The existing literature on the effect of the COVID-19 pandemic on elementary and secondary school learning losses finds that students have experienced greater learning losses than would be expected in a normal year, but the specific cause of these losses remains an open question. Researchers have demonstrated the existence of the learning losses among American students and explored the moderating effect of remote learning at the district level. In this study, I complement prior analysis by using the SafeGraph phone-tracking database as a proxy for school openness. This provides two advantages: first, a more precise estimation of openness, without having to resort to using the nebulous term "hybrid" to encapsulate all states between open and remote learning, and secondly, allowing for a school-level analysis, which is more granular than the district level.
The result of my study is a group of linear regression analyses of learning loss regressed on school openness with demographic, fiscal, and state fixed effects of approximately 7,500 schools in 10 states. The results of the analyses indicate that schools which were more open during the 2020-21 school year experienced significantly smaller learning losses in math, but that openness did not have a significant effect on English language arts learning loss when controls were included. Additionally, there were significant racial, economic, and state fixed effects, with schools that have higher percentages of black and Hispanic students, as well as higher percentages of students eligible for free and reduced-price lunch, experiencing larger learning losses. These results are consistent with prior literature suggesting that math skills suffer more than reading during periods in which school is not in session, and that racial minorities and students of low socioeconomic status are disproportionately harmed by interruptions to learning.