Hi!

I am doing a longitudinal analysis on the effects of firm contexts/practices on employees' relationships.
The data contains four waves of data from 2010 to 2016, taken every two years.

My dependent variable is a binary variable on whether an employee has a mentor within his/her organization; independent variables include both context- and individual-level variables related to organizations:

xtlogit mentor x1 x2 x3 ... i.year, re vce(r)


Then, I want to do another analysis on whether employees experienced any "change" in mentorship; that is, whether his/her relationships with their mentors changed over the years (e.g. lost a mentor; gained a mentor; no change).

So the new model would be

xtlogit change_mentor x1 x2 x3 ... i.year, re vce(r)


My question is, is it possible to create a dependent variable measuring the change? Also, is it theoretically and empirically feasible to test such model via random effects model?

Thank you so much in advance for your inputs!