In a publication from Juergen Cox and Matthias Mann (2012) there are test statistics stated for the manova test. However, with the manova() test in R I get different results than i get with their software. Are the test statistics (included as images from the paper) different and how can I implement it into R (simply calculating the formula in the statistic also doesn't solve the problem)
first Part of the test statistics second part
(1D and 2D annotation enrichment: a statistical method integrating quantitative proteomics with complementary high-throughput data, Juergen Cox and Matthias Mann (2012))
example: For each species as one group and the others as the second group i do the test
minExample<-data.frame(dim1=iris$Sepal.Length,
dim2=iris$Sepal.Width,
categorical1=iris$Species=="virginica",
categorical2=iris$Species == "versicolor",
categorical3=iris$Species == "setosa")
test<- apply(minExample[,3:5],2, function(x,minExample){
manovaRes<-manova(cbind(dim1,dim2)~ x ,data=minExample)
manovaResSummary<-summary.manova(manovaRes,test='Pillai')
return(manovaResSummary$stats[1,"Pr(>F)"] )
},minExample=minExample)
test
with this (and the other methods such as wilks etc)I get the results of:
| categorical1 | categorical2 | categorical3 |
|---|---|---|
| 1.301495e-17 | 1.246040e-08 | 3.253074e-50 |
however, Perseus (the software from the publication) produces:
| categorical1 | categorical2 | categorical3 |
|---|---|---|
| 7.48401E-11 | 4.98709E-06 | 0 |
The information on how they do the test are in the publication, i have no code to look at.
