+ Reply to Thread
Results 1 to 2 of 2

Thread: testing regressions

  1. #1
    Points: 9,676, Level: 66
    Level completed: 7%, Points required for next Level: 374

    Posts
    26
    Thanks
    0
    Thanked 0 Times in 0 Posts

    testing regressions




    Hi,

    I have a very simple question:
    when I run a regression what are the things that I should test for to make sure that my regression is solid. Someone asked me whether there is co-integration in my regressions and was upset when i mentioned that i didn't test for them. I hope a practitioner can tell me :
    1-multicollinearity,
    2-Heteroskedasticity , etc... without referring me to a textbook. I will figure out how to test for each by referring to textbooks but simply I want to fix my ideas and adopt the diagnostics that you would suggest as long as I run regressions.
    Regards

  2. #2
    Points: 2,228, Level: 28
    Level completed: 52%, Points required for next Level: 72

    Posts
    65
    Thanks
    0
    Thanked 1 Time in 1 Post

    Re: testing regressions


    Well, you should test for all the things a textbook would mention. Basically, you need to test the Gauss-Markov assumptions.
    It is, however, very data dependent which tests make sense. Usually an inspection of data will reveal the same things a test will tell you.
    For example, if your data is not trending there is no need to test for co-integration since co-integration is a test for equal stochastic trends.
    In practice it is not a very big issue, though, as most software packages will just give you a test summary which includes everything relevant.

+ 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