After the SAT

I have mixed feelings about it. It has worked well for Bowdoin, but it has become harder and harder as we have become more popular. Right now, we have about 4,500 applicants for 435 places. And even that is a challenge, although our staff has expanded. Thirty years ago, it was easier to evaluate student ability and achievement than it is today.

One thing that has changed is that so many schools have abandoned class rank. In 1969 and throughout the '70s, most schools did look at a weighted class rank. That -- how well a student performed in a challenging program -- was a very good predictor of success. Usually, it was a better predictor of success than test scores. Now, many schools have moved away from looking at class rank -- though it may swing back in places like California and Texas.

Does Bowdoin look at class rank when it's available? What do you look at in its absence?

Oh, yes, we consider class rank if we have it. I think we do an excellent job of predicting success, but it involves an awful lot more detective work. We spend an incredible amount of time looking at school profiles, looking at thousands of different school systems across the country, to put each student's achievement in context. To do that, we must understand the grading system -- grade distribution, how much grade inflation or deflation there is at a particular school.

Obviously, we place more and more emphasis on recommendations. In the absence of test scores, we need the other information that allows us to successfully predict achievement.

Thirty years ago, the argument for schools looking at the SAT was that it would lead to admissions based on merit, as opposed to admissions based on privilege.

I'm not one who advocates abandoning testing for all schools. It's not something I go out on the stump to help promote. I feel that it works for us, and I know why it works for us. I think as soon as it didn't work for Bowdoin, we would think seriously about moving back to requiring the SAT for all students.

As it works out, most of our candidates do submit their test results, and we find it somewhat helpful. I think we use the test properly: We use it as a supplement to what we see in the rest of the file. I am not one who says that the tests are bad or flawed. None of them are perfect, but I do think that, in general, they are helpful. People are sometimes surprised to hear me take that position.

So you don't require applicants to submit test scores, but you do take them into consideration if they are submitted. Let's say you have two candidates who are more or less equal, but one has submitted test scores and the other has not. How often do you find yourself using the test scores as a tiebreaker?

I don't think of them that way. We often work with very successful candidates who, for one reason or another, choose not to submit their test scores. We don't assume that because a candidate has not submitted their test scores that their scores are weak. Some do it on philosophical grounds and say that they want to be evaluated on the strength of their performance and so forth.

We have had to develop, in fairness, a pretty elaborate rating system. When we run our validity studies -- which we do fairly regularly -- we try to work from earned GPA here at Bowdoin and work backwards. We do the research to see if we predicted as well for those who submitted their scores as we did for those who did not. We keep testing that.

What we find is that our rating system is by far the best predictor of college success. Every candidate goes through two different readers, and then a committee discussion, so it doesn't just come down to a decision that I would make singlehandedly between one candidate with SAT scores and another without. This system factors in all kinds of criteria: recommendations, writing ability, test scores if we have them and so forth.

Fortunately, we've been very pleased with that result. That's by far our best predictor. If we add in the verbal and the math SAT scores, we gain a little, but not a lot. It's certainly not useless information, but we have other predictors that are better.

How does your office handle such a personalized system of admissions?

Next Page: I think it's going to be quite interesting if the U.C. system does this

• View blog reactions