>> Specifically, I need to test two bivariate samples for
>> a difference in means. Both variables are from continuous
> >distributions but one has a strong ceiling effect and the
> >other I would rather also not trust to be normally distributed.
> >
> >Any hints, please?
cut
>Even in the univariate case _and_ normality you have the Behrens-Fisher
>problem which has no universally accepted answer.
cut
--
>Brian D. Ripley, ripley@stats.ox.ac.uk
and
>Take a look at an interesting alternative analysis in which O'Brien turns
>the problem around using a logistic model to predict group membership.
--
>Frank E Harrell Jr
>http://hesweb1.med.virginia.edu/biostat
I found these two replies interesting particularily as they both implicitly
involve the adoption of a Neyman null hypothesis (testing a main effect,
say mean, in the presence of an interaction or an effect that varies by
subject) rather than a Fisher strict null hypothesis.
Not a central SPlus software issue but one that might be of interest to
some. Possible useful references on this subject are
DB Rubin. Neyman (1923) and causal inference in experiments and
observational studies. Statistical Science 5:472-480, 1990.
DR Cox. The interpretation of the effects of non-additivity in the Latin
square. Biometrika 45:69-73, 1958.
Also not a SPlus software issue, but it would seem to of importance if the
groups were formed by randomization. If they were not, very similar methods
to those of O'Brien ( thanks for the O'Brien reference Frank ) are available
for also removing confounding (and testing in the turned around away of
O'Brien).
These are perhaps best accessed through references in
MM Joffe and PR Rosenbaum. Invited commentary:Propensity scores. American
Journal of Epidemiology 150(4):327-333, 1999.
regards
Keith O'Rourke
Statistician, Department of Surgery and Clinical Epidemiology Unit
University of Ottawa/ Ottawa Hospital
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
|