Major Data Loss on Pavlovia - Online Experiments

URL of the experiment: V2 Gesture&MC [PsychoPy]

Description of the problem: I have been collecting data since September 2021. 63 people participated in my experiment, but I realized that almost 50% of the data was not recorded on Pavlovia! This is a major data loss, and I am very sorry. On the other hand, there were no problems during the experiments, and none of my participants told me they encountered any experimental participation issues.

Do you think there is any way for me to find this data or stop this problem from occurring?? It is urgent and essential.

I appreciate your help.


I think that this is most likely to happen if there is a crash or network outage or the browser window is closed before the data has finished saving. What happens at the end of your experiment? Do you ask participants to wait for the “Thank you for your patience” message?

One way to monitor this would be to bookend your study with my VESPR Study Portal. After the experiment, the participant is redirected back to a debrief page on the portal and the researcher can monitor the number of participants who launch the experiment and whether they reach the debrief or lapse. Pavlovia should save a data file for every participant who reaches the debrief. If this isn’t the case, then there must be something more serious going on.

Thank you for your answer!
Yes, my participants were asked to wait a couple of minutes after the experiment to prevent possible saving issues. Unfortunately, that is what we ended up with. There seems to be no solution for data loss. What huge waste of time and energy.
I can do what you suggested for the rest of the data collection period, but I was curious about possible ways to rescue the data.

Are there log files for the missing participants? If not, then I don’t think the data can be retrieved.

I’m flagging this for @Becca since you’ve said that your participants were asked to wait after the experiment ended.

Hi There,

Sorry that you are experiencing this. Yes @wakecarter is correct - if there is a log file the csv can be “recreated” from that. Unfortunately there are lot’s of factors beyond our control for online studies (like those wakefield highlights). The best ways to avoid data loss I believe is short experiments and emphasising that participants must wait for the green "thank-you message before closing their browsers.


1 Like