Repository navigation
[QTI] - Inconsistent validation for the number of answer options #6167
Description
❌ This issue is not open for contribution. Visit Contributing guidelines to learn about the contributing process and how to find suitable issues.
Observed behavior
If I attempt to create a question of type 'Ordering' with only one answer I'm correctly seeing the following message: "At least 2 items are required for an ordering question." however this is not the case for questions of type 'Single choice' and 'Multiple select' where it's possible to create a question with a single answer:
answers.validation.mp4
Expected behavior
To be further discussed.
Steps to reproduce the issue
- Sign in to Studio at
unstable - Go to Channel > Channel editor > New exercise
- Create an exercise and add several question of different types
- Check the validation for number of answers
Usage Details
- Studio Unstable
- OS: Windows 11
- Browser: Chrome, Firefox
Activity
AlexVelezLl commentedon Sep 24, 2026
Also reflected on whether choice interactions should always have at least 2 options. But there are some small use cases, like "check the option below to confirm...", and having something like "- [ ] I confirm I have ...". Also, the current implementation on master allows questions with a single option, so the new editor would automatically mark these as incomplete.
pcenov commentedon Sep 24, 2026
Yeah I thought that this is the case but at the same time there is inconsistency in the approach so it's better to be further discussed and specified. :)
AlexVelezLl commentedon Sep 29, 2026
Is it inconsistent? 😅 I think these are two different types of questions with different needs -> a choice interaction still needs the learner to interact with the question to get the response right; an ordering interaction would not need any interaction at all, as a single option is always correct. If we had a polygon builder type of question, we would ask the editor to add at least 3 options because we cannot build a polygon with only two.
If an ordering interaction with only one option makes sense from the learner's POV, we can relax the ordering and allow only one option. I'd say we should not ask the editor to have at least two options for choice interaction because that'd invalidate old correct exercises. What do you think?
pcenov commentedon Sep 30, 2026
Hi @AlexVelezLl, by inconsistent I mean that we are applying strict validation in one case and not strict at all in the other - so yes, that's inconsistent. :)
A 'Choice' question implies selecting among alternatives. A question with only 1 choice isn't testing choice—it's essentially a confirmation button. While technically functional, allowing 1-choice questions in the authoring tool usually points to author error (e.g., forgotten distractors) rather than an intentional design strategy.
The same is also valid for a question of type multiple choice.
At the same time I fully appreciate your concern that old correct answers can be invalidated - but if that's the case then we still have multiple possible ways to proceed which brings us back to my initial suggestion: that this needs to be further discussed and specified so that I can test the implementation and confirm that it's implemented as specified. I don't have personal preferences on the matter.
AlexVelezLl commentedon Sep 30, 2026
Good point, @pcenov! I just talked to @rtibbles, and we agreed not to block or mark choice questions as incomplete when they have only one option; instead, we will show a warning message to flag that they might be making a mistake. For ordering questions, we will still require at least two options, since ordering questions with one option have no use case.