Hi Silvia and Stephan,
Using -clogit- or -nlogit- excluding alternatives is easy. Since the
data is structured so that you have one row for each alternative per
person, you can simply drop the observation containing the alternative
you want to remove from the individual's choice set.
If you have a dummy indicating that the person does not have a
driver's licence ("no_licence") and a variable identifying the
different modes ("mode") you could do something like:
drop if mode == "car" & no_licence == 1
Make sure you save a copy of the original dataset though!
Hope this helps.
Arne
On 22/03/07, Stephan Brunow <[email protected]> wrote:
Dear all,
we want to estimate a multinomial logit model for transport mode choice.
There are some restrictions on the availability of alternatives. E.g. if one
does not own a driver licence the person can not choose car. Therefore we
would like to exclude the car alternative for that individual.
Alternatively, we might use clogit to estimate the model.
Is there some way to exclude alternatives for individuals? Finally, we face
a similar problem in nlogit.
The second question concerns interpretation but we would be glad if someone
might answer the first one. Assuming car is the reference category, how
should one interpret the coefficients for a person without driver license,
for instance.
Many thanks,
Silvia and Stephan
*
* 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/
*
* 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/
