The Washington PostDemocracy Dies in Darkness

Did Columbia game the U.S. News college rankings with sketchy data?

The controversy offers yet more evidence that rating colleges objectively is an impossible task

Columbia University graduates take pictures in May 2020 in New York. (Frank Franklin II/AP)
7 min

In the fall, Columbia University celebrated a significant achievement: It tied for second in the much-watched U.S. News & World Report ranking of national universities, behind only Princeton. It was the university’s highest ranking ever. “It is gratifying to receive this confirmation of the success of our undergraduate programs,” the dean of the college crowed. Columbia has been steadily rising in ranking — it was No. 18 when the list debuted in 1988 — sailing by such competitors as Stanford and Yale and enjoying record-high numbers of applicants.

There’s just one problem. According to a long, fiery and rigorously argued blog post by Columbia math professor Michael Thaddeus, the data that Columbia has been supplying U.S. News to attain second place contains a liberal amount of, well, No. 2.

For example, one factor in the numerical bouillabaisse that makes up the U.S. News ranking is instructional spending per student. Thaddeus infers that Columbia gave U.S. News the same figure for instructional spending that it reported to the Education Department: $3.1 billion, the highest in the country. (U.S. News confirmed to Thaddeus that the Education Department figure is the one universities typically submit for the list.) That would amount to more than $100,000 per undergraduate and graduate student. And it’s more than the spending reported to the department by Harvard, Yale and Princeton combined.

But that the number is so big is not actually impressive — it goes beyond that, to fully implausible. Do you really think Columbia spends more on its students than Harvard, Yale and Princeton together spend on theirs? Thaddeus finds suggestive evidence that it does not. In other contexts, such as its consolidated financial statements, Columbia reports a much lower figure for “instruction and educational administration,” $2 billion, despite showing an identical overall budget. And those statements include a large separate expense of $1.2 billion for “patient care.” In other words, the university seems to be telling the government, and U.S. News, that $1.2 billion spent on patient care should count as “instructional expenses” because a medical student in the room while a patient was being treated might have learned something. (Responding to questions from the New York Times, Columbia basically affirmed this interpretation — saying that for its students in medical fields, instruction and care happen at the same time.)

That is just one of Thaddeus’s bullet points; there are many more. Columbia reports that 100 percent of its faculty hold terminal degrees (that is, the highest degree possible in their field). Thaddeus read through the online Columbia College Bulletin and found 69 who didn’t. (Columbia has said that the determination of which degrees count as “terminal,” in a given field, can be a judgment call.)

He casts doubt, too, on Columbia’s claim to U.S. News that 82.5 percent of its undergraduate courses have 20 or fewer students. The magazine lists only four universities with a higher percentage: Gallaudet, Wilmington, Touro and Cardinal Stritch, as you no doubt guessed. Examining publicly available information on courses and their enrollment, Thaddeus found that the proportion of courses with enrollments that small at Columbia was closer to 65 percent. (In a statement to the Chronicle of Higher Education, Columbia said the directory of classes that Thaddeus consulted “is not an official record of enrollment.”)

Harvard and its peers should be embarrassed about how few students they educate

All of this looks bad for Columbia. But it could be worse. What if Columbia had actually tried to hit that 82.5 percent benchmark by rigorously capping enrollment in courses and shunting students to their second- (or third-) choice classes? Or fired Nobel Prize-winning faculty members who don’t hold a terminal degree? Would that make Columbia a better place to go to college? By the rules of the U.S. News game, yes. And that’s the problem. Maybe straight-up fabrication of the numbers is better than distorting your educational practices to conform to the metrics the magazine imposes.

To the extent that universities chase rankings — whether through policy changes or sketchy numbers — they are declaring their real priorities. They are making clear that the imaginary students envisioned by the ranking algorithm are more important than the real-life students already present in the classrooms and the dorms.

That’s why Thaddeus’s beef is not so much with Columbia as with the whole apparatus of college ranking. If you think the relevant question here is, “Was Columbia’s second-place ranking correct, or should they have been lower?” you are already conceding the claim that there is a correct ranking for Columbia to have.

There isn’t! To place U.S. colleges in order from Princeton to umpty-hundred is to declare, whether you say it out loud or not, that colleges are one-dimensional — they can be laid out on a line, from top to bottom. Mathematicians such as Thaddeus, or me, know that most interesting spaces have a lot more dimensions than that. Different colleges have different virtues that mean different things to different students. You could rank colleges by the size of the metro area in which they are based, or the number of research labs employing undergrads, or the quality of the dining hall chicken sandwich; right now, this year’s college applicants are anxiously weighing those factors, and many more, most of them non-quantifiable.

But numbers carry a special sort of imperious authority — there’s a tendency to let differences between numbers outweigh differences of feeling, even when the numerical differences are so small as to be insignificant. This year’s list has the University of Texas at Austin at No. 38 and Brandeis at No. 42. Would you let that slim difference in ranking be the deciding factor between these two wildly different houses of learning?

Of course you wouldn’t. And yet the ranking stays popular: so popular that even though U.S. News & World Report doesn’t exist anymore as a news magazine, the ranking has persisted and thrived, like the dumb grin that outlasted the cat.

Coronavirus vaccines work. But this statistical illusion makes people think they don’t.

A few years back, the National Research Council, which ranks graduate programs, tried a multidimensional approach, rating universities on various factors and then allowing people to “roll their own” rankings by choosing how much weight to place on each factor. Everybody hated it. The U.S. News ranking stays popular, in contrast, because people crave simple, definitive answers: What’s better than what? Where should I go? But one of the key tasks of higher education is to equip us to resist that craving. Simple, definitive answers are the Snickers bars of thinking: satisfying for a moment but dangerous when they displace healthier cognitive options.

Obviously, it’s bad to tinker with numbers to wrestle yourself above your Ivy rivals. But I can’t help wanting to turn toward the bright side of this story. Wouldn’t it be great if the scandal helped loosen our belief in the U.S. News list, the way Jared Kushner’s eyebrow-raising Harvard admission loosened our belief in elite-college access being synonymous with merit? Life’s more complicated than that, and universities should be first in line to assert that complexity. Toward the end of his piece, Thaddeus quotes Columbia president Lee Bollinger, who said in 2003, “Rankings give a false sense of the world and an inauthentic view of what a college education really is.” Thaddeus is even blunter: “No one should try to reform or rehabilitate the [U.S. News] ranking. It is irredeemable.” Aspiring college students should listen to both of them.


An earlier version of this article identified Lee Bollinger as a former president of Columbia University. He is the current president. The article has been corrected.