Hello!
I stiil have problem with Amelia.
I have problem with starting it.
The datafile that I used was an ASCII file. I put it into memory using
load [,] command. It worked perfectly till I run Amelia. Then the error
occured:
run c:\gauss\amelia\amelia.c
*********************
Amelia Begins
*********************
C:\GAUSS\SRC\AMELIA.SRC(233) : error G0014 : 'C:\gauss\DATA1.dht' : File not
found
Currently active call: AMELIA [233]
I tried to save data in Gauss format. I did it (in .dat file).
But when I tried to run Amelia the problem occured again.
Could you help me with it, what may be wrong?
Best regards,
Andrzej
Poland
Suppose the following analysis model:
y = a + b1x1 + b2x2 + b3x3 +b4x1x2 + b5x1x3 + b6x1x2x3 + u
Suppose x1 is a multi item measure (factor scale) consisting of z1 + z2
+ z3 + z4
Suppose that all variables contain some degree of MAR data.
Based upon previous listserve responses I suppose the proper imputation
model would contain the following:
y, z1,z2,z3,z4,x1(built from z
variables),x2,x3,(x1*x2),(x1*x3),(x1*x2*x3) + any related imputation
model variables (which probably should be included in the analysis model
as exogenous terms, but that's another story).
Imputation would fill all missing values in all these variables.
However, to regain the information lost in the constructed variables
(i.e. x1 and interactions) when they are created with variable
containing some (but not all) missing data, (after imputation) you
should delete and rebuild x1 and the interaction variables.
Does this sound correct?
Matthew Vile,
University of New Orleans
Hello!
Some time ago I scientifically began to deal with the problem of missing
data analysis, especially multiple imputation. While trying to use Amelia
for Gauss I approached a real problem. I tried to make some imputations on
my data with five interval variables. Two of them were MCAR. As in the
example in the manual:
new;
library amelia;
_AMepri=3;
dbuf = amelia("miss"); ***** "miss" is my data name
I tried to start Amelia, but instead of imputation I received info:
C:\gauss\miss.dht: (CODE NUMBER: ....) File NOT FOUND.
I verified if data is OK, changed its name, but nothing helped.
What can be wrong?
I use Gauss for Windows 3.2.33.
Thank you for help in advance!
Andrzej Kwiatkowski
Assistant at the Institute of Statistics and Demography
Warsaw School of Economics
Warsaw, Poland
e-mail: akwiatko(a)sgh.waw.pl
below...
On Thu, 13 Feb 2003, SUAZO-GARCIA, BELKIS wrote:
> Dear Professor Gary King:
>
> I am using your AMELIA program for a sample drawn from NELS:88 (National
> Education Longitudinal Study of 1988, administered by the U.S. Department of
> Education, National Center of Education Statistics). The NELS:88 is based
> on a two stage, stratified sample design, with schools as the first-stage
> unit and students within schools as the second-stage unit. Because of the
> sampling design, I must use a statistical program that calculates standard
> errors by taking into account the complex sample design. So, I can't use
> STATA because it does not correct for complex sample designs. Instead, I am
> using AM, a program designed by the American Institutes for Research
> (http://am.air.org).
>
> I want to create one file in SPSS that contains all five Ameliarized data
> sets. I plan to bring this one file into AM and divide the sample weight by
> five to adjust the degrees of freedom to the correct sample size. I will
> like to know if this a correct use of the Ameliarized data sets. Is this
> procedure similar to what happens to the data in STATA? The alternative is
> to run my models separately on each Ameliarized data set and then average
> the results myself. I prefer not to do the latter because it will be an
> enormous task.
>
> Thank you for your assistance. I look forward to your response.
>
> Sincerely,
>
> Belkis Suazo-Garcia
>
>
> ________________________________________________
> Belkis Suazo-Garcia,
> Doctoral Candidate
> Ph.D. Program in Sociology
> CUNY Graduate Center
> 365 Fifth Avenue
> New York, NY 10016
> Office: (212) 817-8786
> Email: bsuazo-garcia(a)gc.cuny.edu
Stata does have options to correct for complex sample designs, and so you
might want to start there. Then you could use Clarify or our MI program
to automate the combination of the data sets.
As to your question, I wouldn't stack the datasets, if that's what you
have in mind. instead, if it is too big a job to run the whole analysis 5
times, I would just use one data set, or possibly use the EM option in
Amelia and do the same. The cost would be that your standard errors would
be too small, but the point estimates would still be about right.
You should also check with Mitch Duneier in your department; he's an
expert in Amelia.
Gary King
: Gary King, King(a)Harvard.Edu http://GKing.Harvard.Edu :
: Center for Basic Research Direct (617) 495-2027 :
: in the Social Sciences Assistant (617) 495-9271 :
: 34 Kirkland Street, Rm. 2 HU-MIT DC (617) 495-4734 :
: Harvard U, Cambridge, MA 02138 eFax (928) 832-7022 :
Wir sind überzeugt, dass Sie unsern SQLArtist testen sollten!
Mit dem SQLArtist verwalten Sie auch komplexe Datenbanken mühelos und sicher.
- Editieren von DB-Strukturen
- SQL-Scripts erstellen (Wizard)
- DB-Versions-Verwaltung
- Möglichkeit der Anbindung an Borland Delphi und PowerDesigner
- und noch viele weitere Funktionen
URL zu Produktebeschreibung und Download Vollversion:
http://www.vid-software.com/prod_sqlartist.php
Mit freundlichen Grüssen
VID-Software GmbH
Stefano Marinello
Hofstrasse 99
8620 Wetzikon
Switzerland
-----------------------------------------------------------------------
Um sich von dieser Mailingliste abzumelden
http://www.vid-software.com/newsletter/lists/?p=unsubscribe&hash=696f46f49a…
Um Ihre Benutzerdaten zu manipulieren
http://www.vid-software.com/newsletter/lists/?p=preferences&uid=696f46f49a4…
below...
On Tue, 11 Feb 2003, Paul Miller wrote:
>
> Dear Dr. King,
>
> Im writing because I have a question about multiple imputation that I
> thought you might be able to answer. I am working on an analysis
> involving an interaction between two latent variables. Specifically, I
> want to use peoples reports of their spouses affection, peoples
> reports of their spouses negativity, and the interaction between these
> variables (all latent) to predict peoples ratings of their partners
> responsiveness. In testing this model, Ive decided to use a 2SLS
> approach developed by Ken Bollen. So my analysis will make use of
> several interaction terms, most of which will function as instrumental
> variables.
>
> In the past, Ive used Joe Schafers program Norm to do multiple
> imputation. However, I was thinking that I might give Amelia a try this
> time since it is supposed to work well with small samples. The
> documentation for Norm indicates that imputations based on a
> multivariate normal model do not preserve interactions among variables.
> However, in Analyzing Incomplete Political Science Data. . . you
> indicate that the user should Include interaction terms that will be
> used in the analysis model and that might explain the data
> distribution. This is the case despite the fact that Amelia also
> appears to be based on a multivariate normal model. To me this suggests
> that it is acceptable to use multiple imputation for an analysis that
> will use interaction terms, provided that these terms are included in
> the dataset that is submitted to the multiple imputation program. Is
> this correct? Or have I misunderstood what you wrote?
>
> Paul
>
What you'd do is to include (say) X, Z and the product of the two as
variables that are input into Amelia, and let it impute. This approach
has two difficulties. The first is that the imputations for Z times the
imputations for X are not constrained by the program to equal the
imputations for the product term. if I were you, i'd run the program and
plot "the imputations of X times the imputations of Z" by the imputations
of "Z*X" to check. If they're far off, you might consider changing the
specification some. But either way, I'd suggest that you only take out of
Amelia the imputed X and the imputed Z and then do with them as you see
fit, including taking their product. I.e., I'd suggest that you discard
the imputations of the product.
The other difficulty is that if X is missing for one observation and Z is
observed, then X*Z will be missing even though part of it is really
observed. This means there is some loss of information, but no one has
figured out a method to deal with this (perhaps because no one has really
tried!).
Both of these issues should affect both Joe's program and Amelia, in the
same way. Amelia should run a good deal faster and it should work more
frequently. But other than that, they give basically the same
imputations.
Incidentally, if you're conducting the survey yourself, I'd suggest you
have a look at the paper at http://gking.harvard.edu/vign/ which may be of
use to you.
Best of luck,
Gary
: Gary King, King(a)Harvard.Edu http://GKing.Harvard.Edu :
: Center for Basic Research Direct (617) 495-2027 :
: in the Social Sciences Assistant (617) 495-9271 :
: 34 Kirkland Street, Rm. 2 HU-MIT DC (617) 495-4734 :
: Harvard U, Cambridge, MA 02138 eFax (928) 832-7022 :
We have been imputing various versions of a gradually changing set of items from an African public opinion data set, with 21,000 cases.
Each time, I have been pushing the limits of the recommended variable limit. I was successfully able to impute data sets of 54 variables, though it has taken as long as 14 hours.
Now after several rounds of preliminary analysis, we have converged on a final model of 54 items, with some new constructs that differ from earlier versions. However, two attempts to impute this final data set have been apparently unsuccessful. I abandoned the first attempt after 24 hours, and over 400 iterations during Stage 1. Amelia never reached State 2. After some tinkering and dropping of some items, the second attempt has been going on for over 15 hours and is now over 200 iterations. For some reason, Amelia is not able to reach convergence. The program did not bomb, but was merely going through an apparently endless series of EM iterations.
I have two questions:
1. Should I simply be patient? Will the program eventually reach convergence, so long as it is still going through iterations, rather than freezing or bombing? Or, has the program gone into a perpetual loop with an unsolvable set of missing values?
2. All the variables entered into this model that have high levels of missingness (at least 3 have missingness on approximately 50 percent of cases) have been in earlier runs that were successful. Can the particular combination of missingness present a solution that Amelia is able to solve, or reach convergence? Or does it seem that there is some other type of problem?
Obviously, any help would be greatly appreciated.
Bob Mattes
Dear Dr. King,
I am using Amelia for Windows.
As I understand it, the usual sequence of steps is imputation - analysis
- combination. Is it possible to combine the datasets before analysis?
What I want to do is to replace the missing values, then combine all
datasets into a single dataset (by averaging) and then perform the
analyses.
I need to do this because I want to create a factor score out of eight
indicators and then use the score in further analyses.
Is this an appropriate way of using Amelia?
I would appreciate any ideas.
Thank you,
Claudiu Tufis