[ARCHIVED] Why do Question IDs change?

[home777]

I have a problem and I'm hoping some of the experts here might have some insight that helps us unscramble things.

Background: We have a course that uses a summative final exam. This exam has a total of three versions. Some of these questions in the 2nd and 3rd exams were created from slightly modifying the stems or answer choices of the first exam. The theory is a student who needs to retake the exam after 180 days should not be able to see the same questions again. Therefore, with a 60 question exam, we have a total of 180 questions, and permit a maximum of two retakes. In theory, a student will have a 33% chance of seeing a repeated question in a future exam, but will not be presented with the exact same questions.

I am trying to come up with an elegant way to manage the questions from these three exams. Whenever a question from the first exam was used to create clones that are used as questions in versions 2 & 3, these questions need to be kept separate. If not, a student could encounter these nearly identical "sibling questions" during one exam testing session.

The problem we're having is that when a course is duplicated in Canvas (for example, creating a Summer Semester course by cloning the Spring Semester), the Question ID numbers assigned to each question change and become different. My understanding is that in Blackboard or other similar LMSes, a static identifier can be associated with these questions.

I have thought about other ways around this, like managing questions in Question Banks at the Account level, and also of having a Sandbox course that contains the master questions. Each of these have some significant downsides. What would elegantly solve the problem is an unchanging question identifier that can be used to evaluate and analyze each item within an assessment.

Does anyone have experience assigning a unique identifier to a question, or how do you work around this problem?

Stumped,

Russ

[Tasha_Biesinger]

I realize that this question is from a year ago, so I'm sure you've moved on, but I had some questions around tracking question data across terms come across my radar today and was browsing topics. So...

I don't have any formal answer for why the Question IDs change or need to change - just that I'm not surprised that the IDs change since I can change the question in this class and not have it change in that class because the questions aren't linked to each other.

I was pondering how to go about tracking the data - and in the Item Analysis CSV that I downloaded from a dummy quiz, I noticed that, in addition to the Question ID, there was a Question Title column.  Since I can assign the Question Title and I don't think it changes as I copy the question from one course to another - I figured I could use this to create a unique ID for the questions I want to track.

I haven't tried this, so definitely room for pitfalls...

[dmalik]

I noticed you asked about question IDs also. We give the same exam in two sections (it's cloned from one class to another). The IDs are different... in fact, sequence of IDs are different too. We would like to consolidate the data from the two sections and get overall statistics, but it doesn't look easy. To add to our problem, the questions have equations in the stem so it is hard to match ID to a question. You said the "question title" appears in the spreadsheet: it's actually the stem. If Canvas used the title, we'd be able to figure out the ID assignments. So... we still have problems discovering how to use the Id better. A static ID assigned at the publishing point would help (maybe different IDs, but same sequence as appearing in exam -- we use groups so it could be helpful to know the match up).

Is there a reference on the question ID and mapping algorithm available?

Addendum added 11/12/17:

How can we get the actual "title" of the question in the spreadsheet?

Is all of this moot because the quiz.next is pending?