st: programming maximum likelihood estimation for specific logistic function

[Miriam Dietrich <miri.dietrich@gmail.com>]

Hello,

I want to run a maximum likelihood estimation of the following
logistic function:

L=ππ(1/(1+e^(-x/variance) ))^y (1/(1+e^(x/variance) ))^(1-y)

And this is what I wrote in Stata:

program define mylogit
1.
. args lnf mu sigma
2.
. quietly replace `lnf' = -ln(1+exp(-`mu'/`sigma')) if $ML_y1==1
3.
. quietly replace `lnf' =-ln(1+exp(`mu'/`sigma')) if $ML_y1==0
4.
. end
. ml model lf mylogit ( bserved= titfortat)
. ml maximize

The problem is that Stata gives me an error message saying invalid name.

I have also tried a couple of variations to understand what the
problem might be. It works if I use:
quietly replace `lnf' = -ln(1+exp(-`Xb')) if $ML_y1==1
but if i use quietly replace `lnf' = -ln(1+exp(-`Xb'/`sigma')) if
$ML_y1==1 it says unknown function.

Does anyone has any ideas?

Thank you very much for your help in advanced.

Miriam

*
*   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/