...
One large configuration issue with this feature is that it cannot easily be used for gap-fill problems without using custom CSS to adjust how the problem input displays. For example (without CSS hacks):
| This removes the selection from the context of the problem, and essentially functions as a multiple choice, resulting in the feature having minimal value pedagogically. | |
| This breaks the sentence in half, affecting readability. With no chevrons (which is necessary for the problem to actually function) the problem’s data becomes harder to analyse, and the problem markup is technically incorrect. |
Problems as a whole have some configuration options that exist in strange places, chiefly:
Advanced settings are a strange place for problem defaults to live, resulting in most authors not knowing they exist
Matlab API keys (Location: Problem Settings) should not be inserted at the problem level - this is poor configuration UX, and also possibly exposes the API key in course exports (verification needed)
Randomization (Location: Problem Settings) is a property specific to Python scripts, despite being something that is possible to configure for non-python problems in the problem syntax. This can lead to author configuration issues
...
While the feature is regularly used in its current form, there's a significant argument to be made that it shouldn’t actually be used the way it is. Dropdown problems do not serve a clear purpose outside of Cloze-type questions, and the platform may be better off if they were unusable outside of this context due to usability and accessibility issues inherent to dropdown menus. This is worthy of further investigation.
Feature Improvement Recommendations
...