Impact of instructional time not equal across studentsSeth Gershenson
Brookings
September 24, 2015
Michael S. Hayes and I investigate the distributional impact of instructional time on achievement gains in a recent IZA Discussion Paper. We do so by estimating the impact of instructional days on various quantiles of the achievement distribution using the same nationally representative data on kindergarteners. We generally find that high-achieving students benefit more from increased instructional time than do lower-achieving students, which is consistent with a Matthew Effect of formal schooling. Figure 1 depicts these results for math achievement, plotting the effect of one school year (250 calendar days) on each percentile of the math achievement distribution (measured in test score SD). While the average effect of a little more than one SD is, of course, constant across the distribution, the quantile effects are monotonically increasing. Moreover, the effect on the 10th percentile of about 0.75 SD is significantly lower than the average effect, while the effect on the 90th percentile of about 1.75 SD is significantly larger than the average effect. The patterns are similar for reading achievement. Interestingly, and consistent with previous research, we find no evidence that the impact of instructional time varies by observed socio-demographic characteristics of the students.
Spending more money on educating the smartest children is efficient, since they tend to be the most educable. Educational systems in the Western world often do the opposite of this, especially at the pre-college level.
