An Idea on Grade Inflation

Grade inflation is back in the news, as the Harvard Crimson reports that the median grade at Harvard is an A-.  This is clearly absurd.  It reminds me of some of the old Olympics judging where they had a 10 point scale but everyone scored between 9.7 and 9.9.  The problem is not necessarily that the mean is skewed, but that there is almost no room left to discriminate between high and low performance.

There is one potential way to combat this, and it was invented by colleges themselves.  Consider grading in high school.  My kids go to a very tough-grading private school where A's are actually hard to get.  The school sends (for Arizona) a fairly high percentage of its students to Ivy and Ivy-level schools, but the school produces someone with a perfect 4.0 only once every four or five years.  Compare that to our local public school, that seems to produce dozens of perfect 4.0's every year -- in fact since it adds a point for honors classes, it produces a bunch of 5.0's.

Colleges understand that a 3.7 from Tough-grading High may be better than a 5.0 from We-have-a-great-football-team High.  They solve this by demanding that when high schools provide them with a transcript, it also provide them with data on things like the distribution of grades.

Employers should demand something similar from colleges.  This is a little harder for employers, since colleges seem to be allowed to legally collude on such issues while employers can get sued over it.  But it seems perfectly reasonable that an employer should demand, say, not only the student's grade for each class but also the median and 90th percentile grades given in that same class.  This will allow an employer to see how the school performed relative to the rest of the class, which is really what the employer cares about.  And schools that have too many situations where the student got an A, the median was an A, and the 90th percentile was an A may get punished over time with less interest from the hiring community.

One way to get this going is for an influential institution to start printing transcripts this way.  The right place to start would be a great institution that feels it has held the line more on grade inflation.  My alma mater Princeton claims to be in this camp, and I would love to see them take leadership on this (the campus joke at Princeton during the Hepatitis C outbreak there was that at Harvard it would have been Hepatitis A).

Postcript - An alternate grading system from Harvard Business School:  When I was at HBS, they did not give A's and B's.  We had three grades called category I, II, and III.   By rule, the professor gave the top 15% of the class category I, the bottom 10% category III, and everyone else got a category II.  I actually thought this was a hell of a system.  It discriminated at the top, and provided just enough fear of failure to keep people from slacking.


  1. Noah Body:

    Yes, lets compete such that the bottom ten percent get shafted without regard to anything they actually learned. Hint: Education has NOTHING to do with learning!

    University educations are for chumps. The whole system is going to blow up in the next ten years and I will laugh at everyone, especially those with huge student loan debts (which cannot be dissolved by bankruptcy, BTW). Fools! Idiots! Bovine led to the slaughter!

  2. NL7:

    Not sure it makes sense to make employers analyze the grade distribution of many schools. It might make sense if they could focus on interpreting just a handful of different grading curves, but making them analyze it every time makes it harder to compare across applicants. Note that a number of colleges have both track records and rubrics for interpreting high school grades, and the top colleges often have an idea of high school quality and which schools are their feeders and in what grade range. So it's less analysis than it seems and a little more habit.

    The simpler way to compare across a large population is standardized testing with universal scoring. Something like the GRE, GMAT, LSAT, MCAT, etc. Those tests have their shortcomings and can only possibly test for a limited number of skills at once. But it would be easy for an employer to compare two individuals directly, which is the way that the SAT works.

    Honors grading is practiced at top schools in certain professions. Law firms are prestige-focused and grade-fixated, but they accept that the top handful of law schools like Yale and Harvard have a loose honors system. The understanding is that the schools already engaged in pre-selection of appropriately qualified students, so law firms sorting through applicants don't have to worry that a 30th percentile student is indistinguishable from a 70th percentile student - both will be fine. But that does not apply to every school, particularly those without elite reputations.

    Being undifferentiated in a class of 10,000 or more students could benefit the students with pretty good marks, who consistently get honors and never get low-pass. But for students in the upper middle (say 70th percentile) who rarely get honors, grades will no longer separate them from the middle and lower middle. So although it would probably produce a more relaxed atmosphere, it sacrifices information and signaling to employers - and to high-volume hiring employers (anybody filtering through a half-dozen or more applicants from the school per year), that lack of information will make them pursue the best students at the expense of the students who are merely above average.

  3. me:

    Based on a small sample size and trying not to come across as biased: I have found graduates from the real big names in my field (including Stanford, the MIT, Havard, Yale and Columbia) to be on average much less skilled and thoughtful in their problem solving than the graduates from overseas universities I am managing. One notable difference is that the Ivy Leaguers tend to be much better in knowing how to appear knowledgeable and in control than the average foreign candidate (Think "Oh, don't worry, I'll get this done by next week" followed by a reasonable excuse or the discovery of a really interesting problem that will require more time as opposed to "Oh, I need to think about this more before I commit"). This is not just a disparity between native and non-native speakers, a similar differential is very visible comparing to US second tier universities. I wouldn't be surprised if the big names churn out more political animals than average given pressure and incentives. And, to be clear, I met some fabulous individuals from lots of institutions, very much including the big ones. It's just that the differential isn't where I'd have expected it to be.

  4. jdgalt:

    It seems to me that universities have always sold a package of three products, and actually learning useful material is the least important of the three (especially in fields like computing, where anything you learn will be obsolete in 5 years anyway).

    #2 in importance is the ability to study, or at least do research, and get results that answer a given question.

    And #1, of course, is being willing to "fit in" by showing up on time and cooperating with endless BS without protest. This is the quality employers most need you to have. (In modern universities this includes enduring outrageous political indoctrination sessions from the prof, just as happens under communism. Some companies and government agencies are the same way.)

    I strongly suspect that at the college level, just as at the grade- and high-school levels, someone who's been homeschooled will do far better than even a Harvard grad at priorities #3 and #2. But a lot of employers won't want him/her because of the lack of #1, so he should probably figure on starting his own company and trying to be the next Sergey Brin.

  5. David Zetland:

    You're right to call attention to this problem (again). Your ps calls for a curve, which helps separate out students, but it's unpopular with professors who like to give As, departments that attract students with "easy A" courses, and universities that turn all their special snowflakes into a A students. After all, if you're paying $60k, don't you DESERVE As?