Re: st: Obtaining predictions with margeff & gologit2

[Richard Williams <richardwilliams.ndu@gmail.com>]

Unfortunately gologit2 does not support factor variables. But, if you 
are careful you can use the -margins- command to get predicted 
probabilities. For example, the following gives you adjusted 
predictions for whites:

webuse nhanes2f,clear
gologit2 health age black orace
margins, at(black = 0 orace=0) predict(outcome(#1))
margins, at(black = 0 orace=0) predict(outcome(#2))
margins, at(black = 0 orace=0) predict(outcome(#3))
margins, at(black = 0 orace=0) predict(outcome(#4))
margins, at(black = 0 orace=0) predict(outcome(#5))

If you prefer to use means for age and other variables, you could add 
the -atmeans- option.

At 01:20 PM 4/19/2013, Tiainen Ossi wrote:
> Hello,
>
> I'm using the gologit2 generalized logistic regression/ partial 
> proportional odds model for ordinal dependent variables. The 
> dependent variable describes firm size and is a categorical variable 
> with three groups, small, medium and large. Independent variables 
> are three dummies that describe categorical firm types + some 
> control variables.
>
> I used margeff to get the average marginal effect coefficients (see 
> output below), but i'm having difficulties in understanding how I 
> could get the predicted probabilities for the three dummies in a 
> convenient way. For mlogit the 'margins' command gives the 
> probabilities directly, so is there some way to obtain them with 
> gologit2 or must I use the predict command and calculate the 
> probabilities by hand for each dummy?
>
> If I use predict, how can I get the base probability level that the 
> three coefficients below are compared to? What about if I use 
> marginal effects at mean instead of average marginal effects?
>
> Thank you for your assistance in advance!
>
> Best regards,
> Ossi Tiainen
>
>
> . margeff, dummies(Iprsv* \ Ipscl* \ Icountry* \ Inace*)
>
> Average partial effects after gologit2
>
> y  =3D Pr(size04)
> --------------------------------------------------------------------------------
>   variable  |      Coef.   Std. Err.      z    P>|z|     [95% Conf. 
> Interval=]
> ---------------+-----------------------------------------------------------------------+---=-----
> Small        |
> Iprsvinn_1 |  -.1368403   .0136474   -10.03   0.000    -.1635887   -.1100918
> Iprsvinn_2 |  -.0130166   .0185622    -0.70   0.483    -.0493979    .0233646
> Iprsvinn_3 |  -.0665578   .0193864    -3.43   0.001    -.1045544   -.0285611
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/

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

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/