Entering edit mode
6.2 years ago
zercchi
•
0
Hi, in a multicentric study, we found that some subjects were recuited twice at different moments. The problem is that for some variables the value is ot the same (ex : age of onset of a specified disorders). The common and easiest way is to remove the two subjects or to keep the discordant variables as NA. Is there an other validated way to merge informations into a single subject. Refrence articles are welcome. Thank you for your help
Can you drop one of the duplicate measurements?
I don't know if dropping one of the measurements is the best way, because statisticians tell me that each evaluation must be taken as a global bloc and mixing values from different evaluations even for the subject haven't sense
Please use
ADD COMMENT/ADD REPLY
when responding to existing posts to keep threads logically organized.Statisticians would also say to have a balanced design. Not being an statistician, I suggest you drop the duplicates if they are few; or contact a statistician about the best approach to deal with them, if there are many duplications.