Learning from learning loss: Bayesian updating in academic universal screening during learning disruptions.

Area of Science:

• Educational Psychology
• Statistics
• Data Science

Background:

• The Covid-19 pandemic caused widespread learning disruptions, impacting student academic needs.
• Assessing the predictive validity of educational screening tools during such disruptions is crucial.

Purpose of the Study:

• To evaluate the predictive strength and accuracy of reading and math screening before and after Covid-19 learning disruptions.
• To investigate the impact of prolonged unplanned school closures (PUSCs) on screening effectiveness.

Main Methods:

• Employed Bayesian ordinal regression models to analyze screening data.
• Utilized a Bayesian updating procedure, incorporating previous year estimates as priors for subsequent years.
• Examined predictive performance across different grade levels.

Main Results:

• Reading and math screening demonstrated consistent predictive power both before and after Covid-19 PUSCs (odds ratios: 15-25).
• Predictive strength and accuracy varied by grade level.
• Grade-level variations in predictive accuracy were not significantly influenced by the learning disruptions.

Conclusions:

• Bayesian updating is a practical method for universal screening prediction, especially during disruptions like PUSCs.
• Educational screening tools maintain predictive validity despite significant learning disruptions.
• Findings support the continued use and adaptation of Bayesian methods in educational assessment.