As a part of TA work, I need to give questions for major exams. I need to evaluate the answer sheets of students later.

Let us consider the following simple examples of questions:

1) Briefly describe natural numbers.

The answer to this question is not unique. It depends totally on the student's perspective on natural numbers. And the answer is generally long.

2) How many natural numbers are there in total? Is 6.6 a natural number?

The answer to this question is too short (Infinite, No). The answer does not depend on the student's perspective on natural numbers. It tests the knowledge of students regarding natural numbers.

I want to give questions of second type, which can test the student's knowledge and intelligence in detail and avoids long answers, ambiguous answers, long evaluation time, partial marking, unnecessary bargaining for marks by students, etc.

But some of my colleagues and faculty are saying that it is similar to multiple choice questions and fill in the blank questions, which are not recommended for major exams, which need long answers. I am confused, but strongly want to give answers of type 2 (single line answers).

Assume that there is no copying during exams.

  • 4
    I think you are mixing up several different ideas and suggesting a false dichotomy. There are at least two orthogonal axes here: short vs long question, and descriptive vs practical/computational question. Oct 16, 2018 at 19:50

2 Answers 2


There's a bit of a false dichotomy in the way you present things. Exam questions can have a long answer without being hopelessly vague and subjective. In college-level math, one usually can't avoid asking some questions with long answers, whether they're problem-solving, proof-based, or conceptual. Partial credit is just part of the deal, as is handling the occasional student seeking more partial credit. (You have to at least hear them out, because it's always possible you've made a mistake.)

However, there are different views on this; I once had a professor who gave an entirely true/false final exam, with no explanations or partial credit, in an abstract algebra class. His reasoning was that over an exam with say 40 questions, the effects of random guessing would average out to a roughly equal small bump for everyone. I was skeptical, but after taking the test I was confident that the same people did well on it, as would have done well on a traditional "solve this problem, prove this small fact, etc." exam. But this may have worked because abstract algebra has a lot of questions with a concise statement and answer, but that require real understanding. (One T/F question I remember: The Galois group of the fifth cyclotomic field is cyclic.) I don't think it would work for every course; at some point in your math education, someone needs to read the gory details of your solutions and give you feedback.

By all means, look for opportunities to make good questions that can be graded quickly and objectively, but a lot of the time, there just aren't many questions like that.


You're starting from the wrong point if you ask whether short questions/answers are good or bad. Instead, start by asking yourself (i) what you want to students to demonstrate in the exam, (ii) what you want students to learn during an exam. (I personally like to ask questions where I hope that students come out of the exam and say "Oh, I don't think I really thought about it this way. I learned something new today.")

When you have answers to these two questions, you can set about writing your questions. In some cases, you will want to test factual knowledge, and for those there are often short questions with short answers like the ones you show in your post. But oftentimes, you will also want to test understanding, for example by asking for counter-examples to a statement that at face value sounds reasonable, along with an explanation of why the statement was wrong, or by posing an "Explain to me why..." question. And then you may also want to test process abilities, like "Show that solutions of the Laplace equation are unique". In both of these latter two categories, many questions will have long answers, and so they will be more work to grade, but they have their value.

So what I would suggest is to not start with "Should I ask questions with short answers?" but with "What is it I want to achieve with this exam?".

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .