From doctorjasonong at gmail.com Fri Jan 15 10:58:48 2021 From: doctorjasonong at gmail.com (Jason Ong) Date: Fri, 15 Jan 2021 10:58:48 +1100 Subject: [Limdep Nlogit List] simulation command in NLOGIT Message-ID: hi, How does the ;simulation command in NLOGIT estimate the "market shares" (I couldn't find the details in the Applied Choice Analysis text book) is it possible to also do a manual calculation using beta coefficients from an RPL model? For example, I have an unlabelled DCE with 3 alternative Alternatives A and B are unlabelled and Alternative C is an opt-out option I ran an RPL model and used the simulation command in NLOGIT to see the change in % of people choosing the opt-out option (alternative C) when I altered one attribute level at a time This worked for all my attributes in the model except for one (it returned unusually high values which did not correspond with the patterns I am seeing from the beta coefficients) so I was wanting to see if it is possible to use a manual calculation to check the output from the simulation? thanks for your help. Let me know if I need to explain further Best, *Jason Ong* From wgreene at stern.nyu.edu Sun Jan 17 05:48:04 2021 From: wgreene at stern.nyu.edu (William Greene) Date: Sat, 16 Jan 2021 13:48:04 -0500 Subject: [Limdep Nlogit List] simulation command in NLOGIT In-Reply-To: References: Message-ID: The market shares are computed by summing the predicted probabilities and dividing the sums by the number of choice tasks used to compute the sums. For an RPL model, the probabilities are computed using the simulation on which the model is based. It would probably be much too complicated to replicate this with CREATE. That is why the ;Simulate option is provided. /B. Greene On Thu, Jan 14, 2021 at 6:59 PM Jason Ong via Limdep < limdep at mailman.sydney.edu.au> wrote: > hi, > > How does the ;simulation command in NLOGIT estimate the "market shares" (I > couldn't find the details in the Applied Choice Analysis text book) > is it possible to also do a manual calculation using beta coefficients from > an RPL model? > > For example, I have an unlabelled DCE with 3 alternative > Alternatives A and B are unlabelled and Alternative C is an opt-out option > I ran an RPL model and used the simulation command in NLOGIT to see the > change in % of people choosing the opt-out option (alternative C) when I > altered one attribute level at a time > This worked for all my attributes in the model except for one (it returned > unusually high values which did not correspond with the patterns I am > seeing from the beta coefficients) > so I was wanting to see if it is possible to use a manual calculation to > check the output from the simulation? > thanks for your help. > Let me know if I need to explain further > > Best, > > > *Jason Ong* > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > https://protect-au.mimecast.com/s/F1oMCK1DvKTGnvvWTMp-pA?domain=limdep.itls.usyd.edu.au > > -- William Greene Department of Economics, emeritus Stern School of Business, New York University 44 West 4 St. New York, NY, 10012 URL: https://protect-au.mimecast.com/s/RFZVCL7EwMfZ1pp9cqg5oc?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Editor in Chief: Journal of Productivity Analysis Editor in Chief: Foundations and Trends in Econometrics Associate Editor: Economics Letters Associate Editor: Journal of Business and Economic Statistics From rturner at colgate.edu Wed Jan 27 04:46:34 2021 From: rturner at colgate.edu (Bob Turner) Date: Tue, 26 Jan 2021 12:46:34 -0500 Subject: [Limdep Nlogit List] rank order random parameter models with varying means Message-ID: Is it impossible to use ranked data with random parameter models with mean-varying parameters? I don't see anything in the manual either way. When I tried to use ;ranks with gmxlogit, the program seemed to ignore the variables in the ;gmx= line. The same thing happens when I use rplogit: if I add ;ranks then the variable in the ;rpl line is ignored. ------------------------- Robert W. Turner Professor of Economics and Environmental Studies Associate Editor, *Journal of Environmental Studies and Sciences* 11 Spear House Colgate University 13 Oak Drive Hamilton, NY 13346 (315)228-7529 From wgreene at stern.nyu.edu Wed Jan 27 06:02:09 2021 From: wgreene at stern.nyu.edu (William Greene) Date: Tue, 26 Jan 2021 14:02:09 -0500 Subject: [Limdep Nlogit List] rank order random parameter models with varying means In-Reply-To: References: Message-ID: Dear Robert. The rank data MNL estimator in NLOGIT is not extended to random parameters or to the GMXL form of the model. It is only supported in the basic MNL form. The ;Ranks switch preempts the RP forms. Regards B. Greene On Tue, Jan 26, 2021 at 12:47 PM Bob Turner via Limdep < limdep at mailman.sydney.edu.au> wrote: > Is it impossible to use ranked data with random parameter models with > mean-varying parameters? I don't see anything in the manual either way. > When I tried to use ;ranks with gmxlogit, the program seemed to ignore the > variables in the ;gmx= line. The same thing happens when I use rplogit: if > I add ;ranks then the variable in the ;rpl line is ignored. > > ------------------------- > Robert W. Turner > Professor of Economics and Environmental Studies > Associate Editor, *Journal of Environmental Studies and Sciences* > 11 Spear House > Colgate University > 13 Oak Drive > Hamilton, NY 13346 > (315)228-7529 > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > https://protect-au.mimecast.com/s/e073Cq71mwf53lEQIZe7bZ?domain=limdep.itls.usyd.edu.au > > -- William Greene Department of Economics, emeritus Stern School of Business, New York University 44 West 4 St. New York, NY, 10012 URL: https://protect-au.mimecast.com/s/ir4fCr81nytvYokqczVxkn?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Editor in Chief: Journal of Productivity Analysis Editor in Chief: Foundations and Trends in Econometrics Associate Editor: Economics Letters Associate Editor: Journal of Business and Economic Statistics