Katherine Hamilton teaches our SRA 365 statistics course. She came to me the other day with a request to build out assessments that randomly generate data sets for use in her quizzes.
As I understand it, there are three use cases that she described to me:
- A randomly generated data set used within a problem. The question is currently manually graded as an essay question.
- A randomly generated data set in the format of a table that is used in multiple questions.
- A randomly generated data set that includes many different formats of data that are used in multiple questions.
We spoke with Chris Gamrat here in my office because he is currently collaborating with Nick Giacobe. They are working on developing a tool that uses PII as random seeds to build questions. That tool is currently being developed outside of Canvas and does not have a clear completion date.
Katherine referred to the formula-format questions in Canvas, but I’m finding little documentation and what appears to be some significant shortcomings for our particular needs, especially for use cases #2 and #3.
My short-term solution was to create assessments using question banks and alternate versions. The question banks could either be configured in a traditional way to provide students with similar but different questions at random, or provide students with related sets of questions at random. Using alternate versions is also possible if we can figure out a scalable way to deliver the assessments each semester. Right now, it looks like we have to pick student individually as opposed to students in particular groups. I have to look into that more.