Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: regression with heavily skewed dependant data (was "tobit?")


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   RE: st: regression with heavily skewed dependant data (was "tobit?")
Date   Wed, 13 Aug 2008 17:18:00 +0100

Others have pointed out that the BMI in effect prejudges the
relationship between weight and height. Naturally, I agree. But it's not
clear whether the original height and weight data are available in this
project. 

I was delighted to see Paul's report here that reciprocating BMI worked
well. I love reciprocal transformations because they are so simple and
they yield variables with well-defined units  An alternative, or a
complement, to transformation would be to work with -glm, link(power
-1)-. 

Nick 
n.j.cox@durham.ac.uk 

Seed, Paul

[...] 

3) When considering possible transformations, I would aim for ones that
give approximate symmetry.  In my own data, 1/BMI = height^2/Weight
performed quite well. Skewness -.0393561, Kurtosis 3.708181.

[...] 


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index