Hi, I am modelling dependent variable y which is a destination choice of residents which has 4 levels: URBAN, SUBURBS, SMALL CITY or RURAL. I use mlogit and run the exact same regression for period A and period B. For independent variable x3 the coefficient is positive and statistically significant for period A but negative and statistically significant in period B.
I would like to say something about how the coefficient for x3 differers between period A and period B but I sense I am running into the Allison (1999) territory with problem of group comparisons for logistic models. I found Williams (2009) paper which discusses a strategy for group comparisons for logistic models and then Stata journal article which shows how this can be implemented with with oglm command but I cannot see how this oglm could be use for multinomial logistic models. Has anyone found a solution for this problem?
References:
Allison, P. D. (1999). Comparing logit and probit coefficients across groups. Sociological methods & research, 28(2), 186-208.
Williams, R. (2009). Using heterogeneous choice models to compare logit and probit coefficients across groups. Sociological Methods & Research, 37(4), 531-559.
Williams, R. (2010). Fitting heterogeneous choice models with oglm. The Stata Journal, 10(4), 540-567.
Code:
mlogit y x1 x2 x3 if period=="A" mlogit y x1 x2 x3 if period=="B"
References:
Allison, P. D. (1999). Comparing logit and probit coefficients across groups. Sociological methods & research, 28(2), 186-208.
Williams, R. (2009). Using heterogeneous choice models to compare logit and probit coefficients across groups. Sociological Methods & Research, 37(4), 531-559.
Williams, R. (2010). Fitting heterogeneous choice models with oglm. The Stata Journal, 10(4), 540-567.
