Date: Fri, 27 Oct 2006 04:24:21 -0400
Reply-To: Jim Groeneveld <jim2stat@YAHOO.CO.UK>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Jim Groeneveld <jim2stat@YAHOO.CO.UK>
Subject: Re: Displaying p values less than 0.0001 using chisq option??
Thank you for your explanation Ron,
So I'll ask again: what are your concrete hypotheses to test?
Which paired interactions to you expect (significantly) different
in advance and why?
If none, then I'm afraid your analysis is merely exploratory
and not hypothesis testing.
That is fine if you would find some pairs of genetic variants
that show a "significant" effect (difference perhaps).
Then you might focus your research further on those pairs,
developing an explanatory theory and deduced hypotheses
that you can test statistically.
This thus may be a two step approach.
Instead of the exploratory "significance" thresholds
you might also use a clinical difference threshold,
based on the relevance of the differences (effects).
Regards - Jim.
--
Jim Groeneveld, Netherlands
Statistician, SAS consultant
home.hccnet.nl/jim.groeneveld
My computer shows magicant effects with miracustical tests on hypodata.
On Thu, 26 Oct 2006 14:41:07 -0400, Ron Do <rondo@HOTMAIL.COM> wrote:
>Maybe I should have been more specific at the beginning, but basically the
>project that we're working on involves looking at 1536 genetic variants that
>are associated with heart disease in a case-control study.
>
>We've done the single-point analysis and are now looking at all the
>pair-wise interactions between the genetic variants with the disease.
>Therefore, we are performing 1536*(1535)/2 tests. We are interested in the
>p values that are lower than a certain threshold (due to multiple testing)
>that is determined by a genetics program that is built to do this sort of
>analysis. This threshold is in the range of 10E-5.
>
>I hope this clears things up. Thank you for the suggestions and I really
>appreciate the help I get from this listserve.
>
>Ron
