What can I do with the missing data at the end of trial
There is a pattern that pupil data may missing at the end of some trails, which could be harmful to pupil reconstruction. Is it acceptable if I left NaN to trails that show this pattern, simply use the nearest valid data before trails ending to reconstruct pupil? I anticipate that there wouldn't be any influence to mean pupil time series cross trails under same condition because missing data won't be taken count.But there might exist some risk in statistic power I can't tell since I'm not good at statistics. Is there any experience you can share me with, please?
Thanks so much
Weizhe Li
Thanks so much
Weizhe Li
Comments
Hi Weizhe,
I would probably just represent the missing data at the end of the trial by
nanvalues. That way it will be silently ignored and doesn't do any real harm. Extending the latest valid pupil-size value to the end is of course also a possibility, but unless there's a real reason to do this (for example because you're doing some analysis that requires all data to be valid) it feels a bit awkward to me.Cheers!
Sebastiaan
Check out SigmundAI.eu for our OpenSesame AI assistant!