Author Topic: Question 2 part (g)  (Read 686 times)

yw213

  • Jr. Member
  • **
  • Posts: 3
    • View Profile
Question 2 part (g)
« on: February 21, 2017, 12:14:36 PM »
Hi,


I have some questions about the interpretation of coefficients. Do we need to include the measurements of every variable? If so, what's the measurement of lwage and exper2? Also, how would you put those variables into actual language instead of math terms? Thank you.


Roric Brown

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 43
    • View Profile
Re: Question 2 part (g)
« Reply #1 on: February 22, 2017, 06:39:26 PM »
For this question, yes, you should interpret, as best that you can, all of the coefficients.  I understand why you were confused by this question, but now that Professor Caetano covered Notes 10: Introducing Nonlinearities  in class today, the interpretations of lwage and exper2 should hopefully make more sense.
Pay particular attention to section 3 of notes 10, where it states "often, there is no economic interpretation at all..."

yw213

  • Jr. Member
  • **
  • Posts: 3
    • View Profile
Re: Question 2 part (g)
« Reply #2 on: February 23, 2017, 10:27:12 AM »
Yes, it makes more sense now. Thanks.


But I am still confused with exper and exper2. When we add exper2, would it affect the interpretation of beta_2 as well? Is exper still a liner variable? Or say it never was linear from the very beginning? If so, what's the difference between beta_2and beta_3 in terms of interpretation? Thank you.




Scott Onestak

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 40
    • View Profile
Re: Question 2 part (g)
« Reply #3 on: February 23, 2017, 07:18:10 PM »
Hey,


So it is difficult to tell you how to interpret variables without giving away the answer.  What I can say is that notes 10 should help and that other students have told me that Professor Caetano's explanation of why wages against experience looks like a quadratic curve from class helped them understand the coefficients of the exper variables.


To somewhat answer the other part of your question, a model that has a quadratic term - as this one does - is still a linear equation.  The regression is still linear in variables.  So, while the variable itself may have some sort of functional change - such as log or ^2, because the model is still linear in form, it is still a linear regression equation.