Wednesday, October 24, 2007

At the elite colleges - dim white kids

Education has been called the key to upward social mobility. It's the great equalizer. If you're smart enough, you can do anything; regardless of money.

For a number of students, they're being denied the opportunity to enter college due to less qualified students being chosen over them. Who are these less qualified admittees? No, it's not minorities due to affirmative action. It's kids who are often being offered second chances despite being wealthy and connected young people who have squandered many of the advantages life has offered them.

Researchers with access to closely guarded college admissions data have found that, on the whole, about 15 percent of freshmen enrolled at America's highly selective colleges are white teens who failed to meet their institutions' minimum admissions standards.

This means opportunities lost for those who at every turn in their youth have had the harder row to hoe, despite that are succeeding, and are just looking for their first break. Sad.


Oneway the Herald said...

Either you espouse a meritocracy, or you don't. If you favor race-based admissions, you have nothing to cry about.

Chairman said...

Well, I'm not sure if you call our alma mater elite. In any case, I have personally observed an overwhelming number of dim white kids. Of course, I have also personally observed an overwhelming number of dim colored kids, as well.

Westy said...

What is a true measure of a meritocracy, though?

Does it reward those who have already achieved the most or does it work to help those who ultimately would?

Chairman said...

Westy. Sort of tangential to your point, but it turns out that, in terms of easily obtained statistics, standardized test scores and GPA are an awfully good predictor of performance. Standardized test scores offer a pretty good measure of general ability. GPA offers a pretty good measure of motivation and effort.

However, what it does not pick up is perspective or social network. Legacies are crucial to have, because one of the factors that drive people to elite schools is the network that can be developed.

With regard to the original point, I think that there is a very important, but unanswered debate about what the role of a university actually is. If the answer it to create knowledge, then I'm all for meritocracy. However, I'd argue that meritocracy is only truly beneficial in areas where problems are the most complex. When it comes to simple to moderate tasks, people are so overeducated that meritocracy is irrelevant. You don't need to pick from the high side. You only need to cut out from the bottom. After that, the difference in performance isn't worth the effort in selecting the best available candidate.

Westy said...

You're probably right, Chairman.

But then what is wrong with preferentially cutting from the bottom those who have achieved those 'just enough above average' standardized test scores and GPAs with lots of help as opposed to those who did it in spite of negative surroundings?

Chairman said...

Nothing. I think that from a macro-level view, it's largely irrelevant. Of course, I wouldn't necessarily worry about cutting from the bottom. Cut where you want. Put in legacies. Put in colored candidates. Put in dwarves and midgets. Whatever.

Let's say that you have 1000 candidates who are more qualified than chimps vying for a spot on a roster of 100. Within that roster of 100, if there are only 25 really crucial roles, while the other 75 will be given roles that could be done by chimps. You want to make sure that you get the top 25 out of the 1000 into your roster. After that, really any of the remaining 975 candidates can be fit into one of the remaining 75 roster spots. Do you really need #26 through #100? Nope. You'd end up just as well by drawing numbers at random. In fact, I'd probably cut #26 through #100, just to make sure that the bitching and moaning is at a minimum. Now, if your buddy's kid is a candidate, and not in the top 25, what does it hurt if you put him in, even if he's #963, out of 1000, so long as he's doing a job that a chimp could do? Put in the black chick that's ranked #814, or the Navajo kid that's ranked #626. Whatever. Nepotism and legacy only hurt when people are put into roles that they are unable to succeed in.

Westy said...

The question remains, how much of a negative impact on future test scores and grades does a substandard upbringing result in? And is that worth adjusting for?

Levitt and Fryer argue that the negative influence begins before kindergarten and that otherwise (without the negative influence of poor upbringings) kids of separate socioeconomic strata would achieve equally.

I would still argue that access to an elite school, if denied, could have very real impacts on a student who needs an elite name on their degree to set them apart from their prior life experiences.

Chairman said...

The Levitt and Fryer blog post (I haven't read the working paper) allows for a lot of intriguing debate. I think that they suggest that until age 2, there's no significant difference in race. I'm assuming that you're suggesting that by kindergarten, there is a difference.

Lots of possibilities emerge here, suggesting that it's not simply education at play, since you wouldn't get divergence until after the age of 5 or 6. It's more likely that there's some difference in the socialization process that has a significant effect on intellectual ability. I don't doubt that.

Now the question of whether it is worth adjusting for is a fascinating one. It depends on how confident you are in your measures. If you believe that SAT score and GPA can predict 100% for your purposes, then no, no need to adjust. If you believe that they can explain, say 49% of the variance, then maybe. Why 49%? Remember that when you have a correlation, a very high value (R) for a correlation is 0.7. Now, your variance explained is R^2. So, even with a very good measure, you're statistically only explaining 49% of the variance. As a benchmark, most psychological measures are very happy when they get correlations of .4, which corresponds to explanation of 16% the variance. And factors with R = 0.1 - 0.2 (R^2 of 1-4%) can still considered to be very important in research.

I believe that I've said that general ability is the single best predictor of performance, wealth, etc. However, those R values come in at around .5, I believe, or only 25% of the variance. Which suggests that there's 75% of the phenomenon that remains. I would guess that SAT score and GPA would be roughly the same.

Meritocracy is fine, in theory, but in practice, how your measure merit may be precise, but there's a definite possibility that it is largely inaccurate. So, methodologically, adjustments from your measure NEED to be done, if you want a "true" measure of merit. However, HOW you do it is up in the air, and adjusting for race may not help, and may even hurt. Measurement issues are not the place for black and white answers. They are places for asking yourself how much uncertainty are you willing to accommodate in your process.

Interestingly, I'm posting more on IJAB than I am in the Board Room.

Westy said...

Good thoughts, and I agree. Not sure what the best way to make an adjustment is either. But it's worth searching for it I do think.

Chairman said...

From a holistic perspective, how you answer your question says a lot about how your view the world in terms of equity and justice. The more I hear from people, the more I'm scared of both the extreme left and the extreme right.

Of course, you know you're in grad school, when you start making analogies between chimps and kids in Harvard.