, publishes student scores by eligibility for subsidized meals. Under the federal No Child Left Behind Act and its successor, the Every Student Succeeds Act, districts have reported scores separately for disadvantaged children, with eligibility for subsidized meals serving as the standard measure of disadvantage.
With Katherine Michelmore, a postdoctoral researcher at the University of Michigan, I have analyzed data held by the Michigan Consortium for Educational Research and found that this measure substantially understates the achievement gap.
In Michigan, as in the rest of the country, about half of eighth graders in public schools receive a free or reduced-price lunch. But when we look more closely, we see that just 14 percent have been eligible for subsidized meals every year since kindergarten. These children are the poorest of the poor — the persistently disadvantaged.
The math scores of these poorest children are far lower than predicted by the standard measure of economic disadvantage. The achievement gap between persistently disadvantaged children and those who were never disadvantaged is about a third larger than the gap that is typically measured.
Education researchers often express test score differences in standard deviations, which allows for a consistent measure of gaps across different tests, populations and contexts. Measured using that conventional approach, the gap in math scores between disadvantaged eighth graders and their classmates in Michigan is 0.69 standard deviations. This places disadvantaged children roughly two grades behind their classmates. By contrast, the gap based on persistent disadvantage is much wider: 0.94 standard deviations, or nearly three grades of learning.

Why American Schools Are Even More Unequal Than We Thought

By SUSAN DYNARSKI