Issues register

EdTech Assessment Toolkit

School Choice (New York City, US)

Algorithm matching model

School choice algorithms are employed to allocate students to schools in situations where there are more students than available school places. These algorithms endeavour to assign students to schools in a fair and efficient manner, considering the preferences and priorities of both students and schools.

However, school choice algorithms can adversely affect marginalised communities. They can restrict access to high-quality public schools for families from predominantly black and low-income areas, result in unjust outcomes, give scant attention to racial diversity, and largely ignore the historical context of racial segregation and housing discrimination, which have historically shaped educational inequalities. This can perpetuate existing disparities and hinder efforts to address them (Swist & Gulson, 2023).

Snapshot (July 2023)

System task/function: School Choice Algorithm
Model: Algorithm Matching
Deployment: Deferred Acceptance Algorithm
Location of application: New York City (US)
Rationale for introduction: Assign students equitable where there are more students than seats
Vendor: NYC Department of Education.
Pricing: N/A
Data and computation: School choice, priority groups, weighting of different values
Inequalities/harms: bias in prioritised groups, bias in weight of algorithm (see Swist & Gulson, 2023)
Status: Active
Authority/regulation: School district
Unintended consequences: The algorithms procedure in weighing certain values encourages new behaviours to achieve the values preferred by the algorithm
Sanction/redress: N/A

References/further reading:

Jones, L. E. (2016). Confronting School Choice, Part I: The NAACP’s Charter Moratorium and the Backlash. Retrieved from https://www.aaihs.org/confronting-school-choice-part-i-the-naacps-charter-moratorium-and-the-backlash/

Swist, T., & Gulson, K. N. (2023). School Choice Algorithms: Data Infrastructures, Automation, and Inequality. Postdigital Science and Education, 5(1), 152-170. doi:10.1007/s42438-022-00334-z