Envisioning a Future That Defies Predictions: Characterizing the Differential Impact of COVID-19 Using Longitudinal Modeling

• In Event: COVID-19 Impacts: The Then, the Now, and Predictions for the Future

Abstract

We present one large, urban district’s approach to characterizing the pandemic’s impact on learning, using six years of achievement data. We detail tensions in developing the approach amid the district’s emphasis on equity and state requirements to quantify “learning loss,” emphasizing potential harms in predictive modeling for instrumental use. Employing a three-level, hierarchical linear model, we compared the gap between actual and predicted achievement before and after 2020, exploring relationships among student demographics, school composition, and achievement. We problematize calls to characterize pandemic impacts with learning-loss frameworks and emphasize the need to use predictive modeling with caution, exploring how it can be applied to justify inequities or, inversely, to catalyze systemic change that prevents unjust outcomes from occurring.

Authors

• Norma Ming, Socos Labs

• Diana Marcela Galvez Bohorquez, San Francisco Unified School District

• Allison Firestone, San Francisco Unified School District