How I should calculate pseudoelasticity in generalized ordered logit model in Stata?

[Pedram Akbari]

Hi,
I have developed a generalized ordered logit model and I want to evaluate the effect of variables. I have two questions.
1) What is the best way to calculate the effect of variables in this kind of models? I should say the variables in the model are both binary, ordinal and scale. My research is about the frequency of transit use and explanatory variables are education, kind of city, waiting time in station, age, the purpose of trips, job, weekday/ weekend, and vehicle ownership.
2) I have seen 2 papers that have calculated pseudoelasticity for the generalized ordered logit model, however, I was not able to find any code about pseudoelasticity in Stata. I would appreciate any help with how I should evaluate the pseudoelasticity of variables in Stata?

[ericmelse]

Dear Pedram,

I assume that you are familiar with this UCLA web page which is good place to start when you are involved with ordered logit models in Stata.

There is a substantial literature on the subject. I recommend the following for both it's practical application and because it facilitates using Stata for your purpose:

Williams, R. (2016). Understanding and interpreting generalized ordered logit models. The Journal of Mathematical Sociology, 40(1), 7-20.

Williams, R. (2006). Generalized Ordered Logit/Partial Proportional Odds Models for Ordinal Dependent Variables. The Stata Journal, 6(1), 58–82.

Williams, R. (2009). Using Heterogeneous Choice Models to Compare Logit and Probit Coefficients Across Groups. Sociological Methods & Research, 37(4), 531–559.

Fullerton, A. S. (2009). A Conceptual Framework for Ordered Logistic Regression Models. Sociological Methods & Research, 38(2), 306–347.

I recommend Richard Williams' user community contributed package gologit2:

ssc install gologit2
h gologit2
* for ANCILLARY FILES, use
findit gologit2

more can be found on his website.

[Pedram Akbari]

Thank you sir for your comprehensive answer, but:
1) Still I have not fully understood. If I want to use pseudoelasticity and evaluate that for all variables, I should write [margine, eyex(*)] after developing results for the model?
2) When I use this code, values of pseudoelasticity for some variables are insignificant. I am confused now, how is it possible that a coefficient is significant but its elasticity is not. what is the interpretation of that? Attatched please find the picture in which a variable's pseudoelasticity is not significant.