http://economics.mit.edu/files/7325Heterogeneity in High Math Achievement Across Schools: Evidence
from the American Mathematics Competitions
by Glenn Ellison, Ashley Swanson - #18277 (ED)
Abstract:
This paper explores differences in the frequency with which students
from different schools reach high levels of math achievement. Data
from the American Mathematics Competitions is used to produce counts
of high-scoring students from more than two thousand public,
coeducational, non-magnet, non-charter U.S. high schools.
High-achieving students are found to be very far from evenly
distributed. There are strong demographic predictors of high
achievement.There are also large differences among seemingly similar
schools. The unobserved heterogeneity across schools includes a
thick tail of schools that produce many more high-achieving students
than the average school. Gender-related differences and other
breakdowns are also discussed.
******************************************************
From the paper:
Our main qualitative findings are that many schools produce high-achieving students at much less than the average rate and that there is a strikingly thick tail of extremely successful schools. For example, we estimate that about 39% of schools produce high-scoring students at less than one-half of the average rate for schools with similar characteristics, and that 2% of schools produce high-scoring students at more than five times the average rate. We also provide estimates of the heterogeneity of school e ects relevant to subpopulations,
including the likelihood of producing high-achieving female students. Here, the estimates suggest that many schools are extremely unlikely to produce high-achieving girls."
...
From a policy perspective, our results suggest that the high-achieving math students we see today in U.S. high schools may be just a small fraction of the number of students who had the potential to reach such levels. This should probably not be surprising. For example, we know that the U.S. is far behind many other countries in the fraction of students who achieve very high scores on tests used in cross-country comparisons. But our finding that there is substantial variation across schools beyond the variation related to demographics could be interpreted as a hopeful one. Substantially increasing the number of high-achieving students nationwide may not be all that hard given that are already many schools that appear to produce high-achieving students at much more than the average
rate for schools with similar demographics. And the existence of upper-tail schools that are much, much more successful than average suggests that there might be programs that could be emulated to produce much larger improvements.
