Hello everyone
Dr. King suggested I send this here. I just signed up to the list. I
am looking forward to joining the conversation.
I have three simple questions that did not become clear from the manual.
When I set the polynomials, setting 0 means time will not be included in
the model or time squared will not be included? For some odd reason I
had a hard time getting the model to converge with time in there.
(Unless I set priors of course.)
Second, I am working in a multilevel modeling framework. I see the
interact with cross-section option. Does this mean that all
coefficients are allowed to vary across cross-sectional units ( which
are countries in my case) or just the impact of time? If only the
latter is true, what are the implication of using the multiply imputed
data in a multilevel model where all variance components are estimated.
This situation would mean that the imputation model is more restrictive
then the analytical model. I thought the imputation model has to be the
analysis model with possible inclusion of additional auxiliary
variables. But I do not really know what happens when this is violated.
And finally, can Amelia be used in non-longitudinal nested data
situations? I guess this is tied in with the previous two questions.
If time is not specified but cross-sectional index is, does the
imputation model has the properties of a multilevel model where the
relationships between the variables are allowed to vary across
cross-sectional units?
Thanks
Levi Littvay
University of Nebraska-Lincoln
-
Amelia mailing list served by Harvard-MIT Data Center
[Un]Subscribe/View Archive: http://lists.gking.harvard.edu/?info=amelia