I have an additional question on the topic of between-level factors scores in a multi-level model.
I am attempting to model variations in the occurrence of a dichotomous dependent variable, using within and between weights (person and agency weights).
When I run an unconditional model and save factor scores, I find that the between level factor scores vary within cluster and seem to be associated with the magnitude of the within level weights. Taking off the weight statements, the factor scores are constant within clusters, as I am used to seeing in multi-level models I have run before.
In an unconditional model, is it common/expected for between level factor score estimates to vary within clusters as a function of the within weights? Why?
Here is a small sample of my SAVE=FSCORES output, just showing that the factor score estimates vary within agencies.