Whether a student receives a free or reduced-price lunch has long been used by policymakers and researchers as an indicator of whether she is from a low-income family. But recent policy changes aimed at increasing the number of students who automatically receive free lunch and reducing administrative burden have eroded the usefulness of the free lunch measure and have led states to explore alternatives. In newly released national data, we get a first glimpse into how states are adapting their methods to get an accurate count of low-income students.

Since 1946, the National School Lunch Program (NSLP) has provided school meals for free or at a reduced cost for children from low-income families. Eligibility for this program is typically determined through a paper-based application, but states have strengthened their ability to “directly certify” students for free lunch on the basis of participation in other social safety net programs, such as the Supplemental Nutrition Assistance Program (SNAP, formerly the Food Stamp Program).

In 2010, Congress enacted the Community Eligibility Provision (CEP), allowing districts and schools with high shares of students identified as low income through direct certification to provide all enrolled students free lunch. This provision expands eligibility and reduces the stigma of free lunch, but it also prevents administrators from getting an application-based count of program eligibility.

The number of students eligible for free or reduced-price lunch has long been used by researchers and policymakers as a proxy for the count of low-income students—a critical number for policymakers trying to target funding to low-income schools or researchers trying to determine a program’s effectiveness—but the utility of that measure has declined as community eligibility has spread.

Last week, the National Center for Education Statistics released new data on the number of students eligible for participation in the NSLP in the 2016–17 school year. This was the first year that schools could report the number of students who were directly certified for free lunch. States could report these numbers in place of, or in addition to, data on the number of students deemed eligible for free and reduced-price lunch through traditional methods (typically, through annual applications).

Schools in Delaware, the District of Columbia, Massachusetts, and Tennessee reported counts of directly certified students only, while schools in 27 states reported free and reduced-price lunch counts only. In 17 states, schools reported both measures, while 3 states had schools that reported either one measure or the other (a “mix”).

Although the free- and reduced-price lunch measure has eroded, the direct certification reporting alternative has introduced new concerns for researchers about cross-state and cross-time comparability of student poverty measures. Compared with the share of students who are eligible for free and reduced-price lunch, direct certification tends to result in smaller shares of students identified as low income.

The new data allow us to see this contrast. In Washington State, one of the states that reported free lunch counts and the direct certification rate, the share of students eligible for free lunch is an average of 12 percentage points higher than the share who are directly certified. In some cases, there is as much as a 30 percentage-point gap between the measures.

The difference between the free lunch measure and the direct certification measure is also a challenge for policymakers. Policymakers must help the public understand how these measures provide valuable, but potentially different, information about students’ needs within a school. States use measures of low-income status for school funding and for accountability, so states must decide which measure to use when quantifying student need and how to implement the transition to direct certification, if they decide to do so.

These new national data on free lunch and direct certification counts reveal how states measure student poverty within their schools and how these measurements align or differ. In particular, researchers can further take up the question of how substantially direct certification varies from the previously used free lunch measure by student demographics, poverty concentration, or CEP participation.

These data are a great first step in helping researchers, policymakers, and the public grow more familiar with the direct certification measure of student poverty and ask tough questions about how best to count and serve students from low-income families.