«IZA DP No. 2072 Comprehensive versus Selective Schooling in England in Wales: What Do We Know? Alan Manning Jörn-Steffen Pischke April 2006 ...»
DISCUSSION PAPER SERIES
IZA DP No. 2072
Comprehensive versus Selective Schooling
in England in Wales: What Do We Know?
zur Zukunft der Arbeit
Institute for the Study
Comprehensive versus Selective
Schooling in England in Wales:
What Do We Know?
London School of Economics
London School of Economics and IZA Bonn Discussion Paper No. 2072 April 2006 IZA P.O. Box 7240 53072 Bonn Germany Phone: +49-228-3894-0 Fax: +49-228-3894-180 Email: email@example.com Any opinions expressed here are those of the author(s) and not those of the institute. Research disseminated by IZA may include views on policy, but the institute itself takes no institutional policy positions.
The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center and a place of communication between science, politics and business. IZA is an independent nonprofit company supported by Deutsche Post World Net. The center is associated with the University of Bonn and offers a stimulating research environment through its research networks, research support, and visitors and doctoral programs. IZA engages in (i) original and internationally competitive research in all fields of labor economics, (ii) development of policy concepts, and (iii) dissemination of research results and concepts to the interested public.
IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion.
Citation of such a paper should account for its provisional character. A revised version may be available directly from the author.
IZA Discussion Paper No. 2072 April 2006
ABSTRACTComprehensive versus Selective Schooling in England in Wales: What Do We Know?* British secondary schools moved from a system of extensive and early selection and tracking in secondary schools to one with comprehensive schools during the 1960s and 70s. Before the reform, students would take an exam at age eleven, which determined whether they would attend an academically oriented grammar school or a lower level secondary school.
The reform proceeded at an uneven pace in different areas, so that both secondary school systems coexist during the 1960s and 70s. The British transition therefore provides an excellent laboratory for the study of the impact of a comprehensive versus a selective school system on student achievement. Previous studies analyzing this transition have typically used a value-added methodology: they compare outcomes for students passing through either type of school controlling for achievement levels at the time of entering secondary education. While this seems like a reasonable research design, we demonstrate that it is unlikely to successfully eliminate selection effects in who attends what type of school. Very similar results are obtained by looking at the effect of secondary school environment on achievement at age 11 and controlling for age 7 achievement. Since children only enter secondary school at age 11, these effects are likely due to selection bias. Careful choice of treatment and control areas, and using political control of the county as an instrument for early implementation of the comprehensive regime do not solve this problem.
JEL Classification: I21, I28 Keywords: tracking, selective secondary schooling, comprehensive schools
Jörn-Steffen Pischke CEP London School of Economics Houghton Street London WC2A 2AE United Kingdom Email: firstname.lastname@example.org * We thank Ghazala Azmat and Michele Pellizzari for excellent research assistance. We thank seminar participants at the NBER Education Program Meetings and particularly Esther Duflo for helpful comments on a previous version of this paper. Pischke thanks the NBER for their hospitality during a visit when much of the work for this project was performed.
Introduction National school systems differ widely in the amount of ability tracking of students they provide in secondary school. Some systems (e.g. the US) are based on comprehensive schools, where students of all abilities attend the same school, although there is typically some tracking within schools. Other systems (e.g. Germany) channel students at an early age into different types of schools based on academic ability. British schools moved from a system of extensive tracking to one with comprehensive schools in the 1960s and 70s. The British experience is interesting, because it involved a major and well defined change in terms of the ability grouping of secondary school students. Hence, it offers potentially very promising research designs in order to assess the impact of the secondary school regime on student achievement. It is unsurprising that numerous studies in the education and economics of education literatures analyze the British experience.
In the traditional British school system, students were tracked into either an academically selective grammar school at age 11, or they would attend a secondary modern school, which was academically less demanding. Starting in the 1950s, there was dissatisfaction with selection at the local level, and some local authorities began to experiment with comprehensive schools. In 1965, the central government asked the Local Education Authorities (LEAs) to draw up plans to switch to a comprehensive system. The implementation proceeded slowly, with faster growing, more Labour leaning LEAs moving to comprehensive schools more quickly, while those without expanding numbers of students, and more Conservative leaning Authorities implemented the change more slowly. In fact, there are still a number of LEAs to date which provide grammar schools as an option.
Much of the research on comprehensive education in Britain uses the National Child Development Study (NCDS), a panel which tracks members of the 1958 birth cohort. The sample members entered secondary school in 1969, at a time when some LEAs in Britain had started to provide comprehensive schools already, while others continued to offer the traditional selective schools. Hence, the comparisons using the NCDS are essentially crosssectional. While Kerckhoff et al. (1996) claim that comprehensive areas differ little from selective areas, we find that comprehensive areas are systematically poorer, and have students with lower previous achievement. A raw comparison of students attending comprehensive and selective schools is therefore not possible. Most previous studies rely on some form of a value added specification, where student performance at age 16 is a function of student performance at age 11, the secondary school environment, and possibly other control variables.
This seems an entirely reasonably research design, particularly since the NCDS offers a large set of controls on student ability and family background. We demonstrate in this paper that this methodology is nevertheless unlikely to be completely successful in removing the selection bias between comprehensive and selective school students. The NCDS provides test scores at ages 7, 11, and 16. It is therefore possible to run an analogous specification for student performance at age 11 on student performance at age 7, and the secondary school environment. Since students at age 11 have not started attending secondary school yet, the secondary school environment should have no influence on age 11 outcomes if the specification successfully removes the selection bias. This exercise is therefore a falsification test for the value added specification.
OLS value added specifications tend to lead to a small negative average effect of comprehensive schooling compared to selective schooling in our sample. We demonstrate that effects of a similar size or larger are obtained when using age 11 test scores as outcomes.
This suggests that the methodology is unable to remove the selection bias between students attending different schools. A more interesting finding of the previous literature (particularly Kerkhoff et al., 1996) may be that comprehensive schools tend to be equalizing, in that they lower the performance of the most able students, and raise the performance of the least able.
We show that this result is also likely due to selection bias.
We also demonstrate that various approaches to improve on the methodology used previously are not more successful. Most studies on this question compare students attending comprehensive and selective schools. Since many LEAs were in the middle of comprehensive reorganization in the 1970s, these schools often coexisted within the same LEA. A better use of the policy variation, possibly removing some of the selection, is the comparison of LEAs which are either purely comprehensive or purely selective in 1969, when the NCDS cohort enters secondary school. We show that this does not help in removing the selection bias. Another approach is the use of instrumental variables (IV) for comprehensive reorganization (see, for example, Galinda-Rueda and Vignoles, 2004).
Political control of the county is a good predictor of early reorganization, and conditional on county socio demographic characteristics is a plausible instrument. We show that the IV strategy similarly fails.
At a theoretical level, there are good arguments for selection as well as for comprehensive education. The main argument for selection or tracking is presumably that it is much easier to teach lower variance classes. Since teachers can focus on the ability level of particular groups of students, students of all ability levels might benefit from selection. One argument against selection is that there might be positive peer effects from the most able students. By tracking these students into separate classrooms, the most able students may benefit from being with each other. However, the lower ability ranges loose from not having this peer group around. We know very little about the different impact of peer group effects on different types of students empirically, so it is difficult to judge a priori whether this leads to lower or higher average performance in a selective system.
Another argument, particularly against early selection as in post-war Britain, is that eventual ability levels are difficult to predict at an age as early as 10 or 11. Moreover, secondary selection is based on a single exam, clearly a noisy mechanism. This may result in some kids ending up in the wrong track. This suggests that particularly middle ability kids may loose in the British style selective system. Since there are arguments going either way, the issue eventually is an empirical one.
The findings in the empirical literature about selection differ widely. Many researchers looking at the British secondary reorganization conclude that the evidence does not support claims to the superiority of either system (see, for example, Crook, Power, and Whitty, 1999).
On the other hand, there are also studies which find more pronounced effects going one way or the other. Jesson (2000) uses data from the 1990s on those LEAs which still remain selective. Also using a value added approach, he argues that comprehensive LEAs systematically outperform selective LEAs in Britain. Based on the findings from the NCDS, these results are equally suspect. We conclude that little can be learned from the existing literature on the British reorganization on the performance of comprehensive versus selective schools.
Two recent papers by Hansushek and Wößmann (2006) and Waldinger (2006) apply an approach roughly similar to ours to educational inequality across countries. Both papers investigate whether inequality in educational outcomes is related to the amount of secondary school tracking and selection across countries. These papers apply a differences-indifferences approach comparing outcomes between secondary and primary schooling.
Hanushek and Wößmann find evidence that tracking raises educational inequality, while Waldinger does not. However, Waldinger shows that the Hanushek and Wößmann results are not very robust to alternative sample and variable specifications.
The remainder of this paper is organized as follows. The next section briefly describes the institutional background of British secondary education, and the history of comprehensive reorganization. Section 3 discusses the empirical framework, and the approaches used in some of the existing literature. The following section describes the data and key variables.
Results are discussed in section 5, and the final section concludes.