Diverging performances: the detrimental effects of early educational selection on equality of opportunity in Hungary

https://doi.org/10.1016/j.rssm.2013.01.002Get rights and content

Abstract

The Hungarian system is ideal to test the effect of early-selection on inequality of opportunity, since students are selected at three different ages. The early-selective academic tracks skim off the best students first at age 10, then at age 12, and finally at age 14 all students enter secondary level. The paper first shows that higher socioeconomic status students are more likely to attend the early-selective academic tracks, even if previous test scores are controlled for. The second part of the empirical analysis looks at the value-added of the separate tracks between 6th and 8th grade, and between 8th and 10th grade, and shows that their mathematical and reading performance diverges, even if skill and status selection is taken into account. Since higher socioeconomic status students are more likely to attend academic tracks than their lower status peers, the divergence in test scores translates to increasing inequality of opportunity. The Section 3 of the empirics looks at whether this process is a Pareto improvement, or whether there are groups in society that lose by the early-selection. It is shown that those who are left in general schools in areas where the best students can opt-out to early-selective tracks perform worse in mathematics than similar students in general tracks with no option of leaving. That is, selection harms those who are left behind. The paper speculates that these results are due to the different peer and teacher quality of the different tracks.

Introduction

Whether or not early selection exacerbates inequality in education is a recurrent question in social science. Most countries, when changing the age at which the first selection in education is made, increase it, rather than decrease it, hoping thus to make the system more equal.2 There are very few countries in which the age of first selection is decreased, and Hungary is one of them.

Hungary is an obvious choice for a test of the effect of early selection on inequality of opportunity for two important reasons. First, inequality of opportunity in the Hungarian education system is especially high. In 2009, the variance in reading performance, explained by family factors, was the highest among all PISA3 participant countries (OECD, 2010a). In other words, the role of the family is significant in how students perform in schools. The difference in literacy between the academic and vocational routes at secondary level is striking: students on the academic track score close to the best PISA performers, while students in the lowest level track of vocational training score near the bottom of the PISA rank. Naturally, this difference stem mainly from the fact that tracks have very different student intake (Horn, Balázsi, Takács, & Zhang, 2006). Second, during the transition, Hungary in effect decreased the first age of selection; two new types of academic track were introduced, which select the best children at ages 10 and 12, as opposed to the traditional age of selection at 14. This paper utilizes this unique multi-level selective feature of the Hungarian system, as well as the spatial variation in these early-selective academic tracks,4 to test whether early selection has contributed to the high inequality of opportunity in the Hungarian system.

Inequality is usually understood either as inequality of outcome or as inequality of opportunity. While the variance in educational outcomes is unquestionably an important aspect, I focus on the dimension of inequality of opportunity in this paper. I believe that giving a fair opportunity in life is a more commonly accepted goal of educational policy, rather than decreasing the variance in outcomes. Henceforth inequality is understood as inequality of opportunity, unless otherwise noted.

But what are the possible reasons behind inequality increasing as an effect of early selection? Why would a lower age of selection increase inequality of opportunity?

The most likely reasons are that teachers matter and peers matter. More precisely: since teachers matter, selecting students and forming them into groups, and allowing the quality of teachers to differ across groups, affects the performance of the students differently in the selected groups. If there are differences in teacher quality between schools, this should affect the mean performance of the students differently between schools. This is precisely the starting point of Rivkin, Hanushek and Kain (2005), who test this assumption. They test the effects of teachers on students, using a large, matched, teacher-student dataset from Texas. They arrive at the unsurprising conclusion that, “teachers, and therefore schools matter importantly for student achievement” (p. 449). Another way of looking at teacher quality, besides rich teacher-student matched datasets, is to qualitatively assess the reasons behind superior education performance. Barber & Mourshed (2007) looked at the best PISA performers and concluded that the most important reason behind the quality of these education systems is the high quality of teachers. The best systems, “get the right people to become teachers”, because the “quality of an education system cannot exceed the quality of its teachers.”

Peers also have an important effect, and several papers have attempted to identify the effects of peers on student outcomes. Although this problem is loaded with methodological difficulties (see Manski, 1993), some have shown, in an especially convincing way, that peers indeed matter (see Sacerdote, 2001 on college roommates). Hoxby (2000), as well as Hanushek, Kain, Markman and Rivkin (2003), in the case of public education, have shown that peer effects are important. Higher achieving schoolmates can improve others, but these effects are likely to be reciprocal, less bright peers can be a hindrance. Thus, again, in a system where students are selected into homogeneous groups, peer effects could increase differences between schools, and thus increase inequality of outcome.5

But selection is usually not “status blind”6. Higher status students are much more likely to attend academic tracks, and continue on to tertiary education, than their lower status peers. It has been accepted for some time that social origin is one of the most important factors in individual educational attainment (Shavit & Blossfeld, 1993). A student's socioeconomic status is also much more important in determining educational outcome differences between schools than the measured differences in school resources (Coleman, 1966, Hanushek, 1998).

Hence, if teacher quality and peer effects increase differences between tracks that have a significantly different composition in socioeconomic status, then the increasing inequality of outcome becomes increasing inequality of opportunity, and the gap between high-status and low-status students increases.

This paper goes through this logic step-by-step to argue that early selection has a causal and detrimental effect on equality of opportunity. The first empirical question is whether the early-selective tracks are status selective. The paper looks at whether the observed differences in socioeconomic status between the different tracks are due to skill selection, or to other non-observed factors, related to individual status, which also determine who gets into the early-selective tracks. The second part of the empirical section looks at the “value-added” of the different tracks. It tests whether the type of track has a significant effect on the student's numeracy or literacy, even if individual status and previous test scores are controlled for. Logically, if the early-selective tracks are composed of higher status students, and they also add to the literacy scores of the individual students, then the inequality of opportunity of the Hungarian system is larger than it would be without the early-selection.

Note, however, that outcome differences between tracks can increase either by increasing performance on the top or by decreasing performance at the bottom. While increasing the performance only of students in academic tracks might be considered a Pareto improvement, if lower status students lose, the process of early selection is clearly undesirable. Early selection skims off the best students and the best teachers from the general schools, and thus decreases the important peer and teacher effects. Thus I expect the performance of general schools to be lower, due to the negative effect of early selection. The final part of the paper studies this problem. It tests whether unselected students, whose classmates move to early-selective tracks, lose out in numeracy or literacy as compared to students who had no classmates leaving early.

Section snippets

The effect of early selection on inequality – state of the art

Studies looking at the effect of early selection (or tracking) on educational inequality use either the geographical variance of educational systems across countries (e.g. Ammermüller, 2005, Hanushek and Woessmann, 2006), geographical variance within countries (e.g. Bauer & Riphahn, 2006), time variance within countries (e.g. Aakvik et al., 2010, Meghir and Palme, 2005, Pekkarinen et al., 2009), or pilot studies (Hall, 2012).7

The early-selective tracks in Hungary8

The Hungarian system selects children quite early, first at age 10. It has not always been like this. Before 1989 the system was a typical “soviet” system, with 8 years of general training and three types of secondary track, in which students could study after the age of 14. There were two vocational tracks, a relatively more academically oriented vocational secondary/technikum (szakközépiskola), and a more practical vocational/apprentice training track (szakmunkásképzés), and an academic track

The NABC database and descriptive statistics

The National Assessment of Basic Competencies (NABC) is a standard-based assessment designed similarly to the OECD PISA survey, but conducted annually in May.11 It measures literacy and numeracy of the 6th, 8th and 10th

Are early-selective tracks status selective?

The conditional student composition of the early-selective tracks shows the status selectivity of the early-selective tracks. By controlling for previous test scores the coefficient of the SES index on track choice shows how important status is. I could not look at the 8-yr-ac choice in 6th grade, because controlling for the 6th grade test score would introduce an endogeneity problem: students in 8-yr-ac have already studied there for two years, hence 6th grade test scores are affected by the

Do early-selective tracks have a higher value-added?

Based on the literature, I expect 8-yr-ac and 6-yr-ac to have a superior performance compared to general tracks. Similarly, 8-yr-ac should perform better compared to the 6-yr-ac track, because students have two additional years of higher quality teachers and peers.

Track differences are estimated by a simple OLS regression for the 2008/6th grade cohort (Table 7) and for the 2008/8th grade cohort (Table 8). The dependent variable is the 2010 reading and mathematics test scores. I standardized the

Do others lose?

The spatial variation in early-selective tracks allows for a comparison of those students who remained in general tracks, in regions where early-selective tracks are available, and those who had effective spatial barriers to entry. As above, I assume that performance differences between tracks are due to differences in teacher quality or differences in peer effects. I hypothesize that in areas where 6-yr-ac or 8-yr-ac tracks skim off the best students, those left in general schools have a

Conclusion

Educational selection is not “status blind”. Higher socioeconomic status students are more likely to attend academic tracks than their lower status peers, thus academic tracks tend to have higher average status. Higher status students also attract higher quality teachers (Rivkin et al., 2005, Varga, 2011). Peer effects and teacher quality are two very important factors that affect student performance (Hanushek et al., 2003, Hoxby, 2000, Barber and Mourshed, 2007). If academic tracks select

Acknowledgements

I would like to thank Zoltán Hermann, Balázs Muraközy, Gábor Kertesi, Gábor Kézdi, Edwin Leuven, Fabian Pfeffer, Álmos Telegdy, Balázs Váradi, Júlia Varga and three anonymous referees for their helpful insights. Comments from participants at the SIMlife conference in Mannheim, NEPS seminar in Bamberg, the CERS HAS seminar in Budapest, the annual meeting of the Hungarian Society of Economics in Budapest and the Workshop on Educational Governance and Finance in Oslo are warmly acknowledged. I

References (44)

  • J.S. Coleman

    Equality of educational opportunity

    (1966)
  • Dolton, P. J. (2002). Improving Educational Quality: How Best to Evaluate Our Schools? (Discussion). Education in the...
  • J. Dronkers et al.

    Has educational sector any impact on school effectiveness in Hungary? A comparison of the public and the newly established religious grammar schools

    European Societies

    (2004)
  • J. Dronkers et al.

    School Choice in the Light of the Effectiveness Differences of Various Types of Public and Private Schools in 19 OECD Countries

    Journal of School Choice

    (2008)
  • T. Fuchs et al.

    What accounts for international differences in student performance?. A re-examination using PISA data

    Empirical Economics

    (2006)
  • F. Galindo-Rueda et al.

    The heterogeneous effect of selection in secondary schools: understanding the changing role of ability

    SSRN eLibrary

    (2004)
  • F. Galindo-Rueda et al.

    The declining relative importance of ability in predicting educational attainment

    Journal of Human Resources

    (2005)
  • C. Hall

    The effects of reducing tracking in upper secondary school evidence from a large-scale pilot scheme

    Journal of Human Resources

    (2012)
  • E.A. Hanushek et al.

    Does peer ability affect student achievement?

    Journal of Applied Econometrics

    (2003)
  • E.A. Hanushek

    Conclusions and controversies about the effectiveness of school resources

    FRBNY Economic Policy Review

    (1998)
  • E.A. Hanushek et al.

    Does educational tracking affect performance and inequality? Differences-in-differences evidence across countries

    The Economic Journal

    (2006)
  • Hermann, Z., & Molnár, T. L., 2008. ‘Országos Kompetenciamérési adatbázis’....
  • Cited by (30)

    • The non-cognitive returns to vocational school tracking: South Korean evidence

      2019, International Journal of Educational Research
      Citation Excerpt :

      Specifically, students in CTE schools have lower levels of academic culture (Van Houtte & Stevens, 2016; Van Houtte, 2006b), sense of belonging (Van Houtte & Van Maele, 2012), self-esteem (Van Houtte, 2005), self-control (Malmberg & Trempała, 1997), and teacher trust (Van Maele & Van Houtte, 2011) than academic school students. The observed detrimental effects of between-school vocational tracking are also often reported for student cognitive outcomes (e.g., Horn, 2013; Kim, 2014). Yet, Salmela‐Aro, Kiuru, and Nurmi (2008) showed that students in an academic track report higher levels of school exhaustion and tiredness compared to vocational track students.

    • Intergenerational transmission of education in Europe: Do more comprehensive education systems reduce social gradients in student achievement?

      2016, Research in Social Stratification and Mobility
      Citation Excerpt :

      Even where students are assigned on the basis of formal tests or administrative decisions, children from better-educated families are more likely to be selected into more demanding educational tracks because less-educated parents tend to have lower expectations regarding their children's educational performance, less knowledge about the educational requirements of more demanding tracks and less confidence that their children will fulfil these requirements, and they are consequently less likely to encourage their children to aspire to such demanding tracks or to invest in the education of their children (e.g., in private lessons or preparatory courses for tests). Research indicates that track placement reinforces the intergenerational persistence of educational achievements in particular if it takes place early and persists over the course of schooling (Bauer & Riphahn, 2013), because in tracking systems peer effects and curricular differences between different tracks may increase differences in educational performance that had already existed prior to the selection (Horn, 2013). Hence I hypothesise that more extensive tracking during compulsory education increases the transmission of education across generations (hypothesis 3).

    • School segregation, student achievement, and educational attainment in Hungary

      2023, International Journal of Comparative Sociology
    View all citing articles on Scopus
    1

    On leave from the Institute of Economics, Research Center for Economic and Regional Studies, Hungarian Academy of Sciences and from the Economics Department, ELTE, Hungary.

    View full text