Heteroskedasticity and linearity

[Pedro Martinez]

Dear All,

does Heteroskedasticity mean that there is non-linearity ? Also what are some steps to solve Heteroskedasticity?

[Carlo Lazzaro]

Pedro:
1) heteroskedasticity and non-linearity are different issues. The first relates to the relationship between predictors and regressand, the latter implies that the variance of the residual distribution is not constant;
2) -robust- standard errors account for heteroskedasticity.

[Pedro Martinez]

Thank you Carlo !