Daniel Waxman wrote:
This is a statistics question rather than a Stata question.
I am struggling with how to best describe (medical journal manuscript)
an interaction effect.
The overall goal of the study is to describe the continuous variable
'zlog' as a predictor of 'outcome' and to determine the degree to which
the association is independent of other variables. It is indeed
independent of most of them, but there are two dummy variables for which
interaction terms are significant and the odds ratio for zlog changes.
The question is: How do I describe/quantify the interaction in a
succinct way?
Do the odds ratios for the interaction terms have any intuitive meaning?
I can see what is happening (sort of) by dropping observations based
upon the dummy variable, but it is hard to describe quantitatively.
The two dummy variables that interact are different. In the first
example, the odds ratio for the continuous variable increases when the
either the observations with dummy==0 or dummy==1 are dropped. In the
second case, dropping dummy==0 decreases the the OR for the continuous
variable and dropping dummy==1 increases it.
----------------------------------------------------------------------
One of Daniel's commands was:
. xi: logistic outcome i.romi*zlog
I find it easier to interpret and communicate the results of a stratified
analysis; in the case with the predictor of interest (zlog) being
continuous, -mhodds- (see [ST] epitab) is useful:
. mhodds outcome zlog , by(romi)
-mhodds- also gives a test of homogeneity.
Just an idea; perhaps it is useful.
Svend
________________________________________________________
Svend Juul
Institut for Folkesundhed, Afdeling for Epidemiologi
(Institute of Public Health, Department of Epidemiology)
Vennelyst Boulevard 6
DK-8000 Aarhus C, Denmark
Phone, work: +45 8942 6090
Phone, home: +45 8693 7796
Fax: +45 8613 1580
E-mail: [email protected]
_________________________________________________________
*
* 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/