+ Reply to Thread
Results 1 to 2 of 2

Thread: Testing for equality of estimated coefficients between two models

  1. #1
    Points: 58, Level: 1
    Level completed: 16%, Points required for next Level: 42

    Posts
    3
    Thanks
    1
    Thanked 0 Times in 0 Posts

    Testing for equality of estimated coefficients between two models




    Good morning,

    I need an information as I'm estimating two models by OLS :

    - The first one estimates a standard linear model Y = a + B.X + epsilon
    This gives me the estimation of a coefficient (beta) for the variable X with a certain estimated standard deviation S0

    - The second one splits the variable X and the constant into a positive part and a negative part with the help of a dummy variable D, which is equal to 1 when X is positive and 0 when X is negative :
    Y = a_pos*D + a_neg*(1-D) + b_pos*D*X + b_neg*(1-D)*X + e

    Therefore this second model would check for a possible assymetry effect in beta.

    With it, I obtain estimates for the constants, and two coefficients beta_pos and beta_neg with standard deviations S1 and S2 ( and both are different to eachother and to S0)

    My problem here is, I would like to compare the two estimates I get from the second model to the one I got from the first model to check if beta_pos and beta_neg are together equivalent to the beta from the first model or if there's an assymetry effect.

    I suppose that I am supposed to use some sort of t (or F?)-test for mean comparison using the value of the estimated coefficients and standard deviations, but I'm not so sure how to proceed since I'm comparing 2 estimates to only one.

    Could someone tell me which test would check for this? Thanks for your help.

  2. #2
    Omega Contributor
    Points: 38,284, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    6,991
    Thanks
    397
    Thanked 1,185 Times in 1,146 Posts

    Re: Testing for equality of estimated coefficients between two models


    You can compare them. I don't run much linear regression, so I don't recall the actual calculations. Though you can use some type of test based on RSS values between models if I recall. Yeah, likely f-test related.


    Sorry I am not more helpful. If they were nested models you could use a log likelihood test.
    Stop cowardice, ban guns!

  3. The Following User Says Thank You to hlsmith For This Useful Post:

    naelwing (07-29-2017)

+ Reply to Thread

           




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats