statrefs home‎ > ‎Main‎ > ‎Methods‎ > ‎Regression‎ > ‎Regression Supporting Concepts‎ > ‎

Variance Stabilizing Transform

A variance stabilizing transform may be useful if Regression Model Heteroskedasticity is encountered.  Variance stabilizing transforms are applied to a model's response variable (Y).

Common variance stabilizing transforms include log(Y) and sqrt(Y).  Box Cox methods might be useful.

Also consider a Weighted Least Square (WLS) model.

NIST: Transformations to Improve Fit and Equalize Variances