 |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: -graph twoway (function ...)- question
Hi,
Yes, you would need to use the -save- option to save the dataset of
predicted probabilities for each group, then manually combine them and
create your graph from there. Automation of that sort of process is
theoretically on my list to add to -oprobpr-, but don't hold your breath
for it to happen any time soon.
The help file example is a mistake--the adj() option was renamed to
levels() along the way.
- NW
Andrea Bennett wrote:
Thanks for this hint, I didn't know your app!
But, after having played around with it, I fear it is not possible to
have multiple probability lines in one graph? E.g. I would want to show
the change in the probability but for each of the three age categories
separately (but in one graph). Is this possible? Would I need to deal
with the -save- option? And last, it seem not possible to display
confidence intervals, right?
When I would like to display changes in probabilities, what else is
available as an option. I know the prvalue/praccum command which I will
try next. Any good input is welcomed, though!
Kind regards,
Andrea
P.S. In your help file, in the "examples" section, there is a line
"oprobpr mpg, adj(weight=2500, foreign=0)" while the option -adj- is not
discussed (and does not work).
On Jul 24, 2008, at 3:37 PM, Nick Winter wrote:
I don't have an answer to your specific question, but another way to
go is to plot predicted probabilities. My -oprobpr- package might
help for this. (-oprobpr- only works after -oprobit- not -probit-,
but oprobit will estimate the same model as probit if there are only
two response categories; just use the -categories()- option to specify
which response category you want plotted.
-Nick Winter
Andrea Bennett wrote:
Dear all,
I have estimated average marginal effects from a probit regression
with -margeff- and would like to generate a graph which plots the
marginal effects for 3 age categories (dummies) dependent on a
continuous variable called cont.
The model is as such:
y = b0 + b1*age2 + b2*age3 + b3*age2*cont +b4*age3*cont + b5*cont +
controls
As age2 and age3 are dummies, would the following graph command be
correct:
twoway (function age1 = _b[cont]*x, range(0 0.5)) (function age2 =
_b[age2] + (_b[cont] + _b[cont_age2])*x, range(0 0.5)) (function age3
= _b[age3] + (_b[cont] + _b[cont_age3])*x, range(0 0.5))
The way I understand it, since I have now marginal effects I can work
equivalently to linear models. Then the above graph should be
correct, right? Further, this would also be correct when introducing
other interactions with -cont- as long as I am looking at the
marginal effects of age groups when -cont- changes?
Many thanks for your considerations,
Andrea
*
* 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/
--
--------------------------------------------------------------
Nicholas Winter 434.924.6994 t
Assistant Professor 434.924.3359 f
Department of Politics [email protected] e
University of Virginia faculty.virginia.edu/nwinter w
PO Box 400787, 100 Cabell Hall
Charlottesville, VA 22904
*
* 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/
*
* 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/
--
--------------------------------------------------------------
Nicholas Winter 434.924.6994 t
Assistant Professor 434.924.3359 f
Department of Politics [email protected] e
University of Virginia faculty.virginia.edu/nwinter w
PO Box 400787, 100 Cabell Hall
Charlottesville, VA 22904
*
* 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/