Dear Brian,
Thank you for your answer.
Another thing that came to my mind: Would it be possible just to
separately rank-transform my 3 dependent variables and then to conduct a
normal MANOVA on this data?
Thanks,
Mike
Eisenring Michael, Msc.
PhD Student
Federal Department of Economic Affairs, Education and Research
EAER
Agroecology and Environment
Biosafety
Reckenholzstrasse 191, CH-8046 Z?rich
Tel. +41 44 37 77181
Fax +41 44 37 77201
michael.eisenring at agroscope.admin.ch<mailto:
michael.eisenring at agroscope.admin.ch>
www.agroscope.ch<http://www.agroscope.ch/>
Von: Cade, Brian [mailto:cadeb at usgs.gov]
Gesendet: Mittwoch, 18. Januar 2017 18:20
An: Eisenring Michael Agroscope <michael.eisenring at agroscope.admin.ch>
Cc: r-help at r-project.org
Betreff: Re: [R] non-parametric manova with post-hoc test
You could try a multi-response permutation procedure (MRPP) for
multivariate hypothesis testing (null is groups come from a common
distribution) without resorting to ranks. There are no automated multiple
comparison procedures, but one could either look at pairwise contrasts of
group (if that is what you are implying by post-hoc testing) with some sort
of correction procedure for multiple comparisons (e.g., Holm's sequential
procedure). Or similarly, comparisons with different subsets of the
multivariate outcome variables (again, adjusting for multiple comparisons)
across the grouping structure. There are several R packages that I think
implement MRPP but the Blossom package might be one of the better
implementations in terms of alternatives provided (including permutation
version of Hotelling's test).
Brian
Brian S. Cade, PhD
U. S. Geological Survey
Fort Collins Science Center
2150 Centre Ave., Bldg. C
Fort Collins, CO 80526-8818
email: cadeb at usgs.gov<mailto:brian_cade at usgs.gov>
tel: 970 226-9326
On Wed, Jan 18, 2017 at 10:00 AM, <michael.eisenring at agroscope.admin.ch
<mailto:michael.eisenring at agroscope.admin.ch>> wrote:
Good day,
I am looking for a way to perform a non parametric manova and to analyze
the result using post-hoc tests (an equivalent of the kruskal wallis test
for anova)
In my book (discovering statistic using R) two tests are described Munzel
and Brunners method (mulrank) and Choi and Mardens test (cmanova). Both are
from the package WRS which unfortunately does not exist anymore (and WRS2
is not containing these tests). Furthermore the test do to my knowledge not
allow post-hoc analyses-
I would be grateful for your help
Best,
Mike
Eisenring Michael, Msc.
PhD Student
Federal Department of Economic Affairs, Education and Research
EAER
Agroecology and Environment
Biosafety
Reckenholzstrasse 191, CH-8046 Z?rich
Tel. +41 44 37 77181
Fax +41 44 37 77201
michael.eisenring at agroscope.admin.ch<mailto:
michael.eisenring at agroscope.admin.ch><mailto:
michael.eisenring at agroscope.admin.ch<mailto:
michael.eisenring at agroscope.admin.ch>>
www.agroscope.ch<http://www.agroscope.ch><http://www.agroscope.ch/>
[[alternative HTML version deleted]]