Optimal power transformation via inverse response plots
Charles Lindsey
Texas A & M University
Department of Statistics
College Station, TX
lindseyc@stat.tamu.edu
|
Simon Sheather
Texas A & M University
Department of Statistics
College Station, smrm
TX
sheather@stat.tamu.edu
|
Abstract. We present a new Stata command, irp, that generates the inverse
response plot (Cook and Weisberg, 1994, Biometrika 81: 731–737) of a response
on its predictors. Using the inverse response plot, an appropriate scaled power
transformation for the positive response variable can be found so that the transformed
response mean is linear in the predictors. The optimal transformation is
displayed in the plot, as are user-specified guesses. By using the graphical display,
the user may determine whether an appropriate transformation exists as well as
determine its value. We demonstrate the irp command using both a generated
and a real example.
View all articles by these authors:
Charles Lindsey, Simon Sheather
View all articles with these keywords:
irp, regress, scaled power transformation
Download citation: BibTeX RIS
Download citation and abstract: BibTeX RIS
Contact StataCorp
Contact service@stata-journal.com
if you have questions about the Stata Journal.
© Copyright 2001–2010 StataCorp LP. Terms of use. Privacy notice.
|
