Hello everyone,
I would like to perform a 2-way manova test, but I'm having some issues.
I implemented like this
Y<-cbind(Resp1,Resp2,Resp3,....,Respn)
model<-manova(Y "tilda" FactorA*FactorB)
summary.aov(model)
1. I don't know at what level I have to do the Type I error correction. Is
it on p-values returned by "summary.aov(model)? Or is it when I compare
each
subgroup with another subgroup? Or is it for summary(model)?
2. I have a significant interaction on summary(model), but not for
summary.aov(model). Is it normal?
3. Do I have to test homoscedasticity and normality for residuals(model)
like in ANOVA?
4. It's about my response variables. I'm working with mice and we
measure
multiple variables per mice, however these variables are quite dependente.
For example I'm measuring the total number of cells (Resp1) in spleen, and
afterwards among these cells, I assess percentage of subpopulations of other
cells among splenocytes, in order to have the absolute number, I multiply
these percentages by R1 ::: R1 * percentage1 = Resp2. Do I have to do an
additional correction due to this?
You would save my life
Thank you all.
--
View this message in context:
http://r.789695.n4.nabble.com/Two-way-manova-tp4650259.html
Sent from the R help mailing list archive at Nabble.com.
