Estimating Test-Score Growth for Schools and Districts With a Gap Year in the Data

Oct 15, 2026

Uploaded by Adam

Evaluating whether test-score growth can be estimated when one year is missing from testing data, the study examines schools and districts facing a gap year created by spring 2020 COVID-19 disruptions. Using a simulated gap year approach, it censors an otherwise available testing year to compare growth estimates from gap-year data against full-data results. Findings show generally high similarity, enabling useful growth measures even with a gap year under specific conditions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages14

Reading Time35 min

AccessFree to download

Estimating Test-Score Growth With a Gap Year

Motivation and Context

Method: Simulating a Gap Year

Focus on District- and School-Level Estimates

Results and Model Comparisons

Interpreting Differences in Estimates

AERA Open January-December 2021, Vol. 7, No. 1, pp. 1–14 DOI: 10. 1177/23328584211041346 Article reuse guidelines: [sagepub.com/journals-permissions](sagepub.com/journals-permissions)[ ](sagepub.com/journals-permissions)© The Author(s) 2021. [https://journals.sagepub.com/home/ero](https://journals.sagepub.com/home/ero) Estimating Test-Score Growth for Schools and Districts With a Gap Year in the Data Ishtiaque Fazlul Cory Koedel Eric Parsons Cheng Qian University of Missouri We evaluate the feasibility of estimating test-score growth for schools and districts with a gap year in test data. Our research design uses a simulated gap year in testing when a true test gap did not occur, which facilitates comparisons of district-and school-level growth estimates with and without a gap year. We find that growth estimates based on the full data and gap-year data are generally similar, establishing that useful growth measures can be constructed with a gap year in test data. Our findings apply most directly to testing disruptions that occur in the absence of other disruptions to the school system. They also provide insights about the test stoppage induced by COVID-19, although our work is just a first step toward producing informative school-and district-level growth measures from the pandemic period. Keywords: assessment, educational policy, evaluation, policy analysis, regression analyses, research methodology, testing TesT-score growth is a commonly used evaluation tool in education research and policy applications. The abrupt cancellation of testing in spring 2020 due to COVID-19 generated a gap year in test data in states across the United States, a consequence of which is that it will not be possible to estimate traditional growth models in 2021. Motivated by this data condition, we assess the potential for reliably estimating testscore growth over a 2-year period with a gap year in testing.1 Our methodological approach is to simulate a gap year in testing in a year preceding COVID-19 . Specifically, we build a data panel spanning the school years 2016–2017, 2017–2018, and 2018–2019, and censor the data as if the 2017–2018 test was never administered. We estimate models of student test-score growth using the artificially censored data and compare the output with analogous output obtained using the full, uncensored data panel over the same 2-year period. These comparisons allow us to assess the accuracy of gap-year growth estimates relative to the full-data condition. Our simulations are not confounded by complicating factors associated with the test gap that occurred in reality due to the COVID-19 pandemic. This is appealing from the perspective of understanding the prospects for gap-year growth modeling in the absence of other complications. In the context of the pandemic—during which there have been many complications—our work is best viewed as providing evidence on a necessary, but not sufficient, condition for the resumption of useful growth modeling in spring 2021. We focus primarily on determining the accuracy with which we can estimate test-score growth for districts and schools. Districts and schools are natural units of analysis from the perspective of state education agency staff interested in understanding variability in learning rates within their states. Moreover, although high-stakes growth measures from the pandemic period are unlikely to be used for accountability purposes, the historical use of district- and school-level growth estimates in this way has created inertia around metrics at these levels within state education agencies.2 From a research perspective, growth-based analyses atthe district and school levels are also commonly used to evaluate education interventions. In our simulations, we find that gap-year models produce estimates of growth at the district and school levels that are highly correlated with estimates that use all the data. Specifically, the correlations are consi

FAQs

Why does the study focus on a gap year in test data?

The abrupt cancellation of testing in spring 2020 created a year missing from test data, preventing use of traditional growth models in 2021.

How is the gap year modeled in the research design?

The approach simulates a gap year by censoring test data as if a specific testing year was never administered, then comparing model outputs with full uncensored data.

What do the correlations between gap-year and full-data estimates show?

District-level growth estimates are highly correlated with full-data estimates (around 0.90), while school-level correlations range roughly from 0.84 to 0.88 across multiple specifications.

Oct 15, 2026

Uploaded by Adam