Success for my analysis would be finish the program. Actually, really not attached to the definition. Would have to have a way to sort the data given it will be hard to determine who has given up on the program for good. But, that is nothing a good statistician couldn’t handle. Might just assume a two year canidacy gap means your done, if those that take such a break rarely come back. All sorts of great stats could be generated.
Yep, if the correlation is that low, the GMAT score would not be something a candidate should consider. Just another note, I would think GMAT verbal scores are even more highly correlated to Ethics scores than the total score is to the test as a whole. I wouldn’t put any money on that guess though.