Re: [SPAM] st: Ordered logit

[Richard Williams <Richard.A.Williams.5@ND.edu>]

At 02:14 PM 12/9/2006, Josefina Posadas wrote:
> Dear Statalist users,
> I'm trying to replicate the ordered logit by programing a maximum
> likelihood.
> I'm running the program with the fullauto.dta database as in the manual.
> I have these two programs that I think should be equal, but the first one
> works and the second doesn't. Can somebody explain what is going on?
I'm still curious why you are doing this, but in any event I think I 
see the problem.


> * First I create the indicator variable
> tempvar i1 i2 i3 i4 i5
> forvalues x=1/5 {
>   gen i`x'=0
>   replace i`x'=1 if `x'==rep77
> }
Note that rep77 has 8 cases with missing values.  Under the coding 
above, those cases will all be coded 0 on i1 through i5.


>   replace
> `lnf'=$ML_y1*log(`d1')+$ML_y2*log(`d2'-`d1')+$ML_y3*log(`d3'-`d2')+$ML_y4*log(`d4'-`d3')+$ML_y5*log(1-`d4')
> end
Under program 1's method, the rep77 8 missing cases get coded 0 on lnf.

>   replace `lnf'=log(`d1') if $ML_y1==1
>   replace `lnf'=log(`d2'-`d1') if $ML_y2==1
>   replace `lnf'=log(`d3'-`d2') if $ML_y3==1
>   replace `lnf'=log(`d4'-`d3') if $ML_y4==1
>   replace `lnf'=log(1-`d4') if $ML_y5==1
> end
Under program 2's method, the rep77 8 missing cases never get 
assigned a value on lnf, hence the program can't produce feasible values.

So, you have to do something about missing values.  You could do something like

keep if rep77 < .

Or, your ml model command could be

ml model lf jologit (i1 i2 i3 i4 i5= ) ( ) ( ) ( ) if rep77 < .

Or, to keep the program more general, you could rework the code a bit 
to add marksample and markout commands.

Incidentally, note that even in the "correct" program, the signs are 
the opposite of those reported by ologit; and the N reported was 74 
compared to ologit's 66, again because those 8 cases aren't being 
dropped correctly.


-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
FAX:    (574)288-4373
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW (personal):    http://www.nd.edu/~rwilliam
WWW (department):    http://www.nd.edu/~soc 
*
*   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/