Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Amelia object as data.frame


From   William Buchanan <william@williambuchanan.net>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Amelia object as data.frame
Date   Tue, 26 Feb 2013 07:27:06 -0800

Hi Laura,

The issue that Daniel pointed out wasn't that MI doesn't support -msp-, but that -msp- does not support MI data.  For more information, see -help program_properties##mi- which explains some of the information that programmers use to make their packages support MI data.  One reason that the program is likely incompatible with the -mi estimate- command is that -msp- is an rclass program (a point that Nick Cox made yesterday in response to your previous thread).  

That being said, you could always use -msp- as a framework to develop your own procedure (say -msp2-) that would be an eclass program and would support MI commands.

HTH,
Billy



On Feb 26, 2013, at 7:08 AM, Laura Maria Schwirz <schwirzl@tcd.ie> wrote:

> Thanks for your advice. Stata's mi does in fact not support msp and
> appears to allow mainly for various types of regression. Using another
> programme to impute data is just one option although as you very well
> point out I do need to bear in mind assumptions underlying mi and MSP.
> 
> On 26 February 2013 14:09, daniel klein <klein.daniel.81@gmail.com> wrote:
>> Aside from the obvious -- R questions are not the topic to be
>> discussed on Statalist -- based on your earlier question
>> (http://www.stata.com/statalist/archive/2013-02/msg00960.html) I get
>> the impression you switch software, because Stata did not do what you
>> wanted.
>> 
>> That is not necessarily a problem, but just some words of caution.
>> 
>> Chaniging to another software, because Stata does not do what you
>> want, might not always be the best choice, given that there might be
>> good reason why Stat does not do what you want. Just because a(nother)
>> software does something, it does not mean that this something makes
>> any sense, or is statistically "correct".
>> 
>> In your case the appearant reason -msp- (SSC) does not work with -mi-
>> is, that its author did not implement it to work with -mi-. Whether
>> there is a good statistical reason remains unclear to me. My point is,
>> that in any case you should think carefully about about what you want
>> to do. If you, for example,  want to combine Loevinger's H, think
>> about the distribution of this statistic. Is it normal? If not,
>> applying Rubin's combination rules (regardless of software) might not
>> be appropriate. This is discussed here:
>> http://www.stata.com/support/faqs/statistics/combine-results-with-multiply-imputed-data/
>> along with how to get Stata to combine results from commands that do
>> not support -mi-.
>> 
>> btw. Ameliaseems to  assume the data to follow a multivariat normal
>> distribution, which might not be appropriate with Items used for
>> Mokken scale analysis, and you need to think about this, too.
>> 
>> Best
>> Daniel
>> 
>> --
>> Hi Stata Users
>> 
>> I have run multiple imputations using R's Amelia package and would
>> like to use the imputed dataset to analyse Mokken Scale Analysis.
>> But mokken requires the object to be a data frame. I tried
>> as.data.frame(x) and as.matrix(x) but it says that it cannot coerce
>> class amelia into a data frame or matrix.
>> 
>> australia93=as.data.frame(australia93)
>> Error in as.data.frame.default(australia93) :
>>  cannot coerce class '"amelia"' into a data.frame
>> 
>> coefH(australia93)
>> Error in check.data(X) : Data are not matrix or data.frame
>> 
>> Any help would be much appreciated.
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>> *   http://www.ats.ucla.edu/stat/stata/
>> 
> 
> 
> 
> --
> Laura Schwirz
> 
> PhD Candidate and IRCHSS Scholar
> Department of Political Science
> Trinity College Dublin
> Dublin 2
> Republic of Ireland
> Email: schwirzl@tcd.ie
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index