Joseph Coveney:
> I just
> was under the impression that odds ratios were a more "natural" metric for
> this kind of data regardless of whether from a case-control or cohort
> design.
I guess that risk ratios are a more direct measure, that is easier to
interpret compared to conditional logistic OR. However, risk ratios that
expresses the risk of being a case if exposed to something cannot be
directly calculated from a case-control study. For that you need a cohort
study (matched on exposure rather than the outcome).
With my data I have the opportunity to try both designs and as Paul
Visintainer kindly pointed out to me, my first presentation on this list was
really a matched cohort study not a case-control study.
I guess that OR are a more natural metric in at least one way. It is
comparable across designs. The OR (conditional logistic) from the
case-control design exactly matches those from the matched-cohort design
which is natural because they are based on the same information. Risk ratios
however differ, because they describe different things and are based on
different samples from the population (subjects discordant on the exposure
vs. discordant on the outcome).
I guess one could summarize the differences in RR like this: In case-control
studies they describe the risk ratio of having been exposed when being a
case. In cohort studies they describe the risk ratio of being a case when
having been exposed. The latter makes more sense to me and I will stick with
the matched-cohort design.
Michael Ingre
-----------------
PhD-student
Department of Psychology
Stockholm University &
National Institute for
Psychoscial Medicine
*
* 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/
