The validity of grades

Marco Arment writes at (proving that he was prescient when it comes to domain names) and created the awesome web service Instapaper, which I use solely for its Kindle export feature. One of his favorite posts is “School grades are hopelessly broken.” It’s worth reading, and Marco is probably right: school grades are too high and don’t reflect real knowledge, but like healthcare or the military, there’s no easy way to fix them.

Although Marco’s essay is mostly right, it also doesn’t propose any real solutions to the problem he describes—because there aren’t any. The incentive for parents in high school is to want their kid to get the highest grade possible; consequently, they will often fight for their kids, leading to an overall negative equilibrium, and one that I benefited from in late middle and early high school when I decided to effectively fail math as an ill-conceived protest against my parents. At the time I didn’t consciously realize this dynamic, or that protesting in ways that chiefly hurt me aren’t terribly wise, but I was also 13 – 15 at the time and didn’t know any better.

For most teachers, the easiest thing to do on an individual level is inflate grades, which reduces complaints from both parents and students. This isn’t optimal on a societal level, which generates posts like Marco’s, but it is on an individual level, and I don’t see an easy way to generate incentives to change this (more on that later). Marco says:

Grades don’t reflect your aptitude, intelligence, or understanding of the subject matter. You don’t need to actually learn much useful material to get good grades. (And many of those who learn exceptionally well don’t get good grades.)

This is probably true. But if grades don’t reflect all this, then imagine what the people with really low GPAs are like. Grades aren’t good at stratifying the high end of the curve, but they at least show some of where the low end is. And I suspect the really high end, especially in hard college majors like engineering, CS, and so forth, are still reasonably good guides to aptitude. Marco says, “You can understand why I don’t trust the validity of grades.” You shouldn’t trust grades fully—but that’s because grades aren’t supposed to be the full measure of man. Nothing is, except maybe life, and what does that really mean?

I see a lot of comments about colleges, high schools, grades, and how to improve those kinds of issues on sites like Hacker News and Slashdot. Most are good at identifying problems; Chris Smeder, for example, tells us how to improve college teaching in three major ways. He’s probably right about all of those, but he misses an important point: most universities are not set up (or, if you like buzzwords, “incentivized”) to reward teaching.

Smeder misses the main point, which isn’t identifying the problem; a gazillion people in the Chronicle of Higher Education have said virtually the same thing at various times. The real problem is solving the problem, which requires changing the incentives that drive professors. At the moment, hiring and tenure decisions at virtually all universities (and all the big ones you’ve heard of) are made mostly on research and publication. Teaching simply doesn’t count for much. Therefore, the people who succeed in getting hired and getting tenured optimize for what they’re being judged on: research and publication. Teaching is secondary. Heroic individuals and people who want to practice better teaching will help somewhat, but they aren’t enough to change the system as a whole.

Once you’ve realized this incentive problem, the question becomes, “How do you change the incentives?” I have no good answers for that, but it’s the real question you should be asking if you’re genuinely interested. And it might keep you from generalizations like this one, which is back to Marco’s essay:

Most people from my generation can’t really do anything else in the real world except bullshit jobs because nobody ever held them to very high standards.

This probably is true of all generations, and the rhetoric of most people during most generations (consider, for example, the New York Times’ Generation Me vs. You Revisited). I suspect that not all the jobs Marco assumes are bullshit are bullshit. And even if all this is true, schools aren’t going to be offering the kind of information he’d presumably like GPAs to show.

I’d get near-zero homework grades because I’d never do it, so I needed (and usually got) near-100% test grades to make up the difference. I’d barely pull through and get a C most of the time.

This works for some people, especially who start their own businesses. Most people don’t and never will. So their grades count. I’m reminded of Paul Graham’s comment from “What You’ll Wish You’d Known:”

In retrospect this was stupid. It was like someone getting fouled in a soccer game and saying, hey, you fouled me, that’s against the rules, and walking off the field in indignation. Fouls happen. The thing to do when you get fouled is not to lose your cool. Just keep playing.

By putting you in this situation, society has fouled you. Yes, as you suspect, a lot of the stuff you learn in your classes is crap. And yes, as you suspect, the college admissions process is largely a charade. But like many fouls, this one was unintentional. So just keep playing.

Rebellion is almost as stupid as obedience. In either case you let yourself be defined by what they tell you to do. The best plan, I think, is to step onto an orthogonal vector. Don’t just do what they tell you, and don’t just refuse to. Instead treat school as a day job. As day jobs go, it’s pretty sweet. You’re done at 3 o’clock, and you can even work on your own stuff while you’re there.

The right thing to do is your homework, because it’s presumably easy, and then do the rest of your work on your own time. And although GPAs are broken, they’re also the best we’ve got. As Joel Spolsky says in his advice to Computer Science majors:

Never underestimate how big a deal your GPA is. Lots and lots of recruiters and hiring managers, myself included, go straight to the GPA when they scan a resume, and we’re not going to apologize for it. Why? Because the GPA, more than any other one number, reflects the sum of what dozens of professors over a long period of time in many different situations think about your work. SAT scores? Ha! That’s one test over a few hours. The GPA reflects hundreds of papers and midterms and classroom participations over four years. Yeah, it’s got its problems. There has been grade inflation over the years. Nothing about your GPA says whether you got that GPA taking easy classes in home economics at Podunk Community College or taking graduate level Quantum Mechanics at Caltech. Eventually, after I screen out all the 2.5 GPAs from Podunk Community, I’m going to ask for transcripts and recommendations. And then I’m going to look for consistently high grades, not just high grades in computer science.

Marco can be largely right in a micro sense and still be wrong, or at least doesn’t really deal with what should happen in a macro sense. If you’re the principal of a high school, or a college president, or an individual employer, or any number of other positions, what can you do to change the presumed brokenness of grades? How can you transform the system producing said grades? Until you’ve answered that, you’ve done a lot of the work that’s already been done (see, e.g., here for an older view of school problems) without facing the hardest part of the task.

5 responses

  1. I think you make a critical mistake in assuming that the primary value from college comes from quality teaching or that most people learn much of anything in college. True, in the quantatative subjects and for those who are inclined to go on to graduate (but not professional) school it’s different but that’s a small percentage of students and these students are usually the ones well served by a focus on research rather than teaching.

    Frankly most students never use 90% of what they digest in college. Go ask some lawyers to do basic math or see if your doctor (GP) remembers anything from his chemistry class or can tell you what a ribosome is. Frankly for most classes it doesn’t matter if they are taught poorly since conveying knowledge isn’t that important.

    For most students college serves three important goals.

    1) Signals social class and the requisite intelligence (SAT) required to be admitted. Businesses are barred from using IQ tests so must use college prestige as a proxy.

    2) Social development, networking and learning to be an adult. Including learning to struggle through areas without excellent guidance or a teacher’s reminders.

    3) Indication of relative ability to perform under conditions similar to those at most jobs, i.e., how well you can sit still and apply yourself even in imperfect situations.

    In none of these 3 areas is good teaching essential. Yes, it is very important for the engineers and scientists but they tend to benefit more from the expertise held by active researchers than from the enthusiasm and entertaining tricks of the good teacher (ideally both but they are fundamentally in tension).


    • I’ve heard variations on the signaling theories of education before, especially from Robin Hanson, and they’re powerful, I haven’t seen a whole lot of evidence behind them. I’m not sure how one would measure the relative importance of signaling versus knowledge acquisition.

      Frankly most students never use 90% of what they digest in college. Go ask some lawyers to do basic math or see if your doctor (GP) remembers anything from his chemistry class or can tell you what a ribosome is. Frankly for most classes it doesn’t matter if they are taught poorly since conveying knowledge isn’t that important.

      A lot of this is probably true, but I think it more true in some majors than others, and among some people than others; lawyers might not remember a lot of basic math, but I bet they remember a lot of philosophy. But I think the major point is that undergrad academic degrees aren’t professional or pre-professional degrees.

      Still, I suspect that students a) learn how to learn, as they say and b) at least master a body of knowledge that they can then use to bootstrap themselves into other bodies of knowledge. Neither seems to me a lot like pure signaling.

      This isn’t to discount the signal value, or how hard-to-fake it is. But without some kind of evidence (and I’m not sure what it will be), we’re probably arguing about the number of angels who can dance on the head of a pin.


  2. As a student member of a faculty search committee, I can tell you that you’re 100% correct that research is valued infinitely more than teaching in the hiring process. As one professor once told me: “It’s lexicographic.”

    One reason is that evaluating teaching quality is a mirror problem to evaluating students, subject to the same handicaps. Research papers are tangible; we can look at them and judge the merit of the work for ourselves. Teaching evaluations, by contrast, are fuzzy and unreliable.

    We could, of course, ask for a “demo tape,” but this would stack the deck toward the most charming and attractive teachers rather than the ones who most enlighten their students. With nothing more to go on, is it any wonder we go for the brightest researchers and simply cross our fingers that they teach competently?


    • One reason is that evaluating teaching quality is a mirror problem to evaluating students, subject to the same handicaps. Research papers are tangible; we can look at them and judge the merit of the work for ourselves. Teaching evaluations, by contrast, are fuzzy and unreliable.

      Interesting comment. It reminds me of the old joke about a guy who loses his keys on a dark street and is searching for them under a streetlight. A cop asks where he lost them. The man says, “Over there, but the light is better here.” In other words, sometimes we appear to measure things because they’re easy to measure rather than because what they’re what we really want to know.


  3. Pingback: What incentivizes professors to grade honestly? Nothing. « The Story's Story

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: