Re: st: Predicted Odds Following Logistic Regression

[Melissa King <melissakng@gmail.com>]

Hi Maarten --

Thank you for the very helpful advice on using adjust. I originally
thought that holding constant at the mean was the default for adjust,
so now I know to add the additional commands. Interestingly, three of
the medical services in my model (SERVICE1, SERVICE2 and SERVICE3) are
mutually exclusive, meaning that if you receive one of the three, you
cannot be receiving the other two. Do you know if Stata will recognize
this when I ask it to generate predicted odds? In other words, will it
automatically "switch off" SERVICE2 and SERVICE3 in my logistic
regression when SERVICE1 is "switched on?" And if I generate predicted
odds for SERVICE1, and indicate that I want values held constant at
the mean, will Stata fix SERVICE2 and SERVICE3 at zero while holding
all other variables constant at their means?

I ask this because it would not make intuitive sense to look at how
enrolling in SERVICE1 impacts a person's odds of hospital admission
while holding SERVICE2 and SERVICE3 at the mean for all youth in my
sample, since if a person is enrolled in SERVICE1, we know he or she
is not enrolled in SERVICE2 or SERVICE3.

Thank you again for your time and assistance, Maarten!

All best,

Melissa
*
*   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/