[Chua] "Not Being the No. 1 Student"

This is part of several posts in response to Amy Chua’s article “Why Chinese Mothers Are Superior” in the Wall Street Journal. Posts on this topic are tagged as [Chua].

There’s a common joke/meme amongst most of my friends of the “Asian Fail”: a B or even an A-. Heck, there’s even an image macro of the “high expectation Asian father”, which can actually be funny enough to overcome the racism (sometimes). But the thing is, it’s not a joke.

For example, if a child comes home with an A-minus on a test, a Western parent will most likely praise the child. The Chinese mother will gasp in horror and ask what went wrong.

I’ll come back to this and explain why I’m not entirely against Chua’s viewpoint, but if everyone was supposed to get As, why would we have BCDF?

The problem comes down to there being two kinds of evaluations: tests where you should be able to do everything (i.e. expected score = 100%) and tests that actually evaluate where you are (expected score = maybe 50%). The former is useful for checking competence (driving tests, certifications, etc.). The latter is for evaluating how good you are at something (AP tests, perhaps final exams, and SATs in theory).

Read that again. I just pointed out something that’s true about AP tests: the average score is calibrated to be a 3, if maybe a high 3, and very few people should be getting 1s and 5s. The AP tests, BTW, are the “best” standardized tests I’ve ever taken, because they actually have a gradient of questions from easy to hard and covering lots of different topics. (The SATs, on the other hand, have a lot of very easy questions and a couple of very hard questions, and not so many medium questions. The questions are also stupider and less of an assessment of understanding, even in the same subject area.)

Read that again: I just said you should get 50% on your final.

At Berkeley, a lot of people freak out because they’ve been getting As all their lives, and suddenly they’re getting Cs. It’s not just that things have gotten harder, or that expectations are higher; it’s that the very standards for a class have shifted. It’s not the expectation any longer that “everyone gets As”. Students are supposed to have to put in effort, probably extra effort, to get to an A. Or in other words, classes are less and less about certification and more and more about evaluation.

This has been on every CS61A exam for years. (That’s the class I TA for.)

Our expectation is that many of you will not complete one or two of these questions. If you find one question difficult, leave it for later; start with the ones you find easier.

IMHO, the best exams we’ve written have questions that break up the students in different ways. Most questions should separate those who are lost (or ignoring the class) from those who are trying. But there should also be questions that separate the “A students” from the “B students” and the “B students” from the “C students”. Exams with abnormally low or high averages are usually missing one of these: either there’s nothing to challenge someone doing well, or there’s nothing to allow for people who kinda get it but not entirely.

You might still say someone should understand everything in a class, and/or that the class shouldn’t try to teach so much material that so few people get As. To that I say: “Bah!” And then: “What about the people who can handle it? Are they never going to be challenged for their entire life? How can they ever grow?” If you don’t know everything, then you shouldn’t get a perfect score!

Let’s pause for a little background here. I am one of those people who has been able to coast most of my life, and is usually glad to be challenged with a class that poses problems that you aren’t supposed to be able to solve easily. Separately, I’m rather dismissive of grades and standardized evaluations that are supposed to measure knowledge and performance.

I went to an alternative elementary school that didn’t believe in grades. Our report cards were all qualitative, and Academic Performance was just another category along with things like “Social Skills”. The school was extended through 8th grade, and so I didn’t get real grades until high school. Marks on individual assignments, sure. But never were we reduced to numbers or letters.

You might think that this could backfire terribly, and indeed McAuliffe got a reputation as a “party school” for this and our other atypicalities. More concretely, people going from McAuliffe into a regular school often were academically behind their peers. I won’t deny that our science program was pretty bad. But for students like me, already academically focused and driven, McAuliffe was a great place to learn patterns for the rest of life, exploration and creation and respect. It’s entirely possible (though not likely) that a more traditional elementary school would have trampled out a lot of what makes me me.

But okay, I get good grades, so dismissing their importance isn’t saying much, right? Maybe if I had to worry about them? Let’s get back to the Chua article and see why this isn’t exactly true. Here’s the quote from above once more:

For example, if a child comes home with an A-minus on a test, a Western parent will most likely praise the child. The Chinese mother will gasp in horror and ask what went wrong.

What does an A- mean? If the test is an evaluative one, designed to spread people out on a spectrum so they can see how much you know, an A- is pretty darn good. It means you’re in, say, the top 10%. But if the test is a competency one, then the Chinese mother is right. You don’t know how to do something that you’re supposed to be able to do!

(I feel a little funny when I go back to ask about something I got wrong on a test where I already got well above the average. But the grade isn’t what’s important — I got something wrong, and I’d like to know what I did wrong so I can do it right next time. Nothing wrong with that.)

The problem is that it’s really hard to tell one kind of test from the other, though as we get older the tests shift more and more evaluative. Still, in a college CS course, the two kinds of, uh, evaluation are usually mixed. Here’s another snippet from the CS61A introductory packet:

A typical C- (barely passing) student would be one who gets 60% scores on exams (114 points), 75% on projects (60 points), and 90% on homework (27 points) [out of a total of 300 points in the course].

Homeworks are graded on effort, not correctness. Projects are graded on correctness, but you should be able to do everything (certification). And exams are graded on correctness, without (necessarily) being able to do everything (evaluation – though it’s never stuff we haven’t covered!). So the 60% for exams and 75% on projects pretty much match up.

So, an A-, or *gasp* a B might not be the end of the world. But it’s not the grade that matters. The real question is, should your kid have understood the questions they missed? If so, they need to work it out, and they should get an A next time! It doesn’t matter whether you get an A or a C; if you’re in fifth grade and you can’t multiply then you’re behind. The only thing gades do is set expectations, and in this case the expectation was near 100%.

But if not, give your kid a break. You aren’t supposed to get an A in every class, and the effort required to do the extra work is going to cost them something (possibly in another class!). It doesn’t matter whether you get an A or a C; if you’re in fifth grade you don’t need to be able to solve a general quadratic equation. The only thing grades do is set expectations, and in this case the expectation was that most students would not be able to solve all problems.

(As an epilogue, I would advocate the “check” system, which is used, for example, in the Linguistics department at UCB. A correct but unremarkable homework gets a check. Homework that has the idea but is missing a few details gets a check-minus. Homework that really doesn’t have the idea gets a minus. Homework that goes beyond what’s asked for gets a check-plus. And really stellar homework gets a plus – which might never happen in a semester. These grades, plus, check-plus, check, check-minus, and minus could be mapped to A+, A-/B+, B-/C+, C-/D+, and F, but somehow people don’t mind only getting checks.)

(The flip side of this argument is why I dislike bell-curve grading on principle. A topic for a later post.)

There are two kinds of tests. The important thing isn’t the grade, it’s what you missed. And hey, that applies even without parenting coming into the picture.

(It’s too bad, however, that the world still cares about your grades and your scores on crappy standardized tests.)