15 Jul
2008
15 Jul
'08
10 p.m.
Hello,
I have run Amelia and got 10 datasets with imputed values. I have two
questions:
1) although there is an option to get the output in STATA, it's not
working in my version of Amelia (I can only get the output in the
other formats). Is there a way to get the imputed datasets in STATA
format?
2) I have combined and analyzed the imputed datasets with "miest", but
I wonder how to get descriptive statistics and a measure of goodness
of fit? Should this be done for each of the imputed datasets
separately or is there a command to get combined/summary measrues?
Thanks!
Carlos
-
Amelia mailing list served by Harvard-MIT Data Center
[Un]Subscribe/View Archive: http://lists.gking.harvard.edu/?info=amelia
17 Jul
17 Jul
12:04 p.m.
Hi Carlos,
1) Make sure that you have the latest version of R (2.7.1) and Amelia
(1.1-31) installed. You can find out which version of Amelia you have
by running the following command: packageDescription("Amelia"). To
install the latest version, update R to its latest version and run:
install.packages("Amelia", repos="http://gking.harvard.edu")
2) There are a set of rules for combining imputation estimates that
are sometimes called the Rubin rules. For point estimates (means and
things like this), you take the mean of the point estimates from each
imputed dataset. For variances, the rule is slightly more complicated.
See the following cite for the rules and a general overview of how to
handle imputed data:
King, Gary; James Honaker, Anne Joseph, and Kenneth Scheve. Analyzing
Incomplete Political Science Data: An Alternative Algorithm for
Multiple Imputation, American Political Science Review, Vol. 95, No. 1
(March, 2001): Pp. 49-69. http://gking.harvard.edu/files/evil.pdf
good luck,
matt.
On Tue, Jul 15, 2008 at 10:00 PM, Carlos Rodriguez
<carlosrodriguez1993(a)gmail.com> wrote:
Hello,
I have run Amelia and got 10 datasets with imputed values. I have two
questions:
1) although there is an option to get the output in STATA, it's not
working in my version of Amelia (I can only get the output in the
other formats). Is there a way to get the imputed datasets in STATA
format?
2) I have combined and analyzed the imputed datasets with "miest", but
I wonder how to get descriptive statistics and a measure of goodness
of fit? Should this be done for each of the imputed datasets
separately or is there a command to get combined/summary measrues?
Thanks!
Carlos
-
Amelia mailing list served by Harvard-MIT Data Center
[Un]Subscribe/View Archive: http://lists.gking.harvard.edu/?info=amelia
-
Amelia mailing list served by Harvard-MIT Data Center
[Un]Subscribe/View Archive: http://lists.gking.harvard.edu/?info=amelia
1:33 p.m.
2) There are a set of rules for combining imputation
estimates that
are sometimes called the Rubin rules. For point estimates (means and
things like this), you take the mean of the point estimates from each
imputed dataset. For variances, the rule is slightly more complicated.
See the following cite for the rules and a general overview of how to
handle imputed data:
I think what Carlos is asking is what to do with fit stats and
descriptives. I know Mplus averages fit statistics but I have not
seen any justification for this behavior in anything published. Is it
OK to average R squares, Chi-square tests, CFI, TLI, SRMR, etc. ?
The descriptives question is strait forward for means, percentiles
(average) variances, standard errors (the more complicated rule) but
tricky for minimums and maximums, skewness, krutosis, etc. Not sure
what to do with these either. Maybe averaging them is OK. For
minimum/maximum I would use minimum/maximum without imputation.
L
2) I have
combined and analyzed the imputed datasets with "miest",
but
I wonder how to get descriptive statistics and a measure of goodness
of fit? Should this be done for each of the imputed datasets
separately or is there a command to get combined/summary measrues?
Thanks!
Carlos
-
Amelia mailing list served by Harvard-MIT Data Center
[Un]Subscribe/View Archive: http://lists.gking.harvard.edu/?info=amelia
-
Amelia mailing list served by Harvard-MIT Data Center
[Un]Subscribe/View Archive: http://lists.gking.harvard.edu/?
info=amelia
5769
days inactive
5770
days old
amelia@lists.gking.harvard.edu
2 comments
3 participants
participants (3)
-
Carlos Rodriguez -
LevI Littvay (UNL) -
Matt Blackwell