Terms and definitions
Pair vocabulary, roles or concepts with their meanings in one compact question.
Two columns, one job: pair each prompt with its correct answer. A compact way to test several related facts at once. Here's how to write ones that grade themselves.
A matching question presents two columns and asks the participant to pair each item on the left with its correct answer on the right. It's recognition-based, like multiple choice, but one question can check a whole set of related facts instead of a single one.
That density is the appeal. Matching questions suit terms and definitions, roles and responsibilities, or items and the category they belong to, anywhere a batch of facts share a theme. A run of four matched pairs covers the same ground as four separate questions while taking up far less room on a test.
The craft is in the right-hand column. Keep the answers similar in kind so none can be ruled out on sight, and avoid a set where matching all but one leaves the last pair for free. Matching questions are written in the editor rather than drafted by AI, so you control every prompt and answer before publishing.
Match each data protection role to its core responsibility.
Correct pairs: data controller → decides why and how personal data is processed; data processor → acts only on the controller's instructions; data protection officer → monitors internal compliance; supervisory authority → enforces the regulation, the ICO in the UK.
Match each SI unit to the quantity it measures.
Correct pairs: newton → force; pascal → pressure; joule → energy; watt → power. The units are related closely enough that a participant has to know each one, not eliminate by process.
Reach for it to test a set of related facts efficiently, where each prompt has one clear partner.
Pair vocabulary, roles or concepts with their meanings in one compact question.
Sort items into the group, owner or category each one belongs to.
One matching question checks several related facts, saving room on a test.
If you need explanation or working, use a long answer question instead.
Long answer questions →Matching questions grade automatically the moment an attempt is submitted, against the correct pairs you set. By default each correctly matched pair earns an equal share of the marks, so a participant who gets three of four right still scores three quarters. Switch to all-or-nothing and only a fully correct answer earns any points.
A participant can't use the same answer twice, and every right-side answer has to be distinct, so there's always exactly one correct pairing. A matching question holds between two and twenty pairs, though four to six reads best and stays manageable on a phone.
You build matching questions in the editor, set the points, and add an explanation participants see after submitting. The dashboard shows per-question accuracy, and if a pair proves unfair after answers arrive, apply a correction and affected attempts regrade automatically.
A matching question presents two columns and asks the participant to pair each prompt on the left with its correct answer on the right. One question can test several related facts at once.
Automatically, on submission. By default each correctly matched pair earns an equal share of the marks; switch to all-or-nothing so only a fully correct answer scores.
Four to six works well. TestSetGo supports two to twenty pairs, but long lists get fiddly on a phone and tire the participant.
Yes. TestSetGo is a free matching quiz maker: build a matching test in the editor, mix it with other formats, and share it by link.
Write matching questions in the editor, mix them with other formats, and share a self-grading test in minutes.