Statalist The Stata Listserver


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: ml and variable types


From   "Feiveson, Alan H. \(JSC-SK311\)" <[email protected]>
To   <[email protected]>
Subject   st: RE: ml and variable types
Date   Wed, 20 Dec 2006 15:57:47 -0600

The short answer is no -  under the -lf- option (most commonly used) the
reason you need high precision is to enable accurate approximation to
first and second partial derivatives of the likelihood function by
differencing. This means that anytime the likelihood is evaluated by
slightly changing the parameter vector, very precise values are needed.
Your original variables don't change values during the estimation
process so it doesn't matter how accurate they are in an absolute sense
- they remain fixed numbers throughout the iterations.



Al Feiveson

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Rachel
Sent: Wednesday, December 20, 2006 3:39 PM
To: statalist
Subject: st: ml and variable types

I'm wondering whether all the variables used in "ml model" equations
need to be of type double.

All my temporary variables are of type double, but the non-temp
(external) variables used to calculate xb are floats and bytes.  "ml
display" is giving me dots in place of the standard errors and/or the
program seems to stop iterating, so I'm wondering whether the external
variable types are the problem.

Thanks again for all your help.  You all have saved me a lot of time.

Rachel
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index