In 2015, the United Nations adopted the Sustainable Development Goals. Regarding education, the UN identified as a key goal equitable, high-quality education, including the achievement of literacy and numeracy for all youth and adults. To assess country-level progress toward these goals, it is important to monitor trends in educational outcomes over time. The purpose of this talk is to demonstrate how optimally predictive growth models can be constructed to monitor the pace of progress at which countries are progressing toward the education sustainable development goals. Numerous models for the pace of progress can be specified. However, choosing one model and using it for predictive purposes assumes that the chosen model is the one that generated the data, and this choice runs the risk of over-confident inferences and potentially risky policy decisions. To address this problem, we adapt and apply Bayesian stacking to form mixtures of predictive distributions from an ensemble of models specified to predict country-level pace of progress. We demonstrate Bayesian stacking using country-level data from the PISA. Using common forms of forecast evaluation, our results show that Bayesian stacking yields better predictive accuracy than any single ensemble member model.

Please register in advance if you wish to attend the presentation via Zoom. After registering, you will receive a confirmation email containing information about joining the meeting.