You should follow the steps in the example at the bottom of
help(imp.mix)...so yes, you do need to do some pre-processing before you
use imp.mix().
Also, note that I forgot to mention something in section last week: If we
fix tau1 = 0, we need to fix tau2 > 0. So you need to exp(par[k+1]) it to
contstrain it to be greater than zero, but I'm sure that everyone has
figured that out by now... 8)
Olivia.
On Wed, 15 Dec 2004, Jens Hainmueller wrote:
is 'imp.mix' the right command to use in the
mix package for the
imputations? it's the only imputation type command i cound find there.
'imp.mix' apparently requires pre-processing with 'da.mix' or
'dabipf.mix'
do we need to use any of those too?
thanks!
in da mix,
j.
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l