| |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: Obtaining Predicted Probabilities using gologit
At 11:46 AM 6/13/2006, Lisa Powell wrote:
HI,
I am running a gologit model and want to assess the magnitude of the
association between one of my independent variables and my outcome
variable using the predict option similar to a probit/logit
model. I specifically want to set my independent variable to a
given value to test varying scenarios for this association. Below
is the syntax I used after my gologit model statement:
...
When I ran this I received the following error statement:
predict xb1, equation(mleq1) xb at [promo=1];
option at not allowed
r(198);
I've never seen the "at" option on predict before; it would probably
work if you change it to
if promo==1
But, If you use gologit2 (which you can download from SSC) you can
save yourself about 15 steps here. gologit2's predict command can
compute the probabilities directly, e.g.
gologit2 y promo
predict prob1 prob2 prob3 prob4 prob5 if promo==1
This assumes there are cases where promo really does equal 1.
Here is a more complicated trick.
clonevar newpromo = promo
gologit2 y newpromo x2 x3 x4 x5
replace newpromo = 1
predict prob1 prob2 prob3 prob4 prob5 if e(sample)
This is useful if, say, you want to see what the predicted value for
each case would be if promo equaled 1 instead of whatever it does equal.
There are lots of other ways of achieving similar results.
-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
FAX: (574)288-4373
HOME: (574)289-5227
EMAIL: [email protected]
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/