violation of multiple regression assumption

[archived_user]

volume 1, page 345 of curriculum, 3rd paragraph, 6th line … says
‘if we make the mistake of including dummy variables for four rather than three quarters, we have violated assumption 2 of multiple regression model…’
can anyone possibly explain , How have we violated assumption 2 that says independent variables are not random and no exact linear relation exists between them?

 

[archived_user]

if your dummy variables are q1, q2, q3 and q4 than their sum is equal to 1.

 

[archived_user]

alright if we have dummy variables for all four quarters instead of three then their sum is equal to 1, which means………..? *scratching my head..blink blink*

 

[archived_user]

Gypsy wrote:
alright if we have dummy variables for all four quarters instead of three then their sum is equal to 1, which means………..? *scratching my head..blink blink*

If the sum of all the variables = 4.. Then there exists a linear relationship where
q1+q2+q3+q4 = 1. But regression assumes that the variables are independent. But in this case it becomes dependant.. So we got to eliminate atleast 1 variable from this..
Hope I made it a little clearer for you..

 

[archived_user]

ah.. ok
thank you sooraj and maratikus