Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: st: RE: bcskew0 transform back

From   "Lachenbruch, Peter" <>
To   "''" <>
Subject   RE: st: RE: bcskew0 transform back
Date   Tue, 18 May 2010 12:42:49 -0700

We always seem to forget that percentiles transform one-one but means and standard deviations get messed up.  
Regarding the Box-Cox transformation, a few years ago I had an article in which I used it and found that the estimation of the power parameter was very unstable because the likelihood surface was very flat near the optimum, and so you could get no stable value.  So, I support Nick's reaction to it.  Nevertheless, it's a start...


Peter A. Lachenbruch
Department of Public Health
Oregon State University
Corvallis, OR 97330
Phone: 541-737-3832
FAX: 541-737-4001

-----Original Message-----
From: [] On Behalf Of Nick Cox
Sent: Tuesday, May 18, 2010 9:56 AM
Subject: RE: st: RE: bcskew0 transform back

This seems doubly problematic. The fact is that -bcskew0- entails
estimation of a parameter. I don't know how that meshes with the
imputation. It sounds as if you should be doing that on each imputed
dataset. Using the same constant on all sounds wrong. Others can improve
on my visceration. 

Personally I think the whole Box-Cox methodology, despite its splendid
name, to be very oversold. 

-ice- is a user-written program from .... 


raoul reulen

I failed to mention that I am using the transformed variable in a
multiple imputation model (using ice) and impute missing values. After
the imputation I then want to transfer the imputed variable back.

*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index