Millisecond Forums

Are you using Inquisit Lab or Inquisit Web? It's not clear from "I'm testing hundreds of subjects with the same scripts on the same pcs." If you are using Inquisit Lab, <survey> will not create separate data file per subject. I.e., if a data file already exists on the machine from a previous participant, any data from subsequent participants will be appended to that file.
If you are using Inquisit Web, connection problems would be the most likely cause of data file's not showing up, but you should see error messages to that effect (as well as a prompt to save data locally).
Beyond that, it's always possible there's some glitch or mistake in the script, but that I can't tell based on the info you shared so far.

Thanks for your reply.

Oh, right, I'm using Inquisit Lab. I know that the data should be included in the respective data file but sometimes it just doesn't show up.
I'll attach one of the scripts. Sorry, since I don't know where this problem could stem from, I don't know what information could be relevant...

Thanks for the file. There is nothing obviously wrong with it and it runs / saves data just fine for me. Hence I can only make vague guesses about what may be happening:
(1) It's possible that some participants simply skip the entire script by pressing CTRL+B or CTRL+Q at some point. No data would be written in this case.
(2) It's also possible that there's something about your particular local setup that may lead to occasional data loss. E.g., I've heard of a (very) few instances where people ran scripts and saved data to a network share (instead of running / saving locally), which can lead to (a) write conflicts (multiple machines trying to write to a shared data file at the same time) or (b) data loss if the network share becomes suddenly temporarily unavailable.

Note that you can force <survey> elements to create separate files per participant via the /separatefiles attribute:

<survey BWL_ZT>
...
/ separatefiles = true
</survey>

This would prevent the write conflict scenario under (2) and also enable you to see if data was or was not written for a given participant at a single glance. The latter might help to zero in on the specific conditions under which the presumed data loss occurs.

Okay, at least now I know that my script seems to be fine.
We save the data locally so the conflicts regarding the network shouldn't apply. I'd also think that we'd have noticed it if someone skipped the tests but thinking again, it might be possible to do that without getting noticed... If they start the survey and cancel it after a while...

Thanks for the suggestion with the seperate files. I'll try this; maybe it'll give me some inside in the conditions.

And again, many thanks for your help.

If you do come across any additional information / make any new observations, please don't hesitate to add them to this thread!