Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: dilemma of transformation to normal dist.


From   Ronan Conroy <rconroy@rcsi.ie>
To   "statalist hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   Re: st: dilemma of transformation to normal dist.
Date   Mon, 19 May 2003 11:19:25 +0100

on 17/05/2003 3:48 pm, Yoshiro Nagao at ng999@yahoo.co.jp wrote:

> 
> To enable linear regression, this dependent
> was logarithm-transformed.  Since a considerable number
> of records have 0 value, however, they could not be
> log-transformed.  Setting an very small positive arbitrary value
> to these records would enable log-transformation.
> However, the size of this arbitrary value would
> affect the result of regression analysis.

Maybe you should think more about the meaning of your data. A lot of zeros
sounds like you may have either
- counted data: consider negative binomial regression
- data in which many values are too low to be observed: consider interval
regression 

The answer to what sort of transformation is appropriate is rarely simply
statistical. You also need to consider what is being measured, and what sort
of back-transformation will get you results on a scale that makes sense to
the people who are interpreting the findings.

Ronan M Conroy (rconroy@rcsi.ie)
Lecturer in Biostatistics
Royal College of Surgeons
Dublin 2, Ireland
+353 1 402 2431 (fax 2764)

--------------------

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   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