Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Delete outcome in mlogit result

    Dear collegues !

    I'm currently working on a mlogit model. During a consumers survey, I asked to consumers to select their preferred labelling schemes among a pool of nine alternatives.
    Through my model, I search to determine which variables can influence the choice of one label rather than another.


    For the moment, my code line is :

    Code:
     mlogit rangtop i.recode_age i.genre revenumid revenusup deplittoral univ tradi prefgeo risqu_sanit effica_choix_conso connais_subj
    It displays the result compare to my base outcome, here the "quality" scheme, for the eight others alternatives.


    To test the IIA assumption, i would like to delete some outcomes in my mlogit model. For exemple, lets choose the "health" scheme. The variable rangtop contains the choice made by my consumers : Quality, health, nutrition, ecolabel, animal-welfare, local, france, EU, fair-trade. I tested the following code

    Code:
     mlogit rangtop i.recode_age i.genre revenumid revenusup deplittoral univ tradi prefgeo risqu_sanit effica_choix_conso connais_subj if rangtop !="Health":rangtop


    The problem is that it doesn't work. Results displayed still includes the health outcome ... So I wonder if you have any solution for me? I would like to perform an Hausman test right after.


    Thank you for your help, I hope my exemple i clear for you, and I remain available if necessary.

    Regards,

    Jean-François






  • #2
    The variable rangtop contains the choice made by my consumers : Quality, health, nutrition, ecolabel, animal-welfare, local, france, EU, fair-trade.
    If so, then why capitalize the category name?

    rangtop !="Health":rangtop

    This syntax is allowed

    Code:
    rangtop !="health":`:val lab rangtop'
    but it seems easier to search the value itself and enter it directly.

    Code:
    lab list `:var lab rangtop'

    Comment


    • #3
      Hello Andrew,

      Thank a lot for your help, it works !

      Best regards,

      Jean-François

      Comment

      Working...
      X