Statalist


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

Re: st: Log Transform Justification


From   "Austin Nichols" <austinnichols@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Log Transform Justification
Date   Wed, 29 Aug 2007 09:48:35 -0400

Clive--
I think Nick is referring to your claim that "you log-transform (LT)
the variable, transforming the scale from 0-1 to minus infinity-plus
infinity" which is incorrect--you mean to logit-transform the
variable, since the log transform ln(y) would turn a range of (0,1)
for y into (-inf, 0) for ln(y) and a logit transform would turn a
range of (0,1) for y into (-inf, +inf) for ln[y/(1-y)].  The -glm-
link option handles the transformation, so yes, -link(logit)- only
makes sense for models whose dependent variables are in [0,1] and
-link(log)- for y in [0,+inf)
--see also
http://statacorp.com/statalist/archive/2006-11/msg00294.html
http://statacorp.com/statalist/archive/2007-08/msg00177.html
and other messages in the "st: Binomial regression" thread for more options.

On 8/29/07, Clive Nicholas <clivelists@googlemail.com> wrote:
> Nick Cox wrote:
>
> > Some confusion here between logarithms and logits?
>
> If I'm thinking straight, you're arguing that the call to -glm,
> link(logit)- only makes sense for models whose dependent variables are
> already scaled 0-1, since the -link()- option does the transformation.
*
*   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