Grade inflation: why weren’t the instructors all giving all A’s already? | Hacker News Reader

Grade inflation: why weren’t the instructors all giving all A’s already?

andrewgelman.com

30 pointstimgluz25 comments

Grade inflation: why weren’t the instructors all giving all A’s already??

People have discussed why the grades have been going up and whether this is a bad thing.

I (Andrew Gelman, professor of statistics, http://www.stat.columbia.edu/~gelman/ ) have a slightly different take on all this. As a teacher who, like many others, assigns grades in an unregulated environment (that is, we have no standardized tests and no rules on how we should grade), all the incentives to toward giving only A’s.

So the real question is, why have grades been going up so slowly?

Open articleView on HN

I taught freshman calculus at Stanford. I had little idea what I should expect of the students, and certainly I didn't have any absolute standard of what constituted an "A", "B", and so on.

So what did I do? I just got the grade distribution from the last time it was taught and curved my class the same way. But this wasn't perfect... there would be a clump of five students who all got around the same numerical grade around a cutoff, and I'd round up rather than down. Similarly there were a couple of students who showed unusual effort and/or improvement, and I rounded them up if they were borderline. The end result was a grade distribution that was just slightly more generous than the last time. And hence I perpetuated the cycle.

As a professor, if you want to give strict grades, you have to be able to justify your reasoning when students complain. I don't terribly like grading, I don't have any deep insight on how to do it, and my department doesn't have strong opinions (other than that I should get back to writing my grants). So what do I do? I follow my peers.

Do you or does your department care about the quality of the students you promote or attract? If so, having and reflecting on grading standards is relevant to that.

If you just want to be a research department with a minimal or laissez-faire approach to your undergraduates, then yeah, why even bother with grading.

Were the majority of students attending Berkeley in the 30s and getting 2.35s (intellectually) worthless people?

Can a department only care about quality if it says most people don't have it? Even if that is true of the general population, who are you to say that's necessarily true of your classroom?

Grades are not binary. An average grade doesn't indicate worthlessness, intellectual or otherwise.

To answer a slightly different question than the one you asked: I don't think a department can care about quality of its students if it marks differently-qualified students as equal.

Why do professors grade their own students anyway? Why not have it done by a less biased examiner, like someone from another university?

You wouldn't be allowed to review your own journal submissions, or even those of a close peer (I hope), but it seems student assessment is done in a much less rigorous way.

I could see that working for very standardized courses with objective tests, like intro calc, but it'd be more problematic in higher-level courses. There isn't usually 100% consensus on how to do many things, like what constitutes a good project and what things are most important in a write-up (and the higher-level you get, the more sharp disagreements between professors get). So students and professors, fairly reasonably, bridge the gap by communicating about shared expectations. If then a third party grades it, they might be grading in a way that isn't actually consistent with what the professor asked the students to do, and with what he/she led them to believe would carry the most weight in grading.

An analogy might be going back and forth with a client on contract work, delivering a result you're pretty sure they're happy with, but then having it assessed by someone other than the client, who didn't participate in the previous discussions. It's a sort of multiple-bosses problem, which can make it harder to focus and feel confident that you know what's being asked of you.

Why did you curve your class? I mean, I know that we expect grades to follow a statistical curve, and I can almost see it justified (although I disagree with it) in a class with more subjective grading (eg history, languages), but questions in freshman calculus have well-defined scores for different answers and parts thereof.

Certainly everyone got a numerical score at the end... but what letter grade should, say, a 77 be? There were no departmental standards, I don't have any particular personal standards, and I hadn't tried to match the difficulty of the exams to what was done previously.

I could do something like 90-100 = A, 80-89 = B, etc., but that is no less arbitrary than any other system, and in my experience it is customary (as well as, IMHO, a good idea) to write harder exams and then make the cutoffs more generous.

"I'd round up rather than down...The end result was a grade distribution that was just slightly more generous than the last time. And hence I perpetuated the cycle."

Next time, why not begin with a slightly harsher grade distribution? Then selective rounding up would result in the same distribution as the previous year.

The question is: how grade adjustments can be gamed by students? If well done, it can turn into a competitive exam, so that students will fight each other rather that syndicate to game the system. But then, you'll have a very nasty class dynamics, which will favor sociopaths.

That's what happens, for instance, with medical students in France: studies are free (or rather everyone pay for them all their lives through taxes), but by the end of 1st year there's the _numerus clausus_: the number N of MD to be trained is determined, a national exam is set up and the N best grades get to enter 2nd year. As a result, describing the relationship between 1st year students as nasty would be an understatement.

If well done gamification may create a competitive environment but you don't really want that. You want people who solve problems, preferably in teams.

This doesn't really seem like rational self-interested behavior on the part of the students. There are thousands of doctors trained every year in a country as large as France. A student's choice to mean or nice to the tiny subset of his peers which he will have substantial interaction with is going to have a negligible effect on the average score. Therefore, it's very much in his interest to cooperate with his small group of local peers in order that the group can do better than the average.

Since students do end up being nasty to each other, that tells us there's something very irrational going on. Thus, it might be more effective to try to manipulate that irrationality away than changing the fundamental structure of the tests.

A problem with a post-hoc adjustment imo is that it makes the connection between the professor/material and the grade received more confusing. Students typically want to know: what does the professor expect, and what do I have to do to get an A (or a B)? Halfway through the semester, a struggling student might want to know, am I on track to fail (or get a D perhaps), and is it plausible for me to bring it up or should I drop the course? The professor might have specific suggestions for improvement, and give rough feedback on what the student needs to do for the rest of the course to get at least a C (or maybe even a B).

But if the answer is that the professor themselves isn't quite sure, because it'll go through some statistical adjustment process after the course ends, that's a bit less helpful.

I'm admittedly thinking more about smaller courses and especially project- or seminar-based ones. Large lecture courses with homework+tests that are just numerically graded and averaged could more plausibly be done as he proposes.

From the article: "I assume that back in the 1940s, a prof couldn’t really just give all A’s to his or her classes: someone would probably notice and say something. But now we really can, and it’s been that way for awhile."

  US College enrollment, 1940: 1.5Mm [1]
  US College enrollment, 2009: 19.5Mm [2]
  
  Average yearly cost of 4 year Uni., 1940: $6000 (adj. for inflation) [3]
  Average yearly cost of 4 year Uni., 2009: $20-35k [4]

So yeah, none of this has anything do do with admitting unprecedented numbers of students—many of whom are woefully unprepared for the experience—to universities already stretched so thin that you might not even meet a tenured professor until your second or third year.

None of this has anything to do with entire sections being managed, instructed, and graded by TAs who themselves are just figuring out what the hell they're doing. None of this has anything to do with a complete lack of focus on accountability by administrations far more concerned with collecting as much money as possible to bankroll research by a handful of their most notable staff, and occasionally give themselves nice bonuses.

It's probably because professors feel sad when they're forced to objectively evaluate things, and they would just give everyone A's if not for their inclination to protect the mental image of their own GPAs fifty years ago.

Again, from the article: "So, now that we’re giving out the grades, we don’t want to devalue this currency."

Newsflash pal, that currency is already way past devalued. We're not even trading in the same market any longer. And this sort of bizarre, self-absorbed postulation, with absolutely no acknowledgement of the poison coursing through the system which you—dear author—have devoted your (professional) life to, casts serious doubts in my mind that it will ever recoup its lost value.

[1]: http://www.census.gov/apsd/cqc/cqc13.pdf

[2]: http://www.census.gov/compendia/statab/2011/tables/11s0273.p...

[3]: http://www.reducemycollegecosts.com/wp-content/uploads/2008/...

[4]: http://www.usnews.com/opinion/articles/2010/08/24/the-averag...

"These trends may help explain why private school students are disproportionately represented in Ph.D. study in science and engineering and why they tend to dominate admission into the most prestigious professional schools."

This part here is malarkey. They're quoting it mainly to show why the topic is noteworthy, but the premise is faulty. Admissions boards are well aware of grade inflation and do not choose students based on ranking their raw GPAs.

Hate to say it but the guy who wrote that article seems very unprofessional to me. He says he gives As to all his students for everything they do regardless of quality because it makes them feel good about themselves and like him more. Then he suggests that other teachers who don't also do this are elitists who had it easy and don't want newcomers in their A club. Then he bemoans that it is impossible to assign different grades without standardized tests.

Wherever he is teaching must not be a very good school (edit: ah Berkeley; it's overrated), everything he says is bunk.

And yes, I have taught elementary, high school and college classes. I don't do grades for elementary, but for high school and college it is not difficult at all to assign objective grades for work done. Also, the students respect you less not more when you just give out all As. Most students respect fair, impartial and realistic grading. Those who don't probably shouldn't be in college anyway.

It's also OK to do as Reed College does and keep the grades in a private file in the event of subsequent grad school transcripts needed, and instead evaluate each student with written essays and never let them see any grades. No grades at all is better than all As. Giving all As does not make you a "nice guy" any more than giving out free crack samples so people "like you" makes you a nice guy. The people don't really like you because you give them free crack, and you're not really a nice person because you do this. Wow, why isn't this obvious to people. It's amazing I have to even say this.

  ...the guy who wrote that article seems very unprofessional to me

He's a highly respected statistician who's written some very valuable books, including at least one very useful book on teaching statistics. He actually says nothing about the rate at which he himself gives A's, just that all the incentives tend that way. In fact, at the end he cites an approach to post-hoc modification of grades which would make it irrelevant whether he assigned A's to everyone or not.

What were your thoughts about the fact that the world's leading expert on the topic, whom he cited in his article, responded in the comments pointing out his errors, and he responded with personal remarks and no facts? That doesn't sound like a professional statistician to me, it sounds like a child.

Although I'm not a teacher, this topic is near and dear to my heart.

My current firm recruits CS, Engineering and Financial Mathematics grads from top schools. Even in this environment, we need to give everyone a math test as part of the hiring process. We can't trust the GPA. Top software firms make developers write code as part of the interview. Even CS (which seems tougher than most majors) suffers from grade inflation. The beauty of this system is that "grade grubbers" who nag the professors for better grades ultimately don't get rewarded. People who are in school to learn do.

A prior firm that I worked for (large consultancy) did a study on if GPA was predictive of good work performance. They found that there was a pass/fail barrier of a 3.0 for technical degrees, and 3.2 for non-technical degrees. Folks who couldn't find a way to hit the lower barrier on average didn't do as well. Folks who did better were more likely - but there was no benefit for getting much higher. A 3.8 student wasn't any more likely to perform better than a 3.1 student. (Slightly off-topic - the only other predictor of success was people who worked at least part-time while in school. It didn't matter what the job was, but people who had to work doing school on average performed better than those who didn't.)

So what can be done that doesn't crush the professors? My grad school forced every non-Phd class to hit an average GPA. Want to give more As? Then give more Cs. When every professor is forced to live by this, nobody suffers in the evaluations.

The incentive for giving a reasonable spread of grades is that your students have incentive to try harder. Thus you get better students.

The disincentive for giving all A's is that you disincentivize your students from bothering to put in much effort.

Ideally you want to grade hard enough so that even your brightest student has to put in a lot of work to get an A. That can be hard to do in the current climate of grade inflation, but when I grade I at least try to err in that direction.

Since grades are so subjective, I've always thought of them as being basically meaningless. So I'm happy seeing grade inflation because it's gradually making grades obsolete.

When I went to grad school in engineering and became a TA, I found out there was a guideline for the undergraduate class average from the department. It was actually presented as the historical class average for informational purposes, with standard deviation, but the implication was that a professor's class average should be close the the historical average.

I think a major problem with grading is that it causes teachers to conflate two separate but important measures, effort and performance.

Those uses grades as a guide to acceptance (colleges, jobs, etc) really want to know both of those pieces of data. So they use standardized tests to get hints at performance and interviews, recommendations, etc to get hints at effort.

Maybe what we really need is a new grading system with two scores for each class, one for performance and another for effort.

Maybe some of the grade improvement is real. The Flynn Effect has recorded an increase of 3 IQ points per decade.

And universities have become much more academically selective. Once upon a time, university students were selected for wealth and spare time, but the rise of standardized tests means that they are now selected for (mostly) by IQ.

Reply on news.ycombinator.com