Hi Renzo,
You should set the intercs argument to TRUE and include either polytime=0
or 1. This will allow imputations to depend on country specific fixed
effects or time trends. If that isn't enough, you could create your own
interactions of the country variable with other variables and include that
in the imputation.
Cheers,
Matt
On Friday, August 29, 2014, Renzo Carriero <renzo.carriero(a)unito.it> wrote:
Dear Amelia users,
I wish to know what is the treatment Amelia does with cross national
datasets. I'm using the fourth wave of the European Values Study (a
European version of the WVS). THe same question are asked in different
countries. I would like to impute data on household income. The rate of
missingness ranges 8% to 48% across countries and also the predictive power
of my predictors changes across countries. Does Amelia take into account
that? I made some tests with the country variable set as cross section
variable or not, but it didn't change so much.
I read an article that made use of multiple imputation with Amelia on a
multinational dataset (Linos, K. & West, M. 2003 Self-interest, Social
Beliefs, and Attitudes to Redistribution, ESR) but there are not details on
this issue. Does anybody knows how to deal with that?
Thanks
Renzo
--
Renzo Carriero, PhD
Dipartimento di Culture, Politica e Società
Lungo Dora Siena 100
10153 Torino - Italy
+390116702617 (office)
Room 3D310
---
Questa e-mail è priva di virus e malware perché è attiva la protezione
avast! Antivirus.
http://www.avast.com
--
Amelia mailing list served by HUIT
[Un]Subscribe/View Archive:
http://lists.gking.harvard.edu/?info=amelia
More info about Amelia:
http://gking.harvard.edu/amelia
Amelia mailing list
Amelia(a)lists.gking.harvard.edu
To unsubscribe from this list or get other information:
https://lists.gking.harvard.edu/mailman/listinfo/amelia
--
-----
Matthew Blackwell
http://www.mattblackwell.org