Re: st: nonlinear maximum likelihood with ml

[Christopher Baum <kit.baum@bc.edu>]

<>
On May 3, 2011, at 2:33 AM, Tatyana wrote:

> Is there any way to use "ml" to estimate a non-linear equation? I'm
> trying to estimate the parameters of the distribution of an outcome
> (Y) assuming that Y follows a Beta distribution with parameters that
> depend on some X's. It's easy to write down the log likelihood
> function, but the equation for Y is non-linear.


Equations to be estimated with -ml- are almost always nonlinear: even in the case of using -ml- to estimate a linear regression. It can just as well estimate a nonlinear regression, or a model involving other distributions such as you suggest above. But before reinventing the wheel, check out the available 'canned' routines in this area, such as -betafit-, -betaprior-, -dagfit-, -gb2fit-, -gbgfit- from SSC, all of which show up with -findit beta-.

Kit

Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
                              An Introduction to Stata Programming  |   http://www.stata-press.com/books/isp.html
   An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html




*
*   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/