|
Thanks Adrienne
Kandel,
I think my question is different.
To check collinearity, I do regression
analysis on
x1 = a + b1 x2 + b2 x3 + e
x2 = ...
x3 = ...
What is partial regression line? For example, Is it
a line Y against x1 when x2 x3 fixed
What is the use of partial regressional line, I am
confused. Is it for detecting collinearity? (I just want to check
linearity)
I got 200 obs on x1 x2 x3 Y. Since I can not get enough obs when x2 x3 fixed (Actually
no same x2 x3), so how to plot this line.
As your explanation, this line seems
unnecessary since I already use one method to detect collinearity between each
variable of them.
I don't know how to plot partial
regression line, but I plot a line Y against x1 (no fixed x2 x3). It
has no pattern (I can not see a line, somewhere condensed,
somewhere loose). I don't know how to
check linearity (There is no line). I did regression on Y based on x1, R
square is only 0.05. What should I do? And I did regression on Y based on
all three x, R square is only 0.2. Is it a failed regression
analysis?
Regards,
Chushu Gu
|