In my most recent experiment I found two entries with identical Prolific IDs take part two hours apart, but only one was recorded in prolific.
How is this possible?
I have a screenshot for proof.
I’m pretty curious about the topic too. One explanation I gave myself for this is that participants are shown the link of your survey when reserving a place and can store the link in another window. So in the case they fail any attention check or similar, they can re-try the study again. See the thread I started some time ago:
Prolific prevents having multiple records in their study summary but not having multiple records in your survey tool. What I would do in your case is consider only the first observation in time and discard the other one. Of course, also the payment should go accordingly, so check that the observation accounted for in Prolific is the one you are keeping for your data analysis.
What do you think about it?
Good points. This is a weird one.
As a case study, say you’re using Qualtrics and you have it set to “one entry per person”. So a participant signs up on Prolific, starts the Qualtrics survey, and their laptop dies halfway through. As long as they have that initial link (Prolific would not have logged their participation because they never returned to Prolific), they could log on to a different computer (or use other methods to bypass the Qualtrics setting) and restart. So you’d see two entries in the survey with the same Prolific PID, different Response IDs, and in this case one would be less complete than the other.
Sometimes, I have had success asking the participant what happened so I would recommend that for maximal information! The case study above is a real example. Hope that helps.
PS. If someone completed the entire survey and you automatically pass their participation code back to Prolific, this could occur if a participant closed the last browser window before redirect
That’s a good piece of advice. So yes, definitely go ask for clarifications from the participants him/herself. Also, check if one of the two submissions is incomplete and the other one is complete. This is the same pattern that both Paul and I seem to have encountered - for different motives from the participant side possibly, but with the same outcome -.