In her junior year, a friend of Yale senior Joshua Monrad suffered from a mental health crisis, which caused her grades to slip. Later, she looked into the possibility of receiving an institutional endorsement for prestigious fellowships in the United Kingdom. The friend — whom Monrad counts among “the smartest people [he] knows” — was told to forget about it, because her grades weren’t good enough despite still meeting the competitions’ requirements.
“We should reconsider the choice of the grade-point average as the one all-important statistic to care about, as it has considerable weaknesses,” he soon wrote in a Yale Daily News op-ed. He argued that GPAs were statistically flawed metrics that can further be easily derailed by “one personal tragedy.”
Mounting evidence confirms Monrad’s views. Arbitrary factors — in addition to mental health crises — influence GPAs to the point that they don’t necessarily reflect intellect or skill. Despite studies demonstrating their pitfalls, award selection committees — along with employers and graduate schools — still rely upon them to varying degrees to select their winners.
Former Duke professor Stuart Rojstaczer compiled data from over 400 universities showing that in the 1960s, science and engineering (STEM) and the humanities and social sciences (HALS) departments used to grade more equally than they do today. HALS grades rose rapidly in the 1970s, but STEM didn’t keep pace. As of 2009 — his most recent data — the humanities nationally grade about 0.3 point higher than the natural sciences.
“Those trends and differences have been long-standing,” Rojstaczer told me. He confirmed that minimum GPA requirements “favor” humanities students, although these students may not necessarily win more often because of them.
In January, I published data indicating that a similar grading gap has existed at Princeton since at least 1980. It’s even more significant at the departmental level. The course-wide GPA for Slavic Languages and Literatures was 3.75, but that of Chemistry was 3.21. Grades are also higher in advanced classes, a trend that helps students from well-resourced high schools. GPAs in the math department, for example, differed by 0.5 points between classes at the 100–200 and 300–400 level.
“In upper-level classes, the assumption is that if you show up, you do the work, you learn the material, then you’re going to be okay,” said three-time class legacy prize winner Matt Tyler ’19, a math major. Lower-level classes, he said, skew the percentage of A’s in the department, because grades are used as a “signal” to show students how they can improve.
“If a department is awarding, on average, higher GPAs, it would make sense that there would be a larger pool of applicants from that department,” Marshall Scholar Joani Etskovitz ’17, an English major, said. She explained that better grades help fellowship candidates advance to the final round of interviews, where judges look beyond transcripts.
Research suggests that STEM majors aren’t earning lower grades for a lack of intelligence.
A 2010 Cornell study found that STEM students had higher grades on average in science and “non-science” courses than their HALS classmates. But science grades overall were still lower than those given in non-science subjects. The 2011 National Survey of Student Engagement reported that STEM students spend more hours per week preparing for classes than their peers in other fields.
Class size is probably also having an effect on Princeton’s grading. I downloaded from the Registrar’s website a list of every course offered last year and the number of students who actually enrolled in them. The median number of students in engineering classes was double that of humanities classes — 24 versus 10, respectively — with the natural and social sciences falling in between. This divide is even greater in introductory courses. 100–200 level engineering classes had a median of 54 students, compared with 11 in the humanities.
A 2008 Cornell study showed that students in bigger classes tended to earn lower grades than their peers in low-enrollment classes, even after controlling for seven variables that included race, department, and academic ability.
“Average grade point declines as class size increases, precipitously up to class sizes of twenty,” it cautioned. Students’ chances of receiving an A dropped from more than 80 percent in a class of six to about 20 percent for a class of 21. Similarly, a leaked internal report from Dartmouth revealed that “large courses” gave 44 percent more B’s than “small courses.”
The cumulative impact of these overlooked forces — none of which relate to students’ abilities — affects who is competitive for academic awards. Selection committees use grades in different ways.
Valedictorian, Shapiro Prizes, Phi Beta Kappa, and Class Legacy Prizes are at one extreme. Committees choose winners almost exclusively on the basis of their raw GPAs.
The Rhodes and Marshall scholarships occupy a middle ground. Students have to meet a minimum 3.7 GPA requirement to apply for them, though this can be waived for the Rhodes.
American Secretary to the Rhodes Trust Elliot Gerson told me that election to a Rhodes Scholarship doesn’t guarantee admission to Oxford. The Trust thought that a GPA threshold was necessary to ensure that winners are competitive in graduate admissions.
“That’s a rule that I have resisted for many years,” he said. “It remains my belief that there are going to be some students every year with GPAs below 3.7 who have unquestioned ability to excel at Oxford.”
Longtime Rhodes selection committee member Nick Allard ’74 said that students would have “a hell of a time” trying to get admitted with a GPA of 3.5. He too opposes what he sees as an “arbitrary” standard.
“Admissions policies underpin our aim to recruit the very best students worldwide, and therefore evidence of a strong academic record is key,” Oxford spokesperson Lanisha Butterfield wrote in an email.
She said that applicants are expected to have the equivalent of an upper second-class U.K. bachelor’s degree, which is usually the equivalent of a 3.3–3.7 GPA at American universities. Individual departments are free to set higher or lower requirements.
Marshall Scholarship chairman Christopher Fisher explained that the GPA cutoff was a way to evaluate academic achievement. “That’s a level that has served us well in drawing that distinction,” he said.
Grading trends didn’t faze him. He said that regional selection committees are sufficiently knowledgeable on higher education issues to “appreciate what excellence looks like when they see it.”
At the opposite extreme are the Pyne Prize, Sachs Scholarship, and Gates Cambridge Scholarship. While judges look at grades, they don’t have a minimum bar.
STEM students are at a disadvantage for awards with strict grade cutoffs. But they may be at an advantage in competitions that look for near-academic perfection thanks to quirks in grading. They consequently dominate the GPA-focused Phi Beta Kappa, Shapiro, and Class Legacy prizes.
I made a prediction in an earlier column that explains how they are able to pull this off. STEM majors likely outnumber HALS majors among the highest GPA levels, while HALS majors are more common in the GPA range above, say, 3.7. STEM students aren’t as strongly represented among Phi Beta Kappa inductees (the top 10 percent of seniors) compared to Shapiro Prize recipients (roughly the top 3 percent of underclass students).
This trend is perhaps most apparent at Oxford itself. While I was on study abroad last spring, I compiled statistics on the university’s exam scores as a reporter for Cherwell, a student newspaper at Oxford.
Instead of having GPAs, British universities award degrees in several classifications called, in descending order, first class, upper second class, lower second class, and third class. A “first” corresponds to a perfect 4.0 GPA while an “upper second” is 3.3–3.7, and so on down the line.
Students in Oxford’s STEM division were 50 percent more likely to be awarded a first than those in the Social Sciences division. They also barely had an edge over the humanities.
On the other hand, social science students were 52 percent more likely than STEM students to earn upper seconds. The humanities also enjoyed a significant lead. For lower seconds and thirds, STEM reigned by a factor of two. These patterns have existed for the past decade.
I then turned Oxford divisions’ degree classifications into American GPAs using the U.S.-U.K. Fulbright Commission’s conversion table. Humanities students would have had on average a GPA of 3.79 last year; Social Sciences 3.77; and STEM 3.69.
To translate this back into American English, the United Kingdom’s most prestigious university suffers from the same arbitrary grading disparities as its peer universities across the pond. Yet state-side selection committees still expect applicants to defy the trends that British students themselves can’t overcome.
The differing nature of subjects’ coursework may explain why grading occurs in this way.
“As long as you got 100 percent or something close to it [in science classes], it didn’t matter if they curved on a bell. You were going to be fine,” said Rhodes Scholar Landis Stankievech ’08, a mechanical and aerospace engineering major.
Sachs Scholarship adviser Matthew Stewart ’85 had noticed over the past 20 years that transcripts with “bucketfuls of A+’s” often belonged to engineers. Princeton’s own grading data confirms his suspicions. Engineering classes awarded A+’s twice as frequently as classes in the humanities.
In contrast, Joani Etskovitz said that it’s impossible to write “a perfect essay” for the humanities. Students less often receive the highest possible scores in subjective fields of study. Nonetheless, they still get good grades. “In the humanities, it’s easier to get a middling but decent score than it is in STEM,” she said.
Keep in mind that a “middling” score in Princeton’s humanities departments is currently a GPA of 3.6.
* * *
It’s high time to stop using GPAs as a rigid measure of undergraduates’ talents.
STEM students work longer hours for lower grades. But they’re still expected to perform on par with their HALS peers when competing against them. Meanwhile, HALS students are left grasping for a pipe dream of perfection that doesn’t exist in their fields. There are countless studies that highlight the deficiencies of grading. Anyone who ignores them is putting their head in the sand.
Comparing GPAs between departments is like reading temperatures in metric and imperial units. Each uses different numbers to mark the same remarkable points of achievement and heat, respectively. One department’s average GPA is 3.2, while 3.7 is the average in another, just as 100 degrees Celsius and 212 degrees Fahrenheit both mark water’s boiling point. It is absurd to say that 212 degrees Fahrenheit is hotter than 100 degrees Celsius because it's a bigger number. But that is precisely what weeding out applicants by their GPAs does.
A dean explained to The Daily Princetonian in 2006 that grade deflation wasn’t hurting students’ chances in fellowship competitions — quite the opposite. Program directors told him they “took Princeton grades more seriously.” Yet, the article noted, “lower grades mean that some students may not make program cutoffs, such as the Marshall's requirement of a GPA of 3.7.”
In other words, such programs viewed the highest grades more favorably without conceding that slightly lower grades — which might have been higher in another department or in the years before deflation — were equally remarkable.
Transcripts can continue to be a part of processes that evaluate students, and they should be viewed more critically. But leave out GPAs. An applicant to a prestigious math program, for example, should rightly be regarded skeptically if they’ve never earned anything above a B- in math courses. A few C’s in Spanish that drag down a math genius’ otherwise commendable GPA, however, shouldn’t be held too strongly against that student (unless a criterion is perfection in every endeavor).
Joshua Monrad, the Yale senior, told me that his friend eventually went to the London School of Economics and won a full scholarship that didn’t need Yale’s nomination. He added that she is now earning top grades there.
“It does seem like our school makes a considerable effort to maximize [its] rate of Rhodes scholars, even if this means dismissing some remarkable students on the basis of GPA alone,” he said.
This is the fourth article in a series examining the outcomes of academic awards.
Liam O’Connor is a senior geosciences major from Wyoming, Del. He can be reached at firstname.lastname@example.org.