possibly, but the imputation stage just requires a good predictive model.
if the weights help you with prediction, great.
when you go to combine the resulting data+imputations into a small number of
quantities of interest, that's when the weights will be of more use.
Gary
--
*Gary King* - Albert J. Weatherhead III University Professor - Director,
IQSS - Harvard University
GKing.Harvard.edu <http://gking.harvard.edu/> - King(a)Harvard.edu -
@kinggary<http://twitter.com/kinggary>- 617-500-7570 - Asst 495-9271 -
Fax 812-8581
On Wed, Mar 16, 2011 at 8:54 PM, Kurt Smith
<kurt.smith(a)archimedesmodel.com>wrote;wrote:
Thank you for the quick response Gary.
My concern was that omitting the weights could bias the model that is
fit to the data and subsequently used to impute missing values. For
instance Amelia fits a multivariate normal distribution over all
variables. Wouldn't a different distribution result if the sampling
weights were considered when it was determined?
Kurt
On Wed, 2011-03-16 at 20:43 -0400, Gary King wrote:
just impute as if there were no weights (probably
including the
weights themselves as one of the variables used in the imputation
algorithm). then use the weights as part of your analysis model as
usual.
Gary
--
Gary King - Albert J. Weatherhead III University Professor - Director,
IQSS - Harvard University
GKing.Harvard.edu - King(a)Harvard.edu - @kinggary - 617-500-7570 - Asst
495-9271 - Fax 812-8581
On Wed, Mar 16, 2011 at 8:39 PM, Kurt Smith
<kurt.smith(a)archimedesmodel.com> wrote:
I am interested in performing multiple imputation for a data
set that
contains sampling weights. (In other words, each subject in
the data has
a sampling weight that must be taken into account when fitting
regressions, etc) I did not find any information on this in
the Amelia
documentation. Is it possible to do this with Amelia?
-Kurt Smith
--
Kurt Smith, PhD
Scientist II
Archimedes Inc
201 Mission Street, 29th Floor
San Francisco, CA 94105
415.490.0591
-
Amelia mailing list served by Harvard-MIT Data Center
[Un]Subscribe/View Archive:
http://lists.gking.harvard.edu/?info=amelia
More info about Amelia:
http://gking.harvard.edu/amelia