Question: Consider the simple linear regression model y i = 0 + 1x i + i , where the variance of i is proportional to xi
Consider the simple linear regression model y i = β0 + β1x i + εi , where the variance of εi is proportional to xi 2
, that is, Var( ) ε σ i i = x2 2.
a. Suppose that we use the transformations y′ = y / x and x′ = l/ x . Is this a variance - stabilizing transformation?
b. What are the relationships between the parameters in the original and transformed models?
c. Suppose we use the method of weighted least squares with w x i i = 1 2
. Is this equivalent to the transformation introduced in part a?
Step by Step Solution
There are 3 Steps involved in it
Get step-by-step solutions from verified subject matter experts
