Hello everyone:
I'm analysing (actually trying to) a set of data on life-history
traits from a two-way fully factorial design. One of the traits has
an intrinsic hoteroscedasticity, and no transformation (which I
dislike doing anyway) has been able to make the variances decently
equal.
I'm interested both in the interaction between the factors and the
main effects, and therefore would not want to divide the test in two
one-way ANOVAs.
I know there are models for 2-way ANOVAs under unequal variances (for
example Malwane M. A. Ananda and Samaradasa Weerahandi in Statistica
Sinica, 7(1997), 631-646) but I haven't found access to any
information.
Does any of you have some suggestion?
Thanks a lot!
Dita
--
---------------------------------------------------------
Dita Brígida Vizoso
Ecology and Evolution at Fribourg, Switzerland
http://www.unifr.ch/biol/ecology/index.html
---------------------------------------------------------
What has to be fought is the ideas.
Not the people.
---------------------------------------------------------
|