Ranking preferences and modelling - Question(s) & Advice
Dear colleagues,
I'm currently working on the results of a consumer survey. In this survey we ask consumers to rank their first, second and third preferred labels from a pool of 5 alternatives.
I would then like to investigate the determinants of consumer preference for these five labelling schemes.
For the moment, I have run a mlogit model. Unfortunately, this model only takes into account the first ranking assigned by consumers in my analysis.
The second step is now to take into account the ranking of our consumers, to test the influence of this ranking on the expressed preferences. However, i'm still confused about the right model to use...
For the moment, i have identified the rologit model. However, it seems that this model is more about studying the influence of the alternative-specific variables on the ranking. I'm my case, i don't have alternative-specific variable. I have only case-specific variables, related to consumers characteristics (age, gender etc.).
I'm really looking for information on how to implement this model in relation to my objectives. At the moment my code is structured as follows:
The "rank" variable identifies the rank assigned by consumers for one alternative. The other variables are related to our consumers' characteristics.
My problem is that in this code we don't take into account the alternative ranked. Our dependant variable is just related to the rank assigned. Introducing an explanatory variable related to the choice of alternative seems rather limited.
If you have any advice for the development of my model, I'm interested!
Best regards,
Jean-François DEWALS
I'm currently working on the results of a consumer survey. In this survey we ask consumers to rank their first, second and third preferred labels from a pool of 5 alternatives.
I would then like to investigate the determinants of consumer preference for these five labelling schemes.
For the moment, I have run a mlogit model. Unfortunately, this model only takes into account the first ranking assigned by consumers in my analysis.
Code:
mlogit label1 Gen pref_frais pref_env_sauv pref_env_elev pref_france pref_sanit pref_indus pref_sauv, base(0) robust /*Only pref*/ estat ic margins, dydx(*)
For the moment, i have identified the rologit model. However, it seems that this model is more about studying the influence of the alternative-specific variables on the ranking. I'm my case, i don't have alternative-specific variable. I have only case-specific variables, related to consumers characteristics (age, gender etc.).
I'm really looking for information on how to implement this model in relation to my objectives. At the moment my code is structured as follows:
Code:
rologit Rank Gen pref_frais pref_env_sauv pref_env_elev pref_france pref_sanit pref_indus pref_sauv, reverse group (id)
My problem is that in this code we don't take into account the alternative ranked. Our dependant variable is just related to the rank assigned. Introducing an explanatory variable related to the choice of alternative seems rather limited.
If you have any advice for the development of my model, I'm interested!
Best regards,
Jean-François DEWALS
