|
|  |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: ancillary parameters in ml display
Partha Deb <[email protected]> is using -e(k_aux)- to get
-ml display- to report ancillary parameters using the '/' notation:
> I have a question regarding the display of ancillary parameters in -ml
> display- .
>
> When I specify my model (-ml model-) with ancillary parameters at the end of
> the specification, all is well:
>
> e.g.
>
> ml model (one: y1 = x1) (two: y2 = x2) /ancil1 /ancil2
>
> e(k_eq) = 4
> e(k_aux) = 2
>
> But if I specify the model with ancillary parameters between equations one
> and two:
>
> ml model (one: y1 = x1) /ancil1 /ancil2 (two: y2 = x2)
>
> e(k_eq) = 4
> e(k_aux) = 2
>
> -ml display- treats one: and /ancil1 as "equations" and /ancil2 and
> two:_const as ancillary parameters. The coefficient on x2 is not displayed.
>
> Is it possible to specify ancillary parameters between equations? It's
> desirable for my application because in the default specification the
> "equation" labeled "two" is also an ancillary parameter, and "two" is best
> reported at the end.
When you set -e(k_aux)- to 2, -ml display- understands this to mean that the
last 2 equations in -e(b)- are to be treated as ancillary parameters,
regardless of whether those equations contain predictors. Thus Partha will
have to use the first -ml model- specification (above) instead of the second
to get the coefficient on -x2- to be reported.
The only alternative I can think of is to use the second specification and set
-e(k_aux)- to 3 only when there are no predictors in equation 'two:'. This is
reasonable assuming specifying predictors for equation 'two:' is a rare thing
when fitting Partha's model.
--Jeff
[email protected]
*
* 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/
