Prioritization within K-12 School Districts: Parents, Policy Categories, and Predictive Algorithms

Rebecca Johnson

A growing body of research investigates how algorithms impact inequality across several domains, including lending, hiring, and criminal justice. But less is known about the rise of predictive algorithms in K-12 schools; most research focuses on algorithms for surveillance, such as facial recognition software or automated exam proctoring, rather than algorithms that try to predict which students need help most urgently to target supportive resources. This project focuses on a particular K-12 resource shortage: how districts make decisions about which students to prioritize for intensive ?high-dosage? tutors in the wake of the COVID-19 pandemic. The project will combine three methodologies. First is a vignette-based survey experiment fielded in fall of 2021 with a nationally-representative sample of U.S. residents and an oversample of current K-12 parents (N = 5,600 respondents; N = 2,670 parents), which finds that both parents and non-parents perceive predictive algorithms as a significantly fairer way to ascertain student need than existing prioritization methods (e.g., counselor discretion; parent requests; legal categories). The Spencer/NAEd extension delves into the parents? qualitative responses to investigate how past racialized and gendered experiences with schools shape fairness perceptions. Second is a simulation-based study to explore which students are identified as ?needing help? using algorithms versus older methods. Third are qualitative interviews with leadership in K-12 districts about how they selected a prioritization method for high-dosage tutors. The project uses a case of post COVID-19 triage to explore longstanding tensions over which stakeholders?parents; school professionals; judges?ought to influence the distribution of school resources.

About Rebecca Johnson

Rebecca Johnson is an Assistant Professor in Georgetown?s McCourt School of Public Policy, where she teaches computational social science and is affiliated with the Massive Data Institute (MDI). Her research focuses on how K-12 school districts use a mix of data and discretion to decide which students need help most urgently, combining computational methods (field experiments; quasi-experimental designs; natural language processing) with normative analysis. Her early career work, funded by the American Bar Foundation (ABF)/JPB Foundation Access to Justice Scholars program, focused on how school districts with tight budgets struggle with two types of legal pressure: weakly-funded legislative mandates that direct them to prioritize certain groups of students and parents? due process claims pressing for priority for their children. Pilot work for the Spencer/NAEd project, funded by the NSF TESS Program and Dartmouth?s Neukom Institution for Computational Science, investigates shifts from districts using legal categories to prioritize to districts using predictive models/algorithms. Previously, she received her BA/MA from Stanford University, her PhD in Sociology and Social Policy from Princeton University, and served as an Assistant Professor in Dartmouth College?s Program in Quantitative Social Science.