Who’s Hungry? Making Sense of Campus Food Insecurity Estimates

This is a blog jointly written by a group of researchers (see list at the end).

Students are hungry. The newspapers are full of reports documenting a problem with food insecurity and yes, hunger, among college students. A recent New York Times op-ed by Sara Goldrick-Rab declared that an estimated 50% of undergraduates nationally are dealing with this challenge.

A serious problem, to be sure. But is the problem really that widespread, and how can we be sure?

As scientists deeply concerned about supporting students both with programs and policies and with data and evidence, we think these are critically important questions. The truth is, they are very hard to answer. But we are doing the best we can with the tools we have. At the same time, we face some difficulties to creating improvements that the field deserves. Here, in the interest of transparency, and reflecting our commitment to getting this right, we lay out some of the dilemmas.

What is food insecurity? What is hunger? This is not a new issue, but it is especially challenging given the many stereotypes of college students and what they eat. Some readers of our studies suggest that students are not hungry, just picky. Others think they skip meals to lose weight or to have money for drinking and going out. Still others think that any challenges that occur only happen during breaks from school. We try to deal with these questions in a few ways.

1. We are basing estimates on studies that use the USDA food insecurity module for assessing food insecurity in the United States. It is currently the most valid measure of food insecurity among Americans, and around the world. The Wisconsin HOPE Lab has used the 6-item module in the past, but recently switched to the 10-item module for prevalence studies and the 18-item module in studies of interventions. The University of California team used the 6-item in 2015 and the 2-item in 2016 and reached similar estimates. The California State University team uses the 10-item, while the City University of New York team used four items from the USDA module. In other cases, published reports do not appear to use a validated measure of food insecurity, just asking students if they ever find themselves without enough to eat.
2. Validating the USDA scale with college students is also a priority. We have conducted focus groups that suggest the scales are appropriate, but we would like to conduct a more intensive mixed-methods study to learn more about how students manage food, what they eat, and where, and how they fall short.
3. Contextualizing the prevalence estimates with other data on the timing of students’ experiences and the programs that support them is also important. Some of us are experimenting with assessing food insecurity over the past 30 days versus the last year, for example. We are also comparing the estimates to the use of meal plans, SNAP, and the like.

Which students should be examined? All studies of college food insecurity are done with surveys and therefore decisions must be made about which students to include in those surveys. The best surveys have a sample representative of the population they seek to describe and response rates high enough to generate confidence that the estimates based on the sample are accurate. This is easier said than done. Generating a sampling frame in order to select some rather than all students from a college requires access to the necessary data and time to do the analytic work. Laws like FERPA and Institutional Review Board processes mean that such efforts are costly and often impossible, especially if the survey is meant to include multiple institutions. This is why good national surveys of college students cost tens of millions of dollars. We are unaware of any researcher in the field of college food insecurity with such a budget. Rather, our surveys are usually done with tens of thousands of dollars, if that. This is why, for example, the Wisconsin HOPE Lab surveys the entire population of students at each participating college. It is why some surveys, like the Hunger on Campus report issued by CUFBA and colleagues, recruit students by explaining they are trying to understand food insecurity on campus — thus likely gaining their attention but also priming their answers. And it is why most studies of food insecurity focus on a single college or a single system, where researchers have negotiated access and have access to institutional review boards that will coordinate human subjects approvals across campuses.

There is only one published report, by the Urban Institute, that claims to include a “nationally representative sample of students” in its estimates of food insecurity. It is also the only report to date to make claims about trends over time, facilitated by those “nationally representative” samples. While it would be very exciting indeed to have information obtained from such a good sample, it is unfortunately not the case that the Urban Institute study provides it. There are very few nationally representative studies of college students and none of them include measures of food insecurity. The study that the Urban Institute employed a subsample of a nationally representative survey of households, not students specifically, and the two are very different.

Of course, one benefit of a survey like the one the Urban Institute utilized is that it has higher response rates. This is primarily because large national surveys are extremely well-funded and can offer monetary incentives to individuals in order to gain participation. But all surveys, including those with high response rates, may produce biased estimates if the people who do not respond are systematically different from those who do. In the case of assessing food insecurity, for example, we might expect students who do not have enough to eat to be more or less likely to do a survey. The expected direction of the bias is unknown. However, we can make some educated guesses about it. For example:

• If a survey is administered electronically and can only be taken on a laptop or cell phone, food-insecure students may be less likely (than the average student) to take it, since they have less money for such items. The results would probably under-state the incidence of food insecurity.
• If a survey is administered outside of a food pantry, food-insecure students are probably more likely to take it, since they are more likely to visit the food pantry. The results would probably over-state the incidence of food insecurity.
• If a survey’s approach reduces participation among housing-insecure or homeless college students, then given the correlation between food and housing insecurity, food insecure students are less likely to complete it. The results would probably under-state the incidence of food insecurity.

And so on. It is also possible that some surveys could attract disproportionate number of responses from food-insecure students. Ideally, comparisons between the sample in the survey and the population meant to be described, combined with a reported response rate, would be sufficient for assessing likely bias. But they are not. There are too many factors that can’t be observed but could explain variation in the results. The best solutions, therefore, are to be transparent about sample characteristics and response rates, seek to do better whenever funds make it possible to employ representative samples and use incentives, and attempt to do non-response studies (which also cost money). This is why we have published (and updated) guides to measuring food insecurity, and offer methodological appendices with our reports. In addition, it is why some of us proposed — back in 2015 — that the federal government add a food security module to a nationally representative study of college students.

Summing up

We have each reported a range of results in our studies, and so have our colleagues. The preponderance of existing evidence from studies of college students justifies the estimate of 50% food insecurity among undergraduates. Nearly half of all undergraduates are enrolled in community colleges, and estimates suggest that more than half of them deal with food insecurity. We can reasonably anticipate, given their similar characteristics, that students at for-profit colleges do as well. Moreover, estimates at public four-year universities — including some of the largest in the nation — also report food insecurity rates upwards of 40%, with some exceptions. Of course estimates are lower at flagships and private institutions, but it is also the case that those schools have been far more reticent to measure food insecurity on campus.

Are these numbers as firm and trustworthy as we would like them to be? It depends. A better question to ask is whether these numbers generalize to all college campuses nationwide. These are the sorts of issues that keep us plugging away, conferring and collaborating, debating, and striving to do our very best work. We trust that the field will evolve along with the science and allow evidence to drive efforts to improve the lives of all college students.

Katharine Broton, Assistant Professor of Educational Policy & Leadership Studies, University of Iowa

Clare Cady, Director of the College and University Food Bank Alliance and Director of Community Engagement for the HOPE Center, Temple University

Ruben E. Canedo, Co-Chair, University of California Basic Needs Committee

Rashida Crutchfield, Assistant Professor of Social Work, California State University, Long Beach

Nicholas Freudenberg, Distinguished Professor of Public Health and Director CUNY Urban Food Policy Institute, City University of New York

Sara Goldrick-Rab, Professor of Higher Education Policy and Sociology, Temple University

Daphne C. Hernandez, Assistant Professor of Nutrition and Obesity Studies, University of Houston

Jennifer Maguire, Assistant Professor of Social Work, Humboldt State University

Suzanna Martinez, Assistant Researcher, Nutrition Policy Institute, Division of Agriculture and Natural Resources, University of California

Aydin Nazmi, Associate Professor of Public Health Nutrition, California Polytechnic State University, San Luis Obispo, CA

Jed Richardson, Acting Director, Wisconsin HOPE Lab, University of Wisconsin-Madison