Drawing duplicate questions from banks
Hi there, I have read multiple posts and cannot find a clear answer. If I ask canvas to pull say, 5 questions from a bank of 10 questions for a new question group in a new quiz, how to I make sure that Canvas does NOT pull duplicate questions from the bank? This has happened, and not just to me. It seems like it should be a straightforward fix to program Canvas to pick 5 different Qs from a bank that has more than 5 Qs in it. Thanks.
You need to divide your question banks into several smaller ones that none of the smaller question banks contains duplicates. Then pick the questions from these different questions banks.
If I am not mistaken, you are trying to draw questions from a major question bank that contains assessments from several chapters.If it is the case, divide your main question bank into chapters and create a group on each chapter, draw questions from there.
You can go even deeper to divide the questions into different types and make these different question banks.
All in all, if you want draw different questions without duplicates, your question bank cannot contain duplicates.
Thank you, BUT, my question banks do NOT contain duplicates. I have 100+ different question banks each with several unique questions in them. Overall, yes, 1000s of unique questions. When I draw, say, 5 questions from a specific bank that contains several (more than 5) unique (ie., NOT duplicated) questions, canvas will pull randomly from them when I make a new quiz question BUT SOMETIMES CANVAS PULLS THE SAME QUESTION FROM THE SAME BANK TWICE. So the student gets two identical questions. Am I more clear now?
Technically, it is impossible.
I would double check if the question bank contains duplicates, do a filter search if you spot any duplicates in the quiz linked to this question bank, I can be 99% sure it is due to the duplicates.
If not, check your question numbers and ID, make sure they are not identical.
Assuming this cannot be answered (I am not the only one who has asked this) the only work around I can find is to make MORE NUMEROUS banks and only draw one question (NEVER MORE THAN ONE) from each bank when making a group question in a new quiz. I just don't trust it. Tedious.
Each question has a unique identification number, it is called by The XML when the QTI is deployed. that is the ID.
They can be 001, 002, 003 or more complicated strings. I don't know how you created the question banks, if imported from a publisher's cartridges, the possibilities of containing duplicates are very low.
If you generated these questions by compiling them from different sources or imported/reimported from old courses, esp, export and imported multiple times from many previous versions of a same courses, it may contain duplicated content.
Also check if you have duplicated groups in the quiz. and check if the quiz is a combination of individual questions and question groups.
The new quiz will typically be a combo of individual Q and question groups. For the groups I draw questions from my question banks. I have made all these questions myself; they are not imported from anywhere. Seems to me that the unique ID number could be coincidentally duplicated, when the question is entirely different....I don't know. I think I just have to play safe, and just make millions of banks and only draw one question from each bank. What a pain. Thanks anyway.
Yes the individual questions are in their own bank. There is no MAIN question bank. Just lots of banks covering twenty to thirty different aspects of the course I teach. Doesn't seem to be any way to make SURE that when canvas is asked to draw two questions at random from a bank containing 5 unique questions, canvas DOESN'T decide to choose say, question 1 twice. Maybe I am still not being clear.
The only time I've had duplicate questions in a quiz was when I accidently pulled random questions from the same test bank twice for the same quiz. To explain, I had a quiz covering three chapters. Instead of pulling random questions from three test banks: Chapter 1, Chapter 2, Chapter 3. Instead, I pulled from Chapter 1, Chapter 1 again, and chapter 3. This resulted in duplicate questions from Chapter 1.
So, perhaps check that when you created the question groups for the quiz you pulled from the correct test bank. That's the only way I can think that it would pull duplicates.
Thanks, lamplighter II. I don't pull from existing test banks; I write my own Qs. I must confess this is still a bit of a mystery, and I'm just not taking chances. I just won't ever pull more than one question from a given bank. Tedious, but there you go.
Hi @Heterochromatic ,
I'm a little confused as you stated in a previous post that you have more than 100 test banks. By "existing test banks' I meant test banks in your course.
I really do wonder if you are pulling from the same test bank in your course multiple times for the same quiz. If so, that is the issue. Every time you create a question grouping to pull random questions from a test bank the "randomness" is reset. The overall quiz itself doesn't verify that the questions aren't repeated, only the grouping does that.
For example, this could result in duplicate questions.
- Pull 5 questions from Test Bank A
- Manually added question
- Pull 5 questions from Test Bank A.
Instead, you should do this,
- Pull 10 questions from Test Bank A
- Manually added question.
I believe what you want to do is to generate an exam that is truly unique for each student, where no question is the same. At least that is what I want to do, and I think there is a very easy solution, which does not require the generation of a huge question bank.
I my case I have have 42 students taking an exam, and I have written a program that produces 42 unique variations of a particular problem. How could I then distribute these unique and individualized problems randomly between the students taking the exam? All I have to do is to create random perturbation of the numbers 1, 2, 3,... up to 42, perhaps resulting in a sequence like:
35, 4, 7, 41, ..... 3, ....
etc., with all numbers in the range 1 to 42 occurring just once. The first number in the sequence then says that student 1 shall be assigned problem number 35, student 2 problem number 4 etc.
What I would like to do then in Canvas is to upload these 42 questions to a question bank, and that the LMS by itself should produce the random perturbation described above for assigning and distributing the problems so that each exam becomes unique. This should not be very difficult to implement for the Canvas producers. For each student taking the exam the only thing the LMS needs to generate unique lookup tables for which question the student is to be assigned.
Note that this is a different approach than selecting the questions randomly from a question bank. In such an approach there is always a chance that two different students will receive the same question. With the approach above you only need to produce the same number of unique variations as there are students taking the exam, and you will be guaranteed that each exam will be unique.
You can do the unique random perturbation approach manually, but you would then be required to produce a set of unique quizzes yourself in Canvas (in my case 42), which of course one would like to avoid.
Thanks. I do have a number of basic quantitative questions that I formulated in Excel and added to Canvas. But harder to do that with the more qualitative Qs. But thanks for the advice.
I also have this issue. I made a quiz that pulls 10 questions from 3 different question banks. The banks have between 12-20 questions in each with no duplicates.
When the quiz is generated, some students have a random questions pulled twice. It is strange and inconsistent as to what question is pulled twice.
I have imported the files from ExamView, but again have searched to ensure no duplicate questions were generated in the banks during this process. Feedback resolving this would be greatly appreciated.