I have a panel data for discrete choice experiment with repeated measures of the choice-sets at two time points. I’m interested to look at how the preference changes over time. Before using cmxtmixlogit, I have to cmset the data.
My question is, what does a command like "cmset" do to the time variable? For example, is declaring the time variable equal to accounting for time variable as a random effect?
I’m asking this, because to examine whether the preference changes over time, I am also including interaction terms of independent variables with time (random coefficients) in the model. Stata is able to run the model and give findings, but I wonder whether the way I have handled the "time" variable is correct statistically.
My question is, what does a command like "cmset" do to the time variable? For example, is declaring the time variable equal to accounting for time variable as a random effect?
I’m asking this, because to examine whether the preference changes over time, I am also including interaction terms of independent variables with time (random coefficients) in the model. Stata is able to run the model and give findings, but I wonder whether the way I have handled the "time" variable is correct statistically.
