<p>Okay, this would give a context for each student's grade. However, the context is still poor, for multiple reasons. First, just the mean and median really just says if the student was better or worse than most (or the average). For top schools, knowing that a student was above average still won't be sufficient, right? So we should also include standard deviation, and then why not just sent out a little histogram, with all the data? But even then, if you provide absolutely full transparency on the performance of all students in a class, this information still has multiple significant flaws. First, and most in line with previous comments, getting a grade that's only slightly above average in a course where the students are highly self selecting (or selected), this doesn't give a clear picture of the actual abilities of the student. If you perform 1.5 standard deviations above the mean in a course which most people aren't willing to take because it's known for being very difficult, then that 1.5 probably doesn't give a very clear picture of the student's abilities. Secondly, many courses aren't properly designed to give grades that would work with this scheme. Even if you added in a complete histogram of grades for the course, we all know there are professors who design their courses such that most students can get an A -- and thus even the top grades look a lot like the average, just because the grading scale tops out to soon. </p>
<p>Really, I think (and I've spent a lot of time thinking about this), that if we wanted to institute a grading system which provided truly valuable results between institutions, we should establish a system in which all classes are graded in a way that consistently gives rise to a normal distribution, and then test each course against controls. If it were effectively impossible to get anywhere close to the top or the bottom of the grading scale, and you then got a control group that was designed to be representative of the average of all college students, and both histograms (the control group and the class) was reported, then adcoms could see where a student stands relative to his peers, and relative to students nationwide. The costs, both financial, and in terms of the educational opportunities that would be lost because of the difficulty of putting together a course that was truly normalized, would be disastrous. An imperfect, and in fact nearly useless, hodge-podge of grading systems is the price we pay for flexibility within the system.</p>
<p>By the way, what do you guys think of cognitive science as a major? Say, with a computational modeling specialization? I keep wavering on whether it's good, or horrible. On the one hand, the study program I'm planning, heavily supported by related electives, incorporates the analytical and logical abilities required for computer science, a little analytic philosophy, with bits of psychology (case study), and a smattering of analytic linguistics, and formal modeling material, demonstrating an ability to handle formal systems, and think logically about complex systems. However, at the same time, much of the details of what goes on in cogsci is still obscure, and it doesn't include specific knowledge or skill sets which might contextualize portions of the law school curriculumn, in the way that econ, business or history might.</p>