This study examines the determinants of educational outcome in eight selected MENA countries. The complicated structure of the TIMSS data has been considered carefully during all the stages of the analysis employing plausible values and jackknife standard error technique to accommodate the measurement error of the dependant variable and the clustering of students in classes and schools. The education production functions provide broad evidence from mean and quantile analysis of very low returns to schooling; few school variables are significant and none have effects across countries and quantiles. In general, student characteristics were far more important than school factors in explaining test scores, but there was considerable variability across countries in which specific factors were significant. Strikingly, computer usage was found to influence students' performance negatively in six MENA countries. Only Turkey and Iran had a significant positive effect of computer usage on maths achievements.
educational attainment education production functions quantile regression meta-analysis MENA