Linear Modelinghard
0:00.0

In a simple linear regression Yi=β0+β1Xi+ϵiY_i = \beta_0 + \beta_1 X_i + \epsilon_i, suppose the model is estimated using weighted least squares (WLS) where weights wi=1Xi2w_i = \frac{1}{X_i^2}. If the original OLS estimator β^1\hat{\beta}_1 is biased due to heteroscedasticity, what is the primary purpose of this transformation?