Grade Inflation

Where to begin?

Let’s look at some data. The figures below are from “Where A is Ordinary: The Evolution of American College and University Grading, 1940-2009” (Rojstaczer and Healy, Teacher’s College Record, 2012, 114, 070306). The dataset is large and comprises a variety of institutional types, both private and public, and also by region.

The most dramatic trends are the increase of A’s and the decrease of C’s. There’s a dip in D’s but smaller. B’s hover between 35-40% and F’s hold steady at 5%. C was the most common grade circa 1940 but was supplanted by B starting in 1965. In 1997, A became the most common grade and its rise continues. (The small “bump” in A’s in the early ‘70s is partly attributed to instructor attitudes during the Vietnam war.) There are differences in the fine-grained data. Schools in the South tend to give lower grades. Public schools show less inflation than privates. Some specific schools (Central Michigan, Colorado, Princeton) had specific policies to combat grade inflation. The private-public difference is the starkest as seen from data comparing 1960, 1980, and 2007.

Why are grades rising?

Some schools argue that they are getting academically stronger students. Perhaps they are, at least based on increases in average SAT/ACT scores of incoming students. However, this increase is small percentage-wise compared to the increase in grades. Some argue that more women attending college have moved grades up. (It is well documented that on average women perform better GPA-wise.) However, comparing both the period and the rate of increase in percentage of college-attending women once again does not match up with observed grade inflation. Perhaps teaching methods are improving. But there’s little data to suggest any effect even if the statement is (partially) true. Perhaps students are choosing “easier” majors. Perhaps the rise in adjunct faculty with precarious job security. Perhaps the changing attitude of higher education towards student-as-customer.

These and other scenarios are discussed by Rojstaczer and Healy with no firm answer to the why question, although they seem to favor the last “perhaps” in the paragraph above when writing their conclusion: “Our data suggest that in the absence of oversight from leadership concerned about grade inflation, grades will almost always rise in an academic environment where professors sense that there are incentives to please students… the principle incentives likely have been a mix of the ability to enhance students’ postgraduate prospects, and with the rise of importance of student-based evaluations, faculty self interest.”

The problem stated by the article title remains. Once an A becomes ordinary, it loses value as a signaling factor to employers, post-graduate programs, and any other entity using grades as some sort of benchmark. It is not surprising to see the rise of alternate credentialism. Rojstaczer and Healy exhort universities to rein in “grades that bear little relation to actual performance”. This will not be easy, because of larger systemic issues. The issue is complicated, although economists (particularly game theorists) are attempting to model “what if” scenarios to get at the problem. The remainder of today’s blog post will highlight just a few examples. Note that unlike the data-driven model for grade norming previously discussed, most game-theory studies come with necessarily simplified assumptions without a supporting data set. I won’t go into the game-theory models, but I will cite the papers for the math-inclined who’d want to read propositions, proofs, and lemmas.

In “A Signaling Theory of Grade Inflation” by Chan et al. (International Economic Review, 2007, 48, 1065-1090), their game-theory model suggests that when one school inflates grades, other schools will follow suit. Here’s the crux: “When a school gives a lot of good grades, the labor market cannot fully distinguish whether this is due to an overly liberal grading standard or whether the school is blessed with a large proportion of high-ability students.” On the one hand, schools trying to help their students get jobs have an incentive to raise grades. On the other hand, a school’s top students might be less distinguishable from their peers. Employers have some knowledge of a student’s abilities, but less than the school that’s (purposely/strategically) providing coarse-grained signaling.

Essentially, “grade inflation reduces the information content of grades and thus impedes optimal task assignment by employers”. To hedge their bets, employers might choose to hire higher GPA students from colleges that have higher average GPAs because there’s a decent probability of getting a high-ability student. Elite colleges are incentivized to continue grade inflation to keep more of their students in these “virtuous cycles”, and thus the hamster wheels turn. After all, brand-name is a signaling factor. Those of us in academia know this well, unfortunately. Academic snobbery is rife. I’m sure I benefited, at least getting my foot in the door where job offers were concerned. But grade inflation isn’t purely an academia issue; similar phenomena show up in consumer magazines, accounting audits of companies, stock-picking in investment banking, and more. Hence, the interest of economists, I guess.

Schools that perpetrate more grade inflation, might, they just might, have better quality education – at least judged by increased resources funneled into education. That’s the argument by Boleslavsky and Cotton in “Grading Standards and Education Quality” (American Economic Journal: Microeconomics, 2005, 7, 248-279). The situation faced by employers (or evaluators): “transcripts are less informative, hindering the ex post selection of high-ability graduates, but schools invest more in education quality, so graduates are more likely to be high ability ex ante.” The authors’ model finds that this “effect on investment may dominate” setting up a complementarity between school and employer. A look at the U.S. landscape certainly bears this out when considering the brand-name schools with large endowments and operating expenses. A virtuous upward spiral perhaps.

The final article I highlight today is “Student sorting and implications for grade inflation” by Herron and Markovich (Rationality and Society, 2017, 29, 355-386). In their model, departments (majors) are divided into two types: ability-revealing or ability concealing. Their conclusion: grade inflation is (at least partly) driven by how students sort themselves into departments. This grade-related sorting was mentioned in my previous post as to why grade norming might bring more women into STEM. The authors sound a cautionary note to a hypothetical university dean who sees many A’s in a department. In an ability-concealing department, grades aren’t distinguishing the stronger students from the less academically-inclined. In an ability-revealing department, avoided by weaker students, the high grades might be an accurate reflection of student caliber. Thus, the claim “our students are excellent!” might be true and reasonable. How might one tell the difference? The authors suggest indirect measures (because there are confounding variables); I encourage you to read the article for details. There is evidence that introductory course grades in ability-revealing departments (where you see the full distribution of grades including many low ones) are better predictors of future student performance comparted to ability-concealing departments.

There are several interesting offshoots mentioned at the end of the article. The authors make an argument that the emergence of ability-concealing departments in a university, even just one or two, could over time lead to the present situation because of the way students sort themselves based on their own perceived abilities. Weaker students move towards ability-concealing departments, while stronger students, to distinguish themselves, migrate to ability-revealing departments. The (idealized) model assumes students know their own ability and can perfectly rank themselves within their cohort. This, we all know as educators, is not true – thus the actual situation is much messier. The authors write: “This sort of error is particularly pernicious for students because a high-ability student who believes that she is of low ability may prefer an ability-concealing department over an ability-revealing department – even though the latter would be more valuable.” Risk-aversion is complicated, although studies do show that on average women are more risk averse than men, at least where grades are concerned.

What to do?

I close by quoting the authors on the challenges of a department trying to mitigate grade inflation. “A department that by itself wanted to address institution-wide grade inflation can be stymied by the ability-concealing behaviors of other departments. If an ability-revealing department were to make its classes increasingly challenging in an attempt to mitigate inflation, then it would make the overall grade inflation problem worse and in so doing decrease its own enrollments. To the extent that low enrollments are problematic for departments are problematic for departments who might want to use enrollment figures to argue for faculty positions, no department has an incentive on its own to increase the cost associated with its classes. This sort of collective action dilemma means that university administrators should not assume that individual departments will ever be able to coordinate themselves and form a solution to what administrators might consider a grade inflation problem.”