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 at the end of May, and its replacement, statalist.org is already up and running.


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

Re: st: Using natural logs on RHS of maximum likelihood models


From   "JVerkuilen (Gmail)" <jvverkuilen@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Using natural logs on RHS of maximum likelihood models
Date   Wed, 27 Mar 2013 08:50:28 -0400

On Wed, Mar 27, 2013 at 7:31 AM, David Hoaglin <dchoaglin@gmail.com> wrote:
> Jay,
>
> I would say "one idea," rather than "the idea."  Even if the data
> contain no points of high leverage (which may or may not exert
> disproportionate influence), a transformation of an RHS variable may
> remove nonlinearity in its contribution to the model.  Sometimes a
> transformation both reduces the leverage of high-leverage points and
> removes nonlinearity.

I missed the qualifier "usual", but yes that's definitely true.

Sometimes people use transformations of raw variables to turn them
into another form to deal with nonlinearity. These can be quite
radical, too, such as the tendency to bin data (or even median split)
and reduce to a two sample t-test or ANOVA. Fortunately regression
splines (among other techniques) help. I need to start putting them
into class.
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


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