Balancing Performance and Equity in Student Cohort Partitions

The current project utilizes a novel dataset of rich student data spanning more than 150 high schools in Greece to shed light on the optimal trade-off between performance and equity in student cohort partitions, motivated by a recent educational policy debate in Greece.

The public education system in Greece is structured around the principle of equal opportunity and is characterized by many unique features, such as its predominantly public nature, high centralization, uniformity, as well as common, absolute metrics for evaluating student performance. Many of the existing educational practices and policies are designed to align with the principle of equal opportunity and are applicable to all school districts with minor exceptions.

A notable example is the assignment of students in primary and secondary education to schools and subsequently to cohorts within the same grade. Typically, students are first assigned to the nearest school in terms of geographical proximity to their residence. Subsequently, students in the same grade at a school are quasi-randomly divided into cohorts of approximately 20-25 students. These cohorts of students are assigned to potentially different teachers and rarely share a few elective courses. Due to this randomized cohort partition, the present policy is considered more equitable than other common practices such as sorting based on the students’ grades.

However, with the goal of improving the performance of Greek students, the Ministry of Education recently proposed a redesign of the cohort assignment process. The suggested approach aimed to loosely group students by performance (i.e., placing students of similar performance levels together) while still maintaining a degree of diversity and equity. This proposal sparked a public debate regarding the extent to which the new design might infringe upon the principle of equal opportunity and whether it would genuinely improve overall student performance.

Motivated by this ongoing debate, the theoretical part of the project will focus on the analysis and design of cohort partition policies that optimally balance performance and equity. Note that the arising trade-off between equity and performance in group partitions is general and can be prevalent in settings both within education and beyond. For example, professional schools often divide students into smaller sections; in organizational contexts, such a trade-off may also arise in team formation problems. In both cases, the chosen partition may have potentially substantial implications on the individual achievement and homogeneity of outcomes.

For the empirical part of the project, as a first step, we aim to understand how cohort composition impacts the individual student's performance through peer effects and other mechanisms. The research assistant will help us identify and test potential models of how cohort assignment and class composition influence individual student performance throughout high school. Leveraging this dataset alongside additional admissions data from Greece, the research assistant may also aid in exploring other related questions regarding educational equity.