URL of experiment: https://run.pavlovia.org/ClareSutherland/socialmemoryfinal/html
(note you need to enter a ppt number from 1-24 for it to work)
Description of the problem: One of the people pilot testing the experiment hit one item where a text response option “1. Marble” (or “2. Marble” depending on the condition) is appearing as a very specific number (“36952.041”). It’s only happening on his computer (I think a PC desktop, could find out) although it is happening across multiple browsers which seems to rule out a cached issue.
When we download his data, the number appears instead of the text, so Pavlovia is running some other file, but when anyone else runs it, we see the text option in their data. Also, when I look at the code on Pavlovia, the text appears - it looks normal. It’s quite bizarre!
Any idea what could be happening? It’s obviously hard to troubleshoot because when I look at it, it’s fine.
Is this text in one of the condition files? If so, could you tell me which one?
Hi Thomas - any of the participants from 1-24, but in the file, the text looks like “1. Marble” to me … it’s under column R and S of the “Encode” spreadsheets if you can see these.
Note that the experiment is live - we’re collecting data so please don’t delete or modify any files
No worries, only looking what happens with the network requests when your experiment gets loaded. I’m not even starting it. I looked at Encode_15.csv, but no marbles there. Could you tell me which files it contains exactly?
I’m so sorry - it’s in the “retrieve.csv” file - should be in column R or S (in retrieve_15.csv, it should be in row 5, column R).
Found it, thanks! I see nothing strange with retrieve_15.csv, but if you could find out which Operating System and Browser the participant used, I can try it out in that one. Ideally, I’d like to know the version of the OS too (for instance MacOS X Catalina)
My colleague used:
Windows 10, Version 1909 (OS Build 18363.1139)
Chrome: Version 86.0.4240.111 (Official Build) (64-bit)
Edge: Version 86.0.622.58 (Official build) (64-bit)
Thanks very much!
Got one more request. Could you share the experiment repo with me? My username on gitlab.pavlovia.org is tpronk
I’m sorry, I have no idea how to do that - do you mean forking or cloning? And how do I add you?
This post explains the steps you can take for that. Unable to download resource
Can’t see your code yet. Maybe give me another role? Like maintainer?
try again now, sorry about the faff!
Thanks! I’ll need to set up a rather funky test to verify the issue; update somewhat soonish!
Very intriquing! Thanks for all your time on it
Well yes this is quite intriguing :). On the PsychoX side there is definitely something useful to be learned, but in relation to your experiment, another question: how would you feel about converting those CSV files to XLSX (Excel) files if that would quickly solve the problem?
Happy to try that - will probably finish recruitment on this group first and then we have two more groups to run, so we could try with xlsx files for the new ones. Thanks for your help!
Bueno! Keep me posted please; if that doesn’t fix it I’d happily dive into your issue again
Hi thomas. the code that produces the csv files which counterbalance the experiment is written in matlab, which at the moment is writing each generated table of data as a separate .csv file. in principle i could simply rename the file suffix to xlsx, but this would not give it any header or formatting info that psychopy may require to interpret their contents correctly? so could you let me know if the content of the csv files would literally have to be saved as an xlsx file with any associated header/formatting, or if its a matter of just renaming from .csv to .xlsx? best - kevin.
It’s about the xlsx format, so basically the format you’d get if saving the file as a spreadsheet from Excel.
For a bit of context: we suspect that the issue with CSV file is due to PsychoPy/PsychoJS being confused by the fact that it contains both commas inside of fields and as separators. We’re thinking of a nice solution for that, but it could take a while before it’s added to the system; that’s why I think xlsx could be a nice workaround.
ok. there is an xlswrite function in matlab that can produce .xls versions. i will try that - if .xls format can be read…? i dont know how similar xls is to xlsx.