Announcement

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

  • gllamm, ... thresh() to model continuation-ratio logit model (seqlogit with random terms)

    Dear all,

    I am trying to fit a continuation-ratio logit model in a multi-level context. I have an outcome variable that takes the values 1 ... 5, and I am interested in modelling the sequential advancement from 1 to 5 as can be done with a sequential logit (i.e. Tutz & Mare models, as in Maarten Buis (SSC) command seqlogit).
    The data is nested, so I seek to estimate it in a multi-level context (two-level model) to account for statistical dependency.

    As far as I understand, gllamm offers the (only?) possibility to estimate a multi level continuation ratio logit model, and it is possible to not impose a parallel lines assumption by using the thresh() option.

    While I have found the description of what I am looking for in Rabe-Hesketh and Skrondal (2012, 3rd edition) and while in the exercises (Ch11) this use case of a continuation ratio model without parallel lines assumption is explicitely asked for, I cannot seem to find any worked example (neither in solutions package which accompanies the book nor on statalist).

    Does anyone have a worked example of gllamm with for a continuation ratio logit model with higher order random terms and without the parallel lines assumption?

    That'd be absolutely fantastic, I am now struggling quite a bit to get this right...

    Best
    Last edited by Johannes Muller; 17 Feb 2019, 05:43.

  • #2
    The issue is still pending, so any help would be highly appreciated.

    Comment

    Working...
    X