Making Sense of Uncertainty in Academic Screening Prediction Accuracy for Ordinal Outcomes using Bayesian Modeling
This project will advance current methods of estimating academic universal screener accuracy for ordinal outcomes when using Bayesian statistical methods. Using these methods, this work will advance research in universal screening accuracy in the general population as well as among English learners (ELs). In addition, this work will use Bayesian methods (specifically, using prior assumptions and information about screening accuracy to build Bayesian priors for screening models) to better understand how screener accuracy has changed since the onset of the international COVID-19 pandemic for ELs and the general population of students. In this work, I will use Bayesian methods (Gelman et al, 2020; Kaplan, 2014) to understand how different models and prior formulations capture screening accuracy when considering linguistic diversity and the COVID-19 pandemic. The intended outcome of this work is to advance the connection of complex methodological issues with universal screening practices in schools to better understand how to identify students in need of additional academic supports. This work will attend specifically to math screening issues surrounding ELs and their developing English language proficiency (ELP) as well as academic screening issues brought on by the COVID-19 pandemic, such as the pandemic’s influence on academic achievement.