Ricardo Ovaldia wrote:
I am using logistic regression to model the effect of
an ordered categorical variable (at 4 levels) on
mortality while adjusting for other covariates.
Indicator variables for the categorical variable were
used in the model. The estimated odd ratios from the
logistic model for these indicator variables increase
with the order of the variable. Although, there is a
trend I would like to formaly test it.
I tryed using -tabodds- and -mhodds- but it does not
work well when adjusting for continous variables.
I then simply included the ordered categorical
variable as continous and re-ran the model. Is the
test performed on this new coefficient a valid test
for trend, although the model is nolonger the same?
Is there a better way to do this in Stata?
--------------------------------------------------------------------------------
In addition to what Tim Wade mentioned, consider using the same tactic that
you would after ANOVA, viz., orthogonal polynomial contrasts and examining
the linear component (first orthogonal polynomial contrast). You can do
this manually with your dummy (indicator) variables using -lincom-, but it's
easier to use one of the two user-written programs that help set this
up: -xi3- and -desmat-.
Joseph Coveney
*
* 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/
