Hi,
I am currently working on a CE paper and I am trying to run a mixed logit model on stata 17 using mixlogit and mixlogitwtp. My experiment is structured as shown below. The respondents can either choose option 1, option 2 or none of the two alternatives
The model using mixlogit includes a constant representing the opt-out option and all other attributes in the model are assumed to have random coefficients and normal distribution. The stata command I have used is as shown:
mixlogit choosen asc, group(csid) id( sys_respnum) rand( medium stor middels_rød mørk_rød faste milde søte Norsk pris2) nrep(500).
Afterwards I run the wtp module to find the wtp estimates in the preference space. The figures that I get are outrageous as they are way higher than the price levels in the CE.
I assume that the wtp estimates above represent the marginal WTP. I would also love to estimate the mean WTP which I understand is possible through simulation but I do not know how to go about it or which stata commands to use.
Secondly I am also trying to estimate a second mixed logit model using wtp space using mixlogitwtp. The stata command I have used is as follows
. mixlogitwtp choosen asc, group(csid) id( sys_respnum) price(mprice2) rand( medium stor middels_rød mørk_rød faste milde søte Norsk) nrep(500)
In this case the model does not converge even after leaving it overnight.
I tried removing the opt-out constant and using 10 replications and in this case the model converged but I don’t think 10 replications are enough.
How can I estimate the mean wtp after mixlogitwtp and do the mean figures I obtain represent the marginal wtp estimates?
Kindly assist me.
Best Regards
Sarah Muiruri
I am currently working on a CE paper and I am trying to run a mixed logit model on stata 17 using mixlogit and mixlogitwtp. My experiment is structured as shown below. The respondents can either choose option 1, option 2 or none of the two alternatives
The model using mixlogit includes a constant representing the opt-out option and all other attributes in the model are assumed to have random coefficients and normal distribution. The stata command I have used is as shown:
mixlogit choosen asc, group(csid) id( sys_respnum) rand( medium stor middels_rød mørk_rød faste milde søte Norsk pris2) nrep(500).
Afterwards I run the wtp module to find the wtp estimates in the preference space. The figures that I get are outrageous as they are way higher than the price levels in the CE.
I assume that the wtp estimates above represent the marginal WTP. I would also love to estimate the mean WTP which I understand is possible through simulation but I do not know how to go about it or which stata commands to use.
Secondly I am also trying to estimate a second mixed logit model using wtp space using mixlogitwtp. The stata command I have used is as follows
. mixlogitwtp choosen asc, group(csid) id( sys_respnum) price(mprice2) rand( medium stor middels_rød mørk_rød faste milde søte Norsk) nrep(500)
In this case the model does not converge even after leaving it overnight.
I tried removing the opt-out constant and using 10 replications and in this case the model converged but I don’t think 10 replications are enough.
How can I estimate the mean wtp after mixlogitwtp and do the mean figures I obtain represent the marginal wtp estimates?
Kindly assist me.
Best Regards
Sarah Muiruri
