On March 20, Freddie deBoer had a piece here at People’s Policy Project in which he argued that tests like the SAT level the playing field for disadvantaged students better than alternative sorting methods. I would summarize his argument as such: we currently use (1) grades, (2) holistic assessments, and (3) tests to sort students for college admissions processes. Empirical research shows that grades are gamed by the rich. Basic intuition tells us that holistic criteria are more easily gamed by the rich than the poor. Empirical research suggests tests are not that easily gamed by the rich, which means tests are the best sorting method, at least from the perspective of disadvantaged students.

Robinson’s Response

On March 31, Nathan Robinson wrote a rebuttal at Current Affairs in which he argues that we should dial down the amount of selectivity we use in the college admissions process.

In other words: just admit everybody. The whole “competitive” nature of undergraduate admissions is absurd to begin with, and the very fact that students are sorted according to “merit” is socially corrosive. […] Instead of finding the “top ten best people” we should be selecting “anyone who has proved they are capable of doing the expected work.” […] Here’s the admissions parallel: everyone who shows themselves capable of doing the work required of a Harvard undergrad is marked “qualified” for Harvard and allowed to apply. There are a limited number of places, of course, but those places will be filled by selecting a random group of students from among all of those marked “qualified.” You might still get a very low percentage of applicants admitted because space is limited, but it won’t be because those applicants have been deemed worthier, it will be because the lottery happened to favor them.

The argument that we should set minimum bars for admissions and then use lotteries for everyone who clears those bars, as opposed to being ultra-selective, is a good one. But it is not actually a rebuttal to deBoer’s point. The level of selectivity is a completely separate question from how to sort students. For any given level of selectivity, it will still be necessary to sort students into various buckets. Even at the lowest level of conceivable selectivity — e.g. literacy — you will still have to use some method to sort the literate from the illiterate.

Robinson says he would place “everyone who shows themselves capable of doing the work required of a Harvard undergrad” into a lottery in order to determine who gets into the school. But how would you go about showing yourself capable? Some method will need to be used to sort the capable from the incapable (however defined) and so that then raises the question of which sorting method is the fairest to the disadvantaged. And the answer seems again to be tests, a point that Robinson agrees with near the top of his piece: “The conclusion I agree with is that the SAT may be the ‘least bad’ of three options for competitive admissions, when compared with using grades or Mushy Holistic Factors.”

It’s important for a clear debate to emphasize that the pro-tests position is agnostic to the level of selectivity. It is about how to sort into selected and unselected buckets, not about how big or small those buckets should be. Robinson’s rebuttal is thus about an important but ultimately orthogonal question.

Bias

Robinson is the only one who has offered a formal response, but others have chimed in through other channels. The most common response in those other channels appears to be that the college admissions tests are biased, which we know because poor and disadvantaged students do worse on them than rich and otherwise privileged students. This argument seems very confused to me.

Any method of academic evaluation you use is going to sort affluent kids higher than non-affluent kids on average. This is because affluent kids acquire a higher average level of academic competency than non-affluent kids. To believe that all groups have equal average academic competency at age 18 is to believe that poverty has no negative effects on learning, which is clearly wrong.

In addition to having a higher average level of academic competency, affluent kids also have a greater ability to game any kind of assessment that you use. This means that affluent kids can exploit techniques that cause them to be sorted higher than their academic competency, properly measured, would sort them. The argument for one-off tests is that they are the least easy for the affluent to game and thus offer the most level playing field. That the tests don’t generate the (incorrect) result that there are no average academic differences between affluent and non-affluent kids at age 18 does not negate this point.

For these reasons, I think it is very plausible that using tests plus affirmative action is the fairest way to sort students for admissions, as deBoer argues. Since I really do not care that much about this issue, I could easily be convinced otherwise, but none of the counterarguments I’ve seen so far have been very persuasive.