abuse and other indicators of problematic living conditions (e.g. Townsend et al. 2007). The most common and consistent finding is that early school-leavers end up jobless, something found in all countries in this study (Payne et al. 1996; Järvinen & Vanttaja 2001; Rumberger & Lamb 2003; OECD 2005; Murray & Sundin 2008; Allen & Meng 2010). To what extent these results also indicate a causal effect is not always clear but the causal story behind labour market problems is certainly convincing. While leaving school early does not necessarily pose a problem for individuals, group differences in drop-out rates are nevertheless likely to lead to group differences in labour market success. Our descriptive picture of early school-leaving rates thus speaks to the issue of ethnic minority marginalisation—if a substantial proportion of the children of immigrants end up with no educational qualifications beyond the compulsory years, their chances of getting into the labour market are diminished, and they subsequently risk poverty and social exclusion.

Much research has been devoted to describing the antecedents of early school-leaving. Low motivation, low cognitive ability and poor previous performance in school, disadvantaged socioeconomic background, and single-parenthood are some of the characteristics that have been singled out (e.g. Rumberger 1987; Eckstein & Wolpin 1999; Alexander et al. 2001; Traag & Van der Velden 2011). Together, these studies suggest that students who doubt their own capacity to manage further education, and/or who lack the necessary resources and external support, run the risk of leaving school early. In addition, though more difficult to study empirically, cultural characteristics (in a broad sense) are likely to be to the disadvantage of ethnic minorities— the distance from the majority culture can be assumed to impact negatively on minorities' achievement, to spur a feeling that schooling is not ‘for them’, and to lead teachers to evaluate or grade them less favourably.

Previous research has shown that aspirations and belief in their own capacity are high in several, but not all, ethnic minority groups, while resources are generally to the disadvantage of ethnic minority students (see reviews by Kao & Thompson 2003; Heath et al. 2008). Lack of support is naturally connected with parental resources and ambitions, but not confined to such. For example, some studies have pointed to the impact of school characteristics, (p.97) such as their organisation, size and curriculum (Bryk & Yeow 1989; Lee & Burkam 2003; Marks 2005), though the results from these studies are often difficult to interpret because unobserved characteristics of families are often clustered in schools due to segregation.2 Nevertheless, it is highly likely that schools can influence the level of early school-leaving, perhaps via allocating resources to students who struggle and paying attention to those who show dwindling interest in school. Therefore, if schools with a high proportion of ethnic minority students have fewer resources or less committed teachers, this may increase overall drop-out rates for ethnic minority students.

The issue of early school-leaving certainly carries a potential story of enduring misery for ethnic minority children, or at least transmission of adversity. Parents who arrive in the destination country with incompatible skills and credentials that are unknown to employers, and who may also experience prejudice and discrimination, risk ending up in marginalised positions and poverty; and it is quite possible that they transmit such acquired disadvantages to their children who therefore leave school early, only to face similar problems with jobs and income. In addition, direct or indirect ethnic discrimination in school (by teachers, peers or through the curriculum) may further diminish the chances of ethnic minority children to stay on in school. In the worst-case scenario, together this will lead to a sizeable permanent ethnic minority underclass.

Pessimistic views on minorities' structural integration, pointing both to unequal access to resources and to unequal treatment, have indeed flourished in the last decades. The alternative, optimistic, view holds that all minorities get assimilated in the long run (cf. Alba & Nee 2003). Some immigrant groups may also have a head start. There are at least two reasons for assuming this. First, this situation could occur if immigrants are positively selected; for instance, on the grounds that they have more drive and ambition, better qualifications, or higher unobserved skills and abilities than their compatriots who did not emigrate (e.g. Feliciano 2005), and perhaps even in relation to the majority population. In Chapter 2 of this volume, this finding is corroborated for many immigrant groups, albeit to different extents, while only a few—such as labour market immigrants from neighbouring countries—are negatively selected. The second reason for optimism is that, probably partly because of positive selection, some ethnic minority groups hold high aspirations that can act as a springboard for upward mobility among minority children. This has been (p.98) shown to be true especially for Asian groups (Goyette & Xie 1999; Rothon 2007; Jonsson & Rudolphi 2011).

Should we believe in the pessimistic view of enduring misery among ethnic minorities, or should we concur with the optimists who emphasise the likelihood of structural integration? The analysis of early school-leaving is one way of shedding light on this crucial issue. In this chapter we ask: do ethnic minority disadvantages prevail also among those who grew up in the destination society? Given the heterogeneity among ethnic minorities, we also ask the follow-up question: do some groups become structurally integrated whereas others end up in perpetual disadvantage? We also profit from the cross-country perspective of this volume by asking whether there are not only origin-country variations in early school-leaving, but also destination-country differences: does it matter for children of immigrants which Western country their parents migrated to?

Understanding ethnic minority advantage and disadvantage

Our first aim in this chapter is to give a comprehensive picture of early school-leaving across both destination and origin countries. Our second aim is more ambitious: namely, to try to understand ethnic differences in early school-leaving. There are many potential explanations and we can only address them superficially here as our data are not detailed enough to discriminate between competing theories. Nevertheless, we will discuss our findings in the light of choice theories and cultural theories, and also comment on the role of institutional characteristics for ethnic patterns in early school-leaving—though for a more thorough analysis of this, see Chapter 9.

Choice theories

According to choice theories, we would predict that differences in early school-leaving primarily depend on expected costs and benefits with continuing in school, as well as the expected probability of success at the next level of education (e.g. Erikson & Jonsson 1996). Though we cannot claim to test such theories, we will compare estimated non-continuation rates before and after controlling first for social origin (which will pick up much of the differences in expected costs), and then after additionally controlling for pupils' performance, as reflected in grades or test results (which will pick up most of the differences in expected probability of success). Arguably, if ethnic differences disappear after these controls—that is, if there are no remaining ethnic net differences in early school-leaving—it suggests that gross (actual) ethnic differences are intelligible using standard models of (p.99) choice. One could imagine also that children of ethnic minority origin stay on in school if they have lower opportunity costs—this could be the case because their chances of getting a job are comparatively small, perhaps due to a lack of labour market relevant networks or discrimination in the unskilled labour market (of which there is some evidence: Riach & Rich 2002). It is also possible that the benefits of staying on in school are higher for minority students—this would be the case if they would expect to have use for qualifications also in their countries of origin, or if their aspirations were aligned with their parents' socioeconomic status before immigration rather than after, so their goal would be to ‘catch up’ in status (Van de Werfhorst & Van Tubergen 2007).

Cultural theories

When studying ethnic minority differences, cultural explanations may have some force—after all, especially non-European immigrants come from countries where customs, religion and life-styles may be very different from those in modern, rich Western countries. Of course, the question is whether such differences are also of relevance for educational careers. Heath & Brinbaum (2007) point to a ‘cultural dissonance’ that makes those of immigrant origin less able to navigate in the destination country. They take the cultural dimension to include also information, in which case immigrant parents' lack of knowledge about the destination society's school system (e.g. Kristen 2005) may hamper their children's chances to succeed. Maybe they are not able to help their children with their homework at upper secondary level if they speak the language poorly, are used to a different curriculum, or are unaware of how to behave strategically with the school work. They are also likely to lack networks that could counteract such an unfavourable situation, partly because they are less likely to know well-educated natives who possess the best possible information, partly because their children are likely to go to schools where their classmates will also be of ethnic minority origin.3

There could be other accounts of cultural explanations: for example, ‘lighter’ ones pointing to differences in tastes and life-styles (e.g. Bourdieu 1984) that would make ethnic minority children (just like working-class children) feel (p.100) less ‘at home’ in school. A stronger version can be found in studies by Ogbu on black students in the USA (e.g. Fordham & Ogbu 1986). They suggest that minority students may ‘actively’ leave school as a reaction to what they see as an alien, if not hostile, institution (an argument that is akin to explanations of why working-class children leave school early; e.g. Willis 1977). If the norm in school is the well-behaved pupil who does her homework, answers questions when asked and keeps quiet otherwise, and who embraces ‘white’ middle-class values and culture, minority students from disadvantaged circumstances may react by rejecting school altogether and leave at the first possible occasion (cf. Smyth & Hattam 2004). Especially in the USA, fears have been aired that immigrant youth, because of their marginal social status and unfavourable residential location, may be attracted to (black) sub-cultures in which the rejection of mainstream achievement goals is relatively common (e.g. Portes & Zhou 1993). If it is true that a sizeable group of minority students are estranged from school we would expect them to both perform worse and also have a higher risk of dropping out, at given levels of performance.

It would seem a corollary of cultural explanations (although they differ in their content) that ethnic minority students who come from another cultural sphere are disadvantaged in general, and that their disadvantage increases with the ‘cultural distance’—in our case, that early school-leaving, controlling for social origin and grades, would be systematically related to ethnic groups. If discrimination is invoked, we would expect especially those of visible minority origin (e.g. with a skin colour that differs from the majority population) to drop out.

Institutional theories

An important question for policy-makers is whether there are institutional differences in the risks of dropping out of school early, particularly whether the organisation of education or the labour market plays a role. Our countries represent different school systems, and therefore provide an opportunity to address this question although, given the small number of nations, these comparisons will necessarily be tentative. One hypothesis is that early tracked systems (such as the Dutch) will promote early school-leaving because a substantial share of pupils have vocational skills already at the first point where they could leave school, and lower-track graduates may therefore be able to compete for a job without further education (cf. Shavit & Müller 1998). Moreover, lower-class and minority students may have difficulty in making the transition from the graduation in the lowest track to continuing vocational education through cultural and choice processes mentioned above; or they may be discriminated against, in that they are placed in lower tracks at given levels of academic aptitude. Such mechanisms would lead us to expect higher early school-leaving rates among minorities.

(p.101) Also, if minority children are over-represented at the bottom of the performance distribution where recruitment to lower tracks is common, we may find negative effects for second-generation immigrants (and we could see evidence that some minority groups perform quite poorly in Chapter 3). On the other hand, if ethnic minority children have greater difficulty finding a job, perhaps because they lack labour market related networks, their opportunity costs with continuing in school will be lower. The chances of finding a job without having any (post-compulsory-school) qualifications is in turn higher in deregulated labour markets, characterising primarily England4 and the USA among our countries. In such labour markets, youth wages are more fungible downwards, providing greater chances of getting low-skill entry jobs, whereas the more regulated labour markets (characterising our other countries) present higher barriers to such entry jobs and instead lower opportunity costs which gives incentives to stay on in school (cf. Breen 2005). This situation may encourage minority children in England and the USA to leave school early, though it may also facilitate their integration into the labour market. Thus, it is possible that we find higher propensities for ethnic minorities of leaving school early both in highly tracked educational systems (in our study, the Netherlands) and in the much more inclusive systems of the USA and England, though for quite different reasons.

A second hypothesis is that there is a difference in ethnic opportunities between school systems that are choice-driven and those that are selective, i.e. where the transition to the next level is based on previous performance. Because (some groups of) ethnic minority students tend to have poorer school performance whereas many ethnic groups maintain high educational aspirations, a school system based on student choice may be to the advantage of ethnic minorities, which is supported in a study of England and Sweden (Jackson et al. 2012). Ethnic minority disadvantage in early school-leaving may, according to this hypothesis, be more pronounced in the Netherlands and France, where early grades and test results are important, than in the largely choice-driven systems of Sweden, Finland, England and the USA. However, there is an opposing hypothesis that is particularly plausible if majority-minority differences in performance partly are based on (overt or covert) discrimination of minority students. This rival hypothesis maintains that transparent selection processes such as the usage of standardised tests can have a diminishing effect on minority disadvantages because tests incur a more objective basis for track placement (Van de Werfhorst & Mijs 2010).

It has to be remembered that there are many other types of institutional differences across countries that could have an impact on early school-leaving, (p.102) such as resource distributions between schools, so the hypotheses above are mainly exploratory.

Early school-leaving in different school systems5

How do we define early school-leaving? Inevitably, this will have to differ somewhat across educational systems and datasets. The core idea is to distinguish those who progress in the educational system from those who either leave it altogether, attend only partly or intermittently, or who stay in school without progressing because they did not qualify to continue. It could be argued that the latter group, which overall is rather small, have not taken a decision to leave and that including them among school-leavers will therefore ‘artificially’ boost the association between previous performance and school-leaving. While we have to keep this in mind, we still prefer to define those who repeat grades or take remedial education as school-leavers because failing at this level in many cases signals an anticipatory decision to quit school, and because they in fact do not advance in the school system: the hurdle we are interested in is the transition, and they do not make this transition.

For students who continue, another question is how to classify those who take vocational schooling. Our choice here is to include those who go on to take formal full-time vocational studies in the group that make the transition. In general, those with vocational qualifications have considerably better chances in the labour market than those who leave school without (e.g. Shavit & Müller 1998; Müller & Gangl 2003), and in some of our countries vocational studies qualify for higher education as well.6 Because the distinction between vocational and academic (or general) studies is crucial for the transition to tertiary education, ethnic differences in this choice will be studied separately in Chapter 5.

With these principles in mind, we define early school-leaving in Sweden and Finland as not making the transition to any (regular) upper secondary education. In these countries, students change schools, teachers and curriculum, and the choice of programme (or track) is essential for the future occupational career and is therefore surrounded by information, including counseling from their comprehensive school. In this process the final grades from comprehensive school are also of importance, not primarily for making the transition (rather than leaving) but for the opportunity of choosing what (p.103) track to follow (cf. Jonsson & Rudolphi 2011; Kilpi-Jakonen 2011). In both countries, the major dividing line for this choice is the distinction between vocational upper secondary tracks and general (academic) tracks, to be analysed in Chapter 5.7

In England, continuing into the sixth form sometimes, but not always, means changing schools, and the transition is not emphasised to the same extent as in the Nordic countries. But in England too, grades, typically in the form of GCSE scores, are given just before the decision to continue in sixth form or not, and may be decisive for enrolment (Payne et al. 1996).8 Also in France, the choice of continuing or leaving school is a tangible decision, although for French students the choice element is somewhat attenuated because the school makes stronger recommendations on the basis of previous performance (as measured by the brevet des collèges at the end of lower secondary school) (Brinbaum & Cebolla-Boado 2007). France shares with the Nordic countries the importance of tracks at upper secondary level (general, technological or vocational baccalauréat), but unlike those educational systems (and England) there is also tracking before this transition; a tracking that does correlate with the choices at the onset of the baccalauréat (Brinbaum & Kieffer 2009).

The situation is different again in the Netherlands, which maintains a system with early selection, so drop out does not occur at the first transition in the educational system. Instead, students are directed into one of three school types at around age 12: preparatory vocational (VMBO), general (HAVO), or academic (VWO) (Luijkx & de Heus 2008; see Chapter 2 for descriptions). Each of these three school types prepares for a different form of continuing education. We model continuation after compulsory schooling as being in education in the nominally fifth year of secondary schooling. This could either be the first year of MBO after VMBO, the final year in HAVO, or the fifth year in VWO. Therefore, for VMBO students, being enrolled in the fifth year of schooling really involves a transition into a subsequent school type often in a different school organisation (entering either the upper secondary vocational MBO (secondary vocational education) school type or the fourth (p.104)

Table 4.1. School continuation rates at the transition to (upper) secondary education, at approximately age 15/16, by country; description of data.


Age at transition

Birth cohort

Measure of performance (age at measure)

Early school leaving rates (total %)




Average grade teacher assigned (16)





Reading test score 10th grade (16)





Grade sum teacher assigned (16)





Standardised test scores (15–16)





GCSE test scores (15–16)





Test scores (12)


year of HAVO), whereas for HAVO and VWO students it just means still being enrolled in the programme they entered before that year.

In the USA, early school-leaving is defined as dropping out of high school. The US educational system is much more undifferentiated than the European systems in ages up to 18, and pupils are expected to graduate from high school. We consider all those who never complete upper secondary education as well as those who drop out of high school and instead take the GED (General Educational Development tests) as early school-leavers (see Ream & Rumberger 2008).

Table 4.1 provides some basic information on the countries we investigated, including overall non-continuation rates. For all countries, we have data of recent birth cohorts—born in the mid-1980s—and we have performance indicators in all datasets (grades, standardised tests). Early school-leaving rates form an almost bimodal distribution in our sample of countries, with Finland, the USA, France and Sweden having rates around or lower than 10 per cent, whereas the Netherlands and England saw 21 per cent of our cohorts leave school early. The low rates of upper secondary education for England is verified in official statistics (OECD 2009: Chart A1.2), and the relatively high rates of school-leaving for the Netherlands are not surprising given the focus on vocational studies with strong connection to the labour market.

Actual drop-out rates across ethnic majority and minority groups

We begin by asking what the actual non-continuation rates are in the different ethnic groups. As the average rate of early school-leaving differs across countries (as shown in Table 4.1), we expect rates also of sub-populations to differ across countries. Therefore, in Figure 4.1, we display school-leaving proportions for different ethnic groups country-wise. (p.105)

Ethnic Differences in Early School-leaving

Figure 4.1. Proportion of early school-leavers in different ethnic groups across the six destination countries (given as percentages).

(p.106) It is clear that most ethnic minority groups have a higher risk of leaving school early compared to those who belong to the majority population. Several groups, such as the minority groups in the Netherlands and the Turks and the Chileans in Sweden, have school-leaving rates above 20 per cent, the Moroccans in the Netherlands even above 40 per cent. However, it is also clear that even among the English and Dutch majority populations, the non-continuation rates are quite high. The relative differences between the majority and minority groups are greatest in the Netherlands, Finland and Sweden. In England, the white majority actually exhibits a higher school-leaving rate than any of the minority groups (cf. also Middleton et al. 2005).9

An important lesson to be learned from Figure 4.1 is that even if, in several of our countries, minority groups are at a disadvantage in relation to the majority population when it comes to early school-leaving, differences are as great within the minority population. In particular, pupils from many East Asian groups tend to stay on in school. In contrast, groups from the Western part of Asia, such as the Turks, generally have high rates of early school-leaving; this also goes for those of African background.

The importance of social origin

We will first analyse to what extent ethnic differences in early school-leaving, revealed in Figure 4.1, can be accounted for by differences in social origin characteristics. Immigrant parents quite often face financial constraints as a consequence of their weaker connection with the labour market and lower occupational attainment (evident for many countries; see Heath & Cheung 2007). In addition, they are often poorly educated themselves, particularly when they come from Third World countries. Because characteristics such as parental education and socioeconomic status are associated with children's educational attainment (a common finding that also holds for ethnic minorities; see Heath & Brinbaum 2007), we expect that controlling for social origin will reduce differences among ethnic groups.

Next, we fit multivariate probit models to our data, for each country, in order to ‘net out’ the differences in social origin composition across ethnic groups. In these models, we use the majority population as a reference point, and calculate the difference to this group for each of our minority groups. In (p.107) our presentation these are expressed as percentage point differences with positive signs meaning higher drop-out rates.10 The results are shown in Figures 4.24.5, which contrast the gross differences between groups (the Model 1 bars) with the ‘net’ one, controlling for social origin and family status (the Model 2 bars).11 We have chosen to show the parameters as they are, without concern about significance levels, because we are interested in the overarching pattern rather than individual estimates. However, all large estimates are statistically significant at the 5 per cent level.

Ethnic Differences in Early School-leaving

Figure 4.2. Average marginal effects (AMEs) from probit analyses of early school-leaving, contrasting majority population with ethnic minorities. Model 1 is a gross model, Model 2 controls for social origin and Model 3 controls for grades (equivalent). Part I: European-origin groups by region and destination country.


Ethnic Differences in Early School-leaving

Figure 4.3. Average marginal effects (AMEs) from probit analyses of early school-leaving, contrasting majority population with ethnic minorities. Model 1 is a gross model, Model 2 controls for social origin and Model 3 controls for grades (equivalent). Part II: Asian-origin groups by region and destination country.

As we are now comparing school-leaving rates within destination countries, we shift focus from the countries of destination to the countries of origin. Figures 4.24.5 thus make a rather crude geographical division of the countries from which our respondents' parents originate, countries that in turn are subsumed under larger geographical areas, namely European countries (where Eastern and Southern Europe are distinguished)12 in Figure 4.2; Asian countries (dividing Asia, rather inexactly, into West, East and Southeast) in Figure 4.3; African (where again a rough division is made between North African and Sub-Saharan) countries in Figure 4.4; and Latin American and Caribbean countries in Figure 4.5.

It is quite clear in Figures 4.24.5 that controlling for social origin does away with most of the ethnic minority disadvantages that we registered in Figure 4.1 (compare the size of the Model 1 bars with the Model 2 ones). For example, the high non-continuation rates for children whose parents come from the Western part of Asia (the Middle East and Turkey in particular) are to a large extent—though not entirely—due to these parents' low level of education (p.109)

Ethnic Differences in Early School-leaving

Figure 4.4. Average narginal effects (AMEs) from probit analyses of early school-leaving, contrasting majority population with ethnic minorities. Model 1 is a gross model, Model 2 controls for social origin and Model 3 controls for grades (equivalent). Part III: African-origin groups by region and destination country.

and their lower socioeconomic attainments. The same goes for children of most European and African immigrants. For a number of these groups the initial differences are reduced by half or more when controlling for social origin. We should be aware, however, that controlling for occupational attainment (which also includes an indicator of joblessness) wipes out some of the ethnic minority disadvantage that is in fact a consequence of their minority status. That is, the inherent counterfactual of the statistical model that controls for social origin asks us to imagine a child of an immigrant parent who has got the same education, labour market status and social class position as a parent in the native group. However, at least some of the immigrant groups—notably the visible minorities—are at such a disadvantage in destination societies' labour markets that their comparison group among native parents is strongly negatively selected (i.e. they are likely to possess some unobserved characteristics that suppress the educational achievements of their offspring).13 On the other hand, it is possible that for some groups our controls are not (p.110)
Ethnic Differences in Early School-leaving

Figure 4.5. Average marginal effects (AMEs) from probit analyses of early school-leaving, contrasting majority population with ethnic minorities. Model 1 is a gross model, Model 2 controls for social origin and Model 3 controls for grades (equivalent).

Part IV: Latin American and Caribbean origin groups by region and destination country.

good enough: this could be the case, especially, for parents' education—we know that at least for some origin countries (such as those in the Third World) the lowest educational category may conceal illiterate parents.

Even though the inclusion of social origin indicators substantially reduces the disadvantage that ethnic minority groups experience, in some cases there is a remaining difference. For example, children of African, Turkish and ex-Yugoslav origin suffer from relatively high non-continuation rates. However, in absolute terms these additional disadvantages seldom exceed five percentage points—a size which, on the other hand, is not altogether trivial in Finland and Sweden where leaving school early is uncommon. The group with the most unfavourable situation, once we have controlled for social origin, is the Surinamese/Antilles in the Netherlands, who suffer a disadvantage of ten percentage points greater than the majority group, controlling for social origin.

It is worth stressing that the net differences in early school-leaving rates are unsystematic across both origin and destination countries (with the exception of England). In fact, there are several examples of groups who have higher propensities to stay on than the majority groups: Africans in France, Iranians and Southeast Asians in Sweden, and all minority groups in England (p.111) do better than the majority at this early hurdle in the educational career. All in all, there is little to suggest that ethnic minority status itself bestows such great disadvantages upon children of immigrants in our countries—the gross differences that we registered in Figure 4.1 appear to a large extent to be a consequence of the educational and job-related disadvantages that immigrant parents suffer. Thus, it is difficult to discern any systematic pattern suggesting an impact of ‘cultural dissonance’ on early school-leaving among children who grew up in the destination country, unless such cultural effects were captured by our measures of social origin.

The importance of school performance

The remaining ethnic differences that we have seen in Figures 4.24.5, in the models that accounted for social origin, escape easy interpretation. However, a possibility is that they are a result of different school performance across ethnic groups—documented in Chapter 3. As we know that both the majority population and ethnic minority groups are largely advised by their grades, or test results, in their educational decisions, we may believe—in common with typical choice theories—that if we could account for grades, all ethnic differences would vanish. Because we are in the fortunate situation that the data in all our countries contain information on grades or test results, we run a final model that includes these variables. The results are shown by the Model 3 bars in Figures 4.24.5. When we control for grades much of the remaining ethnic differences do in fact disappear entirely or are reduced to trivial sizes. This is especially true of some of the African minorities, and of children to Turkish and other West Asian immigrants who lagged behind in their school achievement; on the other hand, some East and Southern Asian groups now appear to have somewhat less of an advantage over the majority group in their continuation propensities (because they excel in their school performance and/or abilities).

Conclusions and discussion

We have studied early school-leaving among immigrants' children, who themselves were born in the destination country or arrived before school start. Even though several minority groups have higher rates of early school-leaving than the majority population, most of this could be accounted for by the social composition of minority and majority groups—because many of the immigrant (p.112) groups have low education and unfavourable social class positions, they are probably less able to support their children in their school achievements and educational choices.

When we furthermore controlled for previous school performance (by grades or test scores) almost all of the remaining differences between ethnic groups disappeared. This means that once we compare students with similar resources and probability of success at upper secondary level of education, the choice to leave school early does not differ much between ethnic groups (and is equally often to the advantage of minorities). This lends some credibility to choice theories suggesting that children in ethnic minority groups— just like majority-group students—leave school simply as a response to the lack of resources in their family or to poor school performance. The importance of the latter is evident by results showing that previous performance is a very strong predictor of the choice to continue in school in ethnic minority groups as well as in the majority population (e.g. Jonsson & Rudolphi 2011; Kilpi-Jakonen 2011).14

We are not in a position to refute any theories, given the rather crude analyses we have undertaken, but the cultural explanations we outlined above do not find much support in our data. The idea that cultural dissonance would be of importance (Heath & Brinbaum 2007), while highly plausible in theory, does not square well with our empirical results: several minority groups from cultural, religious and language groups that are strikingly different from the destination country do well in school, whereas some culturally closely related groups (even from neighbouring countries) do worse. It is still possible, of course, that there are more specific cultural theories that could explain why some groups do better or worse than others, but such theories need to be much better developed in order not to appear as ex-post constructions.

Another cultural assumption, drawing on the idea of ‘oppositional cultures’ and cultural distinctions, suggests that the school is experienced as an alien place by ethnic minority students who subsequently become discouraged from further studies and withdraw from school. Though a popular theory, some have expressed doubt about the empirical support for it (e.g. Downey 2008), and our new analyses support such scepticism. Likewise, our results do not support the more critical idea that ethnic minority students are ‘pushed (p.113) out’ of school to a large extent due to discrimination of different forms (inherent in the curriculum and textbooks, or performed by teachers or peers) (e.g. Fine 1991; Smyth & Hattam 2004). If Western schools were generally repellent for ethnic minority students—because they embrace oppositional cultures and/or are discriminated against—we would expect that ethnic minorities at large would do poorly and leave school early. This is simply not the case. In particular, we are struck by the lack of any discernable pattern that would show greater non-continuation rates for visible minorities, net of social origin. To the extent that there is a ‘cultural bias’ of schools it appears to take its toll primarily on children from less advantageous social and educational backgrounds, and ethnic minority students suffer from this because their parents often have poorer qualifications and occupy more disadvantageous social positions than majority students' parents. The pervasive nature of class origin differences means that they appear both for school achievement and for educational choice at given levels of achievement (Jackson 2013), and these two types of effects then reflect negatively on most ethnic minorities. However, if we are to look for any cultural explanations, rather than looking for disadvantage, a more pertinent question is why the negative socioeconomic choice effects are counterbalanced in many ethnic minority groups when they have the opportunity of making ambitious educational choices (Jackson et al. 2012).

While choice theories appear to tell a convincing story of early school-leaving, there are remaining differences to account for also when we control for both social origin and previous performance. These are both in terms of ethnic disadvantage (some, though not all, West Asian and Northern African groups) and advantage (mostly minorities in England). In trying to explain these results, it is tempting to invoke particular circumstances, for example, the lower opportunity costs of ethnic minority children in England (as the majority children who leave school have much higher employment rates than ethnic minority children who do so; which in itself is unexpected given the greater flexibility of wages in England). However, just as we felt uncomfortable in applying cultural explanations for specific findings above, we feel that further exploration of the net differences registered in Figures 4.24.5 must await better data and more detailed analyses.

It is of great relevance to compare the ethnic minorities' situation in early school-leaving rates across destination countries, to get a clue to whether institutional differences—in schools or labour markets—make a difference. However, we have few cases and our hypotheses are imprecise because we know little about the way institutional differences impinge on ethnic differences. A conservative conclusion is that we find no clear pattern between destination countries and no clear support for any of our tentative hypotheses. For example, the idea that ethnic minorities would drop out more in countries (p.114) with deregulated labour markets (England and the USA) does not stand up to empirical scrutiny. The ethnic minority disadvantage instead appears to be accentuated in the Netherlands, which may suggest that early selection (but not standardised track placement) is negative—the more rigorous macro-level analysis in Chapter 9 gives more details on this. But the minority disadvantage is also relatively pronounced in Finland, which is not expected. This may, on the other hand, be due to the composition of the minority population; in Finland, immigration is quite recent, which means that children whom we include as second generation are likely to have recently immigrated parents who are not yet integrated into the destination society, and whose (to us unobserved) problems also contribute to the relatively unfavourable position of Finnish ethnic minorities (cf. Kilpi-Jakonen 2011).

Comparing countries is in general difficult, because those we study have different compositions of ethnic minorities. What we could do as a preliminary way of ‘holding constant’ for this is to look at the results for children of Turkish ancestry, because they are a significant minority in Sweden and the Netherlands.15 The disadvantage experienced by children of Turkish origin in these countries is quite similar, which does not support the idea that differences in school systems are of importance for ethnic inequalities.16 Again, the analyses in Chapter 9 will scrutinise this.

To conclude, our results for early school-leaving speak in favour of an optimistic interpretation of structural integration, and mostly rebut the pessimistic views of enduring misery among immigrant groups. We have three caveats to that interpretation. First, our respondents grew up in the destination coun-tries—newly arrived immigrant children fare much worse, so integration may take time, perhaps a whole generation. Second, our results may to some extent be illusory if minority children have higher drop-out rates later on during upper secondary school—perhaps as a result of being ‘over-advised’ by their schools (Driessen 2006; Van de Werfhorst & Van Tubergen 2007). Such an analysis is outside the scope of this chapter, but the apprehension is in fact given some support in other studies for England (Middleton et al. 2005) and for Sweden (Jonsson & Rudolphi 2010), although our impression is that our main conclusions still stand.

Third, notwithstanding the good news from our statistical models in Figures 4.24.5, we want to finish this chapter by revisiting Figure 4.1, showing (p.115) the actual school-leaving rates among different groups. These figures do suggest that several minority groups face substantial gross disadvantages that are real and therefore have real consequences: Although predominantly because of quite general socioeconomic processes, many will follow their parents into joblessness and marginalised positions. As many as one in five of some ethnic minority groups (Turks and Chileans in Sweden; Africans and ex-Yugoslavs in Finland; Pakistanis in England) will enter the modern twenty-first-century labour market largely lacking marketable qualifications— although this is true also for almost every fourth child of majority origin in England. These figures are markedly higher in the Netherlands, but there the school-leavers in many cases have vocational qualifications. Whether children born in the mid-to-late 1980s manage to make a career despite an early exit from school, or at least get a firm footing in the labour market, remains to be seen. While some will get back to school, it is likely that pockets of misfortune will linger on for children of immigrants, but in much reduced form as compared to the vulnerability of many first-generation immigrants. We thus envisage that children of ethnic minority origin, even when born into the destination country, will continue to be over-represented among those with a precarious situation in the labour market; at the same time as many of their brothers and sisters will succeed both in education and employment. This polarised fate of ethnic minority children is both a promise and a challenge for the future.

Proceedings of the British Academy, 196, 95–118. © The British Academy 2014


Bibliography references:

Alba, R. & Nee, V. (2003), Remaking the American Mainstream: Assimilation and Contemporary Immigration (Cambridge, MA, Harvard University Press).

Alexander, K.S., Entwistle, D.R. & Kabbani, N.S. (2001),‘The Dropout Process in Life Course Perspective: Early Risk Factors at Home and School’, Teachers College Record 103(5): 760–822.

Allen, J. & Meng, C. (2010), Voortijdige Schoolverlaters: Aanleiding en gevolgen (Maastricht, Researchcentrum voor Onderwijs en Arbeidsmarkt).

Bourdieu, P. (1984), Distinction: A Social Critique of the Judgement of Taste (Cambridge, MA, Harvard University Press).

Bradley, S. & Lenton, P. (2007), ‘Dropping Out of Post-compulsory Education in the UK: An Analysis of Determinants and Outcomes’, Journal of Population Economics, 20: 299–328.

Breen, R. (2005), ‘Explaining Cross-national Variation in Youth Unemployment: Market and Institutional Factors’, European Sociological Review, 21: 125–34.

Brinbaum, Y. & Cebolla-Boado, H. (2007), ‘The School Careers of Ethnic Minority Youth in France’, Ethnicities, 7: 445–74.

Brinbaum, Y. & Kieffer, A. (2009), ‘Trajectories of Immigrants' Children in Secondary Education in France: Differentiation and Polarization’, Population-E, 64: 507–54.

(p.116) Bryk, A.S. & Yeow, M.T. (1989), ‘The Effects of High School Organization on Dropping Out: An Exploratory Investigation’, American Educational Research Journal, 26: 353–84.

Downey, D.B. (2008), ‘Black/White Differences in School Performance: The Oppositional Culture Explanation’, Annual Review of Sociology, 34: 107–26.

Driessen, G. (2006), ‘Het advies voortgezet onderwijs: is overadvisering over?’ Mens & Maatschappij, 81: 5–23.

Eckstein, Z. & Wolpin, K.I. (1999), ‘Why Youths Drop Out of High School: The Impact of Preferences, Opportunities, and Abilities’, Econometrica, 67: 1295–1339.

Erikson, R. & Jonsson J.O. (1996), ‘Introduction: Explaining Class Inequality in Education: The Swedish Test Case’, in R. Erikson and J.O. Jonsson (eds) Can Education Be Equalized? The Swedish Case in Comparative Perspective (Boulder, CO, Westview Press), 1–64.

Feliciano, C. (2005), ‘Educational Selectivity in US Immigration: How Do Immigrants Compare to those Left Behind?’ Demography, 42: 131–52.

Fine, M. (1991), Framing Dropouts: Notes on the Politics of an Urban Public High School (New York, State University of New York Press).

Fordham, S. & Ogbu J.U. (1986), ‘Black Students' School Success: Coping with the Burden of “Acting White”’, The Urban Review, 18: 176–206.

Goyette, K. & Xie, Y. (1999), ‘Educational Expectations of Asian American Youths: Determinants and Ethnic Differences’, Sociology of Education, 72: 22–36.

Heath, A.F. & Brinbaum, Y. (2007), ‘Explaining Ethnic Inequalities in Educational Attainment’, Ethnicities, 7: 291–305.

Heath, A.F., & Cheung S.-Y. (eds) (2007), Unequal Chances: Ethnic Minorities in Western Labour Markets. Proceedings of the British Academy 137 (Oxford, Oxford University Press).

Heath, A.F., Rothon, C. & Kilpi, E. (2008), ‘The Second Generation in Western Europe: Education, Unemployment, and Occupational Attainment’, Annual Review of Sociology, 34: 211–35.

Jackson, M. (ed.) (2013), Determined to Succeed? Performance versus Choice in Educational Attainment (Stanford, CA, Stanford University Press).

Jackson, M., Jonsson, J.O. & Rudolphi, F. (2012), ‘Ethnic Inequality and Choice-driven Educational Systems: A Longitudinal Study of Performance and Choice in England and Sweden’, Sociology of Education, 85: 158–78.

Järvinen, T. & Vanttaja M. (2001), ‘Young People, Education, and Work: Trend and Changes in Finland in the 1990s’, Journal of Youth Studies, 4: 195–207.

Jonsson, J.O. & Erikson, R. (2000), ‘Understanding Educational Inequality: The Swedish Experience’, L'année sociologique, 50: 345–82.

Jonsson, J.O. & Rudolphi F. (2010), ‘Structural Integration in Multi-ethnic Society: Educational Careers of Children to Immigrants in Sweden’, Manuscript, Previous version presented at the ISA/RC28 Conference at Stanford University, August 2008.

Jonsson, J. O. & Rudolphi, F. (2011), ‘Weak Performance–Strong Determination. School Achievement and Educational Choice among Ethnic Minority Students in Sweden’, European Sociological Review, 27: 487–508.

Kao, G. & Thompson, J.S. (2003), ‘Racial and Ethnic Stratification in Educational Achievement and Attainment’, Annual Review of Sociology, 2: 417–42.

(p.117) Kilpi-Jakonen, E. (2011), ‘Continuation to Upper Secondary Education in Finland: Children of Immigrants and the Majority Compared’, Acta Sociologica, 54: 67–94.

Kristen, C. (2005), School Choice and Ethnic School Segregation: Primary School Selection in Germany (Münster, Waxmann Verlag).

Lee, V.E. & Burkam, D.T. (2003), ‘Dropping Out of High School: The Role of School Organization and Structure’, American Educational Research Journal, 40: 353–93.

Luijkx, R. & de Heus, M. (2008), ‘The Educational System of the Netherlands’, in Schneider, S.L. (ed.) The International Standard Classification of Education (ISCED 97): An Evaluation of Content and Criterion Validity for 15 European Countries (Mannheim, MZES), 47–75.

Marks, G.N. (2005), ‘Do Schools Matter for Early School-leaving? Individual and School Influences in Australia’, School Effectiveness and School Improvement, 18: 429–50.

Middleton, S., Rennison, J., Cebulla, A., Perren, K. & De-Beaman, S. (2005), Young People from Ethnic Minority Backgrounds: Evidence from the Education Maintenance Allowance Pilots Database, Research Report 627 (London, Department for Education and Skills).

Mood, C. (2010), ‘Logistic Regression: Why We Cannot Do What We Think We Can Do and What We Can Do About It’, European Sociological Review, 26: 67–82.

Müller, W.& Gangl, M. (eds) (2003), Transitions from Education to Work in Europe (Oxford, Oxford University Press).

Murray, Å. & Sundin, S. (2008), ‘Student Flows and Employment Opportunities Before and After Implementation of a Third Year in Vocational Programmes at Upper Secondary School’, European Journal of Vocational Training, 44: 110–31.

OECD (2005), From Education to Work: A Difficult Transition for Young Adults with Low Level of Education (Paris, OECD).

OECD (2009), Education at a Glance (Paris, OECD).

Payne, J., Cheng, Y. & Witherspoon, S. (1996), Education and Training for 16–18 Year Olds, Individual Paths and National Trends (Poole, BEBC).

Portes, A. & Zhou, M. (1993), ‘The New Second Generation: Segmented Assimilation and Its Variants’, Annals of the American Academy of Political and Social Sciences, 530: 74–96.

Ream, R.K. & Rumberger, R.W. (2008), ‘Student Engagement, Peer Social Capital, and School Dropout among Mexican American and Non-Latino White Students’, Sociology of Education, 81: 109–39.

Riach, P.A. & Rich, J. (2002), ‘Field Experiments of Discrimination in the Market Place’, The Economic Journal, 112: F480–F518.

Rothon, C. (2007), ‘Can Achievement Differentials be Explained by Social Class Alone? An Examination of Minority Ethnic Educational Performance in England and Wales at the End of Compulsory Schooling’, Ethnicities, 7: 306–22.

Rumbaut, R. (2004), ‘Ages, Life Stages and Generational Cohorts: Decomposing the Immigrant First and Second Generations in the United States’, International Migration Review, 38(3): 1160–1205.

Rumberger, R.W. (1987), ‘High School Dropouts: A Review of Issues and Evidence’, Review of Educational Research, 57: 101–21.

Rumberger, R.W. & Lamb, S.P. (2003), ‘The Early Employment and Further Education Experiences of High School Dropouts: A Comparative Study of the United States and Australia’, Economics of Education Review, 22: 353–66.

(p.118) Shavit, Y. & Müller, W. (eds) (1998), From School to Work: A Comparative Study of Educational Qualifications and Occupational Destinations (Oxford, Oxford University Press).

Smyth, J. & Hattam, R. (2004), ‘Dropping out’, Drifting Off, Being Excluded: Becoming Somebody Without School (New York, Peter Lang).

Townsend, L., Flisher, A.J. & King, G. (2007), ‘A Systematic Review of the Relationship between High School Dropout and Substance Use’, Clinical Child and Family Psychology Review, 10: 295–317.

Traag, T. & Van der Velden R.K.W. (2011), ‘Early School-leaving in the Netherlands: The Role of Family Resources, School Composition and Background Characteristics in Early School-leaving in Lower Secondary Education’, Irish Educational Studies 30: 45–62.

Van de Werfhorst, H.G., & Mijs, J.J.B. (2010), ‘Achievement Inequality and the Institutional Structure of Educational Systems: A Comparative Perspective’, Annual Review of Sociology, 36: 407–28.

Van de Werfhorst, H.G. & Van Tubergen, F. (2007), ‘Ethnicity, Schooling, and Merit in the Netherlands’, Ethnicities, 7: 416–44.

Willis, P. (1977), Learning to Labour. How Working Class Kids get Working Class Jobs (Farnborough: Saxon House).


Proceedings of the British Academy, 196, 95–118. © The British Academy 2014.

(1) Mathieu Ichou kindly prepared the Longitudinal Survey of Young People in England (LSYPE) and ran the analyses. He is currently completing a doctorate at L'Observatoire Sociologique du Changement (Observatory for Social Change, CNRS-Sciences Po), Paris.

(2) For example, the finding that schools with an academic curriculum have lower drop-out rates in the USA may suggest that these schools are well organised and have high expectations on their students, perhaps combined with good teachers and strong general support to students. However, the fact that these schools can offer advanced courses in the first place may be because their students are motivated from home and have high educational aspirations.

(3) It is notoriously difficult to distinguish ‘cultural’ theories from choice theories. Lack of information may for example be seen as adding costs to the decision process (e.g. Jonsson & Erikson 2000), although in practice much of the knowhow when it comes to navigating the school system is in the form of tacit knowledge. One can argue that it is not primarily an issue of lacking a resource (such as knowledge about the school system) but rather that immigrants may have a different kind of knowledge due to the fact that they grew up in a country with a different school system.

(4) In this chapter we refer to England, rather than to England and Wales as in other chapters, since our data source is the Longitudinal Survey of Young People in England, which does not cover young people in Wales.

(5) This section refers to details about national educational systems and data sources that are described in more detail in Chapter 2.

(6) We recognise that there are courses that are ‘light’ vocational, such as apprenticeships with some connection with schools, or shorter training courses, but these are counted as school-leaving.

(7) There is a slight difference in the definition of school-leaving between Finland and Sweden, as in the latter country school-leavers are defined as those who had not begun an upper secondary school track on 15 October the year they finished comprehensive school (graduation is in June, terms normally start in mid-August), while in the Finnish data the cut-off point is approximately one year later.

(8) In England, we use the LSYPE, and classify those as early school-leavers who in wave 4 (grade 12) respond with ‘no’ to the question whether he or she is ‘doing any courses at school or college which lead to qualifications’ and who do not subsequently say that they were studying for GCSEs.

(9) The English results are similar when we instead use the YCS dataset, though the minority advantage is somewhat less pronounced (cf. Bradley & Lenton 2007). We also ran a sensitivity check by studying drop out in wave 3 of the LSYPE—there also, the main results stand, though with slightly attenuated minority advantage.

(10) Because probit coefficients cannot be compared across models or datasets, we follow the advice in Mood (2010) and calculate average marginal effects (AMEs). What we present is then estimated average percentage differences in early school-leaving rates between groups, had these groups been similar in terms of parental education and class, as well as family structure (in addition, the models control for gender).

(11) Because the models we run separately for each destination country are rather complex and would need much space to present, we show only the relevant ethnic origin parameters. The effects of control variables are overall in the expected direction. The full results are available from the authors upon request.

(12) We have, rather arbitrarily, put the second-generation ‘white’ in the USA in the Eastern European category.

(13) An alternative model would not contain endogenous variables such as occupation or family structure. However, we have tested this model on Swedish data and the changes in minority-majority differences were slight.

(14) Jonsson & Rudolphi (2010) tested another choice-inspired idea on Swedish data, namely that the high incidence of early school-leaving among children of Middle East origin could be explained by the fact that so many of their parents are self-employed (around 40 per cent of the Turks, for example), making the choice of early labour market entry, and presumably integration into their parents' business, rational (especially if viewed as a family decision). However, it turned out that early school-leaving was as common among those with employed parents as among those with self-employed parents.

(15) It is quite possible that the ‘Turkish’ group differs across countries: In Sweden, there is a tangible proportion of Kurds and Syrians among the Turks, for example. Controlling for parents' education and class position erases some of these differences.

(16) To gain leverage, we also analysed data from Germany (in a dataset that could not be made comparable enough to include in the chapter), and the negative effects of Turkish origin were very similar to the figures for the Netherlands and Sweden. We are grateful to Cornelia Kristen and Nadia Granato for conducting these analyses.