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

  • Interpretation of xtologit regression

    Dear all,

    I estimated a xtolgit regression using a unbalanced panel. My dependend variable is a cathegorical variable (labor) indicating the labor force status of individuals (4= full employed; 3=part-time employed; 2=mini job; 1=not employed). I want to investigate the effect of getting a heart diseases (heart=1 if yes; 0 otherwise) on the labor force participation.

    My command is:

    xtologit labor heart age agesq ses ever_married educ diabetes high_blood_pres stroke depression i.year, vce (cluster pid)
    margins, dydx(heart) predict(outcome(4))
    margins, dydx(heart) predict(outcome(3))
    margins, dydx(heart) predict(outcome(2))
    margins, dydx(heart) predict(outcome(1))

    For the first margins command (outcome(4)) I get a coeficient of: -.0609702.
    For the second margins command (outcome(3)) I get a coeficient of : .0109574
    For the third margins command (outcome(2)) I get a coeficient of: .0070751
    For the fourth margins command (outcome(2)) I get a coeficient of: .0429377.

    For all p=0.000.

    I would interprete my results as following: Getting a heart disease reduces the probability to be full time employed (outcome 4) on average by 6.09702 % points ceteris paribus.
    From all people who stop working full-time 17.97 % (0.0109574 / 0.0609702) start working part-time time, 11.60 % (0.0070751 / 0.0609702) start working in a mini job and 70.42 % (0.0429377 / 0.0609702.) stop working completely.

    Is this the right way to interprete my model?

    I thank you in advance!