<p>If I recall correctly, the BDM data was mean LSAT vs mean GPA of law school applicants from each college, as pulled from a self reported web site. Is that right? This served to generate a regression of gpa, in completely non standardized curricula that would vary from one college to another, vs performance on the LSAT, which is not a test of what one learned in college. It might indicate that students at the most selective colleges have higher LSAT’s than their grades would predict, but this does not tell us how equivalent work would have been graded at different colleges. The LSAT does not test what they would have done in college, so one cannot use it to norm one college to another.</p>
<p>The problem is that the LSAT tests only certain abilities, which are not universally needed for academic success, although they seem to predict academic success in law school. But it is a poor proxy for overall academic accomplishment.</p>
<p>One could include the mean SAT for these same colleges to attempt to adjust for academic ability of the student body overall. That is, if grading standards were uniform, one would expect the colleges with higher mean SAT’s to have higher mean GPA’s. The SAT is in many ways a test of overall intelligence, with bonus points for coming from a well educated upper middle class background. Both contribute to better college performance. Evidence of “easy” grading would be higher mean GPA than predicted based on SAT.</p>
<p>In comparing across colleges, one would also have to address the question of systematic differences in GPA by broad field (say engineering vs humanities). There is some data that engineering has lower grades overall than humanities, and that this is observed at many colleges. Thus a college with a high proportion of engineering majors may have a lower mean GPA than a college with few or no engineers. This could occur even if every engineering project were graded the same at both colleges, and for that matter if every humanities paper were graded the same way as well.</p>
<p>
</p>
<p>Right. </p>
<p>Either you know what would happen, or you don’t. Since it is untestable, you don’t know. So, as you so well illustrated, any such claims are fiction. </p>
<p>The fact that no one has data does not make all fiction valid. It just means there is no data.</p>