[R] Test for multiple contrasts?
Peter Dalgaard BSA
p.dalgaard at biostat.ku.dk
Thu Feb 8 16:27:42 CET 2001
Kaspar Pflugshaupt <pflugshaupt at cirsium.ethz.ch> writes:
> Now, I'd like to test which of my contrasts are significantly different from
> zero. I assume that the p values given by the summary are not corrected for
> multiple testing. Thus, I might correct them with p.adjust(). But since the
> contrasts are not independent, I'm not sure if the adjustment methods would
> work here.
> On the other hand, I've come across a procedure called "Scheffe's multiple
> comparisons" (or S test), which is said to be appropriate for multiple
> contrasts like these. Before I try to implement it: Has anybody already done
> that, or are there good reasons not to use it?
The p.adjust methods do work for correlated tests, at least the Holm
and Bonferroni tests where I can understand the theory. The Hochberg
test has some amount of handwaving associated with it, but it too is
intended to be applied to testing contrasts.
Scheffe also works with correlated tests. It is more conservative than
the other tests because it protects against multiple testing on any
number of linear contrasts by projecting the (F-based) multivariate
confidence region onto the direction given by the contrast. This
argues both for and against its use...
> BTW, I tried to extract the SEs of the contrasts by se.contrast(), but it
> would not work for survival models. Would they be the same that appear in the
> summary above?
se.contrast works for aov models only. You probably need to massage
the covariance matrix for the estimates.
O__ ---- Peter Dalgaard Blegdamsvej 3
c/ /'_ --- Dept. of Biostatistics 2200 Cph. N
(*) \(*) -- University of Copenhagen Denmark Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk) FAX: (+45) 35327907
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch
More information about the R-help