Sunday, January 23, 2005

Grade Inflation

"Leopold Stotch" at Outside the Beltway discusses Princeton's recent efforts to crack down on grade inflation. In a nutshell, Princeton has limited the number of grades of A- or better to 35 percent of the class in most cases. Leopold doesn't like the idea of an administrator who has never taught telling him how to evaluate his students. As a fellow professor, I cannot help but agree with that (though I acknowledge that my situation is different--my department chair, associate dean, and dean are all first-rate academics). Steven Taylor at Poliblogger agrees and Robert Prather at Signifying Nothing evaluates the policy a bit more sympathetically as it pertains to the way administrators deal with professors.

But in reference to the Princeton case, Leopold states:

The fact of the matter is that students at Ivy League schools should be getting a disproportionate number of As -- otherwise, why were they admitted?

This should be true if they were put in classes alongside students that did not otherwise gain acceptance into Princeton, but not among other students who did. They were admitted because they showed the promise to be able to make the most of Princeton's academic environment. In the context of that environment, expectations should be sufficiently high that only a small minority of the students will merit an A or A-.

The main problem with grade inflation is that, since the maximum grade stays fixed, it it tends to compress the distribution of grades. Some people get A's because they would earn them even if expectations were higher, and some people get A's rather than B's because grades have been inflated. This benefits less capable students at the expense of more capable students, and I see absolutely no reason for this to occur.

Note that this problem of grade compression refers to grades at a point in time. What generally captures people's attention is when the distribution of grades shifts higher over time. This is what got Harvard into trouble during the 2001-2 academic year, when it acknowledged that a disproportionate share of its students were graduating with honors. (The lowest honors threshold was a fixed GPA, and over time, more and more students crossed it. I believe that the problem has been fixed by limiting the number of students who can qualify for honors based solely on a GPA. At Dartmouth, where Latin honors are determined by GPA, the fractions graduating summa, magna, and cum laude have historically been limited to a fixed percent of the class.) If students are arriving better prepared over time (I'm not convinced), or if resources available to them are improving over time (certainly true), then our expectations of them should be increasing over time in a commensurate way.

At Dartmouth, the faculty voted in the 1993-4 academic year (the year before I got here) to display the median grade for each course on the student's transcript alongside the grade awarded. Classes with fewer than 10 students are excepted. The transcript contains a summary of how many classes the student earned a grade above, at, or below the median. This is a useful addition, because it makes the transcript a more honest representation of the student's performance. (Read here to see how that point was lost on the editors of a student paper from a university with the motto, "Veritas.")

However, including the median grade on the transcript is incomplete as a measure to address the problems of grade inflation.

First, it doesn't stop grade inflation over time. We know this because the honors thresholds typically increase each year and because we can analyze the data and see for ourselves that median grades are rising. The reason that the policy doesn't stop grade inflation is that it is not used to change grading policies in any formal way--there is no consequence on campus of having a course with a high median grade.

Second, because the information is collected and presented but not used to change grading behavior, it allows for very large differences to persist across easily identifiable groups. For example, controlling for course size, course number (a proxy for the level of the course), and enrollment, courses in the humanities over the past two years have had median grades that are 0.136 and 0.111 (out of 4.0) higher than those in the sciences and social sciences, respectively. This comparison excludes the language courses, where median grades are even higher.

Third, it is possible that some classes have high median grades because there are a disproportionate number of very talented students in that class, and this policy doesn't do anything to reflect that information. It could be made even better if it did.

With those issues in mind, what changes would I make to Dartmouth's current system? When computing class rank and awarding Latin honors, I would adjust for known and persistent differences across departments in grading practices. I am open to suggestions about precisely how the adjustment would take place. Using the difference between the student's grade and the median grade in the course seems like a useful place to start.

Even better, I suppose, would be to also incorporate a measure of the ability of the students in the course in addition to the grade earned and the median grade. One such metric would be the average combined SAT scores of the students in the course. We would almost have the usual Ratings Percentage Index that is used in many sports like the NCAA basketball and hockey, which are combinations of the team's Winning Percentage, its opponents' winning percentage, and its opponents' opponents' winning percentage.

Other blogs commenting on this post


Leopold Stotch said...

I'm completely unconvinced. First, regarding whether Ivy League students should get a disproportionate number of As: of course they should. Those schools have a disproportionate number of highly intelligent students. You acknowledge this, but imply that they should be graded relative to one another -- am I right in stating that?

My philosophy is that the students are being graded for their mastery of the material, which means that 100 percent A grades is not only possible, but preferable. It will never happen, but that should be the goal. And your post complete neglects this point.

And I think this very well may be the real answer; if my boss feels that I or my colleagues give too many As, then the solution should be a review of course materials to see if the course is designed and executed at the appropriate level. If so, then any "inflation" of grades is attributable to either a smart batch of kids or a professor with low standards. Smart kids should be rewarded, and professors with low standards should be denied tenure.

Robert Schwartz said...

A much more elaborate version of these concepts was used at The Ohio State University College of Law, where I was a student when you were a small child.

The system was invented by a professor named Robert Wills, a man so dull that his students called him "The Chiller" Nonetheless he deserves a footnote at least in the history of information science. He devised the first versions of the database that is now Lexis-Nexis. His grading system, which is stll in use works as follows:

The Law School maintains its transcripts on a 0-100 scale.

The University maintains its transcrits on a ABC A=4pts scale. When the Law School sends grades to the University 92-100=A, 82-91=B etc.

Law School professors rarely gave numeric grades of less than 84, so almost all students carried at least a B average.

The Law School maintaines exact class rankings based on the numeric grades.

Here is where the math comes in. Professors are constrained in handing out grades by the records of the students who are taking their classes. They cannot bid for popularity by handing out easy grades. The Deans office looks at the records of the students coming in to their classes and tells them in effect -- you may hand out no more than 1 100 2 95s 3 92s 6 90s etc. The result is that there are no "guts" taken for easy A's and difficult courses are not shunned. Good students have their class ranks and honors and even mediocre students graduate with B averages.

I think that this controversy, like almost every thing in American society can be understood through d'Tocqueville. The importance of equality is such that it will always be very hard to sell the American people on the proposition that half of them, and even worse, half of their precious children, are below average. Our neighborhood newspaper published a quote from an offical in our school district that amused us, He said that 44% of the children in our district are gifted. If grades bring bad news that half of us are below average, so much the worse for grades.

Should we do without grades? Teachers could communicate with students without grades, even parents, although it might be a much higher burden
on the teachers. But trying to get a synoptic view of the student without grades is much more difficult. If there are no grades, then employers and higher level schools need alternatives. One alternative is standardized testing. I do not think I need to remind you how unhappy that makes a lot of people. Another alternative is feeder schools and favored professors. We usually call this the old boy network. It is not thought to be a socially egalitarian method of allocating opportunities. Maybe grades and class ranks are not so bad.

3 more ideas here.

1. Lets get rid of the current letters, the emotional baggage they carry is too great. My Proposal is:

O = OK, work was acceptable, you passed the course. used to be a C, but the word average is poison in Lake Wobegone.

P = Pat on the back, good work kid. Used to be a B.

W = Wow, way out of the ordinary. Why aren't you teaching this course? What an A should have been, but hasn't been for a generation.

S = So sorry, you did not pass the course. (I do not see why you need a D and an F. Although, I remember my Typing teacher in high school "Bob, you are a nice boy and I do not want to flunk you, but you have not learned to type so I cannot give you a grade higher than a D-).

Numeric values:

O = 1,000,000.1
P = 1,000,000.2
W = 1,000,000.4

2. It is high school students who are hurt the most by these shenanigans. My kid's high school where 44% of the studens are gifted ;-) once claimed that 15% of its graduating class were valedictorians. When that producced guffaws, they clammed up and refused to disclose class ranks at all. Now these are well to do suburban kids, but the trend is worrisome. As I said before, it is the poor and lower classes who get screwed when the objective measures are removed.

3. At the college level, the classic problem is the humannities/science split, which you note. At Northwestern U., my daughters' school, the Engineers are always complaing about the easy grading of the Humanities students. The solution, it seems to me, is to maintain separate rankings for different areas. Perhaps forcing a curve would be a good thing in this context. One last war story. I was an undergraduate at the University of the Chicago in the 1960's. At that time all students had to take about half of their course load in "core cirriculum" courses. There few variations in these courses, although in math and science there were Math 101 and Physical Science 101 courses for the seriously unscientific. Biology 101, however, was the same for everyone, future nobel prize winners, pre-meds and future lawyers like me. I was assigned a lab bench with several very serious guys who did not mind spending their lunch hours messing around with slimy things. I on the other hand felt that lunch was a higher calling. At the end of the course, The Professor called me in to his office, and said: "Bob, 'Usually when a student does not turn any lab reports in, he flunks the final also and I can give him an F with a clear concience. But you had the second highest score [out of about 100 students]. I do not want to flunk you, but I do think you should turn in some lab reports, so if you will get the lab notes of one of your friends from the course and write up 3 lab reports over the break, I will give you a B.'" I said Thank You and left and did as he requested. I did not go to medical school.

Anonymous said...

ATTN: Robert Schwartz

After reading Voxbaby daily and following Andrew Samwick's career I do not think he was ever a "small child"

Anonymous said...

There may be a direct cost to students in cases where employers or graduate schools do not adjust for the overall quality of the schools. Federal employees coming out of college start at a GS-5 ($28,620 a year in DC), but if they have a 3.5 or higher, they start at a GS-7 ($35,452). I had a summer federal job right after graduation, and when I applied in the fall of my senior year, I just had a 3.5. It fell below 3.5 after first semester (and even farther after second semester), but fortunately they never asked for an updated transcript.

Adam Crouch said...

One's position on grade inflation is really determined by the answer to this question: "what is the purpose of giving grades?"

If you believe that it is to measure RELATIVE performance/knowledge, you think that grade inflation is bad. If you believe that it is to measure ABSOLUTE performance/knowledge, then you don't have a problem with grade inflation.

Every time I see a discussion about grade inflation, the two sides just completely miss each other. Each side assumes that the other is starting from the same premise, and so the debate is never over which premise is correct.

marjo moore said...

I've been thinking about this one quite a bit today. I formulated a number of thoughts, but then I realized my experiences in college were all in a humanities arena (I'm now a writer/editor).

I agree with the comment that says different specialties need different thresholds. It is easier in humanities. And teacher subjectivity is a major factor.

I once talked a professor out of an Incomplete and into an A. My hand to god, when I explained my work differently to him, he called it 'thought provoking and innovative.'

Looking at that now, I see what a detrimental move it was to myself and to my learning. A lack of clarity in communicative skills should never, ever be misperceived as artistic licence.

There must always be a set criterion for any course- when x numbers of students meet that, the A grades are the ones that do so either the fastest, the most originally, or the most stylistically.

Jim said...

GPA's can be adjusted to eliminate most of the problems associated with grade inflation. See Jonathan P. Caulkins, Patrick D. Larkey, and Jifa Wei, “Adjusting GPA to Reflect Course Difficulty,” The Heinz School, Carnegie-Mellon University, Working Paper 1996-4, January 1996, available at;jsessionid=2031901109367516906?id=35.