<p>"However, it's hard for me to see how they can be properly measured any better by grades. Just because a person has top grades doesn't mean that he has good teaching ability, nor has he necessarily demonstrated patience/resolve, nor the ability to compose, nor the ability to analyze new information."</p>
<p>I'm curious as to what kind of class could fail to measure these qualities. Except for a few math courses, every one of my undergrad classes assessed research, writing, patience in completing long-term projects, and several measured teaching ability as well. In fact, the graduate courses I've taken in my major are virtually identical to the upper-division undergrad courses except for quantity of work assigned (longer reading lists, etc.). The problem with GPA is variation in the grade curve - between colleges, between fields of study, between individual teachers - but it seems odd to argue that grades do not measure grad-school-relevant qualities.</p>
<p>I am genuinely interested in the substitute test you propose, though. Posting a sample question might be helpful in getting a sense of this alternative.</p>