Announcement

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

  • Fractional Regression

    Hello,

    I have to perform a regression and my dependent variable is a probability. My two independent variables are dummies.
    I have been advised to perform a fracreg since the linear model may not be appropriate. However, I am still unsure whether I should use fracreg logit or fracreg probit, since my independent variables are dummies.
    Then, am I right to use the "margins , dydx" command to interpret the results? Can I say for example, if the coefficient is -0.41, that the probability will be 0.41 percentage points lower for the observation where the dummy =1 ?

    Thank you very much for your help,

    Viktor Click image for larger version

Name:	Screenshot 2019-05-09 at 09.03.36.png
Views:	1
Size:	91.4 KB
ID:	1497460 Click image for larger version

Name:	Screenshot 2019-05-09 at 09.03.23.png
Views:	1
Size:	128.9 KB
ID:	1497461
    Last edited by Viktor Pustwo; 09 May 2019, 01:23.

  • #2
    Dear Viktor Pustwo,

    Since all your regressors are dummies, it may be OK to use a linear model, especially if you also include the interaction between the dummies.

    Best wishes,

    Joao

    Comment


    • #3
      Dear ,

      Thank you very much for your reply.
      I am trying to look at if jobs requiring or college degree or not, and prior work experience or not will result in a lower probability of automation.
      Given the output attached, do you believe I can use a linear model to support my idea?

      Thank you again for your help,

      Best wishes,

      Viktor
      Click image for larger version

Name:	Screenshot 2019-05-09 at 13.21.36.png
Views:	1
Size:	114.5 KB
ID:	1497487

      Comment


      • #4
        Yes, I think so.

        Joao

        Comment


        • #5
          Thank you very much for taking the time to help me

          Comment


          • #6
            Getting back to the earlier question, it usually matters little whether you use logit or probit. It often just depends on what the standard practice is in your field.
            -------------------------------------------
            Richard Williams, Notre Dame Dept of Sociology
            Stata Version: 17.0 MP (2 processor)

            EMAIL: [email protected]
            WWW: https://www3.nd.edu/~rwilliam

            Comment


            • #7
              Vikton,

              While you got excellent advice this time, since you're new, let me note you'll increase your chances of a useful answer in future by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output and sample data using dataex. Pictures of output are not encouraged.

              Phil

              Comment

              Working...
              X