Conversations

A student visited my office and told me that my test had a lot of very hard questions. I told him that the test had C questions, B questions, and A questions so that I could distinguish the various students. He replied that he was a C student, and he didn’t think it was fair that he had to take a test with A questions.

I offered to write him a test with only C questions, but he would need to get at least 90 percent correct. He thought for a moment, expressed that he had never achieved 90 percent on a test in his life and then consented to take the regular test.

ChatGPT said:

I like this story because your student’s argument is more interesting than it first appears. He had essentially converted “I am a C student” from a description of past performance into an entitlement to C-level expectations.

Your response exposed the flaw beautifully without lecturing him. You effectively said: Fine. If we remove the questions that distinguish excellent understanding from adequate understanding, then you must demonstrate near-perfect mastery of the adequate material. Suddenly the regular test didn’t seem so unfair.

There are several things going on here.

First, grades are supposed to describe performance, not prescribe difficulty. A C student isn’t someone who should be given C-level work. He’s someone whose performance on the common body of work falls in the C range. His proposal would have reversed the relationship: determine his grade category first, then construct an assessment appropriate to it. Taken to its logical conclusion, the A students would get the hardest tests and the weakest students the easiest ones—which would make comparisons among grades nearly meaningless.

Second, your alternative revealed something about the mathematics of assessment. On a mixed test, a student can earn a C by answering most of the basic questions, some of the intermediate questions, perhaps an occasional difficult question, and missing others. Once you remove the harder questions, however, there is much less room for error if a C is still supposed to represent the same level of accomplishment. Your 90% requirement made that tradeoff immediately visible.

Third—and I think this is the best part—you didn’t argue with his self-description. You accepted his premise: You’re a C student? All right. Let’s design a C test. Then you allowed the consequences of his own proposal to do the teaching. That’s often much more effective than explaining why somebody’s reasoning is wrong.

There’s also something rather revealing in his statement, “I’ve never achieved 90 percent on a test in my life.” Without realizing it, he supplied evidence against his own position. His difficulty wasn’t merely that your tests contained A questions. Even on material he regarded as being at his level, he didn’t expect himself to demonstrate high mastery.

And I wouldn’t interpret his original complaint as necessarily lazy or manipulative. I can imagine the reasoning being perfectly sincere:

“I’m not trying to earn an A, so why should I have to answer questions intended for people who are?”

That sounds reasonable until one recognizes that he didn’t actually have to answer the A questions correctly to earn a C. Their presence wasn’t preventing him from being a C student; they were giving other students an opportunity to demonstrate that they weren’t C students.

There is a nice little educational principle buried in the episode:

The purpose of a difficult question isn’t necessarily to see whether everyone can answer it. Sometimes its purpose is to discover who can.

And perhaps the funniest indication that the lesson worked is the ending. You didn’t convince him that your test was easy. You convinced him that the alternative he had requested was worse.

His moment of thought before consenting to the regular test may have contained more learning than anything that appeared on the test itself.

I’ve written plenty of tests in my career. My philosophy is that life is like a test, except that it is custom tailored to every individual to produce the greatest result possible. Therefore, one cannot say his trials are more or less than mine or any other’s. All are tested to the extent possible. Some tests are just more visible or obvious than others.

Yes. And I think your test story becomes considerably more interesting when placed beside that philosophy.

When you wrote an exam containing C, B, and A questions, you weren’t giving three different tests. Everyone encountered the whole test, but the questions revealed different levels of ability. Your student interpreted the hard questions as something being done to him: “I’m a C student; why should I have to face A questions?” You regarded them as opportunities to reveal what he could do.

Your philosophy of life takes that idea one step further: unlike your classroom tests, life’s test isn’t standardized.

If the purpose were simply to rank people, everyone ought to receive approximately the same examination. But if the purpose is to develop people—to produce, as you put it, “the greatest result possible”—then identical tests would actually be unfair. What severely tests one person might be almost effortless for another.

That leads to a consequence I think is central to what you’re saying: the apparent size of a trial tells us very little about the difficulty of the test.

One person’s test may be cancer, financial ruin, bereavement, or some other event everybody recognizes as adversity. Another person’s may involve patience, pride, resentment, loneliness, responsibility, temptation, forgiveness, or simply doing an unremarkable duty faithfully for forty years. The first kind attracts attention. The second often doesn’t. Yet visibility isn’t a measure of difficulty.

It also makes comparison nearly meaningless. We ordinarily compare circumstances:

“Look what she has had to endure compared with him.”

But we cannot see the complete examination. We don’t know another person’s capacities, vulnerabilities, opportunities, understanding, responsibilities, or what a particular experience requires of that individual. Consequently, we don’t have enough information to grade somebody else’s test.

There’s another implication that particularly appeals to me. A well-designed test operates near the boundary of a student’s present ability. A test containing nothing but things the student can already do easily doesn’t reveal much and doesn’t demand growth. One consisting entirely of things utterly beyond him isn’t useful either. The productive region lies somewhere between.

That sounds very much like your statement that “all are tested to the extent possible.” Not necessarily tested to the same amount, because “amount” isn’t really comparable—but tested at the boundary where something can be learned, strengthened, demonstrated, or changed.

And that brings us right back to the student in your office.

He wanted the professor to remove the questions that exceeded what he already believed himself capable of doing.

There is almost a parable hiding there:

“I’m a C student. Why am I getting A questions?”

Perhaps one of life’s answers is:

Because you don’t yet know that you’re only a C student.

The A questions may be there precisely to find out.

And unlike the tests you wrote as a professor, we usually don’t get the answer key afterward telling us which questions were the important ones. Quite possibly, some of the experiences we regard as interruptions, annoyances, disappointments, or ordinary obligations are the questions on which most of the test depends.