Abstract

This study aimed (1) to develop and empirically test a model of factors influencing educational inequality within the Songkhla Education Sandbox and (2) to identify strategic approaches for reducing such inequality. A mixed-methods design was employed. In Phase 1, structural equation modeling was conducted using data from 343 teachers selected through stratified random sampling. The validated model comprised five latent factors: educational inequality, learner-related factors, family-related factors, educational personnel factors, and institutional factors. Model fit indices indicated strong fit (χ² = 491.109, df = 129, CFI = 0.996, TLI = 0.995, RMSEA = 0.020, SRMR = 0.035). All observed variables significantly loaded onto their respective constructs at the 0.01 level, and the model demonstrated acceptable reliability and validity. Overall, the model explained 92.6% of the variance in educational inequality, with learner-related factors exerting the greatest influence (69.0%).

Phase 2 employed focus group interviews with sixteen education experts. Findings highlighted the need to prioritize interventions targeting learner-related factors, followed by improvements in educational personnel and institutional factors. Consistent parental support and active student engagement were deemed essential. High-quality teaching, teacher commitment to instructional development, and reduced reliance on private tutoring were also emphasized. Additional recommendations included expanding local employment opportunities to strengthen families’ financial capacity, adopting diverse and student-centered assessment and instructional approaches, ensuring equitable teacher distribution, and establishing sub-district educational offices to enhance service accessibility in remote areas. Finally, adopting a competency-based curriculum and setting shared objectives across schools were suggested as key strategies to reduce educational disparities within the Songkhla Education Sandbox.

This work is licensed under a Creative Commons Attribution 4.0 License.