I just got a multiple choice in review, where one of the answers was “the book are on the chair”. Since the Hebrew phrase I was being quizzed on actually can mean “the books are on the chair”, I selected that one just to see if it would be right, but it was marked wrong. The correct answer was also given as an option, “the books on the chair / the books are on the chair”; but I deliberately picked the partially-right answer with the typo (missing “s”) because I’m testing my own course.
Anyway, after the review I went into edit and tried to find that answer in my database, to see why I had ambiguous items and also to fix the typo (add the “s” back)… but I can’t find it. Not by doing database searches for various substrings and variations (“book”, “chair”, “book chair”, “the book are”, “are on the chair”, etc.), and not by going through the levels where it might be.
I’m pretty sure “the book are on the chair” is a temporary artifact from when I was first adding this item, which was probably last year sometime. I must’ve initially typed that, then amended it soon after to “the books on the chair / the books are on the chair”. So why is that old answer with a typo still showing up in multiple choice such a long time later? And how can I get rid of it?
I don’t know why but it has happened to me too. If I delete smth it will appear on multiple choice test. It is better to overwrite things, less chance for the wrong-right ones to pop up
These ‘ghost answers’ that can’t be found in the database are a recurring problem.
What used to work was to change the status of the course, save it, and then change the status back to what you want. This triggered an update of the course database used to select answers on the MC questions.
I’ve not had any positive feedback about this method recently but it’s worth trying. Maybe leave the status changed for a day or two?
Otherwise there’s no simple fix I’m afraid, and no comment from Memrise about the problem as far as I know. (unless it was the standard ‘‘we’re aware and working on it’’ a long time ago.)
Never seen this on the web version, but the apps definitely do this - there the only way of fixing the problem is to uninstall/reinstall (a bit of a pain); I could imagine that clearing the cache and the cookies relating to Memrisk should do the same for the web version.
it does not work for me… I’ve done that several times, phantom entries still pop up in reviews, and, unfortunately, add to the number of items to review (because, of course, in speed review you click those phantoms)
Sorry, I should’ve specified that I saw this on the web. I only check out the apps every couple of months to see how they’ve changed, I don’t actually use them much (I think they’re awful and they prevent effective learning).
Unfortunately if there’s a possible solution that we don’t know works or not, I don’t have a good way to check whether it works, because I’m unlikely to get a multiple choice review on a similar item again very soon, and even if I do, it’s just a matter of luck whether it picks that typo. I have very few phantoms in my course because I haven’t deleted or changed many items.
I’m reluctant to try that because it’s such a popular and active course, and I have no way to communicate with the users. But maybe I’ll try late at night (assuming most of my users are in the US and Europe?) … though I wouldn’t be able to tell if it actually works.
I believe that those, who are already learning the course will not be affected by your changing it to private – they can still access it from either their dashboards or the direct URL they might have bookmarked.
Just came across a more harmful incidence of this problem in my tropical fish course.
Originally I had a “Mexican hogfish” item that included photos of that species in both initial and terminal phase. After a trip on which I had the opportunity to take photos of lots of Mexican hogfish and finally had enough photos of each phase separately, I created new “Mexican hogfish, initial phase” and “Mexican hogfish, terminal phase” items, then deleted the original “Mexican hogfish” which had photos of both.
Just now, on a multiple choice test, I was shown a photo of a terminal phase Mexican hogfish, and the multiple choice answers included both “Mexican hogfish, terminal phase” (correct), and “Mexican hogfish” (technically also correct, but would be marked wrong).
I guess I’ll try your solution with this course, which is not yet that publicized, and see if that ever happens again!
Well, that was quick! Got an old typo right after I tried it, so I guess that unfortunately it doesn’t work
What I did was edit my tropical fish course to “Unlisted”, and save, then edit it back to “Public”, and save. Then I did some more review, and one of the tests had a multiple choice answer of “Bird wrasse, terminal stage”. That was a typo, it’s supposed to be “terminal phase” not “terminal stage”. I corrected it a while ago. This isn’t a deleted entry, it’s just the former value of an existing entry whose text I edited months ago.
a- either I have not made that many mistakes lately, in adding items, or b- the database is getting slowly in better shape (?) … have very seldom reencountered the pesky phantoms… on the other hand, in the last days i’ve crammed almost only mainland chinese, my own private cramming courses and 2 very accuraste courses (made by two very conscientious creators as it seems )
EDIT: big pufff aua:i dared review other course of mine, and here they are, popping up their ugly heads, those pesky old typos… one strange occurence: one faulty entry, i remember very well - because the original translation is the standard one in the HSK coures - i’ve corrected it, not deleted it. However, the variant before correction appeared in test…
I have the same problem with images. I have a course for the devanagari alphabet which is image-based and I recently replaced the old images by higher resolution ones, because the old images looked awful in the app. But they are still showing up in tests…
@neoncube I was considering trying it, but because your script was obviously intended for text entries, I decided to first try some other approaches that I hoped would update the “secret information”:
I added a dummy column to the database, and then removed it again.
I changed the “test on … prompt with …” settings for each level, and then changed them back again.
Neither of these seemed to work. So then I did something analogous to what someone in another thread suggested (for text entries): to make a new column, move the text there, delete the old column. So…
I added a new image column to the database, uploaded all the images there, and deleted the old image column.
This did work, I suppose, since the old images are definitely gone now and don’t show up anymore. But it also caused every item to be unlearned. And there are hundreds of them, so I’ve caused my course followers a lot of grief (I guess they will drop the course, actually). I think any mems that were based on them are also gone. I had not expected that. So, as a warning to others: don’t do this for a substantial amount of entries. Since I did it for all entries, the effect is the same as if I had deleted the course and made it again from scratch
I think that what probably caused everyone’s progress to be reset was changing the testing direction. There’s actually a note in the popup where one changes the testing direction that says that doing so will reset everyone’s progress.
Anyway, I hope some people stick with your course, regardless You could always make an introductory level letting people know what happened (or edit your course’s description), which might encourage some people to relearn the items in the course.
You’re right. I didn’t pay attention to that
But then at least the approach of adding a new image column and removing the old might still work…
Yes, I did do that. As well as point them to the Memrise auto learn userscript, which (fingers crossed) will hopefully also work for image based courses (after the creator of the script finishes updating it, because it’s currently broken due to some front-end changes that Memrise has applied in recent days).