Estimating equation through a correcting factor to correct heteroskedasticity may cause extra correlation to enter the model, which increase the R^2. This renders our ultimate regression results meaningless. True or false, or uncertain? Please provide an explanation.