What the research says

Why Explaining the Answer Beats Just Showing It

Is it enough to show the right answer after a quiz question? A Duke study found the bare answer did as well on repeated questions, but far worse on new ones.

By the Giga Quizzes team4 min read
A pink neon speech bubble reading “hello” on a black wall
Photo: Drew Beamer, Unsplash, Unsplash License, cropped.
In this post
  1. 1The puzzle
  2. 2The experiment
  3. 3What they found
  4. 4The limits
  5. 5Writing the explanation

When a quiz question is over, the teacher has a few seconds to decide what to show. The right answer, highlighted? Or the right answer and two sentences on why? It seems obvious that more explanation should help, yet when researchers compared the two, they often found no difference. Andrew Butler, Namrata Godbole and Elizabeth Marsh suspected the studies were measuring the wrong thing, and designed an experiment to find out.

The puzzle

Most earlier studies of feedback ended with a final test that repeated the questions from the first one. If the final test asks the same thing, then all a student needs is to remember the right answer, and a plain correct answer does that job perfectly well. The extra explanation could only show its value on a question that needs understanding: a new question about the same idea.

The experiment

Sixty students at Duke University read ten passages of about 500 words each, on topics such as the respiratory system and vaccines. Each passage contained two key concepts. After a five-minute break, they took a short-answer test of 20 definition questions, one per concept, and after each answer received one of three kinds of feedback, depending on their group.

The same question, three kinds of feedback

  • Question: What is the process by which gas exchange occurs in the part of the human respiratory system called the alveoli?
  • No feedback: the next question appears.
  • Correct answer: Gas exchange occurs within the alveoli through diffusion.
  • Explanation: the same sentence, plus two more from the passage on what diffusion is and why oxygen and carbon dioxide cross the alveolar membrane in opposite directions.

Two days later the students came back for a final test. For half the concepts, it repeated the definition question. For the other half, it asked a new inference question, such as: if people are having trouble breathing they are often given pure oxygen; how does that help gas exchange compared with regular air? The explanation never contained that answer. Students had to work it out.

What they found

Two days laterNo feedbackCorrect answerExplanation
Repeated questions43%62%66%
New inference questions28%30%45%
Butler, Godbole & Marsh (2013), Experiment 1

On the repeated questions, both kinds of feedback helped about equally. That matches the earlier studies, and explains why they found no difference. On the new questions, the plain correct answer did almost nothing: 30%, barely above no feedback at all. The explanation lifted students to 45%.

Why would a correct answer be enough for one kind of question and useless for the other? Answering a new inference question takes two steps. First you must remember the relevant idea, and then you must apply it to a situation you have not seen. A bare correct answer gives you something to remember, but not the idea underneath it, so there is nothing to apply. An explanation gives you both.

A second experiment with 24 students gave each person both kinds of feedback on different concepts, and asked only new inference questions. Again explanation feedback won, 37% against 27%. Then the students were allowed to answer each inference question a second time, now with the explanation on screen in front of them. Concepts that had first been explained still came out ahead, 46% against 37%. Having the explanation to hand at the end was not the same as having learned from it earlier: the earlier explanation seems to have built understanding, not just memory.

The limits

Both experiments used university students reading prose passages in a laboratory, and the final test came after two days, not weeks. The explanations were short and taken from the passages, not new information. The quality of the explanations was not varied either, and two well-chosen sentences are not the same as a paragraph of loosely related detail. Still, the pattern explains a long-standing puzzle neatly, and fits the review on transfer, which also found that explanations help learning reach new questions.

Writing the explanation

The study suggests a small habit with a large return: after every quiz question, give the reason, not just the answer. Two sentences were enough in the experiment.

An explanation that does its job

  • Start by restating the right answer in a full sentence.
  • Add one sentence on the underlying idea, in this case what diffusion is.
  • Add one sentence linking the idea to the question, here why the gases move as they do.
  • Leave out side details. The explanation is about the concept, not the topic.
  • If a wrong option is tempting, a few words on why it is wrong can help too.

Writing them takes a little extra time when you build the quiz, but you write each one once and every class that plays the quiz gets it. Explanations also matter when the answer is right. A student who guessed correctly gets no benefit from seeing a green tick, but can still learn from reading why. This is also where teaching in the reveal comes in: the moment after a question, when the whole class is looking at the answer, is the best moment to spend those two sentences.

Sources

  1. Butler, A. C., Godbole, N. & Marsh, E. J. (2013). Explanation feedback is better than correct answer feedback for promoting transfer of learning. Journal of Educational Psychology, 105(2), 290–298.

More from the blog

All posts →