+ Reply to Thread
Results 1 to 11 of 11

Thread: Linear Regression Without intercept ?

  1. #1
    Points: 1,325, Level: 20
    Level completed: 25%, Points required for next Level: 75

    Location
    Pakistan
    Posts
    13
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Linear Regression Without intercept ?




    hello i m little confuse i want to know that when its is suitable to use regression without intercept ? Any advantage over using linear regression with intercept ?

  2. #2
    TS Contributor
    Points: 22,339, Level: 92
    Level completed: 99%, Points required for next Level: 11
    spunky's Avatar
    Location
    vancouver, canada
    Posts
    2,135
    Thanks
    166
    Thanked 537 Times in 431 Posts

    Re: Linear Regression Without intercept ?

    in theory, people will say that if in your regression equation the value of your predictors = 0 implies that your dependent variable should also = 0 then removing the intercept would make sense...

    in reality, removing the intercept tends to bias your results in one way or another because you're forcing the regression to go where you want it in a rather arbitrary way... because, if you think about it, you should leave the data to speak for itself and if it includes the point (0,0) (in the bivariate regression case, please feel free to add as many coordinates as you want) then so be it but if there is no (0,0) point in the collected data to begin with, why force the regression line to include a point that was not observed?

    i think (well, better say "hope") that the general consensus is that the intercept should never be removed from the regression equation..
    for all your psychometric needs! https://psychometroscar.wordpress.com/about/

  3. #3
    Points: 1,325, Level: 20
    Level completed: 25%, Points required for next Level: 75

    Location
    Pakistan
    Posts
    13
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Linear Regression Without intercept ?

    Quote Originally Posted by spunky View Post
    in theory, people will say that if in your regression equation the value of your predictors = 0 implies that your dependent variable should also = 0 then removing the intercept would make sense...

    in reality, removing the intercept tends to bias your results in one way or another because you're forcing the regression to go where you want it in a rather arbitrary way... because, if you think about it, you should leave the data to speak for itself and if it includes the point (0,0) (in the bivariate regression case, please feel free to add as many coordinates as you want) then so be it but if there is no (0,0) point in the collected data to begin with, why force the regression line to include a point that was not observed?

    i think (well, better say "hope") that the general consensus is that the intercept should never be removed from the regression equation..
    correct me if I am wrong
    if I say that average of dependent variable almost approx. (Ybar=0 )then it is better to use regression without intercept ?

  4. #4
    TS Contributor
    Points: 22,339, Level: 92
    Level completed: 99%, Points required for next Level: 11
    spunky's Avatar
    Location
    vancouver, canada
    Posts
    2,135
    Thanks
    166
    Thanked 537 Times in 431 Posts

    Re: Linear Regression Without intercept ?

    Quote Originally Posted by wahibr View Post
    if I say that average of dependent variable almost approx. (Ybar=0 )then it is better to use regression without intercept ?

    mmmmhh... no.
    Last edited by spunky; 08-20-2011 at 01:56 PM. Reason: BAD statistical practice..
    for all your psychometric needs! https://psychometroscar.wordpress.com/about/

  5. #5
    Devorador de queso
    Points: 95,540, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Awards:
    Posting AwardCommunity AwardDiscussion EnderFrequent Poster
    Dason's Avatar
    Location
    Tampa, FL
    Posts
    12,930
    Thanks
    307
    Thanked 2,629 Times in 2,245 Posts

    Re: Linear Regression Without intercept ?

    That's not right. If ybar=0 and xbar=0 then the estimated intercept will be 0 but just because ybar is 0 doesn't mean that you shouldn't include an intercept.

  6. #6
    TS Contributor
    Points: 22,339, Level: 92
    Level completed: 99%, Points required for next Level: 11
    spunky's Avatar
    Location
    vancouver, canada
    Posts
    2,135
    Thanks
    166
    Thanked 537 Times in 431 Posts

    Re: Linear Regression Without intercept ?

    i've never done regression without intercept in practice (but i know it's wrong) however (in the bivariate case), doesnt remove the intercept implies you're forcing it to go through the (0,0) point?
    for all your psychometric needs! https://psychometroscar.wordpress.com/about/

  7. #7
    TS Contributor
    Points: 22,339, Level: 92
    Level completed: 99%, Points required for next Level: 11
    spunky's Avatar
    Location
    vancouver, canada
    Posts
    2,135
    Thanks
    166
    Thanked 537 Times in 431 Posts

    Re: Linear Regression Without intercept ?

    wait... i stand corrected.... removing the intercept also changes the slope of the line... oh god! rookie mistake!
    for all your psychometric needs! https://psychometroscar.wordpress.com/about/

  8. #8
    Devorador de queso
    Points: 95,540, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Awards:
    Posting AwardCommunity AwardDiscussion EnderFrequent Poster
    Dason's Avatar
    Location
    Tampa, FL
    Posts
    12,930
    Thanks
    307
    Thanked 2,629 Times in 2,245 Posts

    Re: Linear Regression Without intercept ?

    Yeah. It's just not a good idea ever in my opinion. I think I wrote a post which had some example data at one point. I liked it. I should try to find it...

  9. #9
    Devorador de queso
    Points: 95,540, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Awards:
    Posting AwardCommunity AwardDiscussion EnderFrequent Poster
    Dason's Avatar
    Location
    Tampa, FL
    Posts
    12,930
    Thanks
    307
    Thanked 2,629 Times in 2,245 Posts

    Re: Linear Regression Without intercept ?

    And here it is...

  10. #10
    Points: 1,325, Level: 20
    Level completed: 25%, Points required for next Level: 75

    Location
    Pakistan
    Posts
    13
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Linear Regression Without intercept ?

    thanks Dason the example helps me a lot

  11. #11
    Points: 1,325, Level: 20
    Level completed: 25%, Points required for next Level: 75

    Location
    Pakistan
    Posts
    13
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Linear Regression Without intercept ?


    The question was asked by my professor i found two more results

    1) Removing the intercept R-square of the model increase

    2) F-Ratio of the model increased too.

+ 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