On The Shopping Theory of Grade Inflation

One of the stranger things about looking at college grading is the bizarre theories people use to try to explain grade inflation.  It's not just the uninformed that do this.  Professors do it too.  They even publish papers about their bizarre theories.  I don't understand it.  It's as if a detective were to come upon a crime scene, see a dead body, see a knife in that body, find a man hiding in the kitchen covered in blood whose fingerprints match those on the knife and declare, "Space aliens came and killed this poor victim."  OK, I'm exaggerating, but you get the point.

What has caused grades to rise nationwide?  Professors are grading easier year after year.  Look at the data class by class.  The data are undeniable.

But to admit that this is the case would be admitting that professors are abrogating their responsibility to challenge students.  No one wants to do that, especially the professoriate.  It's a pathology, I guess, to not admit guilt.  It creates a professoriate that believes in fantasies.  For example, there's a chapter in a book that surveys professors about grading practices.  If I remember that chapter correctly (I need to go back to the reference to make sure of the numbers) a minority admit they inflate grades, yet a majority believe that their colleagues do so.  It's funny, this pointing of fingers at "the other guy," but it also causes people to come up with bizarre excuses for grade inflation and maybe even more bizarre grading policies.

One of the stranger theories about grade inflation is that it's mostly the students' fault.  The theory goes like this.  Students shop around and find the classes with the highest grades.  They do more and more careful shopping every year.  Grades go up as a result.  This theory is really an extension of "the other guy is doing it" idea. It says that many professors are truly responsible, but there are weak links in grading and students exploit those links to the maximum (and that maximum keeps rising).  It's a nice theory in that it removes culpability from a good portion of the professorate.  Too bad it isn't true.

Grades are, in fact, going up class by class.  There are differentials between departments, of course, but those differentials have been in place for decades.  Grades are going up in physics, English, psychology, economics, et al.  They are going up almost everywhere.  The last decade saw some plateauing of grades in a small number of departments and universities that defied national trends; there are also a handful of schools that have worked hard to make sure they grade honestly.  There are few strong links in the chain.

If the shopping theory were true, one would expect that students would flock to the humanities, where grades are as high as an elephant's eye, and avoid the sciences in increasing numbers every year.  In fact, science enrollments, in terms of percentage of students enrolled, have been static for decades nationally.  In contrast, humanities enrollments have dropped precipitously.

Here's a link to a paper that extols the shopping theory.  It examines undergraduate grades at Cornell over the 1990s to the early 2000s.  To be fair, it does admit that there is plain old "classic" grade inflation caused by other factors, but it is loathe to admit that professors are simply grading easier (instead it seems to assume that the other principal cause of inflation is that students are getting smarter and smarter).

The paper is mishagos to the third power.  Not only does it say that students are shopping more and more, but also that Cornell is facilitating that shopping by posting median grades of its classes (examples of the postings can be found here and here).  Students are supposedly flocking to these postings and preferentially selecting easy A classes.  But if that were the case, wouldn't grades be going up at a higher rate since the time median grades began to be posted?

Median grades have been posted online since 1998.  Here are the data on grades (in terms of percent A grades awarded) at Cornell from 1988 to 2004.  Where's the acceleration in grade inflation from 1998 onward?  It isn't there.  In fact, Cornell has had garden variety inflation for about 25 years now, the same as just about everywhere else even though it's only one of a small percentage of schools that posts its grades in detail online.  None of the other schools that post their grades, some with far more detail than Cornell, have particularly high rates of inflation relative to national averages either.  Yes, students shop for easy grading classes to some extent.  They shop by word of mouth in some schools.  In others, they shop by looking online.  But are they shopping more than in the past?  The data - both national and at Cornell - say that if they are, it's a minor effect at best.

The paper admits that not all of the observed rise in grades is due to student shopping.  It tries to make estimates of the contribution of the shopping effect and, surprise, surprise, finds that effect to be significant.  In other words, Cornell would hardly have any grade inflation compared to other similar schools (who don't post their grades) if students didn't shop.

How does it make that claim?  The paper states that students are enrolling in high grading classes at a rate faster than the rate that those high grading classes are being created.  Here are the data for high grading Cornell classes in liberal arts (classes in the humanities, social sciences and natural sciences excluding cross listed classes because quite frankly I got bored with cleaning up the data files past about 500 classes for each year; it's still a very good highly representative sample); by high grading, I mean classes that report a median grade of A- or better.

The percentage of easy grading classes - and there are 1500 classes represented in this chart total - keeps going up at essentially the same rate as the number of students who enroll in easy grading classes.  Students weren't preferentially selecting easy grading classes in 2010 with any significantly greater frequency than in 1998; the lines are essentially parallel.  Over the 13 years encompassed by the data, the percentage of students who enroll in high grading courses do outpace the creation of those courses but the effect is tiny; the two lines were 23 percent apart in 1998.  In 2010, they were 20 percent apart.  That narrowing could be due to course shopping.  It could be due to noise.  It's a minor secondary influence at best.

Let's say that 3 percentage point differential is indeed all due to increased shopping.  What does that mean in terms of a rise in GPA over 1998 to 2010?  Probably about a 0.01 increase in GPA or less in terms of university averages.  It's a big nothing.

So what's the real story at Cornell?  Grades are indeed going up.  There are a ton of easy grading classes nowadays, more than ever before.  Professors keep lowering their standards, and have made A the most common grade awarded by far.  Students keep enrolling in about the same assortment of classes as before.  At least in the liberal arts at Cornell, there is no readily observable shopping effect.  Is the shopping effect a major component of grade inflation in other classes like engineering?  I doubt it.    Why would students even need to shop carefully when two thirds of all classes have a median grade of A- or better?  Random selection will find them lots of easy A's.  The shopping theory for Cornell likely is fiction.

Cornell has about the same level of grade inflation as its peer schools, although its GPA's historically have been a little lower than Harvard et al.  Just like at its peer schools (with the exception of Princeton, which has made a real effort to control grade inflation), professors keep grading easier year after year.  At quite a few (if I include elite liberals arts colleges as well as research universities), the percentage of A's has already crossed the 60% threshold.  Cornell probably won't get there until the mid-to-late 2010s.

Lots of dopey papers are written and published every day.  So this paper shouldn't be a big deal.  It should be ignored.  But here's the crazy thing.  It hasn't been ignored.  In fact it has influenced Cornell policy.  As a result of this paper, Cornell  has decided to no longer post median grades.  Apparently, the professors of Cornell really believe that by doing so, they'll quell grade inflation.  Good luck with that.

I have a better idea, of course, for Cornell.  Listen up.  You really want to tackle grade inflation?  Tell your professors to stop being such easy graders.  And what about that shopping theory of grade inflation?  It  may be a minor secondary effect with or without data postings online and is nowhere near as important as professors desperate to blame anyone but themselves for high grades would lead you to believe.  Or maybe space aliens are the real cause of grade inflation.  Anything is possible.