District size affects estimates of equity in K–12 funding, but better measures might be on the way
Assessing whether low-income students receive more education funding than their high-income peers is hampered by substantial variations in the size and economic homogeneity of school districts. But new efforts at collecting school-level data could eventually give researchers a more universal measurement of school funding progressivity.
Most measurements of K–12 funding equity look at the distribution of education dollars by school district. Researchers take this approach because most education money is allocated to districts, rather than schools, and because reliable national data on revenue and expenditures are available only at the district level.
But the size and number of school districts varies substantially across states, making comparisons difficult. Maryland and Florida have some of the largest school districts, with the median district enrolling roughly 17,000 and 12,800 students, respectively. In contrast, Vermont and Montana have the smallest school districts, typically enrolling about 100 and 200 students. While Texas has more than 1,000 school districts, Nevada has only 17.
Funding equity measures change when funding is aggregated at the county level
To illustrate how the size and number of districts matter for equity analyses, I conducted a thought exercise, looking at what happens if we considered education spending at the county, rather the district, level.
County sizes are more consistent across the US than school district sizes, with less variation in the number of students (the coefficient of variation, a measure of the enrollment distribution at the county level, is 3.3, compared with 4.2 for school districts) and schools (with a coefficient of variation of 2.5 versus 3.7).
My colleague Matthew Chingos and I have developed a measure of progressivity that looks at the difference in average funding for low-income students and students who are not low income. We’ve previously noted that district size affects our measure; it is more difficult for states with large, homogenous districts to distribute dollars progressively, compared with states with more variation in student poverty by school district.
If we were to compute our measure for funding at the county level, we would estimate a decline in progressivity for most states. This decline is particularly evident in states that have many small school districts, such as Ohio and New Jersey. For states like Florida, Maryland, Nevada, and Virginia, where school district borders tend to closely mirror county lines, our measure doesn’t change much, if at all.
Under this blunter, experimental, county-level measure, most states stay on the progressive side of funding for poor versus nonpoor students. Just five states flip from regressive to progressive (or vice versa), and there’s strong correlation between our district- and county-level estimates (r = 0.85).
Other researchers have taken a different approach to measuring progressivity, using a regression to estimate the relationship between funding and student poverty at the district level within each state, controlling for such factors as enrollment size and district population density. Progressivity is calculated as the ratio of predicted funding for students in a low-poverty district versus a high-poverty district.
Similar to the weighted progressivity measure, states like Florida and Nevada see little change in their regression-based funding equity measure if measured at the county level. But for most states, the experimental county-level aggregation reduces the number of observations for each state, making the regression estimate more unpredictable. Thirteen states flip their predicted funding ratios (for example, from a ratio below 1 to a ratio above 1), and there’s a weaker correlation between the district and county estimates (r = 0.43).
Data on school-level funding could change education funding debates
My results show that district-level estimates of funding equity could be sensitive to the state’s definition of a district, as district size affects both the variation in student poverty across districts, and the number of observations available for analysis.
Fortunately, there might be new data on the horizon. Under provisions in the Every Student Succeeds Act, states are now required to report school-level, per pupil expenditures on school report cards for the 2018–19 school year (PDF). In addition, the National Center for Education Statistics is conducting a school-level reporting pilot, the School-Level Finance Survey, which will further assess whether these data can be standardized and reported across states (PDF).
These data, potentially combined with a more standardized measure of school-level poverty—such as the share of students who were directly certified for school lunch (PDF)—would offer researchers a more consistent unit of analysis—the school—and could substantially change discussions about the distribution of education dollars.
Photo by Monkey Business Images via Shutterstock