Feedback for Card Sort and Matching

9 views
Skip to first unread message

Bradley Miller

unread,
May 11, 2026, 4:11:51 PMMay 11
to PreTeXt development
I’ve heard from Vilma and other Proteus researchers that some students are just randomly trying one match at a time for matching and cardsort problems.  A couple of suggestions to fix this were:

1. Limit the total number of attempts a student can make.
2. Don’t give the student any feedback until X percent of the drops/matches have been made.

I am not a fan of 1 especially for formative exercises.  We have never limited the number of attempts a student can make with any question type.  But maybe this is a feature of other homework systems that people are used to using?

I do like 2, forcing the student to have at least the majority of matches attempted feels like it goes a long way toward getting the student to think a bit harder about their selections.

Does anyone have other ideas?  Other opinions on 1 versus 2?

Brad

Brad Miller
Professor Emeritus, Luther College
Founder, Runestone Academy LTD
Blog: http://reputablejournal.com

Set up a time to meet with me.


Barbara Ericson

unread,
May 11, 2026, 4:14:54 PMMay 11
to prete...@googlegroups.com
I agree with #2.

Dr. Barbara Ericson
Associate Professor, School of Information
University of Michigan


--
You received this message because you are subscribed to the Google Groups "PreTeXt development" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pretext-dev...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/pretext-dev/2C59696C-01DB-4CC6-9EB3-57D1C31DCB73%40runestone.academy.

Mitch Keller

unread,
May 11, 2026, 4:18:41 PMMay 11
to prete...@googlegroups.com
I also like option 2.

Sean Fitzpatrick

unread,
May 11, 2026, 4:34:47 PMMay 11
to PreTeXt development
Agreed on 2. Would there be some sort of generic feedback below the minimum threshold? 

Something like "You need to find X correct matches on your own before a hint will be provided" 

To view this discussion visit https://groups.google.com/d/msgid/pretext-dev/7354419A-75D2-4C8C-824D-EDDD7A55FC2A%40rellek.net.

Bradley Miller

unread,
May 11, 2026, 5:22:34 PMMay 11
to PreTeXt development
What I was anticipating was something more like:  You need to have matched N things before we will provide feedback.  I think saying N correct things would just support the old style behavior.

I’m not entirely clear what the goal would be (not sure if anyone is).  But I think it is to prevent students from trying to match one thing at a time by randomly moving it around.

Then I’m not sure if we should show them which things are matched correctly or only if we should say that you have matched X out of Y things correctly, and you have yet to place Z things.


Brad Miller
Professor Emeritus, Luther College
Founder, Runestone Academy LTD
Blog: http://reputablejournal.com

Set up a time to meet with me.

Andrew Scholer

unread,
May 11, 2026, 5:34:21 PMMay 11
to PreTeXt development
Is the concern that students get 100% for doing it that way? Or that they can do it to complete the question?

If the concern is primarily about questions being used for summative assessments, I think that is best solved at an instructor level by specifying a different grading method (like first attempt). For that purpose, it might be nice to have a "decaying" option that gives reduced credit the more times you get the question wrong to penalize spamming solutions. (Though we might need logic in different question types to figure out how to calculate that decay.)

For formative questions, I think the highest priority should be providing useful and immediate feedback in a way that allows students who are engaging in exercises to verify and fix their understanding. If I make all the matches that I think are correct, and the system won't tell me more than "you are wrong" (and not that I have too many, or too few or what might be wrong), at some point I am going to be reduced to guessing and checking. I don't see any value in that. For formative assessments, protecting students who are intent on bypassing learning from their own worst tendencies is nice if it can be achieved without making the experience worse for those that are trying to do the work as intended. (But "Open Gemini in Chrome" in the sidebar also will probably tell them the solution...)

I could see an argument for not showing the <feedback> in a matching until it is done. Or make it more granular so you can specify feedback for individual wrong parts without giving away the overall picture. (Multiple choice questions allow item by item feedback, allowing you to correct.) But that really should be consistent across problem types and right now the behavior is generally show feedback after the first attempt.

Andrew

Bradley Miller

unread,
May 11, 2026, 6:00:41 PMMay 11
to PreTeXt development
I think what I heard from Vilma is that their analysis of how some(many???) students are approaching the card sort problems is to guess one at a time to complete the question, which indicates they are not really giving the question any thought.  

I don’t  think we can solve the problem of the student that just wants to complete the assignment and doesn’t care if they learn anything.  And I agree that we do want to give good encouraging feedback to the students who genuinely are trying to learn.  So I’m trying to discern how we balance that.  In parsons problems we simply say “Your answer is too short.  Add more blocks.” if the student doesn’t have at least N blocks added to the solution.  After they have N blocks in place then we just show them the blocks that are in the wrong order (without giving away the order). That seems like a good way to do it. 

If we make the students match N blocks and then show them which are right that takes away the systematic guessing approach of just grabbing a block and dropping it on one after the other until it is in the right place.  And maybe encourages the student to think it might be quicker to look for the answer than to just systematically guess.??

Brad Miller
Professor Emeritus, Luther College
Founder, Runestone Academy LTD
Blog: http://reputablejournal.com

Set up a time to meet with me.

Charilaos Skiadas

unread,
May 11, 2026, 8:29:32 PMMay 11
to prete...@googlegroups.com
Unless I’ve misunderstood how the cardsort problems work, I think if a student wants to solve them “one item at a time” then showing them which ones are wrong, regardless of whether we require them to put them all in place or not, isn’t going to help, the student can simply start by putting them all in one place then moving one particular item one place at a time until that is correct, then moving to the next item and so on. If they want to do a search like that then any version of showing them which items are wrong will not prevent it. Even telling them only how many they have wrong won’t affect that, and even just telling them if the whole thing is right or wrong will at best slow down the exhaustive search, from n*k to n^k or something the sort, someone with better combinatorics skills can probably get a better answer.

My thought would be (all of the following):

1. default behavior as we have it now
2. allow a number of attempts as an optional setting the author can set (though admittedly it feels more of an instructor setting so I’m a bit torn)
3. have a setting to specify between two forms of feedback: by-item vs only-holistic. Again this feels more of an instructor setting than an author setting.

I think allowing an instructor to limit the number of attempts is something that should be an option at the runestone level at least. I feel it is a pedagogical choice that I would want to leave to the instructor. Similarly for the kind of feedback provided. Not sure how easy it is to do this just at the runestone level though.

Charilaos Skiadas
Department of Mathematics
Hanover College

Reply all
Reply to author
Forward
0 new messages