Educational Performance and Gaps in Chile: is there a Coverage bias?

The widespread educational coverage combined with a stagnant student performance in Chile, within the context of increased expenditure in education, is a true puzzle. One natural hypothesis is that the increase in coverage, as it progressively includes the more vulnerable groups, explains the stabilization of quality and the wider income-related gap. We explore the existence of a sample bias and we find an overestimation of performance of around one fifth of one standard deviation of the SIMCE test when measured by the average score instead of correcting by selection. We also find that the differential between the mean scores of total population and the educated population narrows as income increases.The consequence is that the mean underestimates the income-related educational gap—as measured by the ratio between the highest and lowest income quintiles— by around 10 percentage points. Finally—and unexpectedly—, we do not find that the increased coverage has resulted in an underestimation of the fall in real gaps over time.