Diagnostics and Robust Estimation When Transforming the Regression Model and the Responses.
Technical rept. Aug 85-Aug 86,
NORTH CAROLINA UNIV AT CHAPEL HILL DEPT OF STATISTICS
Pagination or Media Count:
In regression analysis, the response is often transformed to remove heteroscedasticity andor skewness. When a model already exists for the untransformed response, then it can be preserved by transforming both the model and the response with the same transformation. This methodology, is called transform both sides has been applied in several recent papers, and appears highly useful in practice. When a parametric transformation family such as power transformations is used, then the transformation can be estimated by maximum likelihood. The MLE however is very sensitive to outliers. This article proposes diagnostics which indicate cases influential for the transformation regression parameters. We also propose a robust bounded-influence estimator similar to the Krasker-Welsch regression estimate. Both diagnostics and the robust estimator can be implemented on standard software. Author
- Statistics and Probability