Missing trials in data file

JasperdeWaard

Hello smart people,

I already posted about this in another thread, but I think it deserves it's own discussion so here goes. I'm not sure whether it's an issue in the experiment or in JATOS, but I noticed that for some of my participants (16 out of 50), not all trials are stored in the data file. In most cases, there are just a few trials missing. In the worst case 58 out of 500 trials are missing. The missing trials seem to occur at random points throughout the experiment, and like I said, only for some participants, which makes me think the issue is not in the experiment but in the data transfer. In the attached image (taken from the 'worst case' I mentioned), trial_nr is the second column from the right, and as you can see there's a bunch of trials missing.

Any idea what is going on here?

All the best and stay healthy,

Jasper

kri

Hi Jasper,

I wrote the same answer in a different thread. I just copy it here:

It's seems as if I don't get notifications for answers in threads at the moment. I stumbled over your answered coincidentally ;). I'll try to monitor this thread from now on.

You suggested waiting after each trial for a few seconds. Did you mean adding a longer delay in the OSWeb program or by changing the jatos.js file generated by OpenSesame?

Both is possible. Changing OSWeb's jatos.js file is the safer alternative. If you do not care for the (small chance of) lost data in case a participant doesn't finish the whole study, then here is a version you could use: https://github.com/kristian-lange/opensesame-extension-osweb/blob/345b528b6cad40a6bade0b48bb9ffc6f36702add/osweb/src/js/jatos.js

I wrote this jatos.js version for Sebastiaan and Daniel but in the end we agreed not use it in OSWeb due to the issues already stated: potential data loss. The current OSWeb version (the one you are using) sends result data after each trail to the JATOS server. The one here in the link sends the data only in the end of the study. This has the advantage that data doesn't get lost between trails (like with the current OSWeb version) - but does have the disadvantage that all data might be lost in case the participant does't finish the study.

There is a second option: I'm working on a new version of JATOS were the whole result data sending will be handled by a backgroud worker in the browser and therefore data loss shouldn't happen anymore. But this will take a couple of days (not sure how many - everything is chaotic right now).

Best,

Kristian

JasperdeWaard

Thanks for your answer! That second option sounds like a great and much needed functionality :)

mariosav

Hi Jasper (and Kristian),

Were you able to solve this issue? I experience the same issue with JATOS 3.5.2. Do I understand correctly that it can be solved with a JATOS update? I can see that there is version 3.5.4 but I refrained from updating as we are currently running experiments and didn't want to mess up something.

Best,

marios

kri

Hi mariosav,

JATOS 3.5.3 should fix this. But If I remember correctly the new OpenSesame/OSWeb should fix this independently too (@sebastiaan correct?). But I agree updating JATOS while experiments are running isn't recommended.

Best,

Kristian

mariosav

Thanks Kristian,

I'm using OS 3.3.1 with OSweb 1.3.5. I didn't want to update after I finished the experiment and all seemed to work all just in case something else would come up. But I guess I should have :)

marios