A minute with...

Thad Domina

Editor's note: Thurston “Thad” Domina, associate professor of educational policy and sociology at Carolina’s School of Education, was the lead author of an article published in Educational Researcher that raises questions about a leading measure of socio-economic status of school children: Free and reduced-price lunches.

Thurston Domina, Nikolas Pharris-Ciurej, Andrew M. Penner, Emily K. Penner, Quentin Brummet, Sonya R. Porter, and Tanya Sanabria. Is Free and Reduced-Price Lunch a Valid Measure of Educational Disadvantage. Educational Researcher. Sept. 6, 2018. https://doi.org/10.3102/0013189X18797609.

Why was it important to do this study?

The National School Lunch Program was designed to provide healthy meals to school children whose parents may not be able to afford them. But over the years, the program has taken on a totally unrelated function – as a source of data about socioeconomic disadvantage in American schools. Children must report that their families’ total household income is less than 1.85 times the federal poverty line in order to receive free or reduced-price lunch.

Since we don't routinely collect other data about kids' family backgrounds, free or reduced-price lunch data often the best measure of kid background we have. As a result, the measure is widely used, both in educational research and educational policy.

As we started researching this paper, my colleagues and I found, for example, that 70 percent of the recent articles in the journal Educational Researcher that reported some measure of student socioeconomic background used free or reduced-price lunch. What's more, the measure is hugely important in school finance and school accountability policy. For example, schools qualify for many federal and state-level funding streams based on the proportion of their students who qualify for free lunch.

The problem is, nobody really knows how well free lunch data track on to other measures of student socioeconomic disadvantage. Our study tries to solve that problem by matching kids' school data with their families' IRS income tax filings.

What were the main things you found?

My colleagues at the U.S. Census and the University of California, Irvine and I learned three main things:

  1. The correlation between free and reduced-price lunch status and children’s IRS-reported family income is weak. Many kids from very low-income families do not enroll in free and reduced-price lunch and many kids from relatively high-income families do enroll in the program.
  2. At the school level, the relationship between the proportion of kids enrolled in free or reduced-price lunch and other measures of school-level socioeconomic disadvantage (like the school-level poverty rate) is also very weak.
  3. Free lunch is negatively related to student achievement, even after controlling for their IRS-reported household income and a host of other factors.

The first two findings raise real questions about the extent to which free or reduced-price lunch data proxy for socioeconomic disadvantage.

The third finding, however, suggests that these data are capturing something about students' educational disadvantage that our ostensibly better measures of family income don't capture. We're still puzzling over this third finding, but I suspect that kids from low-income families that have access to other resources might not enroll in the program, while schools may find ways to enroll kids from families with highly volatile household incomes or other challenges.

Why is this important for other educational researchers?

We all know that students' family backgrounds have huge consequences for students' educational experiences, and basically all educational researchers try to account for those consequences as we study what happens inside the schoolhouse. Free lunch data is a handy and widely-used tool for doing so, but our study suggests that it's of questionable reliability. More broadly, I think our study raises important questions about how we conceptualize and understand student disadvantage and its role in educational processes.

What other research questions need to be examined as a result of your findings?

In order to understand schools and students' educational experiences, we need to understand the contexts that students experience outside of school.

For me, our study is a reminder of what a poor job we've done at understanding and measuring those out-of-school contexts. The measurement problems that we're highlighting have huge consequences, and not just for educational research.

For example, when policy-makers measure school or teacher quality, they typically use free and reduced-price lunch data to account for differences in the learning resources available to kids from poor and non-poor families. I think this study raises big questions about the accuracy of these models and how fair it is to use them in school accountability systems.

Furthermore, these measurement problems are likely getting worse over time. Thanks to recent changes in free lunch regulations that are designed to help schools more easily provide meals to students who need them, I suspect the relationship between free lunch and student family background is weakening.

While these changes are wonderful for making sure that kids who need lunch are getting it, it means that we as researchers and policymakers shouldn't be trying to build research and policy on the back of this measure. If we're serious about making sure our schools work for students from across the socioeconomic spectrum, I think it's time for a serious national effort to measure and understand student family background.

Editor's note: You may also listen to Domina discuss his findings, and their ramifications, in this podcast produced by the Consortium for Policy Research in Education.