Re: compare 2 odds ratios

arrayprofile@xxxxxxxxx wrote:
Marc,

Thanks for suggesting CMH test with each subject as a stratum. As far
as I understand, CMH test was testing whether the common odds ratio
across the stratum is equal to 1 or not, assuming the equal odds ratios
across stratum first. My real intention is to test whether the 2 odds
ratios for the 2 diagnostic tests are equal (doesn't matter if they are
equal to 1 or not). So is CMH the one for my purpose?

What about the loglinear models? I am not sure how to set up a model
with my data.

Thanks


I still find an ambiguity here: When you say "odds ratios for the 2
diagnostic tests," I infer that you are looking at the result
of each diagnostic test *in relation to some other outcome,*
such as their odds ratios in relation to some gold
standard test. On the other hand, I could interpret what you
say as meaning that you don't have any other outcome, that the
question just concerns the two tests in relation to one another.
In that case, the McNemar could be relevant (e.g. "Is test 1
showing positives more frequently than test 2?"). However,
I wonder if your actual question might just be about the two
tests and the extent too which they agree with one another, in
which case something like Cohen's Kappa would be relevant.
And yes, depending on what exactly your question and situation
is, some kind of log linear model e.g., conditional logit,
could be relevant.
I'd suggest you give us a bit more context here.
If, on the other hand, no one else is seeing an ambiguity here,
my apologies for an irrelevant response.


--
=-=-=-=-=-=-=-=-=-==-=-=-=
Mike Lacy, Ft Collins CO 80523
Clean out the 'junk' to email me.
.

Relevant Pages

  • Re: compare 2 odds ratios
    ... Thanks for suggesting CMH test with each subject as a stratum. ... across the stratum is equal to 1 or not, assuming the equal odds ratios ... If we have the same patients, upon whom two different diagnostic tests ...
    (sci.stat.consult)
  • Re: compare 2 odds ratios
    ... CMH test was testing whether the common odds ratio ... across the stratum is equal to 1 or not, assuming the equal odds ratios ... across stratum first. ... ratios for the 2 diagnostic tests are equal (doesn't matter if they are ...
    (sci.stat.consult)