Gary

Gary King - Albert J. Weatherhead III University Professor - Director, IQSS - Harvard University

GKing.Harvard.edu - King@Harvard.edu - @kinggary - 617-500-7570 - Asst 495-9271 - Fax 812-8581

On Mon, Nov 19, 2012 at 9:24 AM, Matt Blackwell <m.blackwell@rochester.edu> wrote:

Hi Kamontip,

Unfortunately, Amelia, along with almost all multiple imputation software, requires the data be MAR. In general, if the data is MNAR, you will have to build an application-specific model that incorporates specific assumptions about the missing data mechanism.

Cheers,
matt.

~~~~~~~~~~~
Matthew Blackwell
Assistant Professor of Political Science
University of Rochester
url: http://www.mattblackwell.org

On Sat, Nov 17, 2012 at 11:39 AM, Kamontip Srihaset <Kamontip.S@student.chula.ac.th> wrote:

Dear All,

I am Ph.D student at Chulalongkorn University in Thailand, I use your package to impute missing data assump MAR and MNAR. I don't have problem to impute under MAR, but I don't know how to impute MNAR. My MNAR data generate under IRT model(3-PL);

n<-500 ## number of examinee

I<-20 ## number of items

num.imp<-5 ##number of imputations

p.missing<-c(0.09, 0.01) #prob of missing

theta<-sort(rnorm(n,0,1)) #ability

a<-rnorm(I,0.5,0.1) #discrimination

b<-rnorm(I,0,1) #difficulty

c<-runif(I,0,0.25) #guess

Only item 1-4 have missing data. If the response to each items was a 1 (correct), the probability of missing for each items was 1%. If the

response was a 0 (incorrect), the probability of missing was 9%. Thus, the probability of missing each items was linked to the response of each items itself (an unknown characteristic in real missing data situations).

Could you please tell me function or how to impute data under my situation. I'am looking forward your advice.

Sincerely yours,

Kamontip Srihaset

--
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@lists.gking.harvard.edu

To unsubscribe from this list or get other information:

https://lists.gking.harvard.edu/mailman/listinfo/amelia

--
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@lists.gking.harvard.edu

To unsubscribe from this list or get other information:

https://lists.gking.harvard.edu/mailman/listinfo/amelia