Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


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

RE: st: RE: Poisson Regression


From   Maarten buis <maartenbuis@yahoo.co.uk>
To   statalist@hsphsun2.harvard.edu
Subject   RE: st: RE: Poisson Regression
Date   Tue, 15 Feb 2011 08:26:47 +0000 (GMT)

--- Visintainer, Paul wrote:
> My frustration is that when the outcome is common and
> logistic regression is used, there virtually no discussion
> of clinical relevance -- only statistical significance,
> (e.g., is a significant odds ratio of 2.5 clinically
> relevant?  Perhaps if the base risk is 2%; perhaps not
> if the base risk 73%. 

This is where Stata has a problem: it automatically suppresses
the display of the baseline odds when you ask for odds ratios
(and baseline hazard when you ask for hazard ratios, and
baseline incidence rate when you ask for incidence rate ratios,
etc.).

To me the baseline odds serve two purposes:

First, is it really helps in communicating the results. By 
discussing it first in the results section of a paper you 
refresh the readers memory on what an odds is. It also makes 
the model less "magical" if you frame it in number of high 
status jobs, deaths, or successes per low status job, survivals, 
or failures. You frame the model in terms that the reader care 
about.

Second, it helps when trying to determine the size of of an 
effect. An odds ratio of 2 is not very impressive if the baseline
odds is small 2 times a small number is still a small number, 
while is much more impressive if the baseline odds is large.
Alternatively, if your baseline odds is already 50 successes per
failure, then any increase is not going to have much substantive
meaning.

There is a trick you can use to display the baseline odds, which 
is discussed in a slighly different context in: 

Roger Newson (2003) "Stata tip 1: The eform() option of regress"
The stata Journal, 3(4), 445.
<http://www.stata-journal.com/article.html?article=st0054> 

-- Maarten

Ps. those who have followed the list a while will have noticed that
I made this point before. I hope I did not bore them too much.

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------



      

*
*   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 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index