Announcement

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

  • Oaxaca-Blinder with multiple groups

    Dear all,
    Now I am doing a research on how education years affect one's annual income. According to my data, I have to divide samples into three different groups. Is it possible to use Oaxaca-Blinder method to deal with multiple groups? Or are there launched commands already? Thanks for all.

  • #2
    Neither
    decompoSition can only be done comparing two groups at a time
    perhaps you can choose a base group and compare that one to the other two
    hth

    Comment


    • #3
      Originally posted by FernandoRios View Post
      Neither
      decompoSition can only be done comparing two groups at a time
      perhaps you can choose a base group and compare that one to the other two
      hth
      Ok. I get it. Thanks!

      Comment


      • #4
        Originally posted by FernandoRios View Post
        Neither
        decompoSition can only be done comparing two groups at a time
        perhaps you can choose a base group and compare that one to the other two
        hth
        Dear Professor,
        Now I am facing another annoying problem. Educaton, gender and race can all affect one's annual income in an invisible way. If I just want the 'pure treatment effect' of education, how can I handle with other two factors in the same time? Thank you very much for help.

        Comment


        • #5
          Well
          Just adding them as controls would allow you to identify the Education treatment effect, conditional on the other s being fixed (as in Linear Regression).
          However, you may also use reweigthing for the same purpose (although that is a bit of an advanced approach.
          HTH
          F

          Comment


          • #6
            Originally posted by FernandoRios View Post
            Well
            Just adding them as controls would allow you to identify the Education treatment effect, conditional on the other s being fixed (as in Linear Regression).
            However, you may also use reweigthing for the same purpose (although that is a bit of an advanced approach.
            HTH
            F
            I really appreciate your help, professor. Now I am quite clear.

            Comment


            • #7
              Originally posted by FernandoRios View Post
              Well
              Just adding them as controls would allow you to identify the Education treatment effect, conditional on the other s being fixed (as in Linear Regression).
              However, you may also use reweigthing for the same purpose (although that is a bit of an advanced approach.
              HTH
              F
              Dear Professor,
              Now I am working with a panel data to observe whether teacher-school communication can help improve children's academic performance.
              When using xtoaxaca, I was bothered by annoying categorical variable.

              This is my command:
              quietly eststo est1: xtreg stdchn stdcog chnshadow year $xlist ecolevel talkstudy, cluster(ids)
              xtoaxaca stdcog chnshadow year $xlist ecolevel talkstudy, groupvar(talkstudy) groupcat(0 1) normalize(ecolevel) timevar(year) times(2014 2015 2016) timeref(2014) timebandwidth(1) model(est1) change(interventionist) forcesample

              where ecolevel is a 5-level variable indicating a family's economic level.

              After running, it shows:
              ecolevel is not a categorical factor variable. Only categorical factor variables may be specified in normalize().

              So where might be the problem? Are there any differences to add categorical factor among command oaxaca, oaxaca_rif and xtoaxaca?

              Thanks for help.

              Li
              Last edited by Li Tianhao; 30 Jun 2023, 02:53.

              Comment

              Working...
              X