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 on April 23, and its replacement, statalist.org is already up and running.


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

Re: Re: st: NLLS w/ normal distribution - some parameters not estimated


From   Maarten buis <maartenbuis@yahoo.co.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: Re: st: NLLS w/ normal distribution - some parameters not estimated
Date   Mon, 14 Jun 2010 06:38:16 +0000 (GMT)

--- On Sun, 13/6/10, Denis Kalugin wrote:
> I was thinking about ml, but it's beyond me how to input
> two sets of parameters (xb1 and xb2) into ml model
> estimation. 

That is pretty standard, see for example the example
below. 

*------------- begin example -----------------------
program drop _all
program define mynormal_lf
	args lnf ln_xb ln_zb
	qui replace `lnf' =  ///
            lnnormalden($ML_y1, exp(`ln_xb'), exp(`ln_zb'))
end
sysuse auto, clear
ml model lf mynormal_lf (xb: price = foreign rep78 mpg gear_ratio) ///
                        (ln_zb: gear_ratio)
ml check
ml search
ml maximize
*---------------- end example -----------------------

> On top of that, several authors commented that in this
> setting ml estimation often results in -infinity 
> loglikelihood value (which i got on a certain stage of
> my trials) hence using this approach is not advisable.

I would say that the model is not advisable: as I said 
earlier, the error variances are probably not identified.
So the algorithm does exactly what it should do, not 
give you an answer when none exists. Searching for an 
algorithm that does give you an answer, even when none
exists is obviously the wrong way to go. 

> I might give gmm a try, but it sounds like a rather
> complicated programming task. Can you suggest a good
> reading on introductory gmm in Stata?

The -gmm- command is new to Stata 11, so the users haven't had
time yet to start writing about it, they are now figuring it
out for themselves. However, the StataCorp people are now 
talks on it at the Stata Users' Group meetings, e.g.:
<http://www.stata.com/meeting/mexico10/mex10sug_drukker.pdf>

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

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


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