Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

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

RE: st: Ado File for MI Estimation Stata 12

From   Nick Cox <>
To   "''" <>
Subject   RE: st: Ado File for MI Estimation Stata 12
Date   Fri, 11 Nov 2011 09:45:00 +0000

I think you are bundling together different problem reports which I have flagged (a) (b) (c) in your posting. 

You will get pertinent advice faster -- and avoid misleading list members with what turn out to be side issues -- by posting in more detail with concrete (not schematised) examples of your real commands. See the Statalist FAQ for advice and especially note the injunction: 

"Say exactly what you typed and exactly what Stata typed (or did) in response. N.B. exactly! If you can, reproduce the error with one of Stata's provided datasets or a simple concocted dataset that you include in your posting."

Addressing (a): 

The -cmdok- option is thus revealed to be necessary but not sufficient if -mi- does not explicitly support an estimation command. That is, -factor- is an estimation command but of idiosyncratic kind and in particular it does not leave behind certain e-class results that are essential for -mi- to work. The way ahead is closed to you on this route. It is not clear quite what you expect -factor- to impute, in any case. 

At a wild guess, you are misreading -mi- as a kind of all-purpose framework for doing any model estimation even if you have missing values. It is very general, but not that general, at least at present. 

What to do next depends on your goal. Perhaps -sem- will allow you to formulate a similar if not identical factor model. Others are better placed to make constructive suggestions but I think they would need to know much more about your substantive goals. 

Addressing (b) and (c): 

The full commands would help people provide better advice. 


Donna L Quadri [edited] 


When using the syntax mi estimate, cmdok: factor x1 x2 x3..... The
error message I receive is:

matrix e(b) is not set
matrix e(V) is not set

when using the syntax mi estimate, cmdok: sem ... it worked fine.


When performing estat gof on the sem estimation, received the error

invalid subcommand gof


same response with mi estimate, cmdok: worked fine but no
estat commands did. Received same error message.

On Thu, Nov 10, 2011 at 2:19 PM, Nick Cox <> wrote:

> You don't show us any of your syntax so it is difficult to comment. -ok- is an option of -mi estimate-, so your call would start
> mi estimate, ok: <whatever>
> except that there may be other options too.

Donna L Quadri

> I must not have the correct syntax for the cmdok option because I
> receive errors that I did not set it correctly. That would be
> preferable because I have never set up a program in Stata before and
> having challenges doing so correctly.
> Does the cmdok option require the program? Wasn't able to find a
> sample of the correct syntax to follow. Again, thanks,

On Thu, Nov 10, 2011 at 2:01 PM, Richard Williams

>> I'm not totally sure I understand the question, but you can set up the
>> properties for a program to allow mi. See
>> help program_properties##mi
>> With -mi estimate- you can also use the -cmdok- option for unsupported
>> commands. See
>> help mi estimate

At 01:36 PM 11/10/2011, Donna L Quadri wrote:

>>> Do you know of an existing ado-file that permits estimation commands
>>> not currently available with mi estimate? If not, any advice to a
>>> first time programming effort in Stata 12. would be much appreciated.
>>> I am using Stata IC/12.0 and wish to use anova, factor, and the new
>>> sem functions. It would be a program beginning with
>>> mi_cmd.....according to the Technical note.

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index