Dear collegues,
I am currently working on a consumer survey. In this survey, we asked our consumers to rank their 3 favourite labels out of 9 alternatives (and an opt-out option).
Lets see an exemple :
For exemple, our first consumer ranked at rank1 the heatlh allegation, at rank2 the ecolabel and at rank3 the EU origin scheme.
My problem :
Now, i want to implement a mix logit model with panel data (cmxtmixlogit) to study the preferences expressed according to several explanatory variable (age, gender ...)
Before doing so, I would like to give weight to the ranks granted. So, for example, rank 1 would weight 3, rank 2 the weight of 2 and rank 1 the weight of 1. Do you know if its possible ?
For the moment, when i realise my cmxtmixlogit, all ranks are considered equal (which is not the case in reality). I tried to use the cmrologit, but unfortunately, it does not work on my computer (problem of computing power.). I would therefore like to overcome this problem by weighting my answers by rank before realising my mixed logit for panel data.
I don't know if this is possible, considering that each indivduals ranked 3 schemes.
I remain available for further information if necessary,
Regards,
Jean-François DEWALS
I am currently working on a consumer survey. In this survey, we asked our consumers to rank their 3 favourite labels out of 9 alternatives (and an opt-out option).
Lets see an exemple :
Code:
list record rang1 rang2 rang3
For exemple, our first consumer ranked at rank1 the heatlh allegation, at rank2 the ecolabel and at rank3 the EU origin scheme.
My problem :
Now, i want to implement a mix logit model with panel data (cmxtmixlogit) to study the preferences expressed according to several explanatory variable (age, gender ...)
Before doing so, I would like to give weight to the ranks granted. So, for example, rank 1 would weight 3, rank 2 the weight of 2 and rank 1 the weight of 1. Do you know if its possible ?
For the moment, when i realise my cmxtmixlogit, all ranks are considered equal (which is not the case in reality). I tried to use the cmrologit, but unfortunately, it does not work on my computer (problem of computing power.). I would therefore like to overcome this problem by weighting my answers by rank before realising my mixed logit for panel data.
I don't know if this is possible, considering that each indivduals ranked 3 schemes.
I remain available for further information if necessary,
Regards,
Jean-François DEWALS
