New York City’s school-choice system is designed to turn thousands of individual preferences and limited school seats into orderly assignments. Yet a new study suggests that the outcome is shaped not only by which schools admit students, but also by how families rank their options. Analyzing application data from 58,901 students, researchers found that Black and Hispanic students were substantially more likely than Asian and white students to end up at schools below the level they might realistically have obtained. This gap was especially pronounced among students with highly competitive applications, revealing how small differences in application strategy can translate into major differences in educational opportunity.
The study, published in Nature Cities, examines a problem known as undermatching. In school-choice research, undermatching occurs when a student is assigned to a school of lower quality or selectivity than another school for which the student could have been eligible. The concept is different from simple rejection. A student may have qualifications strong enough for a more competitive program but fail to receive it because of the way their preferences were submitted, the order in which schools were listed, or the interaction between their choices and the city’s admissions rules. In a centralized assignment system, a preference list is not merely a statement of desire; it is an input into an algorithm that determines how scarce seats are distributed.
New York City provides a particularly important setting for studying this process because its high school admissions system combines centralized matching with student-submitted rankings. Applicants can list schools in order of preference, while each school applies its own admissions criteria within the rules established by the city. The assignment mechanism then processes applicants and available seats, attempting to give students their highest-ranked feasible option. Because every student’s list affects the opportunities available to others, the system creates a complex strategic environment. Families must decide not only which schools they like, but also how to position ambitious choices, realistic options and safer alternatives.
That decision-making burden can be difficult to manage even for families with extensive information and time. Applicants may not know how competitive a particular program will be, how many seats it has, how other students will rank it or how their own academic profile compares with the likely applicant pool. They may also avoid placing a highly desirable but difficult-to-enter school near the top of their list, believing that doing so could jeopardize their chances elsewhere. Such concerns can lead families to submit conservative rankings. The researchers describe this as an administrative burden: the effort required to understand, navigate and optimize participation in a complicated public system.
The new analysis was designed to separate the effects of admissions policies from the effects of ranking behavior. Those two forces are often intertwined. A school may use criteria that advantage some groups over others, while applicants may respond to the system in ways that reflect unequal access to guidance and information. By modeling what the researchers considered an optimal application strategy, and by calculating admissions counterfactuals, the study estimates what might have happened if students had made different choices under the same underlying admissions conditions. This approach allows the researchers to ask whether a student’s outcome was constrained primarily by eligibility and competition, or whether the submitted preference list itself helped produce a weaker match.
The results point to a pronounced disparity. Black and Hispanic students undermatched at 1.5 times the rate of their Asian and white peers. Among the most competitive students, the difference grew to roughly twofold. That finding is significant because it suggests that disparities cannot be explained solely by a lack of qualifications or by unequal access to the most selective programs. Even students with strong applications may lose opportunities when they navigate the ranking process less effectively. In practical terms, a student who could plausibly have reached a highly sought-after school may instead be assigned to a less-preferred option because the list did not place an ambitious choice where it could have been useful.
The researchers also identified a potential intervention with striking theoretical power. Within their framework, changing a single position on an applicant’s list, using information available before the assignment process, could reduce undermatching by 24%. The intervention would not require rewriting the entire admissions system or changing every school’s criteria. Instead, it would focus on the point where individual strategy meets the matching algorithm. A student with a competitive application could be advised to place an appropriate “reach” program higher on the list, rather than relying exclusively on options perceived as safe. If the model accurately identifies cases in which an applicant is unnecessarily cautious, a targeted adjustment could improve the student’s final placement.
The word “targeted” is central to the study’s implications. The researchers do not suggest that every applicant should simply rank more selective schools. A universal campaign encouraging families to add ambitious programs could produce unintended consequences. If many students simultaneously moved reach schools upward, competition for those seats would intensify. Some applicants who might otherwise have secured a place at a less selective program could instead remain unmatched, particularly if they fail to include enough realistic alternatives. In a centralized matching market, a strategy that benefits one student in isolation may have different effects when adopted widely. The system’s limited capacity means that behavioral advice must account for responses by all participants, not just the individual receiving it.
The findings therefore point toward a new generation of interventions that combine algorithmic analysis with personalized guidance. Rather than offering generic advice to all families, a city could identify disadvantaged students whose applications appear competitive but whose rankings are unusually conservative. Those students might receive clear, individualized information about how a revised list could affect their chances, including warnings about the importance of retaining fallback options. Such support could reduce the role of unequal access to counselors, informal networks and specialized admissions knowledge. At the same time, the analysis underscores that information alone cannot eliminate structural inequities embedded in school capacity, admissions rules or differences in family resources.
The New York City study ultimately reframes school choice as both a policy problem and an information problem. Admissions systems determine who can compete for desirable seats, but application behavior determines how those opportunities are pursued. The researchers’ results show that an apparently minor decision—a single position on a preference list—can have consequences large enough to widen group-level disparities. They also demonstrate why reform must be tested carefully: an intervention that improves assignments for targeted students may create new risks if applied indiscriminately. As cities increasingly rely on centralized algorithms to allocate public resources, the challenge will be to make these systems not only efficient, but understandable, strategically fair and responsive to the families least equipped to navigate them.
Subject of Research: Centralized school assignment, application-ranking behavior and educational inequities in New York City.
Article Title: Connecting application behavior to undermatching in New York City school choice
Article References: Peng, K., Ryu, E., Kleinberg, J. et al. Connecting application behavior to undermatching in New York City school choice. Nat Cities (2026). https://doi.org/10.1038/s44284-026-00486-0
Image Credits: AI Generated
DOI: https://doi.org/10.1038/s44284-026-00486-0
Keywords: New York City school choice, centralized school assignment, undermatching, admissions algorithms, application behavior, administrative burden, educational inequality, student preferences, school admissions, matching markets
