Is 'Undermatching' Overrated?

Few educational theories have taken off as quickly in recent years as that of "undermatching." The idea is that many academically talented, low-income students who could succeed at top colleges are not applying to, enrolling in or graduating from them. Research on the topic has attracted widespread attention not only from colleges but from the White House, where administration officials have urged higher education leaders to do more on the issue.

But an analysis published Friday in Educational Researcher (abstract available here) argues that some key assumptions behind much undermatching research are flawed -- and that new studies are needed to determine how much of the theory holds. The authors are Michael N. Bastedo, director of the Center for the Study of Higher and Postsecondary Education at the University of Michigan, and Allyson Flaster, a doctoral student at Michigan.

A key part of undermatching theory is that the disadvantaged students who enroll at less competitive colleges are missing the chance at institutions with greater resources, higher graduation rates and more prestige. But Bastedo and Flaster question whether the researchers have in fact identified the "margins that matter" to student success.

They argue that the much increased opportunity that comes from attending a "top" institution is truly evident only at the very top, the wealthiest institutions that don't require students to borrow. But much of the undermatching research isn't looking at the top 50 colleges, but the top 200 or so, a group so wide that it doesn't focus on the institutions that really have exceptional resources compared to all others.

Further, the new article says that undermatching studies largely ignore a match that truly matters: whether a student enrolls at a community college or four-year institution. This choice, the authors write, is a crucial one (and perhaps far more important than whether a student attends a more or less competitive four-year institution) if the goal is to have more disadvantaged students earn bachelor's degrees because of the relatively low rates at which community college students go on to do so.

"The distinctions that matter are thus at the extremes, a fact easily mistaken when we create ladders of selectivity with seemingly equal rungs," the authors write.

A second assumption the paper questions is whether the undermatching researchers can accurately predict that students would be admitted to competitive colleges.

The paper notes that much of the research relies on patterns based on high school grades and SAT or ACT test scores. Bastedo and Flaster note that most competitive colleges engage in "holistic" admissions, which uses many factors in addition to grades and test scores. While some factors in holistic admissions (such as noting a disadvantaged background) may help disadvantaged students, others may not -- particularly those who, like many of the subjects of undermatching research, don't have much money.

They may not have access to the full range of educational and extracurricular activities that benefit wealthier applicants. And the authors note the increasing use of policies favoring "full pay" students, including out-of-state students who pay more to attend public universities -- policies that make it more difficult for disadvantaged students to get in to top colleges and universities.

"Given this complexity, undermatching researchers are simply overconfident in their ability to predict who has access to selective colleges," the authors write, questioning the assumption that if only the undermatched students applied to elite institutions, they would be admitted.

Finally, the authors question the assumption that SAT scores are a good tool to help the talented disadvantaged students get into better colleges. They write that undermatching theory assumes that if students enrolled in a meritocratic way based on SAT scores, more disadvantaged students would end up at elite institutions. But the paper argues that the SAT correlates so much with family income that this is not true -- and that pure meritocratic admissions would not close the gap for disadvantaged students.

That paper cites research done earlier by Bastedo with another co-author to "estimate college destinations for all students if they were 'perfectly matched' based on their academic achievement, including G.P.A., SAT, and curriculum rigor. They find there would be no change in the percentage of low-SES [socioeconomic status] students admitted to colleges in the top three Barron’s categories (very, highly, or most selective institutions). If perfectly matched purely on G.P.A. and SAT scores, students in the highest SES quartile would actually increase their access to the most selective colleges. This is because the majority of low-SES students attending highly selective colleges are actually overmatched to their institutions based on traditional indicators. Thus, even if low-SES students who are undermatching were to be admitted, they might simply replace low-SES students who are currently overmatching."

Caroline M. Hoxby, a professor of economics at Stanford University, and the co-author of several of the leading studies on undermatching, was highly critical of the Bastedo-Flaster analysis. Via email, Hoxby said: "Our studies are definitive. We not only study 100 percent (I said 100 percent and I am not kidding) of low-income high achievers, but we also have causal impacts (we have studies that rely on randomized controlled trials in which students are induced by our interventions to apply to more selective colleges)."

She suggested that Inside Higher Ed "simply ignore this low quality study," which she characterized as "a 1 on a scale of 1 to 10," noting that "one of the great faults of the media is to give similar weight to studies" without being able to evaluate their quality. (Hoxby is a highly respected researcher on higher education, as are some of the others who work on undermatching, but so is Bastedo, and it may be relevant that this new analysis is being published in the flagship journal of the American Educational Research Association.)

Christopher Avery, a professor of public policy at Harvard University who has written pieces with Hoxby about undermatching, said via email that the "ultimate test" of the theory would be whether interventions have an impact. If the Educational Researcher analysis is accurate, he said, then interventions wouldn't have much of an impact. But, he noted, a study by Hoxby and another co-author found that interventions do appear to work, and that evidence is "pretty compelling," he said.

Sara Goldrick-Rab, associate professor of educational policy studies and sociology at the University of Wisconsin at Madison, whose research frequently focuses on low-income students, praised the Bastedo-Flaster analysis.

She said that undermatching theory assumes that it knows what's best for low-income students. In many cases, she said, the more elite college does not in fact meet full costs (however much it claims to do so) and so some students may be making a decision entirely appropriate for them to avoid loans and related stress of going to the more prestigious college rather than a more affordable one.

At more expensive colleges, the disadvantaged student may also need to hold a job, or to work longer hours on the job, decreasing chances of "being able to be fully engaged," Goldrick-Rab said.

She also praised the Educational Researcher article for its focus on a range of limited opportunities for disadvantaged students, not just the question of whether they can go to an elite institution. She noted that there is a "capacity issue" at elite institutions, many of which are intentionally small or have limits on their ability to grow.

Assuming that the only choice that matters is whether to go to an elite college will help "the cream of the crop of low-income students," while diverting attention from the lack of resources for the schools and colleges the vast majority of low-income students attend, and are likely to continue to attend.

The focus on undermatching may well help some individual students, she said, "but it's going to be a very small number."