I have two questions about the mixed command and aweights. I have multilevel data where the level 1 units are themselves geographical units (municipalities) with widely varying populations, nested within regions at level 2. My outcome variable consists of means at the municipality level. As a result I would like to use aweights to adjust for unequal sizes of level 1 units in a multilevel model. However, mixed does not accept aweights. As a result I have the following questions:
1. Is there a good reason why I shouldn't be trying to use weights of this kind at the first level of a linear mixed model?
2. My proposed solution to this problem is to reverse engineer aweights in the following way. Rescale my weight variable so its' total adds up to the total N in my model, and then enter this into the mixed command as an fweight. Are there any problems with my idea?
Thanks in advance!
1. Is there a good reason why I shouldn't be trying to use weights of this kind at the first level of a linear mixed model?
2. My proposed solution to this problem is to reverse engineer aweights in the following way. Rescale my weight variable so its' total adds up to the total N in my model, and then enter this into the mixed command as an fweight. Are there any problems with my idea?
Thanks in advance!
