A recent column that ran in The Washington Post’s Grade Point blog raised questions about an analysis I conducted that found that nearly two-thirds of selective public universities have reduced the share of students they enroll from families in the bottom 40 percent of the income scale since the late 1990s. Scholars at the American Enterprise Institute, a conservative think tank, conducted their own analysis of the data I used and found that, in the aggregate, there has been only a small decline in the enrollment of low-income students at these institutions during that time.

The conservative scholars are missing the forest for the trees. As I wrote, a substantial number of the most prestigious public universities — which historically have helped low-income and working-class students rise up the economic ladder — have become significantly less accessible over time, forcing low-income students to go to public regional colleges, as well as community colleges and for-profit colleges, that tend to have worse outcomes for students than the universities from which they are increasingly being shut out. This is a significant public policy concern that needs bold solutions — an outcome that the American Enterprise Institute scholars fear. By downplaying the problem, they are trying to give cover to the Trump administration and Republican lawmakers who are unwilling to make significant investments in higher education that would help keep the doors of the best public universities open to all students, rather than just to those wealthy enough to afford them.

My findings came from an analysis I conducted of data that the Equality of Opportunity Project released in January as part of a study examining the role colleges play in providing social mobility for low-income students. The project’s Mobility Report Cards provide the clearest picture we’ve ever had of the family income of students at individual colleges.

Stanford economist Raj Chetty and his colleagues at the Equality of Opportunity Project worked with the U.S. Treasury Department to get tax returns — stripped of identifying information — that they could link to college attendance records. By doing so, they were able to get family earnings data for nearly all traditional students — those between the ages of 18 and 22 — who attended college between 1999 and 2013.

In examining the Mobility Report Cards, I was able to see how the socioeconomic makeup of students at individual colleges changed during that period of time. The news wasn’t all bad. About a quarter of selective public institutions increased the share of low-income students they serve at the same time that they reduced the share of wealthy ones. Still, the results confirmed that the vast majority of selective public universities have become less accessible for the most financially needy students since the late 1990s. The beneficiaries of this shift have largely been students from upper-middle-income and wealthy backgrounds, as the overall share of middle-income students at these institutions dropped, too.

The American Enterprise Institute scholars reanalyzed the data, looking at enrollment data at these schools. They found that, overall, the share of students at these schools “from the bottom 40 percent of the income distribution declined by less than a percentage point, from 20.6 percent to 20.1 percent” and that “the share of students from the top income quintile did increase, from 37 percent to 38.4 percent.”

They note that our findings about the overall enrollment patterns “are accurate.” But they say that we, and the reporters who covered our report, are making a mountain out of a molehill. When looking at the data in aggregate, they say that these are very small changes and that policymakers shouldn’t “overreact” to them.

The problem with their analysis, however, is that the scholars treat all the universities as if they are the same. In part, that’s because the Equality of Opportunity Project makes only small distinctions among the schools, labeling 375 schools as “selective,” even though they are of varying quality.

A closer look at the data, however, suggests that in many states, low-income students are increasingly being pushed out of public flagship and research universities, leading them to enroll in less-prominent regional state colleges, community colleges and for-profit schools.

Take Arkansas, for example. At the University of Arkansas at Fayetteville, the state’s flagship campus, the share of students from families in the bottom 40 percent of the income scale dropped from about one-fifth of the student body in 1999 to a little more than one-tenth in 2013. In the meantime, the share of low-income students at regional state colleges such as Southern Arkansas University increased significantly. Why is this significant? Because according to the U.S. Department of Education’s College Scorecard, undergraduates at the University of Arkansas are nearly twice as likely to graduate within six years as those at Southern Arkansas.

Similarly, in Oklahoma, the two leading public universities — the University of Oklahoma and Oklahoma State University — saw sharp reductions in their shares of low-income students. Over the same time period, low-income enrollment skyrocketed at Langston University, a historically black school.  More than two-thirds of students graduate within six years at the University of Oklahoma, and 61 percent at Oklahoma State. In contrast, only 8 percent of Langston students graduate within six years.

The American Enterprise Institute analysis doesn’t deal with such nuances. Because if it did, it might have to acknowledge that a significant problem exists that needs addressing. And that might mean that policymakers need to increase higher education spending to deal with the problem.

So instead, they present the data in a way that obscures the problem. As they express in their conclusion, “Of course, some policymakers might not be persuaded to do anything after seeing how little things have changed.” And that’s exactly the outcome they want.

Stephen Burd is a senior policy analyst with the Education Policy program at New America.