Combining Two Decades of Evidence to Study the Effects of Tracking on Inequality and Segregation

• In Event: Combining International Student Assessments for Country-Level Longitudinal Analyses

Abstract

The objective of the present study is to use country-level data to estimate the effect of early between-school tracking on different conceptions of inequality in educational achievement and segregation between schools, specifically on (1) educational inequality, (2) social inequality, (3) educational inadequacy, (4) gender segregation, and (5) social segregation. In the context of the symposium, this study showcases how combining the accumulated evidence of large-scale assessments can be useful to identify effects of education-level policies.
In the debate on between-school tracking, a trade-off between efficiency and equity is discussed. Proponents of tracking argue that it enables schools to adapt better to the abilities of each learner. On the other hand, critics of such a policy are concerned that tracked systems have different curricula, varying expectations and an unequal allocation of resources. Such differences may lead to a larger dispersion of achievement (inequality) and detrimental effects for lower performance students who do not reach higher academic tracks (inadequacy). Furthermore, it is argued that the selection into the more prestigious tracks might not only base on achievement but also social background and therefore foster social inequality and segregation (e.g., Maaz et al., 2008).
Previous empirical studies with robust designs indicate inconclusive or contradictory findings. A possible explanation for the mixed findings is that the effects are dependent on the sample of countries and the inequality measurement analyzed, in addition to the inefficient low power effects found in small sample designs.
The study uses all available cycles from PIRLS, PISA and TIMSS and 75 countries. Data from primary and secondary schooling are combined to 45 separate datasets (e.g., PIRLS 2011 & PISA 2015).
Following Hanushek and Wößmann (2006), we estimate a simple difference-in- differences model where we regress the degree of inequality in secondary school (e.g., PISA 2015) on a variable that indicates whether the country is an early tracking country while controlling degree of inequality in primary school (e.g., PIRLS 2011). This approach circumvents confounding through unobserved contextual differences between early and late tracking countries. Separate models are run for educational inequality, social inequality, educational inadequacy, gender and social segregation for each of the 45 datasets. The results were combined with meta- analytical approaches (Card, 2012).
The main contribution of the present study is that we use all available information from international large-scale assessments to investigate effects of early tracking on various types of inequality and segregation. This circumvents the problems found in previous studies, suffering from the different sample of countries considered in each analysis and the small analytical samples that lead to unreliable effect estimates.
The results indicate that early tracking increases educational inequality and social inequality. However, the effect on social inequality is more robust, looking at both standardized effects and differences between domains. Early tracking also presents a weak effect on educational inadequacy, decreasing the performance of lower achievement students. We also observe that tracking increases gender and social segregation between schools. Further thoughts are presented on the methodological advantages and limitations of the analysis as well as the consequences on public policy.

Authors

• Andrés Strello, IEA Hamburg

• Rolf Strietholt, IEA Hamburg

• Isa Steinmann, Oslo Metropolitan University

• Charlotte Siepmann, TU Dortmund University