Written answers vs multiple choice: what each one actually tells you about a learner

The argument in every staffroom

One side: multiple choice is lazy. It rewards guessing, it tests recognition, and a monkey gets a quarter of them right.

The other side: written answers are unmarkable. They take forever, two markers never agree, and the student who writes well scores better than the student who knows more.

Both sides are half right, and both are arguing about the wrong thing. The question isn't which one is better. It's what each one measures, because they measure different things, and a check that knows the difference uses each for what it's good at.

Check: What does the post say the real question is, instead of "which type is better"?

What a choice question measures

A multiple-choice question asks a learner to recognise the right answer among wrong ones. That's the whole of it, and it's more useful than it sounds, because of the word "among".

The wrong options, the distractors, are where the information is. If the distractors are the mistakes learners actually make, then the option a learner picks tells you which mistake they made. "I have seen him yesterday" as a wrong option isn't a trap; it's a diagnosis. A learner who picks it has a specific, nameable gap, and you can see it in one tap.

Choice questions are also fast, and marked the same way for everyone, instantly, with no judgement call. Six of them take two minutes and give you a reading on six distinct misconceptions. No other format does that.

So a choice question measures: can they tell the right thing from the specific wrong things I planted. When the planting is good, that's a lot.

Check: Where does the information in a choice question live, and what does a well-chosen wrong option tell you?

What a choice question can't tell you

Four things, and they're the four that matter most to a tutor.

Whether they could produce it. Recognising the past simple among four options is not the same as producing it in a sentence of their own. Every language teacher has watched a student ace the grammar quiz and then say "I have seen him yesterday" in the corridor.

Why they chose. A right answer might be knowledge, elimination, or luck. The tap looks the same.

Partial understanding. A learner who has the idea three-quarters right and picks the wrong option looks identical to a learner who has no idea.

Anything a bad distractor hides. If the wrong options are obviously wrong, the question measures nothing. A choice question is exactly as good as its worst distractor.

Check: Name the four things a choice question can't tell you.

What a written answer measures

A written answer asks a learner to pull the idea out of their own head and put it into words, with nothing to recognise. That's retrieval and production at once, and it's the closest thing on a screen to what they'll have to do in the world.

What it shows you that a tap can't:

• The shape of their understanding. "The present perfect is for things that happened before now" is half the idea. You can see which half, and the score out of ten with its reason says so.
• Their reasoning. A learner who writes "past simple, because yesterday is finished" has shown you the rule in use. One who writes "past simple" has shown you a guess or a memory.
• The vocabulary they've actually got. Their words, not yours.
• Partial credit, honestly. Seven out of ten with a reason is a real measurement. Choice questions can't give you a seven.

The marking is against an expected answer the tutor writes, in plain words, and the same marker applies it to every learner the same way. That takes the "two markers never agree" objection off the table, and puts the responsibility where it belongs: on writing an expected answer that says what you'd accept.

Check: What does a written answer show that a choice question can't, and what is it marked against?

What a written answer can't tell you

It's not a universal instrument either.

Exact facts, efficiently. "Which chamber pumps blood to the lungs?" is a choice question. Asking it as a written answer makes a learner compose a sentence to deliver one word, and the marking has to read around the sentence to find it.

Lots of things quickly. Six written answers is a twenty-minute task. You'll get depth on six ideas and a tired learner. For breadth, choice questions win.

Anything the expected answer gets wrong. The marking follows the expected answer. If it's too long, good learners score low. If it's too narrow, a correct answer in different words scores low. The written question is only as good as the standard behind it, which is why the expected answer deserves more editing time than the question.

Confidence from a nervous writer. Some learners know the thing and freeze at a text box. A choice question lets them show what they know without composing.

Check: Name three situations where a written answer is the wrong instrument.

The guessing problem, in numbers

The objection to choice questions that gets quoted most is guessing, so here it is honestly.

With four options and no knowledge at all, a learner gets a question right one time in four. Over six choice questions, that's one and a half right by pure chance. A learner who scores two out of six on choice questions may know nothing. A learner who scores five has almost certainly got most of it.

A written answer has no floor. You can't guess a sentence. A learner who scores seven out of ten on a written answer has produced seven-tenths of the idea, full stop.

What this means in practice: don't read a low choice score as "knows a bit". Read it as "might know nothing". And don't read a high choice score as "knows it". Read it as "can recognise it", and go to the written answer to find out whether they can use it.

Check: Over six four-option choice questions, how many does a learner with no knowledge get right by chance, and why does a written answer have no floor?

How I mix them

The shape I use in almost every check, for one idea, in ten minutes.

Four choice questions. Each one targets a mistake I've actually seen learners make, as a distractor. Together they cover the four ways this idea goes wrong.

Two written answers. One asks them to explain the idea in a sentence, in their own words. One asks them to produce it about something real: their own work, their own weekend, their own site. The second is the one I read first.

The ratio isn't sacred. It's that the choice questions find the misconception and the written answers show whether the idea can be used. Four and two gets both in ten minutes.

Check: What do the four choice questions do, and what do the two written answers do, in the post's standard check?

Reading the two kinds of result differently

A choice result and a written result need different eyes.

For choice questions, I don't look at right or wrong first. I look at which wrong option was chosen. Three learners who missed the same question by picking the same distractor share one misconception, and that's tomorrow's five minutes. Three who picked three different wrong options is three conversations.

For written answers, I read the reason under the score before the score. "Describes the finished-time use, leaves out the still-true use" is a sentence I can teach from. A seven on its own isn't. And across attempts, I watch the reason change: a learner whose reason goes from "describes one use" to "describes both" across two days has learnt something, and the text says exactly what.

Check: What does the post look at first in a choice result, and what does it read first in a written result?

When one kind is enough

Sometimes you only want one, and that's fine if it's deliberate.

• A four-minute exit ticket in a class of thirty: choice questions, maybe one written. You want breadth and speed.
• A procedure check for a crew: scenario choice questions, plus one written "what do you do first". The scenarios find the wrong habit; the written one shows the right one can be produced.
• Language production: written answers carry the load. Recognising a tense isn't the goal; using it is.
• Early learners, or anyone who freezes at a text box: choice questions, and get the distractors right.
• A quick retake at home: at least one written answer, for the reason in the next section.

Check: For a procedure check with a crew, what mix does the post suggest and why?

The second attempt changes the answer

One last thing, because it decides more than people expect.

Choice questions can be memorised. Four options, the same order, an hour later: a learner can remember "the third one was B" without remembering why. A retake made only of choice questions can flatter.

A written answer can't be memorised that way. The learner has to produce the idea again, in words, and if they produce it better the second time, they've learnt something and you can see it in the text.

So if learners will retake, and they should, put at least one written answer in. It's the part of the second attempt you can trust without checking the clock.

Check: Why does the post say every check that will be retaken should contain at least one written answer?

If you're in Australia

If your learners are heading for a local exam, look at what its own papers use. Many use both, and the balance tells you what the examiners think each one measures; a check that mirrors that balance is good practice for the format as well as the content. Swap this section for your own country's exams; nothing above depends on it.

Check: Why might a tutor mirror a local exam's balance of question types in a check?

Try it with your own notes →