Checklists and criteria
Safety requirements, document sets, eligibility rules: anywhere “which of these apply” is the real question.
One prompt, several options, and more than one correct answer. Also called multiple answer or select-all-that-apply questions. Here's how to write and score them fairly.
A multiple response question presents a stem and a list of options, of which two or more are correct. The participant must select every correct option and leave the wrong ones unchecked. It is the natural format whenever the honest answer is “more than one of these”, commonly labelled “select all that apply”.
Multiple response is harder than multiple choice, both to answer and to score. A participant can be partly right, correct on three options but wrong on a fourth, so you decide in advance whether partial knowledge earns partial marks. That decision changes what the question measures, so it deserves thought rather than a default.
The format's strength is that it resists lucky guessing far better than single-answer multiple choice: there are many more wrong combinations than right ones. Its weakness is that a poorly written item becomes a set of independent true/false judgements in disguise, so keep the options related to one another and to the stem.
Which of the following are classified as personal data under UK GDPR? Select all that apply.
Reach for it when a correct answer is a set of things, not a single one.
Safety requirements, document sets, eligibility rules: anywhere “which of these apply” is the real question.
With several options in play, wrong combinations dwarf the right one, so a confident guess rarely lands.
Requiring every correct option and no wrong ones reveals whether someone knows the full picture.
If exactly one option is correct, use multiple choice: clearer for participants and simpler to score.
Multiple choice questions →Multiple response questions grade automatically on submission, using the scoring mode you choose. By default they are all-or-nothing: the participant earns the points only if their selection matches the correct set exactly, every right option ticked and every wrong one left alone. This is the strictest, most common approach, and it keeps a partly-correct answer from reading as a pass.
When you would rather reward partial knowledge, switch to per-item partial credit. Each option then contributes to the score independently, so selecting three of four correct answers earns proportional marks rather than zero. Choose deliberately: all-or-nothing measures whether someone has the complete answer, partial credit measures how much of it they have.
As with all choice types, option order can be shuffled per participant, you can attach an explanation shown after submission, and each question carries its own points. If a question proves flawed after answers arrive, apply a correction and affected attempts regrade automatically.
Multiple choice has exactly one correct option; multiple response has two or more, and participants must select all of them. Multiple response needs a scoring decision, all-or-nothing or partial credit, that single-answer questions don't.
All-or-nothing when a partly-correct answer isn't good enough, like safety checklists or compliance criteria. Partial credit when you want to measure how much of the full answer someone knows; TestSetGo supports both.
Four to six works well, with at least two correct: enough to make guessing hard without turning the item into a tiring hunt. Keep every option relevant to the stem.
Usually not. “Select all that apply” without a count is the more honest test of understanding, since stating the number turns it into a narrower puzzle and makes guessing easier.
Write multiple response questions with the scoring that fits, and share a self-grading test in minutes.