Multiple choice
One prompt, several options, exactly one correct answer. The auto-graded workhorse of professional testing.
Multiple choice →A test is only as good as its questions. This guide covers each format, what it measures, and how to choose.
Two decisions do most of the work: what you want to measure, and how you want it graded.
The first decision is between objective and constructed-response questions. Objective questions (multiple choice, multiple response, true or false) give the participant options to choose from, so there is one predetermined correct answer. Constructed-response questions ask them to produce the answer, from a typed word to a full paragraph. The more they construct, the richer the reasoning it reveals, and the harder it is to grade.
The second decision follows from the first: auto-graded or judged. Choice types, true or false, short answer and fill in the blank all grade automatically against a correct option or an accepted-answer list. That gives instant, consistent results at any scale, which is why objective formats dominate professional testing. Long answers have no single correct answer to match, so a person marks them against a guide.
Most good tests mix formats rather than lean on one. Objective questions cover facts, rules and thresholds quickly and cheaply. An open-ended question or two captures the reasoning no tick box can. The nine formats below are the building blocks, each with a page of examples and guidance.
TestSetGo supports nine question types. Every one auto-grades on submission except long answer, which a person marks against a guide.
One prompt, several options, exactly one correct answer. The auto-graded workhorse of professional testing.
Multiple choice →Several options with more than one correct answer. Auto-graded, all-or-nothing or per-item partial credit.
Multiple response →A single statement judged true or false. Quickest to grade, and best used in numbers.
True or false →A typed word, number or phrase. Tests recall, auto-graded against a list of accepted answers.
Short answer →A sentence with a word removed. Cloze-style recall in context, graded against accepted alternatives.
Fill in the blank →An essay-style response in the participant's own words. Measures reasoning; marked by a person.
Long answer →A typed number graded on its value. Accepts an exact answer, a range or a tolerance.
Numeric →Items arranged into the correct sequence. Also called sequencing; auto-graded per position or all-or-nothing.
Ordering →Two columns to pair up, each prompt with its answer. Auto-graded per pair or all-or-nothing.
Matching →Build a test from any mix of the nine types, each with its own points — from true/false myth-busting to a reasoned long-answer scenario.
As many as the material calls for, and no more. Most good tests mix two or three formats: objective questions for facts, a long answer or two for reasoning.
Multiple choice, multiple response, true or false, short answer and fill in the blank all auto-grade on submission. Long answer is the exception: a person marks it against a guide.
Objective questions give the participant options to choose from, so there's one predetermined correct answer that grades automatically. Open-ended questions ask them to construct the answer, which reveals reasoning but needs a person to grade.
There's no single best type. It depends on what you're measuring: multiple choice is the reliable default for facts at scale, short answer tests recall, and long answer shows reasoning.
Mix any of the nine question types — including numeric, ordering and matching — share by link, and let the objective ones grade themselves. Free, no card required.