3 ms·

> Many universities have decided that they value diversity (ever how they define that) higher than or equal to pure academic results. This is a rational thing

by geofft 7y ago

> Many universities have decided that they value diversity (ever how they define that) higher than or equal to pure academic results. This is a rational thing to do, because "pure academic results" are hard to get a useful measure from. Here's an analogy. Suppose you are looking for the most secure encryption algorithm. One thing you could measure is the number of unbroken messages generated from the algorithm. This seems to make sense, because good algorithms will be broken less easily than bad ones. But a bit of thinking will demonstrate this is a bad idea. There are a ton of DES messages in the world, most of which nobody has bothered to break. So by sheer numbers we would conclude DES is more secure than AES, and both are much more secure than any brand née algorithm. Since this is untrue, our heuristic must be wrong. Meanwhile, another heuristic of giving intentional and honest consideration to new, unproven algorithms seems likely to work better. In the same way, if you want to select for academic potential or * aptitude, measuring prior academic results* may not be the highest-value signal.

zaroth 7y ago
The point is to select for diversity for the purpose of diversity, not for the purpose of academic potential. Selecting for prior academic success is a good way to predict future academic success — your analogy about encryption makes no sense at all. Academic success doesn’t fail in any of the same ways that encryption does. But the problem is that selecting for prior academic success does not select for diversity, because the environments which support early adolescent academic success are not as diverse as colleges want their own student populations to be. So colleges are left trying to increase the diversity of incoming freshman by passing over more academically successful candidates which don’t meet their diversity quotas. There is tremendous value in having a diverse campus population, but the best way to get there is to have diverse academically successful applicants. Colleges aren’t willing to wait. One problem is that these applicants are also generally paying a lot of money to attend these colleges, and if they ultimately aren’t going to be successful those colleges are causing significant harm by admitting them. Note that one of the primary purposes of the SAT is to predict success in the first two years of college. The not very good alternative is to lower academic standards in the classes themselves, and certainly there is anecdotal evidence of this but I’m not sure about rigorous studies.

by geofft 7y ago

zaroth 7y ago

The point is to select for diversity for the purpose of diversity, not for the purpose of academic potential. Selecting for prior academic success is a good way to predict future academic success — your analogy about encryption makes no sense at all. Academic success doesn’t fail in any of the same ways that encryption does. But the problem is that selecting for prior academic success does not select for diversity, because the environments which support early adolescent academic success are not as diverse as colleges want their own student populations to be. So colleges are left trying to increase the diversity of incoming freshman by passing over more academically successful candidates which don’t meet their diversity quotas. There is tremendous value in having a diverse campus population, but the best way to get there is to have diverse academically successful applicants. Colleges aren’t willing to wait. One problem is that these applicants are also generally paying a lot of money to attend these colleges, and if they ultimately aren’t going to be successful those colleges are causing significant harm by admitting them. Note that one of the primary purposes of the SAT is to predict success in the first two years of college. The not very good alternative is to lower academic standards in the classes themselves, and certainly there is anecdotal evidence of this but I’m not sure about rigorous studies.