Announcement

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

  • Exclude constant in Stata, Generalized Ordered Logistic Regression

    Hi,

    I don't know how i cun run an ordered logistic regression or a generalized ordered logistic regression in stata that the resultst, don't show me a constant in the model. It is always included... How can I get an estimation without this constant?

  • #2
    You can use the "noconstant" option at the end:

    Code:
    logit y x, nocons
    PS: Please note the preference for real names here. Just use the "contact us" tab and request admin for a name change.
    Last edited by Roman Mostazir; 01 Feb 2015, 05:20. Reason: Name change reminder
    Roman

    Comment


    • #3
      Roman advice holds for logit and mlogit but for ordered logit I think it is impossible to omit the constant. The only way to make it possible, and I didn't try to do that, is to use constraints and set the constant to zero, for example.

      Comment


      • #4
        For an ordered logit no constants are reported and they cannot exist, so there is nothing to drop. You can think of the cut values as constants. In fact, that is the reason why there cannot be a constant in the main equation: they already exist in the form of cut-values. You definately do not want to ommit those, as they are crucial to the logic of an ordered model. A generalized order logit (as implemented by Rich Williams in gologit2, available from SSC) uses a different but equivalent parameterization with constants and no cut-values. You could use the constraint command to ommit the constants, but you should not do so.
        ---------------------------------
        Maarten L. Buis
        University of Konstanz
        Department of history and sociology
        box 40
        78457 Konstanz
        Germany
        http://www.maartenbuis.nl
        ---------------------------------

        Comment


        • #5
          My apology for considering a logit example. A releavant example here: http://www.stata.com/support/faqs/st...probit-models/
          Roman

          Comment

          Working...
          X