In my dataset, different observations should have different weights to account for the sampling procedure.
Could you elaborate a bit on this? My first thought would be to incorporate your weights through the observation variance, but I’d like to figure out whether this is a reasonable thing to do or not.