Mohmmad Nasim Wafa
Herat University

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

Bayesian hierarchical modeling of factors associated with students’ educational status in less developed areas: A case study of Ghor Province, Afghanistan Mohmmad Nasim Wafa
Polyhedron International Journal in Mathematics Education Vol. 4 No. 2 (2026): pijme
Publisher : Nashir Al-Kutub Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.59965/pijme.v4i2.318

Abstract

Students’ educational status in remote, data-scarce regions is shaped by several interrelated factors, yet evidence that quantifies these associations together with their uncertainty is scarce. This study examined factors associated with students’ educational status in less developed areas of Ghor Province, Afghanistan, using Bayesian regression and Bayesian hierarchical modeling. A cross-sectional correlational survey was conducted in 2019 with 372 students in Grades 11 and 12 from 32 schools in the Firozkoh, Dawlatyar, and Lal wa Sarjangal districts. A 35-item questionnaire measured economic status and family support, educational facilities and infrastructure, teacher quality and teaching process, geographical accessibility, environmental conditions, and perceived educational status. A single-level Bayesian regression and a Bayesian hierarchical model with school-specific random intercepts and district indicators were fitted, and inference was based on posterior distributions, 95% credible intervals, and Markov chain Monte Carlo diagnostics. All five factors were positively associated with educational status; teacher quality and teaching process showed the largest posterior mean coefficient (0.31; 95% credible interval [0.21, 0.41] in the single-level model), followed by educational facilities and infrastructure (0.26). The hierarchical model indicated between-school heterogeneity (posterior mean between-school standard deviation = 0.28) and showed better expected predictive performance than the single-level model (LOOIC lower by 36.8 points). Because the design was cross-sectional and the sample was a convenience sample, the findings indicate associations rather than causal effects. They nevertheless suggest that teacher quality, school infrastructure, and household, geographical, and environmental conditions deserve attention in educational planning for such regions.