Friday, July 15, 2011

A's for All

Anyone who has ever TA'd, taught, or taken a college course knows about grade inflation, but it turns out there is actually really good data on the subject.  From the a paper by by Stuart Rojstaczer and Christopher Healy published in the Teachers College Record (via the NY Times):

In astronomy we'd call that a really clean trend, fit a power law to it, and conclude that by 2100 all grades will be A's.


  1. Pretty.  The first part fits (I think) with what I've been told about grade inflation -- namely, that grades inflated in the 60s in order to keep students out of Vietnam.  If you gave a student a poor grade in the 60s, you might be giving a death sentence.

    But a couple of things bother me about that story.  First, I'm not sure why the As and Cs are trading places.  Why not a bump in Bs?  Second, the inflation started to go back down in the 70s, but then something happened in the early 80s that sent it back up again.  Not a draft or a hot war.  So ... what happened?

  2. I'm puzzled by the steadiness of the B's as well.  It seems like one should expect them to rise as well.  I've heard a lot of ideas floated to explain grade inflation, but never why the switch is primarily from C's to A's.

  3. The best way I've seen for a school to counter grade inflation without hurting their students' ability to get into grad school is to list on the transcript the mean grade in that course next to the student's grade.  That way somebody looking at the transcript can calibrate the student's grades.

    Of course it would make more sense if everyone were simply graded the same way, but that's never going to happen.

  4. In defense of the change I think college now has a different function.  If most of the people in the class learn the material shouldn't they get A's?  Don't A's reflect understanding the material?  Why should there be a curve?

    The problem is when you have most people doing A = knows the material and B = knows most of the material down and then you get one or two professors who only give out two or three A's.  

  5. I still think grades should be based on standard deviations from the mean of your class placement.  A means one standard deviation better then the average kid in the class.  B is between the mean of the class and an A, C between -1 standard deviation and the mean and D/F below.  

    Then you *immediately* know what a A in differential equations from BYU means.  It means if you took the scores from the while class your score was at least a standard deviation higher from the mean of the scores.

    Instead an A can mean a variety of things.

  6. And I should add teachers should also have things like B-, B and B+'s as well, just that the B- should start at the mean and the B+ should go up to one standard deviation, etc...

  7. "and conclude that by 2100 all grades will be A's"

    You know what's funny is I wouldn't be surprised it that could happen.  I think society progressively has a harder and harder time saying things that suggest "Booby is better than Sally".

  8. That only seems fair if your class distribution is about the same as every other class distribution.  But I've been in classes where half the students are idiots and I've been in classes where half the class was ridiculously smart.  It seems quite unfair to suggest an A based upon std dev means the same thing in both classes.  The grades, as I said, ought to be tied to a more objective measure of what you should know when you complete the class.


To add a link to text:
<a href="URL">Text</a>