Re: st: Estimating MNL model

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

At 03:57 PM 10/11/2013, Nabin Kafle wrote:
> I have a data set which has four choices, people choosing between car,
> rail, auto and air. The choice of an alternate is affected by the
> availability of the alternate (if air, rail, bus and car is available
> to them or not), travel time and travel cost and age of the person. I
> need to estimate the utility function for a person for choosing each
> alternate. I could do it using biogeme but I am not sure how to do it
> in stata.

Just to follow up, the help for asclogit (which came out oafter Long 
& Freese's book) says

"asclogit fits McFadden's choice model, which is a specific case of 
the more general conditional logistic regression model (McFadden 
1974). asclogit requires multiple observations for each case 
(individual or decision), where each observation represents an 
alternative that may be chosen. The cases are identified by the 
variable specified in the case() option, whereas the alternatives are 
identified by the variable specified in the alternatives() option. 
The outcome or chosen alternative is identified by a value of 1 in 
depvar, whereas zeros indicate the alternatives that were not chosen. 
There can be multiple alternatives chosen for each case. asclogit 
allows two types of independent variables: alternative-specific 
variables and case-specific variables. Alternative-specific variables 
vary across both cases and alternatives and are specified in 
indepvars. Case-specific variables vary only across cases and are 
specified in the casevars() option."

This sounds to me like what you want (unless you need the more 
general clogit) but I have never run such a model so I can't swear to it.


> The utility function would be
> Vi,bus= constant+ B1*time+ B2*cost (assuming this is the base choice)
> Vi, car= constant + B3*time + B4*cost + B5* age
> Vi, air = constant + B6* time + B7* cost + B8*age
> Vi, rail = constant + B9* time + B10* cost + B11* age
>
> On Fri, Oct 11, 2013 at 4:15 PM, Richard Williams
> <richardwilliams.ndu@gmail.com> wrote:
> > Like Maarten, I am not totally sure what you want. But, Long and Freese do
> > have an entire chapter on "Models for nominal outcomes with alternative
> > specific data." Among other things they talk about the clogit command. They
> > also describe various utilities for setting the data up and 
> interpreting the
> > results.
> >
> > For more, see
> >
> > 
> http://www.stata.com/bookstore/regression-models-categorical-dependent-variables/
> >
> > See especially chapter 7 and maybe chapter 6. My best guess is that you
> > especially want
> >
> > 7.2.5 Using clogit with case- and alternative-specific variables
> >
> >
> > If you type -findit spost9- you can get the programs used in the book.
> >
> > If this isn't the sort of thing you are looking for, then please provide
> > additional details.
> >
> > As a sidelight I have read various descriptions of the clogit command --
> > Long & Freese's, and using clogit for fixed effects models -- and 
> it sort of
> > amazes me that they are talking about the same command, as the uses seem
> > radically different.
> >
> >
> > At 11:23 AM 10/11/2013, Nabin Kafle wrote:
> >>
> >> How can I estimate MNL moedl in stata with alternate specific and
> >> generic variables. What would be the command to describe them and run
> >> the model. Thanks in advance
> >>

-------------------------------------------
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/