March 2019
Beginner to intermediate
448 pages
13h 14m
English
What happens to this test when the covariance matrices are different? In theory, the results should not be reliable, but in practice, it has been found that this is fairly robust:
library(MASS)f1 = matrix(c(2,1,1,1,2,1,1,1,2),3,3)f2 = matrix(c(3,2,1,2,3,1,1,1,3),3,3)f3 = matrix(c(2,1,-0.5,1,2,1,-0.5,1,2),3,3)x1 = mvrnorm(50,c(10,10,10),f1)x1 = cbind(x1,1)x2 = mvrnorm(50,c(10,10,10),f2)x2 = cbind(x2,2)x3 = mvrnorm(50,c(30,10,10),f3)x3 = cbind(x3,3)total_data = data.frame(rbind(x1,x2,x3))colnames(total_data) = c("History","Math","Biology","class")result = manova(cbind(History,Math,Biology) ~ class,data=total_data)summary(result)
The preceding code generate the following output of MANOVA tests:
We still reject the null with ...
Read now
Unlock full access