GPA margin of error

posted: July 6, 2019

tl;dr: Too much emphasis is placed on GPA figures, without considering the margin of error...

A GPA is a measurement. What it measures is debatable, but most people interpret it as a measure of how well a particular student did in high school or college, and therefore how smart that student is. But as any scientist will tell you, all measurements come with a margin of error. A detail about GPAs that most omit is: what is the margin of error in a GPA?

Looking around at how my fellow students were going about their studies at Cornell when I was there, I concluded that GPAs, on a 4-point scale, were no more accurate than +/- 1 point at best. I said this at the time and I still believe it. Far too much emphasis is placed upon GPAs by others, and the emphasis has only gotten greater over the decades since I graduated, as society’s competitiveness has increased.

Here are my reasons why GPAs should be taken with a grain of salt:

First, you can’t compare GPAs across institutions, or even different programs within an institution, as they all have different policies, sometimes radically different, for calculating grades. At the time I attended Cornell it followed a pretty traditional model, but it did award a GPA figure of 4.3 if you got an A+ grade in a class as opposed to a 4.0 for a straight A. A+ grades were rare (back then), but other schools maxed out at 4.0. But there were more radical tweaks to the grading system than this one.

MIT, arguably as good of an Engineering school as Cornell😉, was and still is pass/fail for the entire freshman year. This was done to reduce pressure on students, but it has the effect of completely wiping out the freshman year from the overall GPA calculation, which many freshman at other schools would gladly accept. The well-regarded small liberal arts college that Steve Jobs dropped out of, Reed College in Portland, is effectively pass/fail for all four years; they don’t even tell the students their grades unless the students ask. When I attended Cornell, fellow Ivy League school Brown supposedly had a policy of having students, after they finished their four years at the school, telling the registrar which course grades they wanted to appear on their official transcripts. The current policy document describes something a bit different, but with no Fs or Ds and the option of taking courses pass/fail, a GPA figure at Brown would be so different from anywhere else that they don’t bother calculating one.

Another policy which affects GPAs is the withdrawal deadline for classes. Cornell had a relatively early one, so you needed to figure out pretty quickly if you were struggling in a class and wanted to withdraw before a bad grade appeared on your permanent record. We heard rumors of other schools with much later deadlines, including (supposedly) some schools where the way you withdrew from a class was by not showing up to take the final exam: nothing would appear on your record.

Then there is the issue of different grading standards for different institutions, which extends to different schools or departments within the same institution, or even different professors teaching the same class in the same institution. Harvard has a well-documented reputation as the hardest university to get into, but once a student gets in, the graduation rate is very high. Back when I was in school it was easier to earn an A at Harvard, on average, than Cornell. Grade inflation over the decades since is also well-documented, which makes it hard to compare GPAs across time from the same program at the same institution.

So far I’ve been describing institutional issues that affect GPAs, but I also saw plenty of individual student behavior that dramatically impacted GPA figures.

Put simply, some students cared a lot more about their GPA than others, and would take steps to boost their GPAs that others wouldn’t, up to and including cheating. I’ve already described a student I knew, nicknamed The Scribe, who was solely focused on GPA, did everything he could to maximize his grade, and eventually ended up as a salesperson instead of a practicing design engineer.

In contrast I knew other students who didn’t care much about grades at all. This may have changed over the years as college has gotten much more expensive and the competitiveness has increased, but I and many of my fellow students felt we were there to learn what we wanted to learn. I had a friend who believed this quite strongly: if he discovered something that interested him which wasn’t on the syllabus, he would spend hours in the library studying it, and would read books that had nothing to do with any course he was taking. These many sidetracks reduced the time available for studying the material that was actually going to be on upcoming tests.

I had another friend who was a person I felt was one of the smartest folks in the EE School. We took more than a few classes together, studied and did projects together, and shared an apartment for a year. He had one problem that prevented him from getting good grades: he was dyslexic, and would often reverse digits in his calculations, which is almost always going to produce the wrong answer on an engineering, math, or physics test.

The graphics on our oscilloscope version of Kaboom! were even more primitive

I saw his dyslexia firsthand when we were doing a project together freshman year which involved typing hexadecimal machine language instructions into a microcomputer. To speed up the entry of all these digits, we started with him reading off the digits and me typing them. He kept reversing digits, which caused our program to crash, so we tried it the other way, with me reading off the digits and him typing them. But he still reversed some of them while typing them, so we gave up and I read and typed the digits solo. We produced a great project: we turned an oscilloscope into a video game called Kaboom!, which so impressed the professor that he burned our code into one of his few available ROM chips to show to future classes. My friend was really smart, but he couldn’t enter hexadecimal digits on a keypad without making errors.

Needless to say, my friend’s dyslexia caused problems on tests, and as a result his GPA wasn’t great. Technology companies hiring from the senior class placed a lot of emphasis on GPA back then, so he didn’t get any good offers. IBM, then the premier computer company in the world, enforced a GPA cutoff to even be eligible to interview. My friend decided to stay on campus for a fifth year, to earn a one-year master’s degree and hopefully boost his GPA enough to get a job. His grades were much better in his fifth year, because a lot of the grade was based on the projects that he did, not on how well he calculated answers to problems on tests.

He got a job offer from Digital Equipment Corporation, then one of the most desirable places for computer engineers and scientists to work. He did well there, and joined a startup. He did well at that startup, and then had the opportunity to cofound his own startup company. That company did so well that it was bought during the late 1990s Internet/telecom bubble for enough money that he never had to work another day in his life. He bought himself a Ferrari and has been racing cars ever since.

If you went by GPA, the numbers would tell you to hire The Scribe, not my friend. But who actually knew more about computers, and who did better in his career? In the computer industry, several of the most successful people dropped out of college, effectively earning a 4-year GPA of “Incomplete”. Microsoft founder Bill Gates dropped out of Harvard, Apple founder Steve Jobs dropped out of Reed, Oracle founder Larry Ellison dropped out of both the University of Illinois and the University of Chicago. These are the reasons why I myself try not to place undue emphasis on GPA numbers.