Question types

Three types, with very different authoring costs and very different things to get right. This page is the chooser; each type has its own guide.

All three share the same basics: a rich-text question body, a points value, an optional per-question time limit, a difficulty level and tags.

Choosing

Multiple choice is for knowledge you can enumerate — which of these is a valid index, what does this snippet print. It scales to any funnel size at no review cost, and it is the easiest type to guess your way through, which is what the scoring settings on the assessment exist to counter.

Essay is for reasoning you cannot enumerate — why would you choose one approach over another, what would you do about this incident. It is the only type that cannot be answered by a lookup, and the only one that costs you review time per candidate. On a large funnel, budget that time before you add one.

Code is the type most people come here for, and the one where a mistake is most expensive: a wrong expected value in a test case scores a competent candidate zero and looks to them like their own code is broken. Everything about getting that right is in Test case format.

What is set where

Per-question settings — points, time limit, difficulty, tags — live on the question.

How answers are scored is set on the assessment, not the question. Partial credit versus all-or-nothing, negative marking and the penalty per wrong answer apply to every multiple-choice question in the assessment at once. See Creating assessments.

Preview before you invite

Every assessment has a Preview action that runs the real candidate experience, including code execution against your real test cases, without creating an attempt or consuming your plan's allowance.

For code questions this is not optional advice. Preview is the only thing that catches a broken test case before a candidate does.