A Bayesian Model for Estimating Sustainable Development Goal Indicator 4.1.2: School Completion Rates

Abstract

Estimating school completion is crucial for monitoring SDG 4 on education. The recently introduced SDG indicator 4.1.2, defined as the percentage of children aged 3-5 years above the expected completion age of a given level of education that have completed the respective level, differs from enrolment indicators in that it relies primarily on household surveys. This introduces a number of challenges including gaps between survey waves, conflicting estimates, age misreporting, and delayed completion. Our Adjusted Bayesian Completion Rates (ABC) model addresses these challenges to produce the first complete and consistent time series for SDG indicator 4.1.2, by school level and sex, for 157 countries. The ABC model estimates unobserved true completion rates using a latent ARIMA(1,1,0) with drift process. The model adjusts observations for late completion and age misreporting effects, and also accounts for survey level differences in bias and non-sampling variance. Validation exercises indicate that the model appears well-calibrated and offers a meaningful improvement over simpler approaches in predictive performance.

A Bayesian cohort model for estimating out-of-school rates and populations

“…In contrast, school completion rates, a related education indicator also produced by household surveys, were found to suffer from material levels of structural bias (Dharamshi et al 2022). This suggests that survey bias must be considered in modelling out-of-school rates.…”

“…In this paper, we introduce a new Bayesian hierarchical cohort-based model that fills this need and estimates out-of-school rates for all countries with available data. A similar Bayesian hierarchical model estimated school completion rates in Dharamshi et al (2022). The proposed out-of-school rate model estimates latent out-of-school rates using a spline-based cohort setup that mirrors the natural progression of students through a school cycle.…”

“…Survey-based estimates of education indicators, however, are expected to suffer from some amount of survey bias and are subject to sampling and non-sampling variances (Dharamshi et al 2022). The sampling and non-sampling variances are estimated in the real space, then transformed to the probit space using a delta method approximation that uses the observed location of S j to adjust for the probit curvature.…”

“…This is a recurrent challenge in education modelling projects. However, as we noted in Dharamshi et al (2022), simple linear models are applied in practice for SDG 4 indicator forecasting (EFA Global Monitoring Report 2015; UNESCO Institute for Statistics, Global Education Monitoring Report Team 2019). As such, we consider a 'Simple' Bayesian model where the out-of-school rate is modelled linearly in the probit space as a function of year and age.…”

A positive consequence of the COVID-19 pandemic: how the counterfactual experience of school closures is accelerating a multisectoral response to the treatment of neglected tropical diseases

“…MDA campaigns for STH and schistosomiasis are largely conducted through schools, reflecting the feasibility and cost-effectiveness of using existing infrastructure to reach this cohort during some of their most vulnerable years and the near universality of children enrolled in primary school. It is important to note that while the out-of-school rate among primary school-age children has halved globally from 19% in 2000 to 9% in 2020, universal enrolment has not yet been achieved [62]. According to a new cohort-based estimation model, the out-of-school rate in low-income countries only fell from 26% to 20% over this period.…”