Fig. 9

Estimates of life expectancy gains from universal primary education interventions using population models with secular trends are systematically and increasingly higher than from those without secular trends. Life expectancy projections for a simulated population health intervention: the expansion of universal primary education for rural women in the year 2000. The graphs shows a “difference in differences” projection: the incremental benefit of the education intervention according to the SPOKE-I model with demographic and socioeconomic time trends, minus the incremental benefit of the education intervention according to the static model (which assumes no change in demographic or socioeconomic variables over time). The y-axis reflects the difference in incremental life expectancy benefits between the two models. The dark line reflects the median difference, and light gray lines are the 95 % confidence intervals around the result from 10,000 repeated simulations