Wednesday, October 29, 2008
My former colleague Mark Gergen (now at Berkeley) writes a propos our recent post on the Yale 'no grade' system which some other schools appear to be adopting:
Your posting on the Harvard and Stanford grading systems got me thinking. Berkeley has long had a P/H/HH system. I gather 60% of the class gets P, 30% gets H, and 10% HH. In addition, in each course a prize is given for 1^st and 2^nd best performance. Assuming “better” students persistently perform better on exams and that grading is not random, this system will enable employers to identify the very top of the class (meaning students who persistently excel) about as well as more finely grained systems. I expect it also gives employers a good sense of rough place in the class for many students. Where precisely the system becomes opaque depends on how many students get all Ps, which is a function of the distribution of ability in a class and the randomness of grading. For example, if there is a long tail in performance at the bottom and grading is somewhat random, then the very bottom may stand out as having no H’s while students between the 15^th and 40^th percentile may be indistinguishable. If there is a short tail at the bottom and grading is somewhat random, then the bottom 40% may be indistinguishable. And so on. The system should not have the effect you predict. Students reaching for the most selective positions may “suck up” to burnish a lustrous transcript. They will not do so to stand out from an otherwise undifferentiated mass.
Of course, the key here is that Berkeley really uses, according to this presentation, three different grades and they do so on a curve. No one has yet given me a straight account of whether Harvard and Stanford will use more than two gradations, and whether there will be any meaningful curve.