Writing questions worth arguing about
A question everyone gets right taught nobody anything. Here is how to aim for the middle.
The most common mistake in a live quiz is writing questions that are too easy, usually because the writer already knows the answer and cannot unsee it. A question where ninety per cent of the room is correct produces a leaderboard and no discussion. The interesting zone is somewhere around half.
Aim for the even split
When the room splits close to evenly, three useful things happen at once. Nobody can coast, because their answer is genuinely uncertain. The reveal actually surprises people, so they remember it. And you have two live factions in the room who can be asked to defend themselves, which is the discussion you wanted.
If you can guess the distribution before you ask, the question is not doing any work.
Where the good distractors come from
Not from invention. The wrong options that actually pull people are the misconceptions you have already heard in your own room, which means the best source of distractors is the marking you did last term. A wrong answer that nobody picks is dead weight on the screen and slows everyone down reading it.
- Use a real misconception, not a nonsense option nobody would pick.
- Keep options roughly the same length. The longest option being correct is a pattern people learn to exploit within about four questions.
- Avoid "all of the above". It rewards partial knowledge and tells you nothing about what someone actually understood.
- One idea per question. Two ideas means you cannot tell which one they got wrong.
Write the explanation first
This is the single change that most improves a question set. Write the one-line explanation of why the correct answer is correct before you write the options. If that explanation is hard to write, or comes out as a restatement of the answer, the question is testing recall rather than understanding, and you will find out in the room rather than on the page.
Timing is a design choice
A short timer suppresses overthinking and rewards instinct, which is the right setting when you are teaching pattern recognition. A longer one rewards working it through, which is right when the question has steps. Setting the same timer for every question in a set is a decision you have made by default rather than on purpose, so it is worth being deliberate about it.