Re: theoretical/practical question about non-ignorable missing processes and how to model them in Amelia
1 Sep
2010
1 Sep
'10
7:12 p.m.
Hi all,
I'm working with a panel data set which contains many missing values for
demographic variable (e.g. mortality rates, life expectancy, etc). Some of
them will be modeled as outcome variables. As it is well know, they are not
missing at random, mostly concentrating around poor and non-democratic
countries. Thus I am assuming I have to model the missing process or at
least provide some information such as the mean of the missing data should
lower or higher for a particular set of countries. I have seen that I can
provide some prior for the imputation procedure in your software but I not
sure whether this is an explicit model of the non-ignorable process. Any
suggestions?
Congratulations on your amazing software!
Help and advice really appreciated,
Antonio.
6 Sep
6 Sep
8:54 p.m.
New subject: theoretical/practical question about non-ignorable missing processes and how to model them in Amelia
Hi Antonio,
Bad news first. Amelia cannot handle non-ignorable missing data
situations. You would have to write a specialized model for that.
Good news. It sounds like you might have ignorable missing data. If
you believe that the missing data depends on only variables that you
observe, then the missing data is still missing at random. Thus, you
could include income and democracy as variables in the imputation
model to satisfy that assumption. Including any other variables on
which the missingness could depend will also help to satisfy MAR.
Cheers,
matt.
On Wed, Sep 1, 2010 at 7:12 PM, Antonio P. Ramos
<ramos.grad.student(a)gmail.com> wrote:
-
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
Hi all,
I'm working with a panel data set which contains many missing values for
demographic variable (e.g. mortality rates, life expectancy, etc). Some of
them will be modeled as outcome variables. As it is well know, they are not
missing at random, mostly concentrating around poor and non-democratic
countries. Thus I am assuming I have to model the missing process or at
least provide some information such as the mean of the missing data should
lower or higher for a particular set of countries. I have seen that I can
provide some prior for the imputation procedure in your software but I not
sure whether this is an explicit model of the non-ignorable process. Any
suggestions?
Congratulations on your amazing software!
Help and advice really appreciated,
Antonio.
8 Sep
8 Sep
4:18 a.m.
New subject: theoretical/practical question about non-ignorable missing processes and how to model them in Amelia
Thank you for your replay. Maybe I'm trying to complicate my life since what
I am trying to do is probably very standard. I have a panel data with more
than 200 countries since 1945. Obviously, I have lost of clustered missing
data for important demographic indicators. So any suggested readings on how
people having been handling such situations recently?
Thank you for you time,
Antonio.
On Mon, Sep 6, 2010 at 5:54 PM, Matt Blackwell <blackwel(a)fas.harvard.edu>wrote;wrote:
Hi Antonio,
Bad news first. Amelia cannot handle non-ignorable missing data
situations. You would have to write a specialized model for that.
Good news. It sounds like you might have ignorable missing data. If
you believe that the missing data depends on only variables that you
observe, then the missing data is still missing at random. Thus, you
could include income and democracy as variables in the imputation
model to satisfy that assumption. Including any other variables on
which the missingness could depend will also help to satisfy MAR.
Cheers,
matt.
On Wed, Sep 1, 2010 at 7:12 PM, Antonio P. Ramos
<ramos.grad.student(a)gmail.com> wrote:
-
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
> Hi all,
>
> I'm working with a panel data set which contains many missing values for
> demographic variable (e.g. mortality rates, life expectancy, etc). Some
of
> them will be modeled as outcome variables. As
it is well know, they are
not
> missing at random, mostly concentrating
around poor and non-democratic
> countries. Thus I am assuming I have to model the missing process or at
> least provide some information such as the mean of the missing data
should
> lower or higher for a particular set of
countries. I have seen that I
can
> provide some prior for the imputation
procedure in your software but I
not
sure
whether this is an explicit model of the non-ignorable process. Any
suggestions?
Congratulations on your amazing software!
Help and advice really appreciated,
Antonio.
10:15 a.m.
New subject: theoretical/practical question about non-ignorable missing processes and how to model them in Amelia
Hi Antonio,
If you have not already, you should check out Honaker and King (2010)
in the American Journal of Political Science. In it, they describe
imputing data such as yours. You can find a copy on Gary's website:
http://gking.harvard.edu/files/abs/pr-abs.shtml
Cheers,
matt.
On Wed, Sep 8, 2010 at 4:18 AM, Antonio P. Ramos
<ramos.grad.student(a)gmail.com> wrote:
Thank you for your replay. Maybe I'm trying to
complicate my life since what
I am trying to do is probably very standard. I have a panel data with more
than 200 countries since 1945. Obviously, I have lost of clustered missing
data for important demographic indicators. So any suggested readings on how
people having been handling such situations recently?
Thank you for you time,
Antonio.
On Mon, Sep 6, 2010 at 5:54 PM, Matt Blackwell <blackwel(a)fas.harvard.edu>
wrote:
-
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
Hi Antonio,
Bad news first. Amelia cannot handle non-ignorable missing data
situations. You would have to write a specialized model for that.
Good news. It sounds like you might have ignorable missing data. If
you believe that the missing data depends on only variables that you
observe, then the missing data is still missing at random. Thus, you
could include income and democracy as variables in the imputation
model to satisfy that assumption. Including any other variables on
which the missingness could depend will also help to satisfy MAR.
Cheers,
matt.
On Wed, Sep 1, 2010 at 7:12 PM, Antonio P. Ramos
<ramos.grad.student(a)gmail.com> wrote:
-
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
Hi all,
I'm working with a panel data set which contains many missing values
for
demographic variable (e.g. mortality rates, life expectancy, etc). Some
of
them will be modeled as outcome variables. As it is well know, they are
not
missing at random, mostly concentrating around poor and non-democratic
countries. Thus I am assuming I have to model the missing process or at
least provide some information such as the mean of the missing data
should
lower or higher for a particular set of countries. I have seen that I
can
provide some prior for the imputation procedure in your software but I
not
sure whether this is an explicit model of the non-ignorable process.
Any
suggestions?
Congratulations on your amazing software!
Help and advice really appreciated,
Antonio.
5002
days inactive
5009
days old
amelia@lists.gking.harvard.edu
3 comments
2 participants
participants (2)
-
Antonio P. Ramos -
Matt Blackwell