URL of experiment:
Description of the problem:
I wanted to download the audio files generated by my participants (length of recording per participant more than 30 min). When I click on my experiment in Pavlovia, I see that 21 recordings should have been submitted, but when I click on “download results”, I see far too few audio files (7 webm files). Does this mean that the other 14 files have not been saved? Thank you very much.
Are some of the data files from incomplete responses?
No, all 7 staved webm files are complete.
I mean the ones that are missing. Are they from complete participants?
Unfortunately I do not know how to check that they have completed the whole task. For those participants who do not have a webm file, I only have a .csv and .log.gz file.
At the moment I have:
submitted: 21 (37.5%)
aborted: 35 (62.5%)
Thank you very much for your responses! Is there anything else I can check?
At the moment my best guess is that the CSV data file gets saved first at the end of the experiment so your participants were closing their browser tab before the sound file had saved.
Please could you test how long it takes for the “Thank you for your patience” message to come up after the end of your experiment with a 30 minute recording and whether you get a recording saved if you:
a) close the browser as soon as the message comes up
b) leave the browser tab open
Might your participants have been using mobile devices and mobile data? If so, then it’s possible that the large upload got rejected or interrupted?
Is there any difference in the data between the participants with a recording and those without?
I’ve asked our team for help with this but haven’t heard anything from anyone yet.
Yes, we had the problem that the audio files were not saved if the time between the recording routine and the end message was too short. Therefore, to save them, we added a 20-second routine in which we asked participants to wait before the experiment automatically ended. So it seems that the other files are lost, right? Just to understand it better, 20 seconds could have been too less time for audio files of such size to be uploaded to Pavlovia (depending on the participants’ internet speed?) or the other reason is that participants aborted the experiment (closed the browser) before the 20 seconds?
Adding 20 seconds within the experiment might be what is causing the issue.
My understanding is that the audio files don’t START saving until the end of the experiment (after the final routine). If you delay that end by 20 seconds you are encouraging participants to close the browser as soon as the experiment ends rather than waiting for the data and audio files to be uploaded.
Have you investigated how long it takes for the “Thank you for your patience” message to appear and whether audio files are saved if you close the browser tab as soon as it appears.
That message can be customised in Experiment Settings / Online / End message so if you need participants to wait longer then you could tell them there.
In our case, it was important to include the final 20-second routine, as the audio file was never saved with a shorter duration. At the end of the experiment, participants were immediately redirected to another page. We did this in the experiment settings > online > complete URL/incomplete URL. Could this have caused the problems?
If (and I still haven’t tested or receive any information about this) the audio files don’t finish saving until AFTER the end of survey message says that the data has been saved then the redirect could be why you are getting a loss of data. Are you redirecting to Qualtrics (in which case you could embed a Pavlovia Survey instead) or back to something like Prolific?
Thank you for this information. Participants in our study were not directed to a survey but to another web page where the study and various other tasks were described.
Hello @M11 ,
Could you share the full path of your experiment so I can have a look at the server logs and at the code? It is difficult to understand what might have happened without it.
Cheers,
Alain
Hi Alain, thank you for your response. Hope this is the information you need:
GitLab Id: #399677
Hello Alain, just wanted to enquire whether you could find something out. I suppose I should assume that the rest of the audio files are lost, right? Thank you very much!
