Hello,

I'm looking for literature discussing a specific complexity in social network analyses - I want to examine the effect of students' motivation at the group level on the individual student's motivation (y). If I use the aggregated mean of the motivation of all students in a social group (a mean of y's) as a level 2 predictor of y, it would create a bias, since y is part of the mean (that would be especially problematic in small groups, creating a false inflated correlation between the predictor and y).

I am looking for literature discussing that problem, even if a solution is not provided.

I hope I managed to explain the issue clearly and will be happy for any direction, Thank you!