either code the NA value as a real value and code and treat it as a number,
or impute it and then ignore the imputation and set it back to NA.
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 Sun, Jan 23, 2011 at 11:13 PM, Blais, Martin <blais.martin(a)uqam.ca>wrote;wrote:
Hi
I need to impute data with two types of missing. Some are real missing that
I want to impute (participants should have answered the questions, but they
skipped it or they were not exposed to it (planned missingness)). Others are
"not applicable" items, so I do not want to impute them...
The two types have different codes in the database (888 for N/A and 999 for
missing). The problem is that I don't know how to prevent the 888s to be
taken as valid value in the imputation model... I want the items to be taken
into account in the imputation model when they have a valid value (that is
not 888) and to be imputed when their value is 999.
How can I avoid the "not applicable" code 888 being taken as a valid value?
Is there an easy way to implement this in AmeliaView? (I do not use R.)
Hope I am clear enough... I could not find any information about this issue
(although it must be very common), but I might not have the right keywords.
Very many thanks,
Martin